clamav/libclamav/pe.c

5966 lines
220 KiB
C
Raw Normal View History

2004-05-11 23:30:57 +00:00
/*
* Copyright (C) 2013-2022 Cisco Systems, Inc. and/or its affiliates. All rights reserved.
* Copyright (C) 2007-2013 Sourcefire, Inc.
*
* Authors: Alberto Wu, Tomasz Kojm, Andrew Williams
*
* Acknowledgements: The header structures were based upon a PE format
* analysis by B. Luevelsmeyer.
*
2004-05-11 23:30:57 +00:00
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
2004-05-11 23:30:57 +00:00
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
* MA 02110-1301, USA.
2004-05-11 23:30:57 +00:00
*/
/*
Portions of Code (i.e. pe_ordinal) Copyright (c) 2014. The YARA Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
2004-05-11 23:30:57 +00:00
#if HAVE_CONFIG_H
#include "clamav-config.h"
#endif
/*
#define _XOPEN_SOURCE 500
*/
2004-05-11 23:30:57 +00:00
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#if HAVE_STRING_H
2004-05-11 23:30:57 +00:00
#include <string.h>
#endif
2004-05-11 23:30:57 +00:00
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#ifdef HAVE_UNISTD_H
2004-05-11 23:30:57 +00:00
#include <unistd.h>
#endif
2004-05-11 23:30:57 +00:00
#include <time.h>
#include <stdarg.h>
2004-05-11 23:30:57 +00:00
#include "clamav.h"
#include "others.h"
#include "pe.h"
#include "petite.h"
#include "fsg.h"
#include "spin.h"
2006-01-14 13:58:35 +00:00
#include "upx.h"
#include "yc.h"
#include "aspack.h"
#include "wwunpack.h"
#include "unsp.h"
#include "scanners.h"
#include "str.h"
#include "entconv.h"
#include "execs.h"
#include "mew.h"
#include "upack.h"
#include "matcher.h"
2011-01-14 16:12:43 +01:00
#include "matcher-hash.h"
2008-07-28 13:23:12 +00:00
#include "disasm.h"
#include "special.h"
2009-07-14 09:36:36 +02:00
#include "ishield.h"
2012-01-08 03:19:29 +01:00
#include "asn1.h"
2004-05-11 23:30:57 +00:00
2014-05-07 10:56:35 -04:00
#include "json_api.h"
#include "clamav_rust.h"
#define DCONF ctx->dconf->pe
#define PE_IMAGE_DOS_SIGNATURE 0x5a4d /* MZ */
#define PE_IMAGE_DOS_SIGNATURE_OLD 0x4d5a /* ZM */
#define PE_IMAGE_NT_SIGNATURE 0x00004550
#define PE32_SIGNATURE 0x010b
#define PE32P_SIGNATURE 0x020b
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#define OPT_HDR_SIZE_DIFF (sizeof(struct pe_image_optional_hdr64) - sizeof(struct pe_image_optional_hdr32))
#define UPX_NRV2B "\x11\xdb\x11\xc9\x01\xdb\x75\x07\x8b\x1e\x83\xee\xfc\x11\xdb\x11\xc9\x11\xc9\x75\x20\x41\x01\xdb"
#define UPX_NRV2D "\x83\xf0\xff\x74\x78\xd1\xf8\x89\xc5\xeb\x0b\x01\xdb\x75\x07\x8b\x1e\x83\xee\xfc\x11\xdb\x11\xc9"
#define UPX_NRV2E "\xeb\x52\x31\xc9\x83\xe8\x03\x72\x11\xc1\xe0\x08\x8a\x06\x46\x83\xf0\xff\x74\x75\xd1\xf8\x89\xc5"
#define UPX_LZMA1_FIRST "\x56\x83\xc3\x04\x53\x50\xc7\x03"
#define UPX_LZMA1_SECOND "\x90\x90\x90\x55\x57\x56\x53\x83"
#define UPX_LZMA0 "\x56\x83\xc3\x04\x53\x50\xc7\x03\x03\x00\x00\x00\x90\x90\x90\x55\x57\x56\x53\x83"
2008-07-31 02:17:25 +00:00
#define UPX_LZMA2 "\x56\x83\xc3\x04\x53\x50\xc7\x03\x03\x00\x02\x00\x90\x90\x90\x90\x90\x55\x57\x56"
2004-06-12 00:16:01 +00:00
#define PE_MAXNAMESIZE 256
#define PE_MAXIMPORTS 1024
2016-07-12 11:53:59 -04:00
#define EC64(x) ((uint64_t)cli_readint64(&(x))) /* Convert little endian to host */
#define EC32(x) ((uint32_t)cli_readint32(&(x)))
2012-01-09 12:45:20 +02:00
#define EC16(x) ((uint16_t)cli_readint16(&(x)))
Spelling Adjustments (#30) * spelling: accessed * spelling: alignment * spelling: amalgamated * spelling: answers * spelling: another * spelling: acquisition * spelling: apitid * spelling: ascii * spelling: appending * spelling: appropriate * spelling: arbitrary * spelling: architecture * spelling: asynchronous * spelling: attachments * spelling: argument * spelling: authenticode * spelling: because * spelling: boundary * spelling: brackets * spelling: bytecode * spelling: calculation * spelling: cannot * spelling: changes * spelling: check * spelling: children * spelling: codegen * spelling: commands * spelling: container * spelling: concatenated * spelling: conditions * spelling: continuous * spelling: conversions * spelling: corresponding * spelling: corrupted * spelling: coverity * spelling: crafting * spelling: daemon * spelling: definition * spelling: delivered * spelling: delivery * spelling: delimit * spelling: dependencies * spelling: dependency * spelling: detection * spelling: determine * spelling: disconnects * spelling: distributed * spelling: documentation * spelling: downgraded * spelling: downloading * spelling: endianness * spelling: entities * spelling: especially * spelling: empty * spelling: expected * spelling: explicitly * spelling: existent * spelling: finished * spelling: flexibility * spelling: flexible * spelling: freshclam * spelling: functions * spelling: guarantee * spelling: hardened * spelling: headaches * spelling: heighten * spelling: improper * spelling: increment * spelling: indefinitely * spelling: independent * spelling: inaccessible * spelling: infrastructure Conflicts: docs/html/node68.html * spelling: initializing * spelling: inited * spelling: instream * spelling: installed * spelling: initialization * spelling: initialize * spelling: interface * spelling: intrinsics * spelling: interpreter * spelling: introduced * spelling: invalid * spelling: latency * spelling: lawyers * spelling: libclamav * spelling: likelihood * spelling: loop * spelling: maximum * spelling: million * spelling: milliseconds * spelling: minimum * spelling: minzhuan * spelling: multipart * spelling: misled * spelling: modifiers * spelling: notifying * spelling: objects * spelling: occurred * spelling: occurs * spelling: occurrences * spelling: optimization * spelling: original * spelling: originated * spelling: output * spelling: overridden * spelling: parenthesis * spelling: partition * spelling: performance * spelling: permission * spelling: phishing * spelling: portions * spelling: positives * spelling: preceded * spelling: properties * spelling: protocol * spelling: protos * spelling: quarantine * spelling: recursive * spelling: referring * spelling: reorder * spelling: reset * spelling: resources * spelling: resume * spelling: retrieval * spelling: rewrite * spelling: sanity * spelling: scheduled * spelling: search * spelling: section * spelling: separator * spelling: separated * spelling: specify * spelling: special * spelling: statement * spelling: streams * spelling: succession * spelling: suggests * spelling: superfluous * spelling: suspicious * spelling: synonym * spelling: temporarily * spelling: testfiles * spelling: transverse * spelling: turkish * spelling: typos * spelling: unable * spelling: unexpected * spelling: unexpectedly * spelling: unfinished * spelling: unfortunately * spelling: uninitialized * spelling: unlocking * spelling: unnecessary * spelling: unpack * spelling: unrecognized * spelling: unsupported * spelling: usable * spelling: wherever * spelling: wishlist * spelling: white * spelling: infrastructure * spelling: directories * spelling: overridden * spelling: permission * spelling: yesterday * spelling: initialization * spelling: intrinsics * space adjustment for spelling changes * minor modifications by klin
2018-02-21 15:00:59 -05:00
/* lower and upper boundary alignment (size vs offset) */
#define PEALIGN(o, a) (((a)) ? (((o) / (a)) * (a)) : (o))
#define PESALIGN(o, a) (((a)) ? (((o) / (a) + ((o) % (a) != 0)) * (a)) : (o))
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Replace all of these with static inline functions
#define CLI_UNPSIZELIMITS(NAME, CHK) \
if (cli_checklimits(NAME, ctx, (CHK), 0, 0) != CL_CLEAN) { \
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo); \
return CL_CLEAN; \
}
#define CLI_UNPTEMP(NAME, FREEME) \
if (!(tempfile = cli_gentemp(ctx->sub_tmpdir))) { \
cli_exe_info_destroy(peinfo); \
cli_multifree FREEME; \
return CL_EMEM; \
} \
if ((ndesc = open(tempfile, O_RDWR | O_CREAT | O_TRUNC | O_BINARY, S_IRUSR | S_IWUSR)) < 0) { \
cli_dbgmsg(NAME ": Can't create file %s\n", tempfile); \
free(tempfile); \
cli_exe_info_destroy(peinfo); \
cli_multifree FREEME; \
return CL_ECREAT; \
}
#define CLI_TMPUNLK() \
if (!ctx->engine->keeptmp) { \
if (cli_unlink(tempfile)) { \
free(tempfile); \
return CL_EUNLINK; \
} \
}
#ifdef HAVE__INTERNAL__SHA_COLLECT
#define SHA_OFF \
do { \
ctx->sha_collect = -1; \
} while (0)
#define SHA_RESET \
do { \
ctx->sha_collect = sha_collect; \
} while (0)
#else
#define SHA_OFF \
do { \
} while (0)
#define SHA_RESET \
do { \
} while (0)
#endif
#define FSGCASE(NAME, FREESEC) \
case 0: /* Unpacked and NOT rebuilt */ \
cli_dbgmsg(NAME ": Successfully decompressed\n"); \
close(ndesc); \
if (cli_unlink(tempfile)) { \
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo); \
free(tempfile); \
FREESEC; \
return CL_EUNLINK; \
} \
free(tempfile); \
FREESEC; \
found = 0; \
upx_success = 1; \
break; /* FSG ONLY! - scan raw data after upx block */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#define SPINCASE() \
case 2: \
free(spinned); \
close(ndesc); \
if (cli_unlink(tempfile)) { \
cli_exe_info_destroy(peinfo); \
free(tempfile); \
return CL_EUNLINK; \
} \
cli_dbgmsg("cli_scanpe: PESpin: Size exceeded\n"); \
free(tempfile); \
break;
#define CLI_UNPRESULTS_(NAME, FSGSTUFF, EXPR, GOOD, FREEME) \
switch (EXPR) { \
case GOOD: /* Unpacked and rebuilt */ \
if (ctx->engine->keeptmp) \
cli_dbgmsg(NAME ": Unpacked and rebuilt executable saved in %s\n", tempfile); \
else \
cli_dbgmsg(NAME ": Unpacked and rebuilt executable\n"); \
cli_multifree FREEME; \
cli_exe_info_destroy(peinfo); \
lseek(ndesc, 0, SEEK_SET); \
cli_dbgmsg("***** Scanning rebuilt PE file *****\n"); \
SHA_OFF; \
if (CL_VIRUS == cli_magic_scan_desc(ndesc, tempfile, ctx, NULL, LAYER_ATTRIBUTES_NONE)) { \
close(ndesc); \
SHA_RESET; \
CLI_TMPUNLK(); \
free(tempfile); \
return CL_VIRUS; \
} \
SHA_RESET; \
close(ndesc); \
CLI_TMPUNLK(); \
free(tempfile); \
return CL_CLEAN; \
\
FSGSTUFF; \
\
default: \
cli_dbgmsg(NAME ": Unpacking failed\n"); \
close(ndesc); \
if (cli_unlink(tempfile)) { \
cli_exe_info_destroy(peinfo); \
free(tempfile); \
cli_multifree FREEME; \
return CL_EUNLINK; \
} \
cli_multifree FREEME; \
free(tempfile); \
}
#define CLI_UNPRESULTS(NAME, EXPR, GOOD, FREEME) CLI_UNPRESULTS_(NAME, (void)0, EXPR, GOOD, FREEME)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO The second argument to FSGCASE below should match what gets freed as
// indicated by FREEME, otherwise a memory leak can occur (as currently used,
// it looks like dest can get leaked by these macros).
#define CLI_UNPRESULTSFSG1(NAME, EXPR, GOOD, FREEME) CLI_UNPRESULTS_(NAME, FSGCASE(NAME, free(sections)), EXPR, GOOD, FREEME)
#define CLI_UNPRESULTSFSG2(NAME, EXPR, GOOD, FREEME) CLI_UNPRESULTS_(NAME, FSGCASE(NAME, (void)0), EXPR, GOOD, FREEME)
#define DETECT_BROKEN_PE (SCAN_HEURISTIC_BROKEN && !ctx->corrupted_input)
extern const unsigned int hashlen[];
struct offset_list {
uint32_t offset;
struct offset_list *next;
};
struct pe_image_import_descriptor {
union {
uint32_t Characteristics;
uint32_t OriginalFirstThunk;
} u;
uint32_t TimeDateStamp;
uint32_t ForwarderChain;
uint32_t Name;
uint32_t FirstThunk;
};
#define PE_IMAGEDIR_ORDINAL_FLAG32 0x80000000
#define PE_IMAGEDIR_ORDINAL_FLAG64 0x8000000000000000L
struct pe_image_thunk32 {
union {
uint32_t ForwarderString;
uint32_t Function;
uint32_t Ordinal;
uint32_t AddressOfData;
} u;
};
struct pe_image_thunk64 {
union {
uint64_t ForwarderString;
uint64_t Function;
uint64_t Ordinal;
uint64_t AddressOfData;
} u;
};
struct pe_image_import_by_name {
uint16_t Hint;
uint8_t Name[1];
};
static void cli_multifree(void *f, ...)
{
void *ff;
va_list ap;
free(f);
va_start(ap, f);
while ((ff = va_arg(ap, void *))) free(ff);
va_end(ap);
}
2010-01-05 01:01:48 +01:00
struct vinfo_list {
uint32_t rvas[16];
unsigned int count;
};
static int versioninfo_cb(void *opaque, uint32_t type, uint32_t name, uint32_t lang, uint32_t rva)
{
2010-01-05 01:01:48 +01:00
struct vinfo_list *vlist = (struct vinfo_list *)opaque;
2010-01-05 02:04:14 +01:00
cli_dbgmsg("versioninfo_cb: type: %x, name: %x, lang: %x, rva: %x\n", type, name, lang, rva);
2010-01-05 01:01:48 +01:00
vlist->rvas[vlist->count] = rva;
if (++vlist->count == sizeof(vlist->rvas) / sizeof(vlist->rvas[0]))
2014-11-07 11:12:30 -05:00
return 1;
2009-12-22 04:16:30 +01:00
return 0;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* Given an RVA (relative to the ImageBase), return the file offset of the
* corresponding data */
2010-02-15 18:12:02 +02:00
uint32_t cli_rawaddr(uint32_t rva, const struct cli_exe_section *shp, uint16_t nos, unsigned int *err, size_t fsize, uint32_t hdr_size)
2004-06-11 20:17:18 +00:00
{
int i, found = 0;
uint32_t ret;
2004-06-11 20:17:18 +00:00
if (rva < hdr_size) { /* Out of section EP - mapped to imagebase+rva */
2014-11-07 11:12:30 -05:00
if (rva >= fsize) {
*err = 1;
2014-11-07 11:12:30 -05:00
return 0;
}
*err = 0;
2014-11-07 11:12:30 -05:00
return rva;
}
2004-06-11 20:17:18 +00:00
for (i = nos - 1; i >= 0; i--) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (shp[i].rsz && shp[i].rva <= rva && shp[i].rsz > (rva - shp[i].rva)) {
2014-11-07 11:12:30 -05:00
found = 1;
break;
}
2004-06-11 20:17:18 +00:00
}
if (!found) {
2014-11-07 11:12:30 -05:00
*err = 1;
return 0;
2004-06-11 20:17:18 +00:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ret = (rva - shp[i].rva) + shp[i].raw;
*err = 0;
return ret;
2004-06-11 20:17:18 +00:00
}
/*
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
void findres(uint32_t by_type, uint32_t by_name, fmap_t *map, struct cli_exe_info *peinfo, int (*cb)(void *, uint32_t, uint32_t, uint32_t, uint32_t), void *opaque)
2009-11-11 03:58:22 +01:00
callback based res lookup
by_type: lookup type
by_name: lookup name or (unsigned)-1 to look for any name
res_rva: base resource rva (i.e. dirs[2].VirtualAddress)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
map, peinfo: same as in scanpe
2009-11-11 03:58:22 +01:00
cb: the callback function executed on each successful match
opaque: an opaque pointer passed to the callback
the callback proto is
int pe_res_cballback (void *opaque, uint32_t type, uint32_t name, uint32_t lang, uint32_t rva);
the callback shall return 0 to continue the lookup or 1 to abort
*/
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
void findres(uint32_t by_type, uint32_t by_name, fmap_t *map, struct cli_exe_info *peinfo, int (*cb)(void *, uint32_t, uint32_t, uint32_t, uint32_t), void *opaque)
{
2009-11-11 03:58:22 +01:00
unsigned int err = 0;
uint32_t type, type_offs, name, name_offs, lang, lang_offs;
const uint8_t *resdir, *type_entry, *name_entry, *lang_entry;
2009-11-11 03:58:22 +01:00
uint16_t type_cnt, name_cnt, lang_cnt;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
uint32_t res_rva;
if (NULL == peinfo || peinfo->ndatadirs < 3) {
return;
}
2009-11-11 03:58:22 +01:00
if (0 != peinfo->offset) {
cli_dbgmsg("findres: Assumption Violated: Looking for version info when peinfo->offset != 0\n");
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
res_rva = EC32(peinfo->dirs[2].VirtualAddress);
if (!(resdir = fmap_need_off_once(map, cli_rawaddr(res_rva, peinfo->sections, peinfo->nsections, &err, map->len, peinfo->hdr_size), 16)) || err)
return;
2009-11-11 03:58:22 +01:00
type_cnt = (uint16_t)cli_readint16(resdir + 12);
type_entry = resdir + 16;
if (!(by_type >> 31)) {
type_entry += type_cnt * 8;
type_cnt = (uint16_t)cli_readint16(resdir + 14);
2009-11-11 03:58:22 +01:00
}
while (type_cnt--) {
if (!fmap_need_ptr_once(map, type_entry, 8))
return;
type = cli_readint32(type_entry);
type_offs = cli_readint32(type_entry + 4);
if (type == by_type && (type_offs >> 31)) {
type_offs &= 0x7fffffff;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!(resdir = fmap_need_off_once(map, cli_rawaddr(res_rva + type_offs, peinfo->sections, peinfo->nsections, &err, map->len, peinfo->hdr_size), 16)) || err)
return;
name_cnt = (uint16_t)cli_readint16(resdir + 12);
name_entry = resdir + 16;
if (by_name == 0xffffffff)
name_cnt += (uint16_t)cli_readint16(resdir + 14);
else if (!(by_name >> 31)) {
name_entry += name_cnt * 8;
name_cnt = (uint16_t)cli_readint16(resdir + 14);
}
while (name_cnt--) {
if (!fmap_need_ptr_once(map, name_entry, 8))
return;
name = cli_readint32(name_entry);
name_offs = cli_readint32(name_entry + 4);
if ((by_name == 0xffffffff || name == by_name) && (name_offs >> 31)) {
name_offs &= 0x7fffffff;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!(resdir = fmap_need_off_once(map, cli_rawaddr(res_rva + name_offs, peinfo->sections, peinfo->nsections, &err, map->len, peinfo->hdr_size), 16)) || err)
return;
lang_cnt = (uint16_t)cli_readint16(resdir + 12) + (uint16_t)cli_readint16(resdir + 14);
lang_entry = resdir + 16;
while (lang_cnt--) {
if (!fmap_need_ptr_once(map, lang_entry, 8))
return;
lang = cli_readint32(lang_entry);
lang_offs = cli_readint32(lang_entry + 4);
if (!(lang_offs >> 31)) {
if (cb(opaque, type, name, lang, res_rva + lang_offs))
return;
}
lang_entry += 8;
}
}
name_entry += 8;
}
return; /* FIXME: unless we want to find ALL types */
}
type_entry += 8;
2009-11-11 03:58:22 +01:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
static void cli_parseres_special(uint32_t base, uint32_t rva, fmap_t *map, struct cli_exe_info *peinfo, size_t fsize, unsigned int level, uint32_t type, unsigned int *maxres, struct swizz_stats *stats)
{
unsigned int err = 0, i;
const uint8_t *resdir;
const uint8_t *entry, *oentry;
uint16_t named, unnamed;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
uint32_t rawaddr = cli_rawaddr(rva, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
uint32_t entries;
if (level > 2 || !*maxres) return;
*maxres -= 1;
if (err || !(resdir = fmap_need_off_once(map, rawaddr, 16)))
return;
named = (uint16_t)cli_readint16(resdir + 12);
unnamed = (uint16_t)cli_readint16(resdir + 14);
entries = /*named+*/ unnamed;
if (!entries)
return;
rawaddr += named * 8; /* skip named */
/* this is just used in a heuristic detection, so don't give error on failure */
if (!(entry = fmap_need_off(map, rawaddr + 16, entries * 8))) {
cli_dbgmsg("cli_parseres_special: failed to read resource directory at:%lu\n", (unsigned long)rawaddr + 16);
return;
}
oentry = entry;
/*for (i=0; i<named; i++) {
uint32_t id, offs;
id = cli_readint32(entry);
offs = cli_readint32(entry+4);
if(offs>>31)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_parseres( base, base + (offs&0x7fffffff), srcfd, peinfo, fsize, level+1, type, maxres, stats);
entry+=8;
}*/
for (i = 0; i < unnamed; i++, entry += 8) {
uint32_t id, offs;
if (stats->errors >= SWIZZ_MAXERRORS) {
cli_dbgmsg("cli_parseres_special: resources broken, ignoring\n");
return;
}
id = cli_readint32(entry) & 0x7fffffff;
if (level == 0) {
type = 0;
switch (id) {
case 4: /* menu */
case 5: /* dialog */
case 6: /* string */
case 11: /* msgtable */
type = id;
break;
case 16:
type = id;
/* 14: version */
stats->has_version = 1;
break;
case 24: /* manifest */
stats->has_manifest = 1;
break;
/* otherwise keep it 0, we don't want it */
}
}
if (!type) {
/* if we are not interested in this type, skip */
continue;
}
offs = cli_readint32(entry + 4);
if (offs >> 31)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_parseres_special(base, base + (offs & 0x7fffffff), map, peinfo, fsize, level + 1, type, maxres, stats);
else {
offs = cli_readint32(entry + 4);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rawaddr = cli_rawaddr(base + offs, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (!err && (resdir = fmap_need_off_once(map, rawaddr, 16))) {
uint32_t isz = cli_readint32(resdir + 4);
const uint8_t *str;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rawaddr = cli_rawaddr(cli_readint32(resdir), peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err || !isz || isz >= fsize || rawaddr + isz >= fsize) {
cli_dbgmsg("cli_parseres_special: invalid resource table entry: %lu + %lu\n",
(unsigned long)rawaddr,
(unsigned long)isz);
stats->errors++;
continue;
}
if ((id & 0xff) != 0x09) /* english res only */
continue;
if ((str = fmap_need_off_once(map, rawaddr, isz)))
cli_detect_swizz_str(str, isz, stats, type);
}
}
}
fmap_unneed_ptr(map, oentry, entries * 8);
}
static unsigned int cli_hashsect(fmap_t *map, struct cli_exe_section *s, unsigned char **digest, int *foundhash, int *foundwild)
{
const void *hashme;
if (s->rsz > CLI_MAX_ALLOCATION) {
cli_dbgmsg("cli_hashsect: skipping hash calculation for too big section\n");
return 0;
}
if (!s->rsz) return 0;
if (!(hashme = fmap_need_off_once(map, s->raw, s->rsz))) {
cli_dbgmsg("cli_hashsect: unable to read section data\n");
return 0;
}
if (foundhash[CLI_HASH_MD5] || foundwild[CLI_HASH_MD5])
cl_hash_data("md5", hashme, s->rsz, digest[CLI_HASH_MD5], NULL);
if (foundhash[CLI_HASH_SHA1] || foundwild[CLI_HASH_SHA1])
cl_sha1(hashme, s->rsz, digest[CLI_HASH_SHA1], NULL);
if (foundhash[CLI_HASH_SHA256] || foundwild[CLI_HASH_SHA256])
cl_sha256(hashme, s->rsz, digest[CLI_HASH_SHA256], NULL);
return 1;
}
/* check hash section sigs */
static int scan_pe_mdb(cli_ctx *ctx, struct cli_exe_section *exe_section)
{
struct cli_matcher *mdb_sect = ctx->engine->hm_mdb;
unsigned char *hashset[CLI_HASH_AVAIL_TYPES];
const char *virname = NULL;
int foundsize[CLI_HASH_AVAIL_TYPES];
int foundwild[CLI_HASH_AVAIL_TYPES];
cli_hash_type_t type;
int ret = CL_CLEAN;
unsigned char *md5 = NULL;
/* pick hashtypes to generate */
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
foundsize[type] = cli_hm_have_size(mdb_sect, type, exe_section->rsz);
foundwild[type] = cli_hm_have_wild(mdb_sect, type);
if (foundsize[type] || foundwild[type]) {
hashset[type] = cli_malloc(hashlen[type]);
if (!hashset[type]) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("scan_pe_mdb: cli_malloc failed!\n");
for (; type > 0;)
free(hashset[--type]);
return CL_EMEM;
}
} else {
hashset[type] = NULL;
}
}
/* Generate hashes */
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
cli_hashsect(ctx->fmap, exe_section, hashset, foundsize, foundwild);
/* Print hash */
if (cli_debug_flag) {
md5 = hashset[CLI_HASH_MD5];
if (md5) {
cli_dbgmsg("MDB hashset: %u:%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x\n",
exe_section->rsz, md5[0], md5[1], md5[2], md5[3], md5[4], md5[5], md5[6], md5[7],
md5[8], md5[9], md5[10], md5[11], md5[12], md5[13], md5[14], md5[15]);
} else if (cli_always_gen_section_hash) {
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
const void *hashme = fmap_need_off_once(ctx->fmap, exe_section->raw, exe_section->rsz);
if (!(hashme)) {
cli_errmsg("scan_pe_mdb: unable to read section data\n");
ret = CL_EREAD;
goto end;
}
md5 = cli_malloc(16);
if (!(md5)) {
cli_errmsg("scan_pe_mdb: cli_malloc failed!\n");
ret = CL_EMEM;
goto end;
}
cl_hash_data("md5", hashme, exe_section->rsz, md5, NULL);
cli_dbgmsg("MDB: %u:%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x\n",
exe_section->rsz, md5[0], md5[1], md5[2], md5[3], md5[4], md5[5], md5[6], md5[7],
md5[8], md5[9], md5[10], md5[11], md5[12], md5[13], md5[14], md5[15]);
free(md5);
} else {
cli_dbgmsg("MDB: %u:notgenerated\n", exe_section->rsz);
}
}
/* Do scans */
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
if (foundsize[type] && cli_hm_scan(hashset[type], exe_section->rsz, &virname, mdb_sect, type) == CL_VIRUS) {
ret = cli_append_virus(ctx, virname);
if (ret != CL_CLEAN) {
if (ret != CL_VIRUS)
break;
else if (!SCAN_ALLMATCHES)
break;
}
}
if (foundwild[type] && cli_hm_scan_wild(hashset[type], &virname, mdb_sect, type) == CL_VIRUS) {
ret = cli_append_virus(ctx, virname);
if (ret != CL_CLEAN) {
if (ret != CL_VIRUS)
break;
else if (!SCAN_ALLMATCHES)
break;
}
}
}
end:
for (type = CLI_HASH_AVAIL_TYPES; type > 0;)
free(hashset[--type]);
return ret;
}
/* imptbl scanning */
2016-07-12 11:53:59 -04:00
static char *pe_ordinal(const char *dll, uint16_t ord)
{
char name[64];
name[0] = '\0';
if (strncasecmp(dll, "WS2_32.dll", 10) == 0 ||
strncasecmp(dll, "wsock32.dll", 11) == 0) {
switch (ord) {
case 1:
sprintf(name, "accept");
break;
case 2:
sprintf(name, "bind");
break;
case 3:
sprintf(name, "closesocket");
break;
case 4:
sprintf(name, "connect");
break;
case 5:
sprintf(name, "getpeername");
break;
case 6:
sprintf(name, "getsockname");
break;
case 7:
sprintf(name, "getsockopt");
break;
case 8:
sprintf(name, "htonl");
break;
case 9:
sprintf(name, "htons");
break;
case 10:
sprintf(name, "ioctlsocket");
break;
case 11:
sprintf(name, "inet_addr");
break;
case 12:
sprintf(name, "inet_ntoa");
break;
case 13:
sprintf(name, "listen");
break;
case 14:
sprintf(name, "ntohl");
break;
case 15:
sprintf(name, "ntohs");
break;
case 16:
sprintf(name, "recv");
break;
case 17:
sprintf(name, "recvfrom");
break;
case 18:
sprintf(name, "select");
break;
case 19:
sprintf(name, "send");
break;
case 20:
sprintf(name, "sendto");
break;
case 21:
sprintf(name, "setsockopt");
break;
case 22:
sprintf(name, "shutdown");
break;
case 23:
sprintf(name, "socket");
break;
case 24:
sprintf(name, "GetAddrInfoW");
break;
case 25:
sprintf(name, "GetNameInfoW");
break;
case 26:
sprintf(name, "WSApSetPostRoutine");
break;
case 27:
sprintf(name, "FreeAddrInfoW");
break;
case 28:
sprintf(name, "WPUCompleteOverlappedRequest");
break;
case 29:
sprintf(name, "WSAAccept");
break;
case 30:
sprintf(name, "WSAAddressToStringA");
break;
case 31:
sprintf(name, "WSAAddressToStringW");
break;
case 32:
sprintf(name, "WSACloseEvent");
break;
case 33:
sprintf(name, "WSAConnect");
break;
case 34:
sprintf(name, "WSACreateEvent");
break;
case 35:
sprintf(name, "WSADuplicateSocketA");
break;
case 36:
sprintf(name, "WSADuplicateSocketW");
break;
case 37:
sprintf(name, "WSAEnumNameSpaceProvidersA");
break;
case 38:
sprintf(name, "WSAEnumNameSpaceProvidersW");
break;
case 39:
sprintf(name, "WSAEnumNetworkEvents");
break;
case 40:
sprintf(name, "WSAEnumProtocolsA");
break;
case 41:
sprintf(name, "WSAEnumProtocolsW");
break;
case 42:
sprintf(name, "WSAEventSelect");
break;
case 43:
sprintf(name, "WSAGetOverlappedResult");
break;
case 44:
sprintf(name, "WSAGetQOSByName");
break;
case 45:
sprintf(name, "WSAGetServiceClassInfoA");
break;
case 46:
sprintf(name, "WSAGetServiceClassInfoW");
break;
case 47:
sprintf(name, "WSAGetServiceClassNameByClassIdA");
break;
case 48:
sprintf(name, "WSAGetServiceClassNameByClassIdW");
break;
case 49:
sprintf(name, "WSAHtonl");
break;
case 50:
sprintf(name, "WSAHtons");
break;
case 51:
sprintf(name, "gethostbyaddr");
break;
case 52:
sprintf(name, "gethostbyname");
break;
case 53:
sprintf(name, "getprotobyname");
break;
case 54:
sprintf(name, "getprotobynumber");
break;
case 55:
sprintf(name, "getservbyname");
break;
case 56:
sprintf(name, "getservbyport");
break;
case 57:
sprintf(name, "gethostname");
break;
case 58:
sprintf(name, "WSAInstallServiceClassA");
break;
case 59:
sprintf(name, "WSAInstallServiceClassW");
break;
case 60:
sprintf(name, "WSAIoctl");
break;
case 61:
sprintf(name, "WSAJoinLeaf");
break;
case 62:
sprintf(name, "WSALookupServiceBeginA");
break;
case 63:
sprintf(name, "WSALookupServiceBeginW");
break;
case 64:
sprintf(name, "WSALookupServiceEnd");
break;
case 65:
sprintf(name, "WSALookupServiceNextA");
break;
case 66:
sprintf(name, "WSALookupServiceNextW");
break;
case 67:
sprintf(name, "WSANSPIoctl");
break;
case 68:
sprintf(name, "WSANtohl");
break;
case 69:
sprintf(name, "WSANtohs");
break;
case 70:
sprintf(name, "WSAProviderConfigChange");
break;
case 71:
sprintf(name, "WSARecv");
break;
case 72:
sprintf(name, "WSARecvDisconnect");
break;
case 73:
sprintf(name, "WSARecvFrom");
break;
case 74:
sprintf(name, "WSARemoveServiceClass");
break;
case 75:
sprintf(name, "WSAResetEvent");
break;
case 76:
sprintf(name, "WSASend");
break;
case 77:
sprintf(name, "WSASendDisconnect");
break;
case 78:
sprintf(name, "WSASendTo");
break;
case 79:
sprintf(name, "WSASetEvent");
break;
case 80:
sprintf(name, "WSASetServiceA");
break;
case 81:
sprintf(name, "WSASetServiceW");
break;
case 82:
sprintf(name, "WSASocketA");
break;
case 83:
sprintf(name, "WSASocketW");
break;
case 84:
sprintf(name, "WSAStringToAddressA");
break;
case 85:
sprintf(name, "WSAStringToAddressW");
break;
case 86:
sprintf(name, "WSAWaitForMultipleEvents");
break;
case 87:
sprintf(name, "WSCDeinstallProvider");
break;
case 88:
sprintf(name, "WSCEnableNSProvider");
break;
case 89:
sprintf(name, "WSCEnumProtocols");
break;
case 90:
sprintf(name, "WSCGetProviderPath");
break;
case 91:
sprintf(name, "WSCInstallNameSpace");
break;
case 92:
sprintf(name, "WSCInstallProvider");
break;
case 93:
sprintf(name, "WSCUnInstallNameSpace");
break;
case 94:
sprintf(name, "WSCUpdateProvider");
break;
case 95:
sprintf(name, "WSCWriteNameSpaceOrder");
break;
case 96:
sprintf(name, "WSCWriteProviderOrder");
break;
case 97:
sprintf(name, "freeaddrinfo");
break;
case 98:
sprintf(name, "getaddrinfo");
break;
case 99:
sprintf(name, "getnameinfo");
break;
case 101:
sprintf(name, "WSAAsyncSelect");
break;
case 102:
sprintf(name, "WSAAsyncGetHostByAddr");
break;
case 103:
sprintf(name, "WSAAsyncGetHostByName");
break;
case 104:
sprintf(name, "WSAAsyncGetProtoByNumber");
break;
case 105:
sprintf(name, "WSAAsyncGetProtoByName");
break;
case 106:
sprintf(name, "WSAAsyncGetServByPort");
break;
case 107:
sprintf(name, "WSAAsyncGetServByName");
break;
case 108:
sprintf(name, "WSACancelAsyncRequest");
break;
case 109:
sprintf(name, "WSASetBlockingHook");
break;
case 110:
sprintf(name, "WSAUnhookBlockingHook");
break;
case 111:
sprintf(name, "WSAGetLastError");
break;
case 112:
sprintf(name, "WSASetLastError");
break;
case 113:
sprintf(name, "WSACancelBlockingCall");
break;
case 114:
sprintf(name, "WSAIsBlocking");
break;
case 115:
sprintf(name, "WSAStartup");
break;
case 116:
sprintf(name, "WSACleanup");
break;
case 151:
sprintf(name, "__WSAFDIsSet");
break;
case 500:
sprintf(name, "WEP");
break;
default:
break;
}
} else if (strncasecmp(dll, "oleaut32.dll", 12) == 0) {
switch (ord) {
case 2:
sprintf(name, "SysAllocString");
break;
case 3:
sprintf(name, "SysReAllocString");
break;
case 4:
sprintf(name, "SysAllocStringLen");
break;
case 5:
sprintf(name, "SysReAllocStringLen");
break;
case 6:
sprintf(name, "SysFreeString");
break;
case 7:
sprintf(name, "SysStringLen");
break;
case 8:
sprintf(name, "VariantInit");
break;
case 9:
sprintf(name, "VariantClear");
break;
case 10:
sprintf(name, "VariantCopy");
break;
case 11:
sprintf(name, "VariantCopyInd");
break;
case 12:
sprintf(name, "VariantChangeType");
break;
case 13:
sprintf(name, "VariantTimeToDosDateTime");
break;
case 14:
sprintf(name, "DosDateTimeToVariantTime");
break;
case 15:
sprintf(name, "SafeArrayCreate");
break;
case 16:
sprintf(name, "SafeArrayDestroy");
break;
case 17:
sprintf(name, "SafeArrayGetDim");
break;
case 18:
sprintf(name, "SafeArrayGetElemsize");
break;
case 19:
sprintf(name, "SafeArrayGetUBound");
break;
case 20:
sprintf(name, "SafeArrayGetLBound");
break;
case 21:
sprintf(name, "SafeArrayLock");
break;
case 22:
sprintf(name, "SafeArrayUnlock");
break;
case 23:
sprintf(name, "SafeArrayAccessData");
break;
case 24:
sprintf(name, "SafeArrayUnaccessData");
break;
case 25:
sprintf(name, "SafeArrayGetElement");
break;
case 26:
sprintf(name, "SafeArrayPutElement");
break;
case 27:
sprintf(name, "SafeArrayCopy");
break;
case 28:
sprintf(name, "DispGetParam");
break;
case 29:
sprintf(name, "DispGetIDsOfNames");
break;
case 30:
sprintf(name, "DispInvoke");
break;
case 31:
sprintf(name, "CreateDispTypeInfo");
break;
case 32:
sprintf(name, "CreateStdDispatch");
break;
case 33:
sprintf(name, "RegisterActiveObject");
break;
case 34:
sprintf(name, "RevokeActiveObject");
break;
case 35:
sprintf(name, "GetActiveObject");
break;
case 36:
sprintf(name, "SafeArrayAllocDescriptor");
break;
case 37:
sprintf(name, "SafeArrayAllocData");
break;
case 38:
sprintf(name, "SafeArrayDestroyDescriptor");
break;
case 39:
sprintf(name, "SafeArrayDestroyData");
break;
case 40:
sprintf(name, "SafeArrayRedim");
break;
case 41:
sprintf(name, "SafeArrayAllocDescriptorEx");
break;
case 42:
sprintf(name, "SafeArrayCreateEx");
break;
case 43:
sprintf(name, "SafeArrayCreateVectorEx");
break;
case 44:
sprintf(name, "SafeArraySetRecordInfo");
break;
case 45:
sprintf(name, "SafeArrayGetRecordInfo");
break;
case 46:
sprintf(name, "VarParseNumFromStr");
break;
case 47:
sprintf(name, "VarNumFromParseNum");
break;
case 48:
sprintf(name, "VarI2FromUI1");
break;
case 49:
sprintf(name, "VarI2FromI4");
break;
case 50:
sprintf(name, "VarI2FromR4");
break;
case 51:
sprintf(name, "VarI2FromR8");
break;
case 52:
sprintf(name, "VarI2FromCy");
break;
case 53:
sprintf(name, "VarI2FromDate");
break;
case 54:
sprintf(name, "VarI2FromStr");
break;
case 55:
sprintf(name, "VarI2FromDisp");
break;
case 56:
sprintf(name, "VarI2FromBool");
break;
case 57:
sprintf(name, "SafeArraySetIID");
break;
case 58:
sprintf(name, "VarI4FromUI1");
break;
case 59:
sprintf(name, "VarI4FromI2");
break;
case 60:
sprintf(name, "VarI4FromR4");
break;
case 61:
sprintf(name, "VarI4FromR8");
break;
case 62:
sprintf(name, "VarI4FromCy");
break;
case 63:
sprintf(name, "VarI4FromDate");
break;
case 64:
sprintf(name, "VarI4FromStr");
break;
case 65:
sprintf(name, "VarI4FromDisp");
break;
case 66:
sprintf(name, "VarI4FromBool");
break;
case 67:
sprintf(name, "SafeArrayGetIID");
break;
case 68:
sprintf(name, "VarR4FromUI1");
break;
case 69:
sprintf(name, "VarR4FromI2");
break;
case 70:
sprintf(name, "VarR4FromI4");
break;
case 71:
sprintf(name, "VarR4FromR8");
break;
case 72:
sprintf(name, "VarR4FromCy");
break;
case 73:
sprintf(name, "VarR4FromDate");
break;
case 74:
sprintf(name, "VarR4FromStr");
break;
case 75:
sprintf(name, "VarR4FromDisp");
break;
case 76:
sprintf(name, "VarR4FromBool");
break;
case 77:
sprintf(name, "SafeArrayGetVartype");
break;
case 78:
sprintf(name, "VarR8FromUI1");
break;
case 79:
sprintf(name, "VarR8FromI2");
break;
case 80:
sprintf(name, "VarR8FromI4");
break;
case 81:
sprintf(name, "VarR8FromR4");
break;
case 82:
sprintf(name, "VarR8FromCy");
break;
case 83:
sprintf(name, "VarR8FromDate");
break;
case 84:
sprintf(name, "VarR8FromStr");
break;
case 85:
sprintf(name, "VarR8FromDisp");
break;
case 86:
sprintf(name, "VarR8FromBool");
break;
case 87:
sprintf(name, "VarFormat");
break;
case 88:
sprintf(name, "VarDateFromUI1");
break;
case 89:
sprintf(name, "VarDateFromI2");
break;
case 90:
sprintf(name, "VarDateFromI4");
break;
case 91:
sprintf(name, "VarDateFromR4");
break;
case 92:
sprintf(name, "VarDateFromR8");
break;
case 93:
sprintf(name, "VarDateFromCy");
break;
case 94:
sprintf(name, "VarDateFromStr");
break;
case 95:
sprintf(name, "VarDateFromDisp");
break;
case 96:
sprintf(name, "VarDateFromBool");
break;
case 97:
sprintf(name, "VarFormatDateTime");
break;
case 98:
sprintf(name, "VarCyFromUI1");
break;
case 99:
sprintf(name, "VarCyFromI2");
break;
case 100:
sprintf(name, "VarCyFromI4");
break;
case 101:
sprintf(name, "VarCyFromR4");
break;
case 102:
sprintf(name, "VarCyFromR8");
break;
case 103:
sprintf(name, "VarCyFromDate");
break;
case 104:
sprintf(name, "VarCyFromStr");
break;
case 105:
sprintf(name, "VarCyFromDisp");
break;
case 106:
sprintf(name, "VarCyFromBool");
break;
case 107:
sprintf(name, "VarFormatNumber");
break;
case 108:
sprintf(name, "VarBstrFromUI1");
break;
case 109:
sprintf(name, "VarBstrFromI2");
break;
case 110:
sprintf(name, "VarBstrFromI4");
break;
case 111:
sprintf(name, "VarBstrFromR4");
break;
case 112:
sprintf(name, "VarBstrFromR8");
break;
case 113:
sprintf(name, "VarBstrFromCy");
break;
case 114:
sprintf(name, "VarBstrFromDate");
break;
case 115:
sprintf(name, "VarBstrFromDisp");
break;
case 116:
sprintf(name, "VarBstrFromBool");
break;
case 117:
sprintf(name, "VarFormatPercent");
break;
case 118:
sprintf(name, "VarBoolFromUI1");
break;
case 119:
sprintf(name, "VarBoolFromI2");
break;
case 120:
sprintf(name, "VarBoolFromI4");
break;
case 121:
sprintf(name, "VarBoolFromR4");
break;
case 122:
sprintf(name, "VarBoolFromR8");
break;
case 123:
sprintf(name, "VarBoolFromDate");
break;
case 124:
sprintf(name, "VarBoolFromCy");
break;
case 125:
sprintf(name, "VarBoolFromStr");
break;
case 126:
sprintf(name, "VarBoolFromDisp");
break;
case 127:
sprintf(name, "VarFormatCurrency");
break;
case 128:
sprintf(name, "VarWeekdayName");
break;
case 129:
sprintf(name, "VarMonthName");
break;
case 130:
sprintf(name, "VarUI1FromI2");
break;
case 131:
sprintf(name, "VarUI1FromI4");
break;
case 132:
sprintf(name, "VarUI1FromR4");
break;
case 133:
sprintf(name, "VarUI1FromR8");
break;
case 134:
sprintf(name, "VarUI1FromCy");
break;
case 135:
sprintf(name, "VarUI1FromDate");
break;
case 136:
sprintf(name, "VarUI1FromStr");
break;
case 137:
sprintf(name, "VarUI1FromDisp");
break;
case 138:
sprintf(name, "VarUI1FromBool");
break;
case 139:
sprintf(name, "VarFormatFromTokens");
break;
case 140:
sprintf(name, "VarTokenizeFormatString");
break;
case 141:
sprintf(name, "VarAdd");
break;
case 142:
sprintf(name, "VarAnd");
break;
case 143:
sprintf(name, "VarDiv");
break;
case 144:
sprintf(name, "DllCanUnloadNow");
break;
case 145:
sprintf(name, "DllGetClassObject");
break;
case 146:
sprintf(name, "DispCallFunc");
break;
case 147:
sprintf(name, "VariantChangeTypeEx");
break;
case 148:
sprintf(name, "SafeArrayPtrOfIndex");
break;
case 149:
sprintf(name, "SysStringByteLen");
break;
case 150:
sprintf(name, "SysAllocStringByteLen");
break;
case 151:
sprintf(name, "DllRegisterServer");
break;
case 152:
sprintf(name, "VarEqv");
break;
case 153:
sprintf(name, "VarIdiv");
break;
case 154:
sprintf(name, "VarImp");
break;
case 155:
sprintf(name, "VarMod");
break;
case 156:
sprintf(name, "VarMul");
break;
case 157:
sprintf(name, "VarOr");
break;
case 158:
sprintf(name, "VarPow");
break;
case 159:
sprintf(name, "VarSub");
break;
case 160:
sprintf(name, "CreateTypeLib");
break;
case 161:
sprintf(name, "LoadTypeLib");
break;
case 162:
sprintf(name, "LoadRegTypeLib");
break;
case 163:
sprintf(name, "RegisterTypeLib");
break;
case 164:
sprintf(name, "QueryPathOfRegTypeLib");
break;
case 165:
sprintf(name, "LHashValOfNameSys");
break;
case 166:
sprintf(name, "LHashValOfNameSysA");
break;
case 167:
sprintf(name, "VarXor");
break;
case 168:
sprintf(name, "VarAbs");
break;
case 169:
sprintf(name, "VarFix");
break;
case 170:
sprintf(name, "OaBuildVersion");
break;
case 171:
sprintf(name, "ClearCustData");
break;
case 172:
sprintf(name, "VarInt");
break;
case 173:
sprintf(name, "VarNeg");
break;
case 174:
sprintf(name, "VarNot");
break;
case 175:
sprintf(name, "VarRound");
break;
case 176:
sprintf(name, "VarCmp");
break;
case 177:
sprintf(name, "VarDecAdd");
break;
case 178:
sprintf(name, "VarDecDiv");
break;
case 179:
sprintf(name, "VarDecMul");
break;
case 180:
sprintf(name, "CreateTypeLib2");
break;
case 181:
sprintf(name, "VarDecSub");
break;
case 182:
sprintf(name, "VarDecAbs");
break;
case 183:
sprintf(name, "LoadTypeLibEx");
break;
case 184:
sprintf(name, "SystemTimeToVariantTime");
break;
case 185:
sprintf(name, "VariantTimeToSystemTime");
break;
case 186:
sprintf(name, "UnRegisterTypeLib");
break;
case 187:
sprintf(name, "VarDecFix");
break;
case 188:
sprintf(name, "VarDecInt");
break;
case 189:
sprintf(name, "VarDecNeg");
break;
case 190:
sprintf(name, "VarDecFromUI1");
break;
case 191:
sprintf(name, "VarDecFromI2");
break;
case 192:
sprintf(name, "VarDecFromI4");
break;
case 193:
sprintf(name, "VarDecFromR4");
break;
case 194:
sprintf(name, "VarDecFromR8");
break;
case 195:
sprintf(name, "VarDecFromDate");
break;
case 196:
sprintf(name, "VarDecFromCy");
break;
case 197:
sprintf(name, "VarDecFromStr");
break;
case 198:
sprintf(name, "VarDecFromDisp");
break;
case 199:
sprintf(name, "VarDecFromBool");
break;
case 200:
sprintf(name, "GetErrorInfo");
break;
case 201:
sprintf(name, "SetErrorInfo");
break;
case 202:
sprintf(name, "CreateErrorInfo");
break;
case 203:
sprintf(name, "VarDecRound");
break;
case 204:
sprintf(name, "VarDecCmp");
break;
case 205:
sprintf(name, "VarI2FromI1");
break;
case 206:
sprintf(name, "VarI2FromUI2");
break;
case 207:
sprintf(name, "VarI2FromUI4");
break;
case 208:
sprintf(name, "VarI2FromDec");
break;
case 209:
sprintf(name, "VarI4FromI1");
break;
case 210:
sprintf(name, "VarI4FromUI2");
break;
case 211:
sprintf(name, "VarI4FromUI4");
break;
case 212:
sprintf(name, "VarI4FromDec");
break;
case 213:
sprintf(name, "VarR4FromI1");
break;
case 214:
sprintf(name, "VarR4FromUI2");
break;
case 215:
sprintf(name, "VarR4FromUI4");
break;
case 216:
sprintf(name, "VarR4FromDec");
break;
case 217:
sprintf(name, "VarR8FromI1");
break;
case 218:
sprintf(name, "VarR8FromUI2");
break;
case 219:
sprintf(name, "VarR8FromUI4");
break;
case 220:
sprintf(name, "VarR8FromDec");
break;
case 221:
sprintf(name, "VarDateFromI1");
break;
case 222:
sprintf(name, "VarDateFromUI2");
break;
case 223:
sprintf(name, "VarDateFromUI4");
break;
case 224:
sprintf(name, "VarDateFromDec");
break;
case 225:
sprintf(name, "VarCyFromI1");
break;
case 226:
sprintf(name, "VarCyFromUI2");
break;
case 227:
sprintf(name, "VarCyFromUI4");
break;
case 228:
sprintf(name, "VarCyFromDec");
break;
case 229:
sprintf(name, "VarBstrFromI1");
break;
case 230:
sprintf(name, "VarBstrFromUI2");
break;
case 231:
sprintf(name, "VarBstrFromUI4");
break;
case 232:
sprintf(name, "VarBstrFromDec");
break;
case 233:
sprintf(name, "VarBoolFromI1");
break;
case 234:
sprintf(name, "VarBoolFromUI2");
break;
case 235:
sprintf(name, "VarBoolFromUI4");
break;
case 236:
sprintf(name, "VarBoolFromDec");
break;
case 237:
sprintf(name, "VarUI1FromI1");
break;
case 238:
sprintf(name, "VarUI1FromUI2");
break;
case 239:
sprintf(name, "VarUI1FromUI4");
break;
case 240:
sprintf(name, "VarUI1FromDec");
break;
case 241:
sprintf(name, "VarDecFromI1");
break;
case 242:
sprintf(name, "VarDecFromUI2");
break;
case 243:
sprintf(name, "VarDecFromUI4");
break;
case 244:
sprintf(name, "VarI1FromUI1");
break;
case 245:
sprintf(name, "VarI1FromI2");
break;
case 246:
sprintf(name, "VarI1FromI4");
break;
case 247:
sprintf(name, "VarI1FromR4");
break;
case 248:
sprintf(name, "VarI1FromR8");
break;
case 249:
sprintf(name, "VarI1FromDate");
break;
case 250:
sprintf(name, "VarI1FromCy");
break;
case 251:
sprintf(name, "VarI1FromStr");
break;
case 252:
sprintf(name, "VarI1FromDisp");
break;
case 253:
sprintf(name, "VarI1FromBool");
break;
case 254:
sprintf(name, "VarI1FromUI2");
break;
case 255:
sprintf(name, "VarI1FromUI4");
break;
case 256:
sprintf(name, "VarI1FromDec");
break;
case 257:
sprintf(name, "VarUI2FromUI1");
break;
case 258:
sprintf(name, "VarUI2FromI2");
break;
case 259:
sprintf(name, "VarUI2FromI4");
break;
case 260:
sprintf(name, "VarUI2FromR4");
break;
case 261:
sprintf(name, "VarUI2FromR8");
break;
case 262:
sprintf(name, "VarUI2FromDate");
break;
case 263:
sprintf(name, "VarUI2FromCy");
break;
case 264:
sprintf(name, "VarUI2FromStr");
break;
case 265:
sprintf(name, "VarUI2FromDisp");
break;
case 266:
sprintf(name, "VarUI2FromBool");
break;
case 267:
sprintf(name, "VarUI2FromI1");
break;
case 268:
sprintf(name, "VarUI2FromUI4");
break;
case 269:
sprintf(name, "VarUI2FromDec");
break;
case 270:
sprintf(name, "VarUI4FromUI1");
break;
case 271:
sprintf(name, "VarUI4FromI2");
break;
case 272:
sprintf(name, "VarUI4FromI4");
break;
case 273:
sprintf(name, "VarUI4FromR4");
break;
case 274:
sprintf(name, "VarUI4FromR8");
break;
case 275:
sprintf(name, "VarUI4FromDate");
break;
case 276:
sprintf(name, "VarUI4FromCy");
break;
case 277:
sprintf(name, "VarUI4FromStr");
break;
case 278:
sprintf(name, "VarUI4FromDisp");
break;
case 279:
sprintf(name, "VarUI4FromBool");
break;
case 280:
sprintf(name, "VarUI4FromI1");
break;
case 281:
sprintf(name, "VarUI4FromUI2");
break;
case 282:
sprintf(name, "VarUI4FromDec");
break;
case 283:
sprintf(name, "BSTR_UserSize");
break;
case 284:
sprintf(name, "BSTR_UserMarshal");
break;
case 285:
sprintf(name, "BSTR_UserUnmarshal");
break;
case 286:
sprintf(name, "BSTR_UserFree");
break;
case 287:
sprintf(name, "VARIANT_UserSize");
break;
case 288:
sprintf(name, "VARIANT_UserMarshal");
break;
case 289:
sprintf(name, "VARIANT_UserUnmarshal");
break;
case 290:
sprintf(name, "VARIANT_UserFree");
break;
case 291:
sprintf(name, "LPSAFEARRAY_UserSize");
break;
case 292:
sprintf(name, "LPSAFEARRAY_UserMarshal");
break;
case 293:
sprintf(name, "LPSAFEARRAY_UserUnmarshal");
break;
case 294:
sprintf(name, "LPSAFEARRAY_UserFree");
break;
case 295:
sprintf(name, "LPSAFEARRAY_Size");
break;
case 296:
sprintf(name, "LPSAFEARRAY_Marshal");
break;
case 297:
sprintf(name, "LPSAFEARRAY_Unmarshal");
break;
case 298:
sprintf(name, "VarDecCmpR8");
break;
case 299:
sprintf(name, "VarCyAdd");
break;
case 300:
sprintf(name, "DllUnregisterServer");
break;
case 301:
sprintf(name, "OACreateTypeLib2");
break;
case 303:
sprintf(name, "VarCyMul");
break;
case 304:
sprintf(name, "VarCyMulI4");
break;
case 305:
sprintf(name, "VarCySub");
break;
case 306:
sprintf(name, "VarCyAbs");
break;
case 307:
sprintf(name, "VarCyFix");
break;
case 308:
sprintf(name, "VarCyInt");
break;
case 309:
sprintf(name, "VarCyNeg");
break;
case 310:
sprintf(name, "VarCyRound");
break;
case 311:
sprintf(name, "VarCyCmp");
break;
case 312:
sprintf(name, "VarCyCmpR8");
break;
case 313:
sprintf(name, "VarBstrCat");
break;
case 314:
sprintf(name, "VarBstrCmp");
break;
case 315:
sprintf(name, "VarR8Pow");
break;
case 316:
sprintf(name, "VarR4CmpR8");
break;
case 317:
sprintf(name, "VarR8Round");
break;
case 318:
sprintf(name, "VarCat");
break;
case 319:
sprintf(name, "VarDateFromUdateEx");
break;
case 322:
sprintf(name, "GetRecordInfoFromGuids");
break;
case 323:
sprintf(name, "GetRecordInfoFromTypeInfo");
break;
case 325:
sprintf(name, "SetVarConversionLocaleSetting");
break;
case 326:
sprintf(name, "GetVarConversionLocaleSetting");
break;
case 327:
sprintf(name, "SetOaNoCache");
break;
case 329:
sprintf(name, "VarCyMulI8");
break;
case 330:
sprintf(name, "VarDateFromUdate");
break;
case 331:
sprintf(name, "VarUdateFromDate");
break;
case 332:
sprintf(name, "GetAltMonthNames");
break;
case 333:
sprintf(name, "VarI8FromUI1");
break;
case 334:
sprintf(name, "VarI8FromI2");
break;
case 335:
sprintf(name, "VarI8FromR4");
break;
case 336:
sprintf(name, "VarI8FromR8");
break;
case 337:
sprintf(name, "VarI8FromCy");
break;
case 338:
sprintf(name, "VarI8FromDate");
break;
case 339:
sprintf(name, "VarI8FromStr");
break;
case 340:
sprintf(name, "VarI8FromDisp");
break;
case 341:
sprintf(name, "VarI8FromBool");
break;
case 342:
sprintf(name, "VarI8FromI1");
break;
case 343:
sprintf(name, "VarI8FromUI2");
break;
case 344:
sprintf(name, "VarI8FromUI4");
break;
case 345:
sprintf(name, "VarI8FromDec");
break;
case 346:
sprintf(name, "VarI2FromI8");
break;
case 347:
sprintf(name, "VarI2FromUI8");
break;
case 348:
sprintf(name, "VarI4FromI8");
break;
case 349:
sprintf(name, "VarI4FromUI8");
break;
case 360:
sprintf(name, "VarR4FromI8");
break;
case 361:
sprintf(name, "VarR4FromUI8");
break;
case 362:
sprintf(name, "VarR8FromI8");
break;
case 363:
sprintf(name, "VarR8FromUI8");
break;
case 364:
sprintf(name, "VarDateFromI8");
break;
case 365:
sprintf(name, "VarDateFromUI8");
break;
case 366:
sprintf(name, "VarCyFromI8");
break;
case 367:
sprintf(name, "VarCyFromUI8");
break;
case 368:
sprintf(name, "VarBstrFromI8");
break;
case 369:
sprintf(name, "VarBstrFromUI8");
break;
case 370:
sprintf(name, "VarBoolFromI8");
break;
case 371:
sprintf(name, "VarBoolFromUI8");
break;
case 372:
sprintf(name, "VarUI1FromI8");
break;
case 373:
sprintf(name, "VarUI1FromUI8");
break;
case 374:
sprintf(name, "VarDecFromI8");
break;
case 375:
sprintf(name, "VarDecFromUI8");
break;
case 376:
sprintf(name, "VarI1FromI8");
break;
case 377:
sprintf(name, "VarI1FromUI8");
break;
case 378:
sprintf(name, "VarUI2FromI8");
break;
case 379:
sprintf(name, "VarUI2FromUI8");
break;
case 401:
sprintf(name, "OleLoadPictureEx");
break;
case 402:
sprintf(name, "OleLoadPictureFileEx");
break;
case 411:
sprintf(name, "SafeArrayCreateVector");
break;
case 412:
sprintf(name, "SafeArrayCopyData");
break;
case 413:
sprintf(name, "VectorFromBstr");
break;
case 414:
sprintf(name, "BstrFromVector");
break;
case 415:
sprintf(name, "OleIconToCursor");
break;
case 416:
sprintf(name, "OleCreatePropertyFrameIndirect");
break;
case 417:
sprintf(name, "OleCreatePropertyFrame");
break;
case 418:
sprintf(name, "OleLoadPicture");
break;
case 419:
sprintf(name, "OleCreatePictureIndirect");
break;
case 420:
sprintf(name, "OleCreateFontIndirect");
break;
case 421:
sprintf(name, "OleTranslateColor");
break;
case 422:
sprintf(name, "OleLoadPictureFile");
break;
case 423:
sprintf(name, "OleSavePictureFile");
break;
case 424:
sprintf(name, "OleLoadPicturePath");
break;
case 425:
sprintf(name, "VarUI4FromI8");
break;
case 426:
sprintf(name, "VarUI4FromUI8");
break;
case 427:
sprintf(name, "VarI8FromUI8");
break;
case 428:
sprintf(name, "VarUI8FromI8");
break;
case 429:
sprintf(name, "VarUI8FromUI1");
break;
case 430:
sprintf(name, "VarUI8FromI2");
break;
case 431:
sprintf(name, "VarUI8FromR4");
break;
case 432:
sprintf(name, "VarUI8FromR8");
break;
case 433:
sprintf(name, "VarUI8FromCy");
break;
case 434:
sprintf(name, "VarUI8FromDate");
break;
case 435:
sprintf(name, "VarUI8FromStr");
break;
case 436:
sprintf(name, "VarUI8FromDisp");
break;
case 437:
sprintf(name, "VarUI8FromBool");
break;
case 438:
sprintf(name, "VarUI8FromI1");
break;
case 439:
sprintf(name, "VarUI8FromUI2");
break;
case 440:
sprintf(name, "VarUI8FromUI4");
break;
case 441:
sprintf(name, "VarUI8FromDec");
break;
case 442:
sprintf(name, "RegisterTypeLibForUser");
break;
case 443:
sprintf(name, "UnRegisterTypeLibForUser");
break;
default:
break;
}
}
2009-11-11 03:58:22 +01:00
if (name[0] == '\0')
sprintf(name, "ord%u", ord);
return cli_strdup(name);
}
2016-06-27 17:55:16 -04:00
static int validate_impname(const char *name, uint32_t length, int dll)
{
uint32_t i = 0;
2016-06-27 17:55:16 -04:00
const char *c = name;
if (!name || length == 0)
2016-07-12 11:53:59 -04:00
return 1;
2016-06-27 17:55:16 -04:00
while (i < length && *c != '\0') {
if ((*c >= '0' && *c <= '9') ||
(*c >= 'a' && *c <= 'z') ||
(*c >= 'A' && *c <= 'Z') ||
(*c == '_') ||
(dll && *c == '.')) {
c++;
i++;
} else
2016-07-12 11:53:59 -04:00
return 0;
2016-06-27 17:55:16 -04:00
}
2016-07-12 11:53:59 -04:00
return 1;
2016-06-27 17:55:16 -04:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
static inline int hash_impfns(cli_ctx *ctx, void **hashctx, uint32_t *impsz, struct pe_image_import_descriptor *image, const char *dllname, struct cli_exe_info *peinfo, int *first)
{
uint32_t thuoff = 0, offset;
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
fmap_t *map = ctx->fmap;
size_t dlllen = 0, fsize = map->len;
unsigned int err = 0;
int num_fns = 0, ret = CL_SUCCESS;
const char *buffer;
cli_hash_type_t type;
#if HAVE_JSON
json_object *imptbl = NULL;
#else
void *imptbl = NULL;
#endif
if (image->u.OriginalFirstThunk)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
thuoff = cli_rawaddr(image->u.OriginalFirstThunk, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err || thuoff == 0)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
thuoff = cli_rawaddr(image->FirstThunk, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err) {
cli_dbgmsg("scan_pe: invalid rva for image first thunk\n");
2016-07-12 11:53:59 -04:00
return CL_EFORMAT;
}
#if HAVE_JSON
if (ctx->wrkproperty) {
imptbl = cli_jsonarray(ctx->wrkproperty, "ImportTable");
if (!imptbl) {
cli_dbgmsg("scan_pe: cannot allocate import table json object\n");
return CL_EMEM;
}
}
#endif
#define UPDATE_IMPHASH() \
do { \
if (funcname) { \
size_t i, j; \
char *fname; \
size_t funclen; \
\
if (dlllen == 0) { \
char *ext = strstr(dllname, "."); \
\
if (ext && (strncasecmp(ext, ".ocx", 4) == 0 || \
strncasecmp(ext, ".sys", 4) == 0 || \
strncasecmp(ext, ".dll", 4) == 0)) \
dlllen = ext - dllname; \
else \
dlllen = strlen(dllname); \
} \
\
funclen = strlen(funcname); \
if (validate_impname(funcname, funclen, 1) == 0) { \
cli_dbgmsg("scan_pe: invalid name for imported function\n"); \
ret = CL_EFORMAT; \
break; \
} \
\
fname = cli_calloc(funclen + dlllen + 3, sizeof(char)); \
if (fname == NULL) { \
cli_dbgmsg("scan_pe: cannot allocate memory for imphash string\n"); \
ret = CL_EMEM; \
break; \
} \
j = 0; \
if (!*first) \
fname[j++] = ','; \
for (i = 0; i < dlllen; i++, j++) \
fname[j] = tolower(dllname[i]); \
fname[j++] = '.'; \
for (i = 0; i < funclen; i++, j++) \
fname[j] = tolower(funcname[i]); \
\
if (imptbl) { \
char *jname = *first ? fname : fname + 1; \
cli_jsonstr(imptbl, NULL, jname); \
} \
\
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) \
cl_update_hash(hashctx[type], fname, strlen(fname)); \
*impsz += strlen(fname); \
\
*first = 0; \
free(fname); \
} \
} while (0)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->is_pe32plus) {
struct pe_image_thunk32 thunk32;
while ((num_fns < PE_MAXIMPORTS) && (fmap_readn(map, &thunk32, thuoff, sizeof(struct pe_image_thunk32)) == sizeof(struct pe_image_thunk32)) && (thunk32.u.Ordinal != 0)) {
char *funcname = NULL;
thuoff += sizeof(struct pe_image_thunk32);
2016-07-12 11:53:59 -04:00
thunk32.u.Ordinal = EC32(thunk32.u.Ordinal);
if (!(thunk32.u.Ordinal & PE_IMAGEDIR_ORDINAL_FLAG32)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
offset = cli_rawaddr(thunk32.u.Function, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (!ret) {
/* Hint field is a uint16_t and precedes the Name field */
if ((buffer = fmap_need_off_once(map, offset + sizeof(uint16_t), MIN(PE_MAXNAMESIZE, fsize - offset))) != NULL) {
funcname = CLI_STRNDUP(buffer, MIN(PE_MAXNAMESIZE, fsize - offset));
if (funcname == NULL) {
cli_dbgmsg("scan_pe: cannot duplicate function name\n");
return CL_EMEM;
}
}
}
} else {
/* ordinal lookup */
funcname = pe_ordinal(dllname, thunk32.u.Ordinal & 0xFFFF);
if (funcname == NULL) {
cli_dbgmsg("scan_pe: cannot duplicate function name\n");
return CL_EMEM;
}
}
UPDATE_IMPHASH();
2016-07-12 11:53:59 -04:00
free(funcname);
if (ret != CL_SUCCESS)
return ret;
}
} else {
struct pe_image_thunk64 thunk64;
while ((num_fns < PE_MAXIMPORTS) && (fmap_readn(map, &thunk64, thuoff, sizeof(struct pe_image_thunk64)) == sizeof(struct pe_image_thunk64)) && (thunk64.u.Ordinal != 0)) {
char *funcname = NULL;
thuoff += sizeof(struct pe_image_thunk64);
2016-07-12 11:53:59 -04:00
thunk64.u.Ordinal = EC64(thunk64.u.Ordinal);
if (!(thunk64.u.Ordinal & PE_IMAGEDIR_ORDINAL_FLAG64)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
offset = cli_rawaddr(thunk64.u.Function, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (!err) {
/* Hint field is a uint16_t and precedes the Name field */
if ((buffer = fmap_need_off_once(map, offset + sizeof(uint16_t), MIN(PE_MAXNAMESIZE, fsize - offset))) != NULL) {
funcname = CLI_STRNDUP(buffer, MIN(PE_MAXNAMESIZE, fsize - offset));
if (funcname == NULL) {
cli_dbgmsg("scan_pe: cannot duplicate function name\n");
return CL_EMEM;
}
}
}
} else {
/* ordinal lookup */
funcname = pe_ordinal(dllname, thunk64.u.Ordinal & 0xFFFF);
if (funcname == NULL) {
cli_dbgmsg("scan_pe: cannot duplicate function name\n");
return CL_EMEM;
}
}
UPDATE_IMPHASH();
2016-07-12 11:53:59 -04:00
free(funcname);
if (ret != CL_SUCCESS)
return ret;
}
}
return CL_SUCCESS;
}
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
static cl_error_t hash_imptbl(cli_ctx *ctx, unsigned char **digest, uint32_t *impsz, int *genhash, struct cli_exe_info *peinfo)
{
cl_error_t status = CL_ERROR;
cl_error_t ret;
struct pe_image_import_descriptor image = {0};
const struct pe_image_import_descriptor *impdes;
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
fmap_t *map = ctx->fmap;
size_t left, fsize = map->len;
uint32_t impoff, offset;
const char *buffer;
void *hashctx[CLI_HASH_AVAIL_TYPES] = {0};
cli_hash_type_t type;
int nimps = 0;
unsigned int err;
int first = 1;
bool needed_impoff = false;
/* If the PE doesn't have an import table then skip it. This is an
* uncommon case but can happen. */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->dirs[1].VirtualAddress == 0 || peinfo->dirs[1].Size == 0) {
cli_dbgmsg("scan_pe: import table data dir does not exist (skipping .imp scanning)\n");
status = CL_BREAK;
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Add EC32 wrappers
impoff = cli_rawaddr(peinfo->dirs[1].VirtualAddress, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err || impoff + peinfo->dirs[1].Size > fsize) {
cli_dbgmsg("scan_pe: invalid rva for import table data\n");
status = CL_BREAK;
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Add EC32 wrapper
impdes = (const struct pe_image_import_descriptor *)fmap_need_off(map, impoff, peinfo->dirs[1].Size);
if (impdes == NULL) {
cli_dbgmsg("scan_pe: failed to acquire fmap buffer\n");
status = CL_EREAD;
goto done;
}
needed_impoff = true;
/* Safety: We can trust peinfo->dirs[1].Size only because `fmap_need_off()` (above)
* would have failed if the size exceeds the end of the fmap. */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
left = peinfo->dirs[1].Size;
if (genhash[CLI_HASH_MD5]) {
hashctx[CLI_HASH_MD5] = cl_hash_init("md5");
2016-07-12 11:53:59 -04:00
if (hashctx[CLI_HASH_MD5] == NULL) {
status = CL_EMEM;
goto done;
2016-07-12 11:53:59 -04:00
}
}
if (genhash[CLI_HASH_SHA1]) {
hashctx[CLI_HASH_SHA1] = cl_hash_init("sha1");
2016-07-12 11:53:59 -04:00
if (hashctx[CLI_HASH_SHA1] == NULL) {
status = CL_EMEM;
goto done;
2016-07-12 11:53:59 -04:00
}
}
if (genhash[CLI_HASH_SHA256]) {
hashctx[CLI_HASH_SHA256] = cl_hash_init("sha256");
2016-07-12 11:53:59 -04:00
if (hashctx[CLI_HASH_SHA256] == NULL) {
status = CL_EMEM;
goto done;
2016-07-12 11:53:59 -04:00
}
}
while (left > sizeof(struct pe_image_import_descriptor) && nimps < PE_MAXIMPORTS) {
char *dllname = NULL;
/* Get copy of image import descriptor to work with */
memcpy(&image, impdes, sizeof(struct pe_image_import_descriptor));
if (image.Name == 0) {
// Name RVA is 0, which doesn't seem right. I guess we skip the rest?
// TODO: Is that right?
break;
}
/* Prepare for next iteration, in case we need to `continue;` */
left -= sizeof(struct pe_image_import_descriptor);
nimps++;
impdes++;
2016-07-12 11:53:59 -04:00
/* Endian Conversion */
image.u.OriginalFirstThunk = EC32(image.u.OriginalFirstThunk);
image.TimeDateStamp = EC32(image.TimeDateStamp);
image.ForwarderChain = EC32(image.ForwarderChain);
image.Name = EC32(image.Name);
image.FirstThunk = EC32(image.FirstThunk);
2016-07-12 11:53:59 -04:00
Spelling Adjustments (#30) * spelling: accessed * spelling: alignment * spelling: amalgamated * spelling: answers * spelling: another * spelling: acquisition * spelling: apitid * spelling: ascii * spelling: appending * spelling: appropriate * spelling: arbitrary * spelling: architecture * spelling: asynchronous * spelling: attachments * spelling: argument * spelling: authenticode * spelling: because * spelling: boundary * spelling: brackets * spelling: bytecode * spelling: calculation * spelling: cannot * spelling: changes * spelling: check * spelling: children * spelling: codegen * spelling: commands * spelling: container * spelling: concatenated * spelling: conditions * spelling: continuous * spelling: conversions * spelling: corresponding * spelling: corrupted * spelling: coverity * spelling: crafting * spelling: daemon * spelling: definition * spelling: delivered * spelling: delivery * spelling: delimit * spelling: dependencies * spelling: dependency * spelling: detection * spelling: determine * spelling: disconnects * spelling: distributed * spelling: documentation * spelling: downgraded * spelling: downloading * spelling: endianness * spelling: entities * spelling: especially * spelling: empty * spelling: expected * spelling: explicitly * spelling: existent * spelling: finished * spelling: flexibility * spelling: flexible * spelling: freshclam * spelling: functions * spelling: guarantee * spelling: hardened * spelling: headaches * spelling: heighten * spelling: improper * spelling: increment * spelling: indefinitely * spelling: independent * spelling: inaccessible * spelling: infrastructure Conflicts: docs/html/node68.html * spelling: initializing * spelling: inited * spelling: instream * spelling: installed * spelling: initialization * spelling: initialize * spelling: interface * spelling: intrinsics * spelling: interpreter * spelling: introduced * spelling: invalid * spelling: latency * spelling: lawyers * spelling: libclamav * spelling: likelihood * spelling: loop * spelling: maximum * spelling: million * spelling: milliseconds * spelling: minimum * spelling: minzhuan * spelling: multipart * spelling: misled * spelling: modifiers * spelling: notifying * spelling: objects * spelling: occurred * spelling: occurs * spelling: occurrences * spelling: optimization * spelling: original * spelling: originated * spelling: output * spelling: overridden * spelling: parenthesis * spelling: partition * spelling: performance * spelling: permission * spelling: phishing * spelling: portions * spelling: positives * spelling: preceded * spelling: properties * spelling: protocol * spelling: protos * spelling: quarantine * spelling: recursive * spelling: referring * spelling: reorder * spelling: reset * spelling: resources * spelling: resume * spelling: retrieval * spelling: rewrite * spelling: sanity * spelling: scheduled * spelling: search * spelling: section * spelling: separator * spelling: separated * spelling: specify * spelling: special * spelling: statement * spelling: streams * spelling: succession * spelling: suggests * spelling: superfluous * spelling: suspicious * spelling: synonym * spelling: temporarily * spelling: testfiles * spelling: transverse * spelling: turkish * spelling: typos * spelling: unable * spelling: unexpected * spelling: unexpectedly * spelling: unfinished * spelling: unfortunately * spelling: uninitialized * spelling: unlocking * spelling: unnecessary * spelling: unpack * spelling: unrecognized * spelling: unsupported * spelling: usable * spelling: wherever * spelling: wishlist * spelling: white * spelling: infrastructure * spelling: directories * spelling: overridden * spelling: permission * spelling: yesterday * spelling: initialization * spelling: intrinsics * space adjustment for spelling changes * minor modifications by klin
2018-02-21 15:00:59 -05:00
/* DLL name acquisition */
offset = cli_rawaddr(image.Name, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err || offset > fsize) {
cli_dbgmsg("scan_pe: invalid rva for dll name\n");
/* TODO: ignore or return? */
/*
continue;
*/
status = CL_EFORMAT;
goto done;
}
buffer = fmap_need_off_once(map, offset, MIN(PE_MAXNAMESIZE, fsize - offset));
2016-07-12 11:53:59 -04:00
if (buffer == NULL) {
cli_dbgmsg("scan_pe: failed to read name for dll\n");
status = CL_EREAD;
goto done;
2016-07-12 11:53:59 -04:00
}
if (validate_impname(dllname, MIN(PE_MAXNAMESIZE, fsize - offset), 1) == 0) {
2016-07-12 11:53:59 -04:00
cli_dbgmsg("scan_pe: invalid name for imported dll\n");
status = CL_EFORMAT;
goto done;
2016-07-12 11:53:59 -04:00
}
dllname = CLI_STRNDUP(buffer, MIN(PE_MAXNAMESIZE, fsize - offset));
2016-07-12 11:53:59 -04:00
if (dllname == NULL) {
cli_dbgmsg("scan_pe: cannot duplicate dll name\n");
status = CL_EMEM;
goto done;
}
/* DLL function handling - inline function */
ret = hash_impfns(ctx, hashctx, impsz, &image, dllname, peinfo, &first);
2016-07-12 11:53:59 -04:00
free(dllname);
dllname = NULL;
if (ret != CL_SUCCESS) {
status = ret;
goto done;
}
}
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
cl_finish_hash(hashctx[type], digest[type]);
hashctx[type] = NULL;
}
status = CL_SUCCESS;
done:
if (needed_impoff) {
fmap_unneed_off(map, impoff, peinfo->dirs[1].Size);
}
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
if (NULL != hashctx[type]) {
cl_hash_destroy(hashctx[type]);
}
}
return status;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
static int scan_pe_imp(cli_ctx *ctx, struct cli_exe_info *peinfo)
{
struct cli_matcher *imp = ctx->engine->hm_imp;
unsigned char *hashset[CLI_HASH_AVAIL_TYPES];
const char *virname = NULL;
int genhash[CLI_HASH_AVAIL_TYPES];
uint32_t impsz = 0;
cli_hash_type_t type;
int ret = CL_CLEAN;
/* pick hashtypes to generate */
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
genhash[type] = cli_hm_have_any(imp, type);
if (genhash[type]) {
2016-07-12 11:53:59 -04:00
hashset[type] = cli_malloc(hashlen[type]);
if (!hashset[type]) {
2016-07-12 11:53:59 -04:00
cli_errmsg("scan_pe: cli_malloc failed!\n");
for (; type > 0;)
2016-07-12 11:53:59 -04:00
free(hashset[--type]);
return CL_EMEM;
}
} else {
2016-07-12 11:53:59 -04:00
hashset[type] = NULL;
}
}
/* Force md5 hash generation for debug and preclass */
#if HAVE_JSON
if ((cli_debug_flag || ctx->wrkproperty) && !genhash[CLI_HASH_MD5]) {
#else
if (cli_debug_flag && !genhash[CLI_HASH_MD5]) {
#endif
genhash[CLI_HASH_MD5] = 1;
hashset[CLI_HASH_MD5] = cli_calloc(hashlen[CLI_HASH_MD5], sizeof(char));
if (!hashset[CLI_HASH_MD5]) {
cli_errmsg("scan_pe: cli_malloc failed!\n");
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++)
2016-07-12 11:53:59 -04:00
free(hashset[type]);
return CL_EMEM;
}
}
/* Generate hashes */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ret = hash_imptbl(ctx, hashset, &impsz, genhash, peinfo);
2016-07-12 11:53:59 -04:00
if (ret != CL_SUCCESS) {
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
2016-07-12 11:53:59 -04:00
free(hashset[type]);
}
if (ret == CL_BREAK) {
ret = CL_SUCCESS;
}
2016-07-12 11:53:59 -04:00
return ret;
}
/* Print hash */
#if HAVE_JSON
if (cli_debug_flag || ctx->wrkproperty) {
#else
if (cli_debug_flag) {
#endif
char *dstr = cli_str2hex((char *)hashset[CLI_HASH_MD5], hashlen[CLI_HASH_MD5]);
cli_dbgmsg("IMP: %s:%u\n", dstr ? (char *)dstr : "(NULL)", impsz);
#if HAVE_JSON
if (ctx->wrkproperty)
cli_jsonstr(ctx->wrkproperty, "Imphash", dstr ? dstr : "(NULL)");
#endif
if (dstr)
free(dstr);
}
/* Do scans */
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++) {
if (cli_hm_scan(hashset[type], impsz, &virname, imp, type) == CL_VIRUS) {
ret = cli_append_virus(ctx, virname);
if (ret != CL_CLEAN) {
if (ret != CL_VIRUS)
break;
else if (!SCAN_ALLMATCHES)
break;
}
2016-07-12 11:53:59 -04:00
}
if (cli_hm_scan_wild(hashset[type], &virname, imp, type) == CL_VIRUS) {
cli_append_virus(ctx, virname);
if (ret != CL_CLEAN) {
if (ret != CL_VIRUS)
break;
else if (!SCAN_ALLMATCHES)
break;
}
}
}
2016-06-22 16:28:02 -04:00
for (type = CLI_HASH_MD5; type < CLI_HASH_AVAIL_TYPES; type++)
2016-07-12 11:53:59 -04:00
free(hashset[type]);
return ret;
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
static struct json_object *get_pe_property(cli_ctx *ctx)
{
struct json_object *pe;
if (!(ctx) || !(ctx->wrkproperty))
return NULL;
if (!json_object_object_get_ex(ctx->wrkproperty, "PE", &pe)) {
2014-05-07 10:56:35 -04:00
pe = json_object_new_object();
if (!(pe))
return NULL;
json_object_object_add(ctx->wrkproperty, "PE", pe);
}
return pe;
}
static void pe_add_heuristic_property(cli_ctx *ctx, const char *key)
{
struct json_object *heuristics;
struct json_object *pe;
struct json_object *str;
pe = get_pe_property(ctx);
if (!(pe))
return;
if (!json_object_object_get_ex(pe, "Heuristics", &heuristics)) {
2014-05-07 10:56:35 -04:00
heuristics = json_object_new_array();
if (!(heuristics))
return;
json_object_object_add(pe, "Heuristics", heuristics);
}
str = json_object_new_string(key);
if (!(str))
return;
json_object_array_add(heuristics, str);
}
static struct json_object *get_section_json(cli_ctx *ctx)
{
struct json_object *pe;
struct json_object *section;
pe = get_pe_property(ctx);
if (!(pe))
return NULL;
if (!json_object_object_get_ex(pe, "Sections", &section)) {
2014-05-07 10:56:35 -04:00
section = json_object_new_array();
if (!(section))
return NULL;
json_object_object_add(pe, "Sections", section);
}
return section;
}
static void add_section_info(cli_ctx *ctx, struct cli_exe_section *s)
{
struct json_object *sections, *section, *obj;
char address[16];
sections = get_section_json(ctx);
if (!(sections))
return;
section = json_object_new_object();
if (!(section))
return;
obj = json_object_new_int((int32_t)(s->rsz));
if (!(obj))
return;
json_object_object_add(section, "RawSize", obj);
obj = json_object_new_int((int32_t)(s->raw));
if (!(obj))
return;
json_object_object_add(section, "RawOffset", obj);
snprintf(address, sizeof(address), "0x%08x", s->rva);
obj = json_object_new_string(address);
if (!(obj))
return;
json_object_object_add(section, "VirtualAddress", obj);
obj = json_object_new_boolean((s->chr & 0x20000000) == 0x20000000);
if ((obj))
json_object_object_add(section, "Executable", obj);
obj = json_object_new_boolean((s->chr & 0x80000000) == 0x80000000);
if ((obj))
json_object_object_add(section, "Writable", obj);
obj = json_object_new_boolean(s->urva >> 31 || s->uvsz >> 31 || (s->rsz && s->uraw >> 31) || s->ursz >> 31);
if ((obj))
json_object_object_add(section, "Signed", obj);
2014-05-07 10:56:35 -04:00
json_object_array_add(sections, section);
}
#endif
int cli_scanpe(cli_ctx *ctx)
2004-05-11 23:30:57 +00:00
{
2014-11-07 10:15:30 -05:00
uint8_t polipos = 0;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
char epbuff[4096], *tempfile;
size_t epsize;
size_t bytes;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
unsigned int i, j, found, upx_success = 0, err;
unsigned int ssize = 0, dsize = 0, corrupted_cur;
2014-11-07 10:15:30 -05:00
int (*upxfn)(const char *, uint32_t, char *, uint32_t *, uint32_t, uint32_t, uint32_t) = NULL;
const char *src = NULL;
char *dest = NULL;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
int ndesc, ret = CL_CLEAN, upack = 0;
2014-11-07 10:15:30 -05:00
size_t fsize;
struct cli_bc_ctx *bc_ctx;
fmap_t *map;
struct cli_pe_hook_data pedata;
#ifdef HAVE__INTERNAL__SHA_COLLECT
2014-11-07 10:15:30 -05:00
int sha_collect = ctx->sha_collect;
#endif
2014-11-07 10:15:30 -05:00
uint32_t viruses_found = 0;
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
int toval = 0;
struct json_object *pe_json = NULL;
2014-05-07 10:56:35 -04:00
#endif
2004-05-11 23:30:57 +00:00
if (!ctx) {
2014-11-07 10:15:30 -05:00
cli_errmsg("cli_scanpe: ctx == NULL\n");
return CL_ENULLARG;
2007-10-05 21:48:57 +00:00
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (cli_json_timeout_cycle_check(ctx, &toval) != CL_SUCCESS) {
return CL_ETIMEOUT;
}
if (SCAN_COLLECT_METADATA) {
2014-05-07 10:56:35 -04:00
pe_json = get_pe_property(ctx);
}
2014-05-07 10:56:35 -04:00
#endif
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
map = ctx->fmap;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
fsize = map->len;
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
struct cli_exe_info _peinfo;
struct cli_exe_info *peinfo = &_peinfo;
2014-11-07 10:15:30 -05:00
uint32_t opts = CLI_PEHEADER_OPT_DBG_PRINT_INFO | CLI_PEHEADER_OPT_REMOVE_MISSING_SECTIONS;
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (SCAN_COLLECT_METADATA) {
opts |= CLI_PEHEADER_OPT_COLLECT_JSON;
2004-05-11 23:30:57 +00:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (DETECT_BROKEN_PE) {
opts |= CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS;
2004-05-11 23:30:57 +00:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_init(peinfo, 0);
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ret = cli_peheader(map, peinfo, opts, ctx);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// Warn the user if PE header parsing failed - if it's a binary that runs
// successfully on Windows, we need to relax our PE parsing standards so
// that we make sure the executable gets scanned appropriately
2014-05-07 10:56:35 -04:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#define PE_HDR_PARSE_FAIL_CONSEQUENCE "won't attempt .mdb / .imp / PE-specific BC rule matching or exe unpacking\n"
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (CLI_PEHEADER_RET_BROKEN_PE == ret) {
if (DETECT_BROKEN_PE) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Handle allmatch
ret = cli_append_potentially_unwanted(ctx, "Heuristics.Broken.Executable");
cli_exe_info_destroy(peinfo);
return ret;
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: PE header appears broken - " PE_HDR_PARSE_FAIL_CONSEQUENCE);
cli_exe_info_destroy(peinfo);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
return CL_CLEAN;
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
} else if (CLI_PEHEADER_RET_JSON_TIMEOUT == ret) {
cli_dbgmsg("cli_scanpe: JSON creation timed out - " PE_HDR_PARSE_FAIL_CONSEQUENCE);
cli_exe_info_destroy(peinfo);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
return CL_ETIMEOUT;
} else if (CLI_PEHEADER_RET_GENERIC_ERROR == ret) {
cli_dbgmsg("cli_scanpe: An error occurred when parsing the PE header - " PE_HDR_PARSE_FAIL_CONSEQUENCE);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->is_pe32plus) { /* PE */
if (DCONF & PE_CONF_UPACK)
upack = (EC16(peinfo->file_hdr.SizeOfOptionalHeader) == 0x148);
}
for (i = 0; i < peinfo->nsections; i++) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->sections[i].rsz) { /* Don't bother with virtual only sections */
// TODO Regarding the commented out check below:
// This used to check that the section name was NULL, but now that
// header parsing is done in cli_peheader (and since we don't yet
// make the section name availabe via peinfo->sections[]) it would
// be a pain to fetch the name here. Since this is the only place
// in cli_scanpe that needs the section name, and since I verified
// that detection still occurs for Polipos without this check,
// let's leave it commented out for now.
if (SCAN_HEURISTICS && (DCONF & PE_CONF_POLIPOS) && /*!*peinfo->sections[i].sname &&*/ peinfo->sections[i].vsz > 40000 && peinfo->sections[i].vsz < 70000 && peinfo->sections[i].chr == 0xe0000060) polipos = i;
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* check hash section sigs */
if ((DCONF & PE_CONF_MD5SECT) && ctx->engine->hm_mdb) {
ret = scan_pe_mdb(ctx, &(peinfo->sections[i]));
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS && !SCAN_ALLMATCHES) {
cli_dbgmsg("------------------------------------\n");
cli_exe_info_destroy(peinfo);
return ret;
} else if (ret != CL_VIRUS) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("cli_scanpe: scan_pe_mdb failed: %s!\n", cl_strerror(ret));
cli_dbgmsg("------------------------------------\n");
cli_exe_info_destroy(peinfo);
return ret;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
2014-11-07 10:15:30 -05:00
}
2004-05-11 23:30:57 +00:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Don't bail out here
if (peinfo->is_pe32plus) { /* Do not continue for PE32+ files */
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
epsize = fmap_readn(map, epbuff, peinfo->ep, 4096);
if ((size_t)-1 == epsize) {
/* Do not continue, all future logic requires at least a partial read into epbuff */
cli_exe_info_destroy(peinfo);
return CL_CLEAN;
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* Disasm scan disabled since it's now handled by the bytecode */
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* CLI_UNPTEMP("cli_scanpe: DISASM",(peinfo->sections,0)); */
/* if(disasmbuf((unsigned char*)epbuff, epsize, ndesc)) */
/* ret = cli_scan_desc(ndesc, ctx, CL_TYPE_PE_DISASM, true, NULL, AC_SCAN_VIR); */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* close(ndesc); */
/* if(ret == CL_VIRUS) { */
/* cli_exe_info_destroy(peinfo); */
/* CLI_TMPUNLK(); */
/* free(tempfile); */
/* return ret; */
/* } */
/* CLI_TMPUNLK(); */
/* free(tempfile); */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->overlay_start && peinfo->overlay_size > 0) {
ret = cli_scanishield(ctx, peinfo->overlay_start, peinfo->overlay_size);
if (ret != CL_CLEAN) {
// TODO Handle allmatch
cli_exe_info_destroy(peinfo);
return ret;
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
pedata.nsections = peinfo->nsections;
pedata.ep = peinfo->ep;
pedata.offset = 0;
memcpy(&pedata.file_hdr, &(peinfo->file_hdr), sizeof(peinfo->file_hdr));
// TODO no need to copy both of these for each binary
memcpy(&pedata.opt32, &(peinfo->pe_opt.opt32), sizeof(peinfo->pe_opt.opt32));
memcpy(&pedata.opt64, &(peinfo->pe_opt.opt64), sizeof(peinfo->pe_opt.opt64));
memcpy(&pedata.dirs, &(peinfo->dirs), sizeof(peinfo->dirs));
// Gross
memcpy(&pedata.opt32_dirs, &(peinfo->dirs), sizeof(peinfo->dirs));
memcpy(&pedata.opt64_dirs, &(peinfo->dirs), sizeof(peinfo->dirs));
pedata.e_lfanew = peinfo->e_lfanew;
pedata.overlays = peinfo->overlay_start;
pedata.overlays_sz = peinfo->overlay_size;
pedata.hdr_size = peinfo->hdr_size;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* Bytecode BC_PE_ALL hook */
bc_ctx = cli_bytecode_context_alloc();
if (!bc_ctx) {
cli_errmsg("cli_scanpe: can't allocate memory for bc_ctx\n");
cli_exe_info_destroy(peinfo);
return CL_EMEM;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_bytecode_context_setpe(bc_ctx, &pedata, peinfo->sections);
cli_bytecode_context_setctx(bc_ctx, ctx);
ret = cli_bytecode_runhook(ctx, ctx->engine, bc_ctx, BC_PE_ALL, map);
switch (ret) {
case CL_ENULLARG:
cli_warnmsg("cli_scanpe: NULL argument supplied\n");
break;
case CL_VIRUS:
case CL_BREAK:
// TODO Handle allmatch
cli_exe_info_destroy(peinfo);
cli_bytecode_context_destroy(bc_ctx);
return ret == CL_VIRUS ? CL_VIRUS : CL_CLEAN;
}
cli_bytecode_context_destroy(bc_ctx);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* Attempt to run scans on import table */
/* Run if there are existing signatures and/or preclassing */
#if HAVE_JSON
if (DCONF & PE_CONF_IMPTBL && (ctx->engine->hm_imp || ctx->wrkproperty)) {
#else
if (DCONF & PE_CONF_IMPTBL && ctx->engine->hm_imp) {
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ret = scan_pe_imp(ctx, peinfo);
switch (ret) {
case CL_SUCCESS:
break;
case CL_ENULLARG:
cli_warnmsg("cli_scanpe: NULL argument supplied\n");
break;
case CL_VIRUS:
if (SCAN_ALLMATCHES)
break;
/* intentional fall-through */
case CL_BREAK:
cli_exe_info_destroy(peinfo);
return ret == CL_VIRUS ? CL_VIRUS : CL_CLEAN;
default:
cli_exe_info_destroy(peinfo);
return ret;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
/* Attempt to detect some popular polymorphic viruses */
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* W32.Parite.B */
if (SCAN_HEURISTICS && (DCONF & PE_CONF_PARITE) && !peinfo->is_dll && epsize == 4096 && peinfo->ep == peinfo->sections[peinfo->nsections - 1].raw) {
const char *pt = cli_memstr(epbuff, 4040, "\x47\x65\x74\x50\x72\x6f\x63\x41\x64\x64\x72\x65\x73\x73\x00", 15);
if (pt) {
pt += 15;
if ((((uint32_t)cli_readint32(pt) ^ (uint32_t)cli_readint32(pt + 4)) == 0x505a4f) && (((uint32_t)cli_readint32(pt + 8) ^ (uint32_t)cli_readint32(pt + 12)) == 0xffffb) && (((uint32_t)cli_readint32(pt + 16) ^ (uint32_t)cli_readint32(pt + 20)) == 0xb8)) {
ret = cli_append_potentially_unwanted(ctx, "Heuristics.W32.Parite.B");
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
cli_exe_info_destroy(peinfo);
return ret;
}
}
}
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* Kriz */
if (SCAN_HEURISTICS && (DCONF & PE_CONF_KRIZ) && epsize >= 200 && CLI_ISCONTAINED(peinfo->sections[peinfo->nsections - 1].raw, peinfo->sections[peinfo->nsections - 1].rsz, peinfo->ep, 0x0fd2) && epbuff[1] == '\x9c' && epbuff[2] == '\x60') {
enum { KZSTRASH,
KZSCDELTA,
KZSPDELTA,
KZSGETSIZE,
KZSXORPRFX,
KZSXOR,
KZSDDELTA,
KZSLOOP,
KZSTOP };
uint8_t kzs[] = {KZSTRASH, KZSCDELTA, KZSPDELTA, KZSGETSIZE, KZSTRASH, KZSXORPRFX, KZSXOR, KZSTRASH, KZSDDELTA, KZSTRASH, KZSLOOP, KZSTOP};
uint8_t *kzstate = kzs;
uint8_t *kzcode = (uint8_t *)epbuff + 3;
uint8_t kzdptr = 0xff, kzdsize = 0xff;
int kzlen = 197, kzinitlen = 0xffff, kzxorlen = -1;
cli_dbgmsg("cli_scanpe: in kriz\n");
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
while (*kzstate != KZSTOP) {
uint8_t op;
if (kzlen <= 6)
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
op = *kzcode++;
kzlen--;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
switch (*kzstate) {
case KZSTRASH:
case KZSGETSIZE: {
int opsz = 0;
switch (op) {
case 0x81:
kzcode += 5;
kzlen -= 5;
break;
case 0xb8:
case 0xb9:
case 0xba:
case 0xbb:
case 0xbd:
case 0xbe:
case 0xbf:
if (*kzstate == KZSGETSIZE && cli_readint32(kzcode) == 0x0fd2) {
kzinitlen = kzlen - 5;
kzdsize = op - 0xb8;
kzstate++;
op = 4; /* fake the register to avoid breaking out */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: kriz: using #%d as size counter\n", kzdsize);
}
opsz = 4;
/* fall-through */
case 0x48:
case 0x49:
case 0x4a:
case 0x4b:
case 0x4d:
case 0x4e:
case 0x4f:
op &= 7;
if (op != kzdptr && op != kzdsize) {
kzcode += opsz;
kzlen -= opsz;
break;
}
/* fall-through */
default:
kzcode--;
kzlen++;
kzstate++;
2014-11-07 10:15:30 -05:00
}
break;
2014-11-07 10:15:30 -05:00
}
case KZSCDELTA:
if (op == 0xe8 && (uint32_t)cli_readint32(kzcode) < 0xff) {
kzlen -= *kzcode + 4;
kzcode += *kzcode + 4;
kzstate++;
} else {
*kzstate = KZSTOP;
}
2014-11-07 10:15:30 -05:00
break;
case KZSPDELTA:
if ((op & 0xf8) == 0x58 && (kzdptr = op - 0x58) != 4) {
kzstate++;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: kriz: using #%d as pointer\n", kzdptr);
} else {
*kzstate = KZSTOP;
}
2014-11-07 10:15:30 -05:00
break;
case KZSXORPRFX:
2014-11-07 10:15:30 -05:00
kzstate++;
if (op == 0x3e) {
break;
}
/* fall-through */
case KZSXOR:
if (op == 0x80 && *kzcode == kzdptr + 0xb0) {
kzxorlen = kzlen;
kzcode += +6;
kzlen -= +6;
kzstate++;
} else {
*kzstate = KZSTOP;
}
2014-11-07 10:15:30 -05:00
break;
case KZSDDELTA:
if (op == kzdptr + 0x48)
kzstate++;
else
*kzstate = KZSTOP;
2014-11-07 10:15:30 -05:00
break;
case KZSLOOP:
if (op == kzdsize + 0x48 && *kzcode == 0x75 && kzlen - (int8_t)kzcode[1] - 3 <= kzinitlen && kzlen - (int8_t)kzcode[1] >= kzxorlen) {
ret = cli_append_potentially_unwanted(ctx, "Heuristics.W32.Kriz");
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
}
}
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: kriz: loop out of bounds, corrupted sample?\n");
kzstate++;
2014-11-07 10:15:30 -05:00
}
}
}
/* W32.Magistr.A/B */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (SCAN_HEURISTICS && (DCONF & PE_CONF_MAGISTR) && !peinfo->is_dll && (peinfo->nsections > 1) && (peinfo->sections[peinfo->nsections - 1].chr & 0x80000000)) {
uint32_t rsize, vsize, dam = 0;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
vsize = peinfo->sections[peinfo->nsections - 1].uvsz;
rsize = peinfo->sections[peinfo->nsections - 1].rsz;
if (rsize < peinfo->sections[peinfo->nsections - 1].ursz) {
rsize = peinfo->sections[peinfo->nsections - 1].ursz;
dam = 1;
2014-11-07 10:15:30 -05:00
}
if (vsize >= 0x612c && rsize >= 0x612c && ((vsize & 0xff) == 0xec)) {
2014-11-07 10:15:30 -05:00
int bw = rsize < 0x7000 ? rsize : 0x7000;
const char *tbuff;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if ((tbuff = fmap_need_off_once(map, peinfo->sections[peinfo->nsections - 1].raw + rsize - bw, 4096))) {
if (cli_memstr(tbuff, 4091, "\xe8\x2c\x61\x00\x00", 5)) {
ret = cli_append_potentially_unwanted(ctx, dam ? "Heuristics.W32.Magistr.A.dam" : "Heuristics.W32.Magistr.A");
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
}
2014-11-07 10:15:30 -05:00
}
}
}
} else if (rsize >= 0x7000 && vsize >= 0x7000 && ((vsize & 0xff) == 0xed)) {
2014-11-07 10:15:30 -05:00
int bw = rsize < 0x8000 ? rsize : 0x8000;
const char *tbuff;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if ((tbuff = fmap_need_off_once(map, peinfo->sections[peinfo->nsections - 1].raw + rsize - bw, 4096))) {
if (cli_memstr(tbuff, 4091, "\xe8\x04\x72\x00\x00", 5)) {
ret = cli_append_potentially_unwanted(ctx, dam ? "Heuristics.W32.Magistr.B.dam" : "Heuristics.W32.Magistr.B");
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
}
2014-11-07 10:15:30 -05:00
}
}
2014-11-07 10:15:30 -05:00
}
}
}
/* W32.Polipos.A */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Add endianness correction to SizeOfStackReserve access
while (polipos && !peinfo->is_dll && peinfo->nsections > 2 && peinfo->nsections < 13 && peinfo->e_lfanew <= 0x800 && (EC16(peinfo->pe_opt.opt32.Subsystem) == 2 || EC16(peinfo->pe_opt.opt32.Subsystem) == 3) && EC16(peinfo->file_hdr.Machine) == 0x14c && peinfo->pe_opt.opt32.SizeOfStackReserve >= 0x80000) {
2014-11-07 10:15:30 -05:00
uint32_t jump, jold, *jumps = NULL;
const uint8_t *code;
unsigned int xsjs = 0;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->sections[0].rsz > CLI_MAX_ALLOCATION)
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->sections[0].rsz < 5)
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!(code = fmap_need_off_once(map, peinfo->sections[0].raw, peinfo->sections[0].rsz)))
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (i = 0; i < peinfo->sections[0].rsz - 5; i++) {
if ((uint8_t)(code[i] - 0xe8) > 1)
2014-11-07 10:15:30 -05:00
continue;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
jump = cli_rawaddr(peinfo->sections[0].rva + i + 5 + cli_readint32(&code[i + 1]), peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err || !CLI_ISCONTAINED(peinfo->sections[polipos].raw, peinfo->sections[polipos].rsz, jump, 9))
2014-11-07 10:15:30 -05:00
continue;
if (xsjs % 128 == 0) {
if (xsjs == 1280)
2014-11-07 10:15:30 -05:00
break;
if (!(jumps = (uint32_t *)cli_realloc2(jumps, (xsjs + 128) * sizeof(uint32_t)))) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
}
j = 0;
for (; j < xsjs; j++) {
if (jumps[j] < jump)
2014-11-07 10:15:30 -05:00
continue;
if (jumps[j] == jump) {
2014-11-07 10:15:30 -05:00
xsjs--;
break;
}
jold = jumps[j];
jumps[j] = jump;
jump = jold;
2014-11-07 10:15:30 -05:00
}
jumps[j] = jump;
2014-11-07 10:15:30 -05:00
xsjs++;
}
if (!xsjs)
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Polipos: Checking %d xsect jump(s)\n", xsjs);
for (i = 0; i < xsjs; i++) {
if (!(code = fmap_need_off_once(map, jumps[i], 9)))
2014-11-07 10:15:30 -05:00
continue;
if ((jump = cli_readint32(code)) == 0x60ec8b55 || (code[4] == 0x0ec && ((jump == 0x83ec8b55 && code[6] == 0x60) || (jump == 0x81ec8b55 && !code[7] && !code[8])))) {
ret = cli_append_potentially_unwanted(ctx, "Heuristics.W32.Polipos.A");
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
free(jumps);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
free(jumps);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
}
2014-11-07 10:15:30 -05:00
}
}
}
free(jumps);
break;
}
/* Trojan.Swizzor.Gen */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (SCAN_HEURISTICS && (DCONF & PE_CONF_SWIZZOR) && peinfo->nsections > 1 && fsize > 64 * 1024 && fsize < 4 * 1024 * 1024) {
if (peinfo->dirs[2].Size) {
2014-11-07 10:15:30 -05:00
struct swizz_stats *stats = cli_calloc(1, sizeof(*stats));
unsigned int m = 1000;
ret = CL_CLEAN;
2014-11-07 10:15:30 -05:00
if (!stats) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return CL_EMEM;
2014-11-07 10:15:30 -05:00
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_parseres_special(EC32(peinfo->dirs[2].VirtualAddress), EC32(peinfo->dirs[2].VirtualAddress), map, peinfo, fsize, 0, 0, &m, stats);
if ((ret = cli_detect_swizz(stats)) == CL_VIRUS) {
ret = cli_append_potentially_unwanted(ctx, "Heuristics.Trojan.Swizzor.Gen");
if (ret != CL_CLEAN) {
if (ret == CL_VIRUS) {
if (!SCAN_ALLMATCHES) {
free(stats);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
} else
viruses_found++;
} else {
free(stats);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return ret;
}
}
2014-11-07 10:15:30 -05:00
}
}
}
}
2010-11-10 16:33:12 +01:00
/* !!!!!!!!!!!!!! PACKERS START HERE !!!!!!!!!!!!!! */
corrupted_cur = ctx->corrupted_input;
2010-11-10 16:33:12 +01:00
ctx->corrupted_input = 2; /* caller will reset on return */
/* UPX, FSG, MEW support */
/* try to find the first section with physical size == 0 */
found = 0;
if (DCONF & (PE_CONF_UPX | PE_CONF_FSG | PE_CONF_MEW)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (i = 0; i < (unsigned int)peinfo->nsections - 1; i++) {
if (!peinfo->sections[i].rsz && peinfo->sections[i].vsz && peinfo->sections[i + 1].rsz && peinfo->sections[i + 1].vsz) {
2014-11-07 10:15:30 -05:00
found = 1;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX/FSG/MEW: empty section found - assuming compression\n");
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonbool(pe_json, "HasEmptySection", 1);
2014-05-07 10:56:35 -04:00
#endif
2014-11-07 10:15:30 -05:00
break;
}
}
}
/* MEW support */
if (found && (DCONF & PE_CONF_MEW) && epsize >= 16 && epbuff[0] == '\xe9') {
2014-11-07 10:15:30 -05:00
uint32_t fileoffset;
const char *tbuff;
// TODO shouldn't peinfo->ep be used here instead? ep is the file
// offset, vep is the entry point RVA
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
fileoffset = (peinfo->vep + cli_readint32(epbuff + 1) + 5);
2014-11-07 10:15:30 -05:00
while (fileoffset == 0x154 || fileoffset == 0x158) {
char *src;
uint32_t offdiff, uselzma;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: found MEW characteristics %08X + %08X + 5 = %08X\n",
cli_readint32(epbuff + 1), peinfo->vep, cli_readint32(epbuff + 1) + peinfo->vep + 5);
if (!(tbuff = fmap_need_off_once(map, fileoffset, 0xb0)))
2014-11-07 10:15:30 -05:00
break;
2014-11-07 10:15:30 -05:00
if (fileoffset == 0x154)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: Win9x compatibility was set!\n");
2014-11-07 10:15:30 -05:00
else
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: Win9x compatibility was NOT set!\n");
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
offdiff = cli_readint32(tbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase);
if ((offdiff <= peinfo->sections[i + 1].rva) ||
(offdiff >= peinfo->sections[i + 1].rva + peinfo->sections[i + 1].raw - 4)) {
cli_dbgmsg("cli_scanpe: MEW: ESI is not in proper section\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
offdiff -= peinfo->sections[i + 1].rva;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i + 1].rsz) {
cli_dbgmsg("cli_scanpe: MEW: mew section is empty\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ssize = peinfo->sections[i + 1].vsz;
dsize = peinfo->sections[i].vsz;
2009-08-31 17:26:22 +02:00
/* Guard against integer overflow */
if ((ssize + dsize < ssize) || (ssize + dsize < dsize)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: section size (%08x) + diff size (%08x) exceeds max size of unsigned int (%08x)\n", ssize, dsize, UINT32_MAX);
break;
}
/* Verify that offdiff does not exceed the ssize + sdiff */
if (offdiff >= ssize + dsize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: offdiff (%08x) exceeds section size + diff size (%08x)\n", offdiff, ssize + dsize);
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: MEW: ssize %08x dsize %08x offdiff: %08x\n", ssize, dsize, offdiff);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: MEW", MAX(ssize, dsize));
CLI_UNPSIZELIMITS("cli_scanpe: MEW", MAX(ssize + dsize, peinfo->sections[i + 1].rsz));
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->sections[i + 1].rsz < offdiff + 12 || peinfo->sections[i + 1].rsz > ssize) {
cli_dbgmsg("cli_scanpe: MEW: Size mismatch: %08x\n", peinfo->sections[i + 1].rsz);
2014-11-07 10:15:30 -05:00
break;
}
/* allocate needed buffer */
if (!(src = cli_calloc(ssize + dsize, sizeof(char)))) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
bytes = fmap_readn(map, src + dsize, peinfo->sections[i + 1].raw, peinfo->sections[i + 1].rsz);
if (bytes != peinfo->sections[i + 1].rsz) {
cli_dbgmsg("cli_scanpe: MEW: Can't read %u bytes [read: %zu]\n", peinfo->sections[i + 1].rsz, bytes);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(src);
return CL_EREAD;
}
cli_dbgmsg("cli_scanpe: MEW: %zu (%08zx) bytes read\n", bytes, bytes);
2014-11-07 10:15:30 -05:00
/* count offset to lzma proc, if lzma used, 0xe8 -> call */
if (tbuff[0x7b] == '\xe8') {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!CLI_ISCONTAINED(peinfo->sections[1].rva, peinfo->sections[1].vsz, cli_readint32(tbuff + 0x7c) + fileoffset + 0x80, 4)) {
cli_dbgmsg("cli_scanpe: MEW: lzma proc out of bounds!\n");
2014-11-07 10:15:30 -05:00
free(src);
break; /* to next unpacker in chain */
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
uselzma = cli_readint32(tbuff + 0x7c) - (peinfo->sections[0].rva - fileoffset - 0x80);
2014-11-07 10:15:30 -05:00
} else {
uselzma = 0;
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "MEW");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: MEW", (src, 0));
CLI_UNPRESULTS("cli_scanpe: MEW", (unmew11(src, offdiff, ssize, dsize, EC32(peinfo->pe_opt.opt32.ImageBase), peinfo->sections[0].rva, uselzma, ndesc)), 1, (src, 0));
2014-11-07 10:15:30 -05:00
break;
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Why do we bail here
if (epsize < 168) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
if (found || upack) {
2014-11-07 10:15:30 -05:00
/* Check EP for UPX vs. FSG vs. Upack */
/* Upack 0.39 produces 2 types of executables
Spelling Adjustments (#30) * spelling: accessed * spelling: alignment * spelling: amalgamated * spelling: answers * spelling: another * spelling: acquisition * spelling: apitid * spelling: ascii * spelling: appending * spelling: appropriate * spelling: arbitrary * spelling: architecture * spelling: asynchronous * spelling: attachments * spelling: argument * spelling: authenticode * spelling: because * spelling: boundary * spelling: brackets * spelling: bytecode * spelling: calculation * spelling: cannot * spelling: changes * spelling: check * spelling: children * spelling: codegen * spelling: commands * spelling: container * spelling: concatenated * spelling: conditions * spelling: continuous * spelling: conversions * spelling: corresponding * spelling: corrupted * spelling: coverity * spelling: crafting * spelling: daemon * spelling: definition * spelling: delivered * spelling: delivery * spelling: delimit * spelling: dependencies * spelling: dependency * spelling: detection * spelling: determine * spelling: disconnects * spelling: distributed * spelling: documentation * spelling: downgraded * spelling: downloading * spelling: endianness * spelling: entities * spelling: especially * spelling: empty * spelling: expected * spelling: explicitly * spelling: existent * spelling: finished * spelling: flexibility * spelling: flexible * spelling: freshclam * spelling: functions * spelling: guarantee * spelling: hardened * spelling: headaches * spelling: heighten * spelling: improper * spelling: increment * spelling: indefinitely * spelling: independent * spelling: inaccessible * spelling: infrastructure Conflicts: docs/html/node68.html * spelling: initializing * spelling: inited * spelling: instream * spelling: installed * spelling: initialization * spelling: initialize * spelling: interface * spelling: intrinsics * spelling: interpreter * spelling: introduced * spelling: invalid * spelling: latency * spelling: lawyers * spelling: libclamav * spelling: likelihood * spelling: loop * spelling: maximum * spelling: million * spelling: milliseconds * spelling: minimum * spelling: minzhuan * spelling: multipart * spelling: misled * spelling: modifiers * spelling: notifying * spelling: objects * spelling: occurred * spelling: occurs * spelling: occurrences * spelling: optimization * spelling: original * spelling: originated * spelling: output * spelling: overridden * spelling: parenthesis * spelling: partition * spelling: performance * spelling: permission * spelling: phishing * spelling: portions * spelling: positives * spelling: preceded * spelling: properties * spelling: protocol * spelling: protos * spelling: quarantine * spelling: recursive * spelling: referring * spelling: reorder * spelling: reset * spelling: resources * spelling: resume * spelling: retrieval * spelling: rewrite * spelling: sanity * spelling: scheduled * spelling: search * spelling: section * spelling: separator * spelling: separated * spelling: specify * spelling: special * spelling: statement * spelling: streams * spelling: succession * spelling: suggests * spelling: superfluous * spelling: suspicious * spelling: synonym * spelling: temporarily * spelling: testfiles * spelling: transverse * spelling: turkish * spelling: typos * spelling: unable * spelling: unexpected * spelling: unexpectedly * spelling: unfinished * spelling: unfortunately * spelling: uninitialized * spelling: unlocking * spelling: unnecessary * spelling: unpack * spelling: unrecognized * spelling: unsupported * spelling: usable * spelling: wherever * spelling: wishlist * spelling: white * spelling: infrastructure * spelling: directories * spelling: overridden * spelling: permission * spelling: yesterday * spelling: initialization * spelling: intrinsics * space adjustment for spelling changes * minor modifications by klin
2018-02-21 15:00:59 -05:00
* 3 sections: | 2 sections (one empty, I don't check found if !upack, since it's in OR above):
2014-11-07 10:15:30 -05:00
* mov esi, value | pusha
* lodsd | call $+0x9
* push eax |
*
* Upack 1.1/1.2 Beta produces [based on 2 samples (sUx) provided by aCaB]:
* 2 sections
* mov esi, value
* loads
* mov edi, eax
*
* Upack unknown [sample 0297729]
* 3 sections
* mov esi, value
* push [esi]
* jmp
*
2014-11-07 10:15:30 -05:00
*/
/* upack 0.39-3s + sample 0151477*/
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
while (((upack && peinfo->nsections == 3) && /* 3 sections */
((
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
epbuff[0] == '\xbe' && cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase) > peinfo->min && /* mov esi */
epbuff[5] == '\xad' && epbuff[6] == '\x50' /* lodsd; push eax */
) ||
/* based on 0297729 sample from aCaB */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
(epbuff[0] == '\xbe' && cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase) > peinfo->min && /* mov esi */
epbuff[5] == '\xff' && epbuff[6] == '\x36' /* push [esi] */
))) ||
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
((!upack && peinfo->nsections == 2) && /* 2 sections */
(( /* upack 0.39-2s */
epbuff[0] == '\x60' && epbuff[1] == '\xe8' && cli_readint32(epbuff + 2) == 0x9 /* pusha; call+9 */
) ||
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
( /* upack 1.1/1.2, based on 2 samples */
epbuff[0] == '\xbe' && cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase) < peinfo->min && /* mov esi */
cli_readint32(epbuff + 1) > EC32(peinfo->pe_opt.opt32.ImageBase) &&
epbuff[5] == '\xad' && epbuff[6] == '\x8b' && epbuff[7] == '\xf8' /* loads; mov edi, eax */
)))) {
2014-11-07 10:15:30 -05:00
uint32_t vma, off;
int a, b, c;
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack characteristics found.\n");
a = peinfo->sections[0].vsz;
b = peinfo->sections[1].vsz;
2014-11-07 10:15:30 -05:00
if (upack) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack: var set\n");
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
c = peinfo->sections[2].vsz;
ssize = peinfo->sections[0].ursz + peinfo->sections[0].uraw;
off = peinfo->sections[0].rva;
vma = EC32(peinfo->pe_opt.opt32.ImageBase) + peinfo->sections[0].rva;
2014-11-07 10:15:30 -05:00
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack: var NOT set\n");
c = peinfo->sections[1].rva;
ssize = peinfo->sections[1].uraw;
off = 0;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
vma = peinfo->sections[1].rva - peinfo->sections[1].uraw;
2014-11-07 10:15:30 -05:00
}
dsize = a + b + c;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: Upack", MAX(MAX(dsize, ssize), peinfo->sections[1].ursz));
if (!CLI_ISCONTAINED_0_TO(dsize, peinfo->sections[1].rva - off, peinfo->sections[1].ursz) || (upack && !CLI_ISCONTAINED_0_TO(dsize, peinfo->sections[2].rva - peinfo->sections[0].rva, ssize)) || ssize > dsize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack: probably malformed pe-header, skipping to next unpacker\n");
2014-11-07 10:15:30 -05:00
break;
}
if ((dest = (char *)cli_calloc(dsize, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
if (fmap_readn(map, dest, 0, ssize) != ssize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack: Can't read raw data of section 0\n");
2014-11-07 10:15:30 -05:00
free(dest);
break;
}
if (upack)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
memmove(dest + peinfo->sections[2].rva - peinfo->sections[0].rva, dest, ssize);
if (fmap_readn(map, dest + peinfo->sections[1].rva - off, peinfo->sections[1].uraw, peinfo->sections[1].ursz) != peinfo->sections[1].ursz) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Upack: Can't read raw data of section 1\n");
2014-11-07 10:15:30 -05:00
free(dest);
break;
}
2004-06-12 00:16:01 +00:00
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "Upack");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: Upack", (dest, 0));
CLI_UNPRESULTS("cli_scanpe: Upack", (unupack(upack, dest, dsize, epbuff, vma, peinfo->ep, EC32(peinfo->pe_opt.opt32.ImageBase), peinfo->sections[0].rva, ndesc)), 1, (dest, 0));
2004-06-12 00:16:01 +00:00
2014-11-07 10:15:30 -05:00
break;
}
}
while (found && (DCONF & PE_CONF_FSG) && epbuff[0] == '\x87' && epbuff[1] == '\x25') {
2014-11-07 10:15:30 -05:00
const char *dst;
uint32_t newesi, newedi, newebx, newedx;
2014-11-07 10:15:30 -05:00
/* FSG v2.0 support - thanks to aCaB ! */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ssize = peinfo->sections[i + 1].rsz;
dsize = peinfo->sections[i].vsz;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: FSG", MAX(dsize, ssize));
if (ssize <= 0x19 || dsize <= ssize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: Size mismatch (ssize: %d, dsize: %d)\n", ssize, dsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
newedx = cli_readint32(epbuff + 2) - EC32(peinfo->pe_opt.opt32.ImageBase);
if (!CLI_ISCONTAINED(peinfo->sections[i + 1].rva, peinfo->sections[i + 1].rsz, newedx, 4)) {
cli_dbgmsg("cli_scanpe: FSG: xchg out of bounds (%x), giving up\n", newedx);
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i + 1].rsz || !(src = fmap_need_off_once(map, peinfo->sections[i + 1].raw, ssize))) {
cli_dbgmsg("cli_scanpe: Can't read raw data of section %d\n", i + 1);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_ESEEK;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
dst = src + newedx - peinfo->sections[i + 1].rva;
if (newedx < peinfo->sections[i + 1].rva || !CLI_ISCONTAINED(src, ssize, dst, 4)) {
cli_dbgmsg("cli_scanpe: FSG: New ESP out of bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
newedx = cli_readint32(dst) - EC32(peinfo->pe_opt.opt32.ImageBase);
if (!CLI_ISCONTAINED(peinfo->sections[i + 1].rva, peinfo->sections[i + 1].rsz, newedx, 4)) {
cli_dbgmsg("cli_scanpe: FSG: New ESP (%x) is wrong\n", newedx);
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
dst = src + newedx - peinfo->sections[i + 1].rva;
if (!CLI_ISCONTAINED(src, ssize, dst, 32)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: New stack out of bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
newedi = cli_readint32(dst) - EC32(peinfo->pe_opt.opt32.ImageBase);
newesi = cli_readint32(dst + 4) - EC32(peinfo->pe_opt.opt32.ImageBase);
newebx = cli_readint32(dst + 16) - EC32(peinfo->pe_opt.opt32.ImageBase);
2014-11-07 10:15:30 -05:00
newedx = cli_readint32(dst + 20);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newedi != peinfo->sections[i].rva) {
cli_dbgmsg("cli_scanpe: FSG: Bad destination buffer (edi is %x should be %x)\n", newedi, peinfo->sections[i].rva);
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newesi < peinfo->sections[i + 1].rva || newesi - peinfo->sections[i + 1].rva >= peinfo->sections[i + 1].rsz) {
cli_dbgmsg("cli_scanpe: FSG: Source buffer out of section bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!CLI_ISCONTAINED(peinfo->sections[i + 1].rva, peinfo->sections[i + 1].rsz, newebx, 16)) {
cli_dbgmsg("cli_scanpe: FSG: Array of functions out of bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
newedx = cli_readint32(newebx + 12 - peinfo->sections[i + 1].rva + src) - EC32(peinfo->pe_opt.opt32.ImageBase);
cli_dbgmsg("cli_scanpe: FSG: found old EP @%x\n", newedx);
if ((dest = (char *)cli_calloc(dsize, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "FSG");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: FSG", (dest, 0));
CLI_UNPRESULTSFSG2("cli_scanpe: FSG", (unfsg_200(newesi - peinfo->sections[i + 1].rva + src, dest, ssize + peinfo->sections[i + 1].rva - newesi, dsize, newedi, EC32(peinfo->pe_opt.opt32.ImageBase), newedx, ndesc)), 1, (dest, 0));
2014-11-07 10:15:30 -05:00
break;
}
2006-01-14 18:21:36 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
while (found && (DCONF & PE_CONF_FSG) && epbuff[0] == '\xbe' && cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase) < peinfo->min) {
2014-11-07 10:15:30 -05:00
int sectcnt = 0;
const char *support;
uint32_t newesi, newedi, oldep, gp, t;
struct cli_exe_section *sections;
2014-11-07 10:15:30 -05:00
/* FSG support - v. 1.33 (thx trog for the many samples) */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ssize = peinfo->sections[i + 1].rsz;
dsize = peinfo->sections[i].vsz;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: FSG", MAX(dsize, ssize));
if (ssize <= 0x19 || dsize <= ssize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: Size mismatch (ssize: %d, dsize: %d)\n", ssize, dsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!(t = cli_rawaddr(cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase), NULL, 0, &err, fsize, peinfo->hdr_size)) && err) {
cli_dbgmsg("cli_scanpe: FSG: Support data out of padding area\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
gp = peinfo->sections[i + 1].raw - t;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: FSG", gp);
if (!(support = fmap_need_off_once(map, t, gp))) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Can't read %d bytes from padding area\n", gp);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EREAD;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* newebx = cli_readint32(support) - EC32(peinfo->pe_opt.opt32.ImageBase); Unused */
newedi = cli_readint32(support + 4) - EC32(peinfo->pe_opt.opt32.ImageBase); /* 1st dest */
newesi = cli_readint32(support + 8) - EC32(peinfo->pe_opt.opt32.ImageBase); /* Source */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newesi < peinfo->sections[i + 1].rva || newesi - peinfo->sections[i + 1].rva >= peinfo->sections[i + 1].rsz) {
cli_dbgmsg("cli_scanpe: FSG: Source buffer out of section bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newedi != peinfo->sections[i].rva) {
cli_dbgmsg("cli_scanpe: FSG: Bad destination (is %x should be %x)\n", newedi, peinfo->sections[i].rva);
2014-11-07 10:15:30 -05:00
break;
}
2014-11-07 10:15:30 -05:00
/* Counting original sections */
for (t = 12; t < gp - 4; t += 4) {
uint32_t rva = cli_readint32(support + t);
if (!rva)
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rva -= EC32(peinfo->pe_opt.opt32.ImageBase) + 1;
2014-11-07 10:15:30 -05:00
sectcnt++;
if (rva % 0x1000)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: Original section %d is misaligned\n", sectcnt);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (rva < peinfo->sections[i].rva || rva - peinfo->sections[i].rva >= peinfo->sections[i].vsz) {
cli_dbgmsg("cli_scanpe: FSG: Original section %d is out of bounds\n", sectcnt);
2014-11-07 10:15:30 -05:00
break;
}
}
if (t >= gp - 4 || cli_readint32(support + t)) {
2014-11-07 10:15:30 -05:00
break;
}
if ((sections = (struct cli_exe_section *)cli_malloc((sectcnt + 1) * sizeof(struct cli_exe_section))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("cli_scanpe: FSG: Unable to allocate memory for sections %llu\n", (long long unsigned)((sectcnt + 1) * sizeof(struct cli_exe_section)));
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
2014-11-07 10:15:30 -05:00
sections[0].rva = newedi;
for (t = 1; t <= (uint32_t)sectcnt; t++)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
sections[t].rva = cli_readint32(support + 8 + t * 4) - 1 - EC32(peinfo->pe_opt.opt32.ImageBase);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i + 1].rsz || !(src = fmap_need_off_once(map, peinfo->sections[i + 1].raw, ssize))) {
cli_dbgmsg("cli_scanpe: Can't read raw data of section %d\n", i);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(sections);
return CL_EREAD;
}
if ((dest = (char *)cli_calloc(dsize, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(sections);
return CL_EMEM;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
oldep = peinfo->vep + 161 + 6 + cli_readint32(epbuff + 163);
cli_dbgmsg("cli_scanpe: FSG: found old EP @%x\n", oldep);
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "FSG");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: FSG", (dest, sections, 0));
CLI_UNPRESULTSFSG1("cli_scanpe: FSG", (unfsg_133(src + newesi - peinfo->sections[i + 1].rva, dest, ssize + peinfo->sections[i + 1].rva - newesi, dsize, sections, sectcnt, EC32(peinfo->pe_opt.opt32.ImageBase), oldep, ndesc)), 1, (dest, sections, 0));
2014-11-07 10:15:30 -05:00
break; /* were done with 1.33 */
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
while (found && (DCONF & PE_CONF_FSG) && epbuff[0] == '\xbb' && cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase) < peinfo->min && epbuff[5] == '\xbf' && epbuff[10] == '\xbe' && peinfo->vep >= peinfo->sections[i + 1].rva && peinfo->vep - peinfo->sections[i + 1].rva > peinfo->sections[i + 1].rva - 0xe0) {
2014-11-07 10:15:30 -05:00
int sectcnt = 0;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
uint32_t gp, t = cli_rawaddr(cli_readint32(epbuff + 1) - EC32(peinfo->pe_opt.opt32.ImageBase), NULL, 0, &err, fsize, peinfo->hdr_size);
2014-11-07 10:15:30 -05:00
const char *support;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
uint32_t newesi = cli_readint32(epbuff + 11) - EC32(peinfo->pe_opt.opt32.ImageBase);
uint32_t newedi = cli_readint32(epbuff + 6) - EC32(peinfo->pe_opt.opt32.ImageBase);
uint32_t oldep = peinfo->vep - peinfo->sections[i + 1].rva;
2014-11-07 10:15:30 -05:00
struct cli_exe_section *sections;
2014-11-07 10:15:30 -05:00
/* FSG support - v. 1.31 */
2006-01-14 18:21:36 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ssize = peinfo->sections[i + 1].rsz;
dsize = peinfo->sections[i].vsz;
if (err) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: Support data out of padding area\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newesi < peinfo->sections[i + 1].rva || newesi - peinfo->sections[i + 1].rva >= peinfo->sections[i + 1].raw) {
cli_dbgmsg("cli_scanpe: FSG: Source buffer out of section bounds\n");
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (newedi != peinfo->sections[i].rva) {
cli_dbgmsg("cli_scanpe: FSG: Bad destination (is %x should be %x)\n", newedi, peinfo->sections[i].rva);
2014-11-07 10:15:30 -05:00
break;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: FSG", MAX(dsize, ssize));
if (ssize <= 0x19 || dsize <= ssize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: FSG: Size mismatch (ssize: %d, dsize: %d)\n", ssize, dsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
gp = peinfo->sections[i + 1].raw - t;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: FSG", gp)
if (!(support = fmap_need_off_once(map, t, gp))) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Can't read %d bytes from padding area\n", gp);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EREAD;
}
2014-11-07 10:15:30 -05:00
/* Counting original sections */
for (t = 0; t < gp - 2; t += 2) {
uint32_t rva = support[t] | (support[t + 1] << 8);
2014-11-07 10:15:30 -05:00
if (rva == 2 || rva == 1)
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rva = ((rva - 2) << 12) - EC32(peinfo->pe_opt.opt32.ImageBase);
2014-11-07 10:15:30 -05:00
sectcnt++;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (rva < peinfo->sections[i].rva || rva - peinfo->sections[i].rva >= peinfo->sections[i].vsz) {
cli_dbgmsg("cli_scanpe: FSG: Original section %d is out of bounds\n", sectcnt);
2014-11-07 10:15:30 -05:00
break;
}
}
if (t >= gp - 10 || cli_readint32(support + t + 6) != 2)
2014-11-07 10:15:30 -05:00
break;
if ((sections = (struct cli_exe_section *)cli_malloc((sectcnt + 1) * sizeof(struct cli_exe_section))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("cli_scanpe: FSG: Unable to allocate memory for sections %llu\n", (long long unsigned)((sectcnt + 1) * sizeof(struct cli_exe_section)));
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
2014-11-07 10:15:30 -05:00
sections[0].rva = newedi;
for (t = 0; t <= (uint32_t)sectcnt - 1; t++)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
sections[t + 1].rva = (((support[t * 2] | (support[t * 2 + 1] << 8)) - 2) << 12) - EC32(peinfo->pe_opt.opt32.ImageBase);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i + 1].rsz || !(src = fmap_need_off_once(map, peinfo->sections[i + 1].raw, ssize))) {
cli_dbgmsg("cli_scanpe: FSG: Can't read raw data of section %d\n", i);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(sections);
return CL_EREAD;
}
if ((dest = (char *)cli_calloc(dsize, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(sections);
return CL_EMEM;
}
gp = 0xda + 6 * (epbuff[16] == '\xe8');
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
oldep = peinfo->vep + gp + 6 + cli_readint32(src + gp + 2 + oldep);
cli_dbgmsg("cli_scanpe: FSG: found old EP @%x\n", oldep);
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "FSG");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: FSG", (dest, sections, 0));
CLI_UNPRESULTSFSG1("cli_scanpe: FSG", (unfsg_133(src + newesi - peinfo->sections[i + 1].rva, dest, ssize + peinfo->sections[i + 1].rva - newesi, dsize, sections, sectcnt, EC32(peinfo->pe_opt.opt32.ImageBase), oldep, ndesc)), 1, (dest, sections, 0));
2014-11-07 10:15:30 -05:00
break; /* were done with 1.31 */
}
if (found && (DCONF & PE_CONF_UPX)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ssize = peinfo->sections[i + 1].rsz;
dsize = peinfo->sections[i].vsz + peinfo->sections[i + 1].vsz;
2006-01-14 18:21:36 +00:00
/*
2014-11-07 10:15:30 -05:00
* UPX support
* we assume (i + 1) is UPX1
*/
/* cli_dbgmsg("UPX: ssize %u dsize %u\n", ssize, dsize); */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: UPX", MAX(dsize, ssize));
2013-03-12 11:11:43 -04:00
if (ssize <= 0x19 || dsize <= ssize || dsize > CLI_MAX_ALLOCATION) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Size mismatch or dsize too big (ssize: %d, dsize: %d)\n", ssize, dsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i + 1].rsz || !(src = fmap_need_off_once(map, peinfo->sections[i + 1].raw, ssize))) {
cli_dbgmsg("cli_scanpe: UPX: Can't read raw data of section %d\n", i + 1);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EREAD;
}
if ((dest = (char *)cli_calloc(dsize + 8192, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
2014-11-07 10:15:30 -05:00
/* try to detect UPX code */
if (cli_memstr(UPX_NRV2B, 24, epbuff + 0x69, 13) || cli_memstr(UPX_NRV2B, 24, epbuff + 0x69 + 8, 13)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Looks like a NRV2B decompression routine\n");
2014-11-07 10:15:30 -05:00
upxfn = upx_inflate2b;
} else if (cli_memstr(UPX_NRV2D, 24, epbuff + 0x69, 13) || cli_memstr(UPX_NRV2D, 24, epbuff + 0x69 + 8, 13)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Looks like a NRV2D decompression routine\n");
2014-11-07 10:15:30 -05:00
upxfn = upx_inflate2d;
} else if (cli_memstr(UPX_NRV2E, 24, epbuff + 0x69, 13) || cli_memstr(UPX_NRV2E, 24, epbuff + 0x69 + 8, 13)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Looks like a NRV2E decompression routine\n");
2014-11-07 10:15:30 -05:00
upxfn = upx_inflate2e;
}
if (upxfn) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
int skew = cli_readint32(epbuff + 2) - EC32(peinfo->pe_opt.opt32.ImageBase) - peinfo->sections[i + 1].rva;
2014-11-07 10:15:30 -05:00
if (epbuff[1] != '\xbe' || skew <= 0 || skew > 0xfff) {
2014-11-07 10:15:30 -05:00
/* FIXME: legit skews?? */
skew = 0;
2014-11-07 10:15:30 -05:00
} else if ((unsigned int)skew > ssize) {
/* Ignore suggested skew larger than section size */
skew = 0;
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: UPX1 seems skewed by %d bytes\n", skew);
2014-11-07 10:15:30 -05:00
}
2014-11-07 10:15:30 -05:00
/* Try skewed first (skew may be zero) */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (upxfn(src + skew, ssize - skew, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep - skew) >= 0) {
2014-11-07 10:15:30 -05:00
upx_success = 1;
}
/* If skew not successful and non-zero, try no skew */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
else if (skew && (upxfn(src, ssize, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep) >= 0)) {
2014-11-07 10:15:30 -05:00
upx_success = 1;
}
if (upx_success)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Successfully decompressed\n");
2014-11-07 10:15:30 -05:00
else
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Preferred decompressor failed\n");
2014-11-07 10:15:30 -05:00
}
if (!upx_success && upxfn != upx_inflate2b) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (upx_inflate2b(src, ssize, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep) == -1 && upx_inflate2b(src + 0x15, ssize - 0x15, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep - 0x15) == -1) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: NRV2B decompressor failed\n");
2014-11-07 10:15:30 -05:00
} else {
upx_success = 1;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Successfully decompressed with NRV2B\n");
2014-11-07 10:15:30 -05:00
}
}
if (!upx_success && upxfn != upx_inflate2d) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (upx_inflate2d(src, ssize, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep) == -1 && upx_inflate2d(src + 0x15, ssize - 0x15, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep - 0x15) == -1) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: NRV2D decompressor failed\n");
2014-11-07 10:15:30 -05:00
} else {
upx_success = 1;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Successfully decompressed with NRV2D\n");
2014-11-07 10:15:30 -05:00
}
}
if (!upx_success && upxfn != upx_inflate2e) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (upx_inflate2e(src, ssize, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep) == -1 && upx_inflate2e(src + 0x15, ssize - 0x15, dest, &dsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep - 0x15) == -1) {
cli_dbgmsg("cli_scanpe: UPX: NRV2E decompressor failed\n");
2014-11-07 10:15:30 -05:00
} else {
upx_success = 1;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: Successfully decompressed with NRV2E\n");
2014-11-07 10:15:30 -05:00
}
}
if (cli_memstr(UPX_LZMA2, 20, epbuff + 0x2f, 20)) {
uint32_t strictdsize = cli_readint32(epbuff + 0x21), skew = 0;
if (ssize > 0x15 && epbuff[0] == '\x60' && epbuff[1] == '\xbe') {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Add EC32
skew = cli_readint32(epbuff + 2) - peinfo->sections[i + 1].rva - peinfo->pe_opt.opt32.ImageBase;
if (skew != 0x15)
2014-11-07 10:15:30 -05:00
skew = 0;
}
if (strictdsize <= dsize)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
upx_success = upx_inflatelzma(src + skew, ssize - skew, dest, &strictdsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep, 0x20003) >= 0;
} else if (cli_memstr(UPX_LZMA1_FIRST, 8, epbuff + 0x39, 8) && cli_memstr(UPX_LZMA1_SECOND, 8, epbuff + 0x45, 8)) {
uint32_t strictdsize = cli_readint32(epbuff + 0x2b), skew = 0;
uint32_t properties = cli_readint32(epbuff + 0x41);
if (ssize > 0x15 && epbuff[0] == '\x60' && epbuff[1] == '\xbe') {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Add EC32
skew = cli_readint32(epbuff + 2) - peinfo->sections[i + 1].rva - peinfo->pe_opt.opt32.ImageBase;
if (skew != 0x15)
2014-11-07 10:15:30 -05:00
skew = 0;
}
2004-06-12 00:16:01 +00:00
if (strictdsize <= dsize)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
upx_success = upx_inflatelzma(src + skew, ssize - skew, dest, &strictdsize, peinfo->sections[i].rva, peinfo->sections[i + 1].rva, peinfo->vep, properties) >= 0;
2014-11-07 10:15:30 -05:00
}
2008-07-31 02:17:25 +00:00
if (!upx_success) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX: All decompressors failed\n");
2014-11-07 10:15:30 -05:00
free(dest);
}
}
2006-01-14 18:21:36 +00:00
if (upx_success) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2004-06-12 00:16:01 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: UPX/FSG", (dest, 0));
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "UPX");
2014-05-07 10:56:35 -04:00
#endif
if ((unsigned int)write(ndesc, dest, dsize) != dsize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX/FSG: Can't write %d bytes\n", dsize);
2014-11-07 10:15:30 -05:00
free(tempfile);
free(dest);
close(ndesc);
return CL_EWRITE;
}
free(dest);
if (lseek(ndesc, 0, SEEK_SET) == -1) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX/FSG: lseek() failed\n");
2014-11-07 10:15:30 -05:00
close(ndesc);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
SHA_RESET;
2014-11-07 10:15:30 -05:00
CLI_TMPUNLK();
free(tempfile);
return CL_ESEEK;
}
if (ctx->engine->keeptmp)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: UPX/FSG: Decompressed data saved in %s\n", tempfile);
2014-11-07 10:15:30 -05:00
cli_dbgmsg("***** Scanning decompressed file *****\n");
SHA_OFF;
if ((ret = cli_magic_scan_desc(ndesc, tempfile, ctx, NULL, LAYER_ATTRIBUTES_NONE)) == CL_VIRUS) {
2014-11-07 10:15:30 -05:00
close(ndesc);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
SHA_RESET;
2014-11-07 10:15:30 -05:00
CLI_TMPUNLK();
free(tempfile);
return CL_VIRUS;
}
SHA_RESET;
2013-02-28 19:32:29 -05:00
close(ndesc);
CLI_TMPUNLK();
free(tempfile);
2014-11-07 10:15:30 -05:00
return ret;
2004-06-12 00:16:01 +00:00
}
2004-06-11 20:17:18 +00:00
/* Petite */
if (epsize < 200) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_CLEAN;
}
found = 2;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (epbuff[0] != '\xb8' || (uint32_t)cli_readint32(epbuff + 1) != peinfo->sections[peinfo->nsections - 1].rva + EC32(peinfo->pe_opt.opt32.ImageBase)) {
if (peinfo->nsections < 2 || epbuff[0] != '\xb8' || (uint32_t)cli_readint32(epbuff + 1) != peinfo->sections[peinfo->nsections - 2].rva + EC32(peinfo->pe_opt.opt32.ImageBase))
2014-11-07 10:15:30 -05:00
found = 0;
else
found = 1;
}
if (found && (DCONF & PE_CONF_PETITE)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Petite: v2.%d compression detected\n", found);
if (cli_readint32(epbuff + 0x80) == 0x163c988d) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Petite: level zero compression is not supported yet\n");
2014-11-07 10:15:30 -05:00
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
dsize = peinfo->max - peinfo->min;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: Petite", dsize);
if ((dest = (char *)cli_calloc(dsize, sizeof(char))) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: Petite: Can't allocate %d bytes\n", dsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (i = 0; i < peinfo->nsections; i++) {
if (peinfo->sections[i].raw) {
unsigned int r_ret;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections[i].rsz)
goto out_no_petite;
if (!CLI_ISCONTAINED(dest, dsize,
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
dest + peinfo->sections[i].rva - peinfo->min,
peinfo->sections[i].ursz))
goto out_no_petite;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
r_ret = fmap_readn(map, dest + peinfo->sections[i].rva - peinfo->min,
peinfo->sections[i].raw,
peinfo->sections[i].ursz);
if (r_ret != peinfo->sections[i].ursz) {
out_no_petite:
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
free(dest);
return CL_CLEAN;
}
}
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "Petite");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: Petite", (dest, 0));
CLI_UNPRESULTS("Petite", (petite_inflate2x_1to9(dest, peinfo->min, peinfo->max - peinfo->min, peinfo->sections, peinfo->nsections - (found == 1 ? 1 : 0), EC32(peinfo->pe_opt.opt32.ImageBase), peinfo->vep, ndesc, found, EC32(peinfo->dirs[2].VirtualAddress), EC32(peinfo->dirs[2].Size))), 0, (dest, 0));
2014-11-07 10:15:30 -05:00
}
}
/* PESpin 1.1 */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if ((DCONF & PE_CONF_PESPIN) && peinfo->nsections > 1 &&
peinfo->vep >= peinfo->sections[peinfo->nsections - 1].rva &&
0x3217 - 4 <= peinfo->sections[peinfo->nsections - 1].rva + peinfo->sections[peinfo->nsections - 1].rsz &&
peinfo->vep < peinfo->sections[peinfo->nsections - 1].rva + peinfo->sections[peinfo->nsections - 1].rsz - 0x3217 - 4 &&
memcmp(epbuff + 4, "\xe8\x00\x00\x00\x00\x8b\x1c\x24\x83\xc3", 10) == 0) {
2014-11-07 10:15:30 -05:00
char *spinned;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPSIZELIMITS("cli_scanpe: PEspin", fsize);
if ((spinned = (char *)cli_malloc(fsize)) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("cli_scanpe: PESping: Unable to allocate memory for spinned %lu\n", (unsigned long)fsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
if (fmap_readn(map, spinned, 0, fsize) != fsize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: PESpin: Can't read %lu bytes\n", (unsigned long)fsize);
2014-11-07 10:15:30 -05:00
free(spinned);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EREAD;
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "PEspin");
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
CLI_UNPTEMP("cli_scanpe: PESpin", (spinned, 0));
CLI_UNPRESULTS_("cli_scanpe: PEspin", SPINCASE(), (unspin(spinned, fsize, peinfo->sections, peinfo->nsections - 1, peinfo->vep, ndesc, ctx)), 0, (spinned, 0));
}
/* yC 1.3 & variants */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if ((DCONF & PE_CONF_YC) && peinfo->nsections > 1 &&
(EC32(peinfo->pe_opt.opt32.AddressOfEntryPoint) == peinfo->sections[peinfo->nsections - 1].rva + 0x60)) {
2014-11-07 10:15:30 -05:00
uint32_t ecx = 0;
int16_t offset;
2014-11-07 10:15:30 -05:00
/* yC 1.3 */
if (!memcmp(epbuff, "\x55\x8B\xEC\x53\x56\x57\x60\xE8\x00\x00\x00\x00\x5D\x81\xED", 15) &&
!memcmp(epbuff + 0x26, "\x8D\x3A\x8B\xF7\x33\xC0\xEB\x04\x90\xEB\x01\xC2\xAC", 13) &&
2014-11-07 10:15:30 -05:00
((uint8_t)epbuff[0x13] == 0xB9) &&
((uint16_t)(cli_readint16(epbuff + 0x18)) == 0xE981) &&
!memcmp(epbuff + 0x1e, "\x8B\xD5\x81\xC2", 4)) {
2014-11-07 10:15:30 -05:00
offset = 0;
if (0x6c - cli_readint32(epbuff + 0xf) + cli_readint32(epbuff + 0x22) == 0xC6)
ecx = cli_readint32(epbuff + 0x14) - cli_readint32(epbuff + 0x1a);
2014-11-07 10:15:30 -05:00
}
2014-11-07 10:15:30 -05:00
/* yC 1.3 variant */
if (!ecx && !memcmp(epbuff, "\x55\x8B\xEC\x83\xEC\x40\x53\x56\x57", 9) &&
!memcmp(epbuff + 0x17, "\xe8\x00\x00\x00\x00\x5d\x81\xed", 8) &&
2014-11-07 10:15:30 -05:00
((uint8_t)epbuff[0x23] == 0xB9)) {
2014-11-07 10:15:30 -05:00
offset = 0x10;
if (0x6c - cli_readint32(epbuff + 0x1f) + cli_readint32(epbuff + 0x32) == 0xC6)
ecx = cli_readint32(epbuff + 0x24) - cli_readint32(epbuff + 0x2a);
2014-11-07 10:15:30 -05:00
}
2014-11-07 10:15:30 -05:00
/* yC 1.x/modified */
if (!ecx && !memcmp(epbuff, "\x60\xe8\x00\x00\x00\x00\x5d\x81\xed", 9) &&
2014-11-07 10:15:30 -05:00
((uint8_t)epbuff[0xd] == 0xb9) &&
((uint16_t)cli_readint16(epbuff + 0x12) == 0xbd8d) &&
!memcmp(epbuff + 0x18, "\x8b\xf7\xac", 3)) {
2014-11-07 10:15:30 -05:00
offset = -0x18;
if (0x66 - cli_readint32(epbuff + 0x9) + cli_readint32(epbuff + 0x14) == 0xae)
ecx = cli_readint32(epbuff + 0xe);
2014-11-07 10:15:30 -05:00
}
2014-11-07 10:15:30 -05:00
if (ecx > 0x800 && ecx < 0x2000 &&
!memcmp(epbuff + 0x63 + offset, "\xaa\xe2\xcc", 3) &&
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
(fsize >= peinfo->sections[peinfo->nsections - 1].raw + 0xC6 + ecx + offset)) {
2006-01-14 13:58:35 +00:00
2014-11-07 10:15:30 -05:00
char *spinned;
2006-01-14 13:58:35 +00:00
if ((spinned = (char *)cli_malloc(fsize)) == NULL) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_errmsg("cli_scanpe: yC: Unable to allocate memory for spinned %lu\n", (unsigned long)fsize);
cli_exe_info_destroy(peinfo);
2014-11-07 10:15:30 -05:00
return CL_EMEM;
}
2006-01-14 13:58:35 +00:00
if (fmap_readn(map, spinned, 0, fsize) != fsize) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: yC: Can't read %lu bytes\n", (unsigned long)fsize);
free(spinned);
cli_exe_info_destroy(peinfo);
return CL_EREAD;
}
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "yC");
#endif
do {
size_t num_alerts = evidence_num_alerts(ctx->evidence);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("%d,%d,%d,%d\n", peinfo->nsections - 1, peinfo->e_lfanew, ecx, offset);
CLI_UNPTEMP("cli_scanpe: yC", (spinned, 0));
CLI_UNPRESULTS("cli_scanpe: yC", (yc_decrypt(ctx, spinned, fsize, peinfo->sections, peinfo->nsections - 1, peinfo->e_lfanew, ndesc, ecx, offset)), 0, (spinned, 0));
if (SCAN_ALLMATCHES && num_alerts != evidence_num_alerts(ctx->evidence)) {
// In ALLMATCH-mode, CLI_UNPRESULTS() will not return CL_VIRUS when something is found.
// We apparently want to return CL_VIRUS here if CLI_UNPRESULTS() found something (preserving previous behavior).
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return CL_VIRUS;
}
} while (0);
}
}
/* WWPack */
while ((DCONF & PE_CONF_WWPACK) && peinfo->nsections > 1 &&
peinfo->vep == peinfo->sections[peinfo->nsections - 1].rva &&
memcmp(epbuff, "\x53\x55\x8b\xe8\x33\xdb\xeb", 7) == 0 &&
memcmp(epbuff + 0x68, "\xe8\x00\x00\x00\x00\x58\x2d\x6d\x00\x00\x00\x50\x60\x33\xc9\x50\x58\x50\x50", 19) == 0) {
uint32_t head = peinfo->sections[peinfo->nsections - 1].raw;
uint8_t *packer;
char *src;
ssize = 0;
for (i = 0;; i++) {
if (peinfo->sections[i].raw < head)
head = peinfo->sections[i].raw;
if (i + 1 == peinfo->nsections)
break;
if (ssize < peinfo->sections[i].rva + peinfo->sections[i].vsz)
ssize = peinfo->sections[i].rva + peinfo->sections[i].vsz;
}
if (!head || !ssize || head > ssize)
break;
CLI_UNPSIZELIMITS("cli_scanpe: WWPack", ssize);
if (!(src = (char *)cli_calloc(ssize, sizeof(char)))) {
cli_exe_info_destroy(peinfo);
return CL_EMEM;
}
if (fmap_readn(map, src, 0, head) != head) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: WWPack: Can't read %d bytes from headers\n", head);
free(src);
cli_exe_info_destroy(peinfo);
return CL_EREAD;
}
for (i = 0; i < (unsigned int)peinfo->nsections - 1; i++) {
if (!peinfo->sections[i].rsz)
continue;
if (!CLI_ISCONTAINED(src, ssize, src + peinfo->sections[i].rva, peinfo->sections[i].rsz))
break;
if (fmap_readn(map, src + peinfo->sections[i].rva, peinfo->sections[i].raw, peinfo->sections[i].rsz) != peinfo->sections[i].rsz)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
break;
}
if (i + 1 != peinfo->nsections) {
cli_dbgmsg("cli_scanpe: WWpack: Probably hacked/damaged file.\n");
free(src);
break;
}
if ((packer = (uint8_t *)cli_calloc(peinfo->sections[peinfo->nsections - 1].rsz, sizeof(char))) == NULL) {
free(src);
cli_exe_info_destroy(peinfo);
return CL_EMEM;
}
if (!peinfo->sections[peinfo->nsections - 1].rsz || fmap_readn(map, packer, peinfo->sections[peinfo->nsections - 1].raw, peinfo->sections[peinfo->nsections - 1].rsz) != peinfo->sections[peinfo->nsections - 1].rsz) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_scanpe: WWPack: Can't read %d bytes from wwpack sect\n", peinfo->sections[peinfo->nsections - 1].rsz);
free(src);
free(packer);
cli_exe_info_destroy(peinfo);
return CL_EREAD;
}
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "WWPack");
#endif
CLI_UNPTEMP("cli_scanpe: WWPack", (src, packer, 0));
CLI_UNPRESULTS("cli_scanpe: WWPack", (wwunpack((uint8_t *)src, ssize, packer, peinfo->sections, peinfo->nsections - 1, peinfo->e_lfanew, ndesc)), 0, (src, packer, 0));
break;
}
/* ASPACK support */
while ((DCONF & PE_CONF_ASPACK) &&
((peinfo->ep + ASPACK_EP_OFFSET_212 < fsize) ||
(peinfo->ep + ASPACK_EP_OFFSET_OTHER < fsize) ||
(peinfo->ep + ASPACK_EP_OFFSET_242 < fsize)) &&
(!memcmp(epbuff, "\x60\xe8\x03\x00\x00\x00\xe9\xeb", 8))) {
char *src;
aspack_version_t aspack_ver = ASPACK_VER_NONE;
if (epsize < 0x3bf)
break;
if (0 == memcmp(epbuff + ASPACK_EPBUFF_OFFSET_212, "\x68\x00\x00\x00\x00\xc3", 6)) {
aspack_ver = ASPACK_VER_212;
} else if (0 == memcmp(epbuff + ASPACK_EPBUFF_OFFSET_OTHER, "\x68\x00\x00\x00\x00\xc3", 6)) {
aspack_ver = ASPACK_VER_OTHER;
} else if (0 == memcmp(epbuff + ASPACK_EPBUFF_OFFSET_242, "\x68\x00\x00\x00\x00\xc3", 6)) {
aspack_ver = ASPACK_VER_242;
} else {
break;
}
ssize = 0;
for (i = 0; i < peinfo->nsections; i++)
if (ssize < peinfo->sections[i].rva + peinfo->sections[i].vsz)
ssize = peinfo->sections[i].rva + peinfo->sections[i].vsz;
if (!ssize)
break;
CLI_UNPSIZELIMITS("cli_scanpe: Aspack", ssize);
if (!(src = (char *)cli_calloc(ssize, sizeof(char)))) {
cli_exe_info_destroy(peinfo);
return CL_EMEM;
}
for (i = 0; i < (unsigned int)peinfo->nsections; i++) {
if (!peinfo->sections[i].rsz)
continue;
if (!CLI_ISCONTAINED(src, ssize, src + peinfo->sections[i].rva, peinfo->sections[i].rsz))
break;
if (fmap_readn(map, src + peinfo->sections[i].rva, peinfo->sections[i].raw, peinfo->sections[i].rsz) != peinfo->sections[i].rsz)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
break;
}
if (i != peinfo->nsections) {
cli_dbgmsg("cli_scanpe: Aspack: Probably hacked/damaged Aspack file.\n");
free(src);
break;
}
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "Aspack");
#endif
CLI_UNPTEMP("cli_scanpe: Aspack", (src, 0));
CLI_UNPRESULTS("cli_scanpe: Aspack", (unaspack((uint8_t *)src, ssize, peinfo->sections, peinfo->nsections, peinfo->vep - 1, EC32(peinfo->pe_opt.opt32.ImageBase), ndesc, aspack_ver)), 1, (src, 0));
break;
}
/* NsPack */
while (DCONF & PE_CONF_NSPACK) {
uint32_t eprva = peinfo->vep;
uint32_t start_of_stuff, rep = peinfo->ep;
unsigned int nowinldr;
const char *nbuff;
src = epbuff;
if (*epbuff == '\xe9') { /* bitched headers */
eprva = cli_readint32(epbuff + 1) + peinfo->vep + 5;
if (!(rep = cli_rawaddr(eprva, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size)) && err)
break;
if (!(nbuff = fmap_need_off_once(map, rep, 24)))
break;
src = nbuff;
}
if (memcmp(src, "\x9c\x60\xe8\x00\x00\x00\x00\x5d\xb8\x07\x00\x00\x00", 13))
break;
nowinldr = 0x54 - cli_readint32(src + 17);
cli_dbgmsg("cli_scanpe: NsPack: Found *start_of_stuff @delta-%x\n", nowinldr);
if (!(nbuff = fmap_need_off_once(map, rep - nowinldr, 4)))
break;
start_of_stuff = rep + cli_readint32(nbuff);
if (!(nbuff = fmap_need_off_once(map, start_of_stuff, 20)))
break;
src = nbuff;
if (!cli_readint32(nbuff)) {
start_of_stuff += 4; /* FIXME: more to do */
src += 4;
}
ssize = cli_readint32(src + 5) | 0xff;
dsize = cli_readint32(src + 9);
CLI_UNPSIZELIMITS("cli_scanpe: NsPack", MAX(ssize, dsize));
if (!ssize || !dsize || dsize != peinfo->sections[0].vsz)
break;
if (!(dest = cli_malloc(dsize))) {
cli_errmsg("cli_scanpe: NsPack: Unable to allocate memory for dest %u\n", dsize);
break;
}
/* memset(dest, 0xfc, dsize); */
if (!(src = fmap_need_off(map, start_of_stuff, ssize))) {
free(dest);
break;
}
/* memset(src, 0x00, ssize); */
eprva += 0x27a;
if (!(rep = cli_rawaddr(eprva, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size)) && err) {
free(dest);
break;
}
if (!(nbuff = fmap_need_off_once(map, rep, 5))) {
free(dest);
break;
}
fmap_unneed_off(map, start_of_stuff, ssize);
eprva = eprva + 5 + cli_readint32(nbuff + 1);
cli_dbgmsg("cli_scanpe: NsPack: OEP = %08x\n", eprva);
#if HAVE_JSON
if (pe_json != NULL)
cli_jsonstr(pe_json, "Packer", "NsPack");
#endif
CLI_UNPTEMP("cli_scanpe: NsPack", (dest, 0));
CLI_UNPRESULTS("cli_scanpe: NsPack", (unspack(src, dest, ctx, peinfo->sections[0].rva, EC32(peinfo->pe_opt.opt32.ImageBase), eprva, ndesc)), 0, (dest, 0));
break;
}
/* to be continued ... */
/* !!!!!!!!!!!!!! PACKERS END HERE !!!!!!!!!!!!!! */
ctx->corrupted_input = corrupted_cur;
/* Bytecode BC_PE_UNPACKER hook */
bc_ctx = cli_bytecode_context_alloc();
if (!bc_ctx) {
cli_errmsg("cli_scanpe: can't allocate memory for bc_ctx\n");
return CL_EMEM;
}
cli_bytecode_context_setpe(bc_ctx, &pedata, peinfo->sections);
cli_bytecode_context_setctx(bc_ctx, ctx);
ret = cli_bytecode_runhook(ctx, ctx->engine, bc_ctx, BC_PE_UNPACKER, map);
switch (ret) {
case CL_VIRUS:
cli_exe_info_destroy(peinfo);
cli_bytecode_context_destroy(bc_ctx);
// TODO Handle allmatch
return CL_VIRUS;
case CL_SUCCESS:
ndesc = cli_bytecode_context_getresult_file(bc_ctx, &tempfile);
cli_bytecode_context_destroy(bc_ctx);
if (ndesc != -1 && tempfile) {
CLI_UNPRESULTS("cli_scanpe: bytecode PE hook", 1, 1, (0));
}
break;
default:
cli_bytecode_context_destroy(bc_ctx);
}
cli_exe_info_destroy(peinfo);
#if HAVE_JSON
if (cli_json_timeout_cycle_check(ctx, &toval) != CL_SUCCESS)
return CL_ETIMEOUT;
#endif
if (SCAN_ALLMATCHES && viruses_found)
return CL_VIRUS;
return CL_CLEAN;
}
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
int cli_pe_targetinfo(cli_ctx *ctx, struct cli_exe_info *peinfo)
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
{
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
return cli_peheader(ctx->fmap, peinfo, CLI_PEHEADER_OPT_EXTRACT_VINFO, NULL);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
/** Parse the PE header and, if successful, populate peinfo
*
* @param map The fmap_t backing the file being scanned
* @param peinfo A structure to populate with info from the PE header. This
* MUST be initialized via cli_exe_info_init prior to calling
* @param opts A bitfield indicating various options related to PE header
* parsing. The options are (prefixed with CLI_PEHEADER_OPT_):
* - NONE - Do default parsing
* - COLLECT_JSON - Populate ctx's json obj with PE header
* info
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
* - DBG_PRINT_INFO - Print debug information about the
* PE file. Right now, cli_peheader is
* called multiple times for a given PE,
* so you don't want to print out the
* same info each time.
* - EXTRACT_VINFO - Parse the PEs VERSION_INFO metadata
* and store it in peinfo->vinfo
* - STRICT_ON_PE_ERRORS - If specified, some cases that
* might be considered a broken
* executable cause RET_BROKEN_PE
* to be returned, but otherwise
* these will be tolerated.
* - REMOVE_MISSING_SECTIONS - If a section exists outside of the
* file, remove it from
* peinfo->sections. Otherwise, the
* rsz is just set to 0 for it.
* @param ctx The overarching cli_ctx. This is required with certain opts, but
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
* optional otherwise.
* @return If the PE header is parsed successfully, CLI_PEHEADER_RET_SUCCESS
* is returned. If it seems like the PE is broken,
* CLI_PEHEADER_RET_BROKEN_PE is returned. Otherwise, one of the
* other error codes is returned.
* The caller MUST destroy peinfo, regardless of what this function
* returns.
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
*
* TODO What constitutes a "broken PE" seems somewhat arbitrary in places.
* I think a PE should only be considered broken if it will not run on
* any version of Windows. We should invest more time to ensure that our
* broken PE detection more closely aligns with this.
*
* TODO Simplify and get rid of CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS if
* possible. We should either fail always or ignore always, IMO.
*
* TODO Simplify and get rid of CLI_PEHEADER_OPT_REMOVE_MISSING_SECTIONS if
* possible. I don't think it makes sense to have pieces of the code work
* off of incomplete representations of the sections (for instance, I wonder
* if this makes any of the bytecode APIs return unexpected values). This
* appears to have been implemented to prevent ClamAV from crashing, though,
* (bb11155) so we need to ensure the underlying issues are addressed.
*
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
* TODO Consolidate when information about the PE is printed (after successful
* PE parsing). This will allow us to simplify the code. Some fail cases,
* then, will cause PE info to not be printed at all, but I think this is
* acceptable. The debug messages generated in the fail cases should point to
* what happened, and that's enough to track down the cause of any issues.
*
* TODO Same as above but with JSON creation
*/
int cli_peheader(fmap_t *map, struct cli_exe_info *peinfo, uint32_t opts, cli_ctx *ctx)
{
uint16_t e_magic; /* DOS signature ("MZ") */
const char *archtype = NULL, *subsystem = NULL;
time_t timestamp;
char timestr[32];
uint32_t data_dirs_size;
uint16_t opt_hdr_size;
uint32_t stored_opt_hdr_size;
struct pe_image_file_hdr *file_hdr;
struct pe_image_optional_hdr32 *opt32;
struct pe_image_optional_hdr64 *opt64;
struct pe_image_section_hdr *section_hdrs = NULL;
size_t i, j, section_pe_idx;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
unsigned int err;
uint32_t salign, falign;
size_t fsize;
ssize_t at;
uint32_t is_dll = 0;
uint32_t is_exe = 0;
int native = 0;
size_t read;
int ret = CLI_PEHEADER_RET_GENERIC_ERROR;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
int toval = 0;
struct json_object *pe_json = NULL;
char jsonbuf[128];
#endif
if (ctx == NULL &&
(opts & CLI_PEHEADER_OPT_COLLECT_JSON ||
opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO)) {
cli_errmsg("cli_peheader: ctx can't be NULL for options specified\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
pe_json = get_pe_property(ctx);
}
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
fsize = map->len - peinfo->offset;
if (fmap_readn(map, &e_magic, peinfo->offset, sizeof(e_magic)) != sizeof(e_magic)) {
cli_dbgmsg("cli_peheader: Can't read DOS signature\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (EC16(e_magic) != PE_IMAGE_DOS_SIGNATURE && EC16(e_magic) != PE_IMAGE_DOS_SIGNATURE_OLD) {
cli_dbgmsg("cli_peheader: Invalid DOS signature\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (fmap_readn(map, &(peinfo->e_lfanew), peinfo->offset + 58 + sizeof(e_magic), sizeof(peinfo->e_lfanew)) != sizeof(peinfo->e_lfanew)) {
/* truncated header? */
cli_dbgmsg("cli_peheader: Unable to read e_lfanew - truncated header?\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->e_lfanew = EC32(peinfo->e_lfanew);
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("e_lfanew == %d\n", peinfo->e_lfanew);
}
if (!peinfo->e_lfanew) {
cli_dbgmsg("cli_peheader: Not a PE file - e_lfanew == 0\n");
goto done;
2006-01-14 13:58:35 +00:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (fmap_readn(map, &(peinfo->file_hdr), peinfo->offset + peinfo->e_lfanew, sizeof(struct pe_image_file_hdr)) != sizeof(struct pe_image_file_hdr)) {
/* bad information in e_lfanew - probably not a PE file */
cli_dbgmsg("cli_peheader: Can't read file header\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
file_hdr = &(peinfo->file_hdr);
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (EC32(file_hdr->Magic) != PE_IMAGE_NT_SIGNATURE) {
cli_dbgmsg("cli_peheader: Invalid PE signature (probably NE file)\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (EC16(file_hdr->Characteristics) & 0x2000) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON)
cli_jsonstr(pe_json, "Type", "DLL");
#endif
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("File type: DLL\n");
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
is_dll = 1;
} else if (EC16(file_hdr->Characteristics) & 0x0002) {
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON)
cli_jsonstr(pe_json, "Type", "EXE");
#endif
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("File type: Executable\n");
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
is_exe = 1;
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!is_dll && !is_exe) {
cli_dbgmsg("cli_peheader: Assumption Violated: PE is not a DLL or EXE\n");
// TODO Don't continue if not an exe or dll?
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->is_dll = is_dll;
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO ||
opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
switch (EC16(file_hdr->Machine)) {
case 0x0:
archtype = "Unknown";
break;
case 0x1:
// New as of Windows 10, version 1607 and Windows Server 2016
archtype = "Target Host";
break;
case 0x14c:
archtype = "80386";
break;
case 0x14d:
archtype = "80486";
break;
case 0x14e:
archtype = "80586";
break;
case 0x160:
archtype = "R3000 MIPS BE";
break;
case 0x162:
archtype = "R3000 MIPS LE";
break;
case 0x166:
archtype = "R4000 MIPS LE";
break;
case 0x168:
archtype = "R10000 MIPS LE";
break;
case 0x169:
archtype = "WCE MIPS LE";
break;
case 0x184:
archtype = "DEC Alpha AXP";
break;
case 0x1a2:
archtype = "Hitachi SH3 LE";
break;
case 0x1a3:
archtype = "Hitachi SH3-DSP";
break;
case 0x1a4:
archtype = "Hitachi SH3-E LE";
break;
case 0x1a6:
archtype = "Hitachi SH4 LE";
break;
case 0x1a8:
archtype = "Hitachi SH5";
break;
case 0x1c0:
archtype = "ARM LE";
break;
case 0x1c2:
archtype = "ARM Thumb/Thumb-2 LE";
break;
case 0x1c4:
archtype = "ARM Thumb-2 LE";
break;
case 0x1d3:
archtype = "AM33";
break;
case 0x1f0:
archtype = "PowerPC LE";
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
case 0x1f1:
archtype = "PowerPC FP";
break;
case 0x200:
archtype = "IA64";
break;
case 0x266:
archtype = "MIPS16";
break;
case 0x268:
archtype = "M68k";
break;
case 0x284:
archtype = "DEC Alpha AXP 64bit";
break;
case 0x366:
archtype = "MIPS+FPU";
break;
case 0x466:
archtype = "MIPS16+FPU";
break;
case 0x520:
archtype = "Infineon TriCore";
break;
case 0xcef:
archtype = "CEF";
break;
case 0xebc:
archtype = "EFI Byte Code";
break;
case 0x8664:
archtype = "AMD64";
break;
case 0x9041:
archtype = "M32R";
break;
case 0xaa64:
archtype = "ARM64 LE";
break;
case 0xc0ee:
archtype = "CEE";
break;
default:
archtype = "Unknown";
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO)
cli_dbgmsg("Machine type: %s\n", archtype);
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON)
cli_jsonstr(pe_json, "ArchType", archtype);
#endif
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->nsections = EC16(file_hdr->NumberOfSections);
if (peinfo->nsections == 0) {
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
pe_add_heuristic_property(ctx, "BadNumberOfSections");
}
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Investigate how corrupted_input is set and whether this
// check is needed
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO &&
!ctx->corrupted_input) {
if (peinfo->nsections == 0) {
cli_dbgmsg("cli_peheader: Invalid NumberOfSections (0)\n");
}
}
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-05-07 10:56:35 -04:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
timestamp = (time_t)EC32(file_hdr->TimeDateStamp);
opt_hdr_size = EC16(file_hdr->SizeOfOptionalHeader);
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("NumberOfSections: %d\n", peinfo->nsections);
cli_dbgmsg("TimeDateStamp: %s", cli_ctime(&timestamp, timestr, sizeof(timestr)));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("SizeOfOptionalHeader: 0x%x\n", opt_hdr_size);
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
cli_jsonint(pe_json, "NumberOfSections", peinfo->nsections);
/* NOTE: the TimeDateStamp value will look like "Wed Dec 31 19:00:00 1969\n" */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_jsonstr(pe_json, "TimeDateStamp", cli_ctime(&timestamp, timestr, sizeof(timestr)));
cli_jsonint(pe_json, "SizeOfOptionalHeader", opt_hdr_size);
}
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// Ensure there are enough bytes to cover the full optional header,
// not including the data directory entries (which aren't all gauranteed
// to be there)
if (opt_hdr_size < sizeof(struct pe_image_optional_hdr32)) {
cli_dbgmsg("cli_peheader: SizeOfOptionalHeader too small\n");
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
pe_add_heuristic_property(ctx, "BadOptionalHeaderSize");
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#endif
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
at = peinfo->offset + peinfo->e_lfanew + sizeof(struct pe_image_file_hdr);
if (fmap_readn(map, &(peinfo->pe_opt.opt32), at, sizeof(struct pe_image_optional_hdr32)) != sizeof(struct pe_image_optional_hdr32)) {
cli_dbgmsg("cli_peheader: Can't read optional file header\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
stored_opt_hdr_size = sizeof(struct pe_image_optional_hdr32);
at += stored_opt_hdr_size;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
opt32 = &(peinfo->pe_opt.opt32);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (EC16(opt32->Magic) == PE32P_SIGNATURE) { /* PE+ */
// The PE32+ optional header is bigger by 16 bytes, so map in the
// additional bytes here
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opt_hdr_size < sizeof(struct pe_image_optional_hdr64)) {
cli_dbgmsg("cli_peheader: Incorrect SizeOfOptionalHeader for PE32+\n");
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
pe_add_heuristic_property(ctx, "BadOptionalHeaderSizePE32Plus");
}
#endif
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (fmap_readn(map, (void *)(((size_t) & (peinfo->pe_opt.opt64)) + sizeof(struct pe_image_optional_hdr32)), at, OPT_HDR_SIZE_DIFF) != OPT_HDR_SIZE_DIFF) {
cli_dbgmsg("cli_peheader: Can't read additional optional file header bytes\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
stored_opt_hdr_size += OPT_HDR_SIZE_DIFF;
at += OPT_HDR_SIZE_DIFF;
peinfo->is_pe32plus = 1;
opt64 = &(peinfo->pe_opt.opt64);
peinfo->vep = EC32(opt64->AddressOfEntryPoint);
peinfo->hdr_size = EC32(opt64->SizeOfHeaders);
peinfo->ndatadirs = EC32(opt64->NumberOfRvaAndSizes);
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("File format: PE32+\n");
cli_dbgmsg("MajorLinkerVersion: %d\n", opt64->MajorLinkerVersion);
cli_dbgmsg("MinorLinkerVersion: %d\n", opt64->MinorLinkerVersion);
cli_dbgmsg("SizeOfCode: 0x%x\n", EC32(opt64->SizeOfCode));
cli_dbgmsg("SizeOfInitializedData: 0x%x\n", EC32(opt64->SizeOfInitializedData));
cli_dbgmsg("SizeOfUninitializedData: 0x%x\n", EC32(opt64->SizeOfUninitializedData));
cli_dbgmsg("AddressOfEntryPoint: 0x%x\n", peinfo->vep);
cli_dbgmsg("BaseOfCode: 0x%x\n", EC32(opt64->BaseOfCode));
cli_dbgmsg("SectionAlignment: 0x%x\n", EC32(opt64->SectionAlignment));
cli_dbgmsg("FileAlignment: 0x%x\n", EC32(opt64->FileAlignment));
cli_dbgmsg("MajorSubsystemVersion: %d\n", EC16(opt64->MajorSubsystemVersion));
cli_dbgmsg("MinorSubsystemVersion: %d\n", EC16(opt64->MinorSubsystemVersion));
cli_dbgmsg("SizeOfImage: 0x%x\n", EC32(opt64->SizeOfImage));
cli_dbgmsg("SizeOfHeaders: 0x%x\n", peinfo->hdr_size);
cli_dbgmsg("NumberOfRvaAndSizes: %u\n", peinfo->ndatadirs);
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
cli_jsonint(pe_json, "MajorLinkerVersion", opt64->MajorLinkerVersion);
cli_jsonint(pe_json, "MinorLinkerVersion", opt64->MinorLinkerVersion);
cli_jsonint(pe_json, "SizeOfCode", EC32(opt64->SizeOfCode));
cli_jsonint(pe_json, "SizeOfInitializedData", EC32(opt64->SizeOfInitializedData));
cli_jsonint(pe_json, "SizeOfUninitializedData", EC32(opt64->SizeOfUninitializedData));
cli_jsonint(pe_json, "NumberOfRvaAndSizes", EC32(opt64->NumberOfRvaAndSizes));
cli_jsonint(pe_json, "MajorSubsystemVersion", EC16(opt64->MajorSubsystemVersion));
cli_jsonint(pe_json, "MinorSubsystemVersion", EC16(opt64->MinorSubsystemVersion));
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", peinfo->vep);
cli_jsonstr(pe_json, "EntryPoint", jsonbuf);
2014-05-07 10:56:35 -04:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt64->BaseOfCode));
cli_jsonstr(pe_json, "BaseOfCode", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt64->SectionAlignment));
cli_jsonstr(pe_json, "SectionAlignment", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt64->FileAlignment));
cli_jsonstr(pe_json, "FileAlignment", jsonbuf);
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt64->SizeOfImage));
cli_jsonstr(pe_json, "SizeOfImage", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", peinfo->hdr_size);
cli_jsonstr(pe_json, "SizeOfHeaders", jsonbuf);
}
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
} else { /* PE */
peinfo->is_pe32plus = 0;
peinfo->vep = EC32(opt32->AddressOfEntryPoint);
peinfo->hdr_size = EC32(opt32->SizeOfHeaders);
peinfo->ndatadirs = EC32(opt32->NumberOfRvaAndSizes);
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("File format: PE\n");
cli_dbgmsg("MajorLinkerVersion: %d\n", opt32->MajorLinkerVersion);
cli_dbgmsg("MinorLinkerVersion: %d\n", opt32->MinorLinkerVersion);
cli_dbgmsg("SizeOfCode: 0x%x\n", EC32(opt32->SizeOfCode));
cli_dbgmsg("SizeOfInitializedData: 0x%x\n", EC32(opt32->SizeOfInitializedData));
cli_dbgmsg("SizeOfUninitializedData: 0x%x\n", EC32(opt32->SizeOfUninitializedData));
cli_dbgmsg("AddressOfEntryPoint: 0x%x\n", peinfo->vep);
cli_dbgmsg("BaseOfCode: 0x%x\n", EC32(opt32->BaseOfCode));
cli_dbgmsg("SectionAlignment: 0x%x\n", EC32(opt32->SectionAlignment));
cli_dbgmsg("FileAlignment: 0x%x\n", EC32(opt32->FileAlignment));
cli_dbgmsg("MajorSubsystemVersion: %d\n", EC16(opt32->MajorSubsystemVersion));
cli_dbgmsg("MinorSubsystemVersion: %d\n", EC16(opt32->MinorSubsystemVersion));
cli_dbgmsg("SizeOfImage: 0x%x\n", EC32(opt32->SizeOfImage));
cli_dbgmsg("SizeOfHeaders: 0x%x\n", peinfo->hdr_size);
cli_dbgmsg("NumberOfRvaAndSizes: %u\n", peinfo->ndatadirs);
2014-11-07 10:15:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
cli_jsonint(pe_json, "MajorLinkerVersion", opt32->MajorLinkerVersion);
cli_jsonint(pe_json, "MinorLinkerVersion", opt32->MinorLinkerVersion);
cli_jsonint(pe_json, "SizeOfCode", EC32(opt32->SizeOfCode));
cli_jsonint(pe_json, "SizeOfInitializedData", EC32(opt32->SizeOfInitializedData));
cli_jsonint(pe_json, "SizeOfUninitializedData", EC32(opt32->SizeOfUninitializedData));
cli_jsonint(pe_json, "NumberOfRvaAndSizes", EC32(opt32->NumberOfRvaAndSizes));
cli_jsonint(pe_json, "MajorSubsystemVersion", EC16(opt32->MajorSubsystemVersion));
cli_jsonint(pe_json, "MinorSubsystemVersion", EC16(opt32->MinorSubsystemVersion));
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", peinfo->vep);
cli_jsonstr(pe_json, "EntryPoint", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt32->BaseOfCode));
cli_jsonstr(pe_json, "BaseOfCode", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt32->SectionAlignment));
cli_jsonstr(pe_json, "SectionAlignment", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt32->FileAlignment));
cli_jsonstr(pe_json, "FileAlignment", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", EC32(opt32->SizeOfImage));
cli_jsonstr(pe_json, "SizeOfImage", jsonbuf);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
snprintf(jsonbuf, sizeof(jsonbuf), "0x%x", peinfo->hdr_size);
cli_jsonstr(pe_json, "SizeOfHeaders", jsonbuf);
}
#endif
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
salign = (peinfo->is_pe32plus) ? EC32(opt64->SectionAlignment) : EC32(opt32->SectionAlignment);
falign = (peinfo->is_pe32plus) ? EC32(opt64->FileAlignment) : EC32(opt32->FileAlignment);
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
switch (peinfo->is_pe32plus ? EC16(opt64->Subsystem) : EC16(opt32->Subsystem)) {
case 0:
subsystem = "Unknown";
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
case 1:
subsystem = "Native (svc)";
native = 1;
break;
case 2:
subsystem = "Win32 GUI";
break;
case 3:
subsystem = "Win32 console";
break;
case 5:
subsystem = "OS/2 console";
break;
case 7:
subsystem = "POSIX console";
break;
case 8:
subsystem = "Native Win9x driver";
break;
case 9:
subsystem = "WinCE GUI";
break;
case 10:
subsystem = "EFI application";
break;
case 11:
subsystem = "EFI driver";
break;
case 12:
subsystem = "EFI runtime driver";
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
case 13:
subsystem = "EFI ROM image";
2014-11-07 10:15:30 -05:00
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
case 14:
subsystem = "Xbox";
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
case 16:
subsystem = "Boot application";
break;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
default:
subsystem = "Unknown";
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("Subsystem: %s\n", subsystem);
cli_dbgmsg("------------------------------------\n");
}
2014-05-07 10:56:35 -04:00
#if HAVE_JSON
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON)
cli_jsonstr(pe_json, "Subsystem", subsystem);
2014-05-07 10:56:35 -04:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!native && (!salign || (salign % 0x1000))) {
cli_dbgmsg("cli_peheader: Bad section alignment\n");
if (opts & CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!native && (!falign || (falign % 0x200))) {
cli_dbgmsg("cli_peheader: Bad file alignment\n");
if (opts & CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
2004-05-11 23:30:57 +00:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// Map in the optional header data directories. The spec defines 16
// directory entries, but NumberOfRvaAndSizes can be less than that
// and the Windows loader will pretend that the data directory does
// not exist. NumberOfRvaAndSizes can be larger than that too, which
// the Windows loader is OK with. To populate peinfo->dirs, we will
// copy in as many data dirs are specified but for a max of 16 (and
// adjust peinfo->ndatadirs accordingly)
2011-04-09 03:58:23 +02:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (peinfo->ndatadirs > 0x10) {
cli_dbgmsg("cli_peheader: Encountered NumberOfRvaAndSizes > 16 (suspicious)\n");
2009-10-02 17:33:11 +03:00
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// In the case where we won't fully populate dirs with file data,
// ensure that the underlying memory is zero so that existing code
// can interact with peinfo->dirs without using peinfo->ndatadirs
if (peinfo->ndatadirs < sizeof(peinfo->dirs) / sizeof(peinfo->dirs[0])) {
memset(&(peinfo->dirs), '\0', sizeof(peinfo->dirs));
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->ndatadirs = MIN(peinfo->ndatadirs, sizeof(peinfo->dirs) / sizeof(peinfo->dirs[0]));
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
data_dirs_size = sizeof(struct pe_image_data_dir) * peinfo->ndatadirs;
if (opt_hdr_size < (stored_opt_hdr_size + data_dirs_size)) {
cli_dbgmsg("cli_peheader: SizeOfOptionalHeader too small (doesn't include data dir size)\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
}
2009-10-02 17:33:11 +03:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
read = fmap_readn(map, peinfo->dirs, at, data_dirs_size);
if ((read == (size_t)-1) || (read != data_dirs_size)) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_peheader: Can't read optional file header data dirs\n");
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
at += data_dirs_size;
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opt_hdr_size != (stored_opt_hdr_size + data_dirs_size)) {
/* Seek to the end of the long header */
cli_dbgmsg("cli_peheader: Encountered case where SizeOfOptionalHeader appears bigger than required\n");
at += opt_hdr_size - (stored_opt_hdr_size + data_dirs_size);
}
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO This level of processing might not be needed in all cases
2014-11-07 10:15:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// Sanity checks
// TODO Also check that salign >= falign
if (peinfo->hdr_size != PESALIGN(peinfo->hdr_size, salign)) {
cli_dbgmsg("cli_peheader: SizeOfHeader is not aligned to the SectionAlignment\n");
}
if (peinfo->hdr_size != PESALIGN(peinfo->hdr_size, falign)) {
cli_dbgmsg("cli_peheader: SizeOfHeader is not aligned to the FileAlignment\n");
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Why align here? -- /* Aligned headers virtual size */
// hdr_size should already be rounded up
// to a multiple of the file alignment.
// TODO in cli_checkpe_fp this aligned to falign, elsewhere it aligned to salign
peinfo->hdr_size = PESALIGN(peinfo->hdr_size, salign);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->sections = (struct cli_exe_section *)cli_calloc(peinfo->nsections, sizeof(struct cli_exe_section));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!peinfo->sections) {
cli_dbgmsg("cli_peheader: Can't allocate memory for section headers\n");
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
section_hdrs = (struct pe_image_section_hdr *)cli_calloc(peinfo->nsections, sizeof(struct pe_image_section_hdr));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!section_hdrs) {
cli_dbgmsg("cli_peheader: Can't allocate memory for section headers\n");
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
read = fmap_readn(map, section_hdrs, at, peinfo->nsections * sizeof(struct pe_image_section_hdr));
if ((read == (size_t)-1) || (read != peinfo->nsections * sizeof(struct pe_image_section_hdr))) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("cli_peheader: Can't read section header - possibly broken PE file\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
at += sizeof(struct pe_image_section_hdr) * peinfo->nsections;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Verify that this performs correctly
// TODO I'm not sure why this is necessary since the specification says
// that PointerToRawData is expected to be a multiple of the file
// alignment. Should we report this is as a PE with an error?
for (i = 0; falign != 0x200 && i < (size_t)peinfo->nsections; i++) {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* file alignment fallback mode - blah */
if (falign && section_hdrs[i].SizeOfRawData && EC32(section_hdrs[i].PointerToRawData) % falign && !(EC32(section_hdrs[i].PointerToRawData) % 0x200)) {
cli_dbgmsg("cli_peheader: Encountered section with unexpected alignment - triggering fallback mode\n");
falign = 0x200;
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
fsize = (map->len - peinfo->offset);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Why do we fix up these alignments? This shouldn't be needed?
for (i = 0, section_pe_idx = 0; i < peinfo->nsections; i++, section_pe_idx++) {
struct cli_exe_section *section = &(peinfo->sections[i]);
struct pe_image_section_hdr *section_hdr = &(section_hdrs[i]);
char sname[9];
// TODO I don't see any documentation that says VirtualAddress and VirtualSize must be aligned
section->rva = PEALIGN(EC32(section_hdr->VirtualAddress), salign);
section->vsz = PESALIGN(EC32(section_hdr->VirtualSize), salign);
section->raw = PEALIGN(EC32(section_hdr->PointerToRawData), falign);
section->rsz = PESALIGN(EC32(section_hdr->SizeOfRawData), falign);
section->chr = EC32(section_hdr->Characteristics);
section->urva = EC32(section_hdr->VirtualAddress); /* Just in case */
section->uvsz = EC32(section_hdr->VirtualSize);
section->uraw = EC32(section_hdr->PointerToRawData);
section->ursz = EC32(section_hdr->SizeOfRawData);
/* First, if a section exists totally outside of a file, remove the
* section from the list or zero out it's size. */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (section->rsz) { /* Don't bother with virtual only sections */
if (section->raw >= fsize || section->uraw >= fsize) {
cli_dbgmsg("cli_peheader: Broken PE file - Section %zu starts or exists beyond the end of file (Offset@ %lu, Total filesize %lu)\n", section_pe_idx, (unsigned long)section->raw, (unsigned long)fsize);
if (opts & CLI_PEHEADER_OPT_REMOVE_MISSING_SECTIONS) {
if (peinfo->nsections == 1) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (j = i; j < (size_t)(peinfo->nsections - 1); j++)
memcpy(&(peinfo->sections[j]), &(peinfo->sections[j + 1]), sizeof(struct cli_exe_section));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (j = i; j < (size_t)(peinfo->nsections - 1); j++)
memcpy(&section_hdrs[j], &section_hdrs[j + 1], sizeof(struct pe_image_section_hdr));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->nsections--;
// Adjust i since we removed a section and continue on
i--;
continue;
} else {
section->rsz = 0;
section->ursz = 0;
}
} else {
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
/* If a section is truncated, adjust it's size value */
if (!CLI_ISCONTAINED_0_TO(fsize, section->raw, section->rsz)) {
cli_dbgmsg("cli_peheader: PE Section %zu raw+rsz extends past the end of the file by %lu bytes\n", section_pe_idx, (section->raw + section->rsz) - fsize);
section->rsz = fsize - section->raw;
}
if (!CLI_ISCONTAINED_0_TO(fsize, section->uraw, section->ursz)) {
cli_dbgmsg("cli_peheader: PE Section %zu uraw+ursz extends past the end of the file by %lu bytes\n", section_pe_idx, (section->uraw + section->ursz) - fsize);
section->ursz = fsize - section->uraw;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 11:12:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
strncpy(sname, (char *)section_hdr->Name, 8);
sname[8] = '\0';
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
add_section_info(ctx, &peinfo->sections[i]);
if (cli_json_timeout_cycle_check(ctx, &toval) != CL_SUCCESS) {
ret = CLI_PEHEADER_RET_JSON_TIMEOUT;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2014-11-07 11:12:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#endif
2014-11-07 11:12:30 -05:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO Why do we do this
// TODO Should this be done before we dump the json
if (!section->vsz && section->rsz)
section->vsz = PESALIGN(section->ursz, salign);
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("Section %zu\n", section_pe_idx);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_dbgmsg("Section name: %s\n", sname);
cli_dbgmsg("Section data (from headers - in memory)\n");
cli_dbgmsg("VirtualSize: 0x%x 0x%x\n", section->uvsz, section->vsz);
cli_dbgmsg("VirtualAddress: 0x%x 0x%x\n", section->urva, section->rva);
cli_dbgmsg("SizeOfRawData: 0x%x 0x%x\n", section->ursz, section->rsz);
cli_dbgmsg("PointerToRawData: 0x%x 0x%x\n", section->uraw, section->raw);
if (section->chr & 0x20) {
cli_dbgmsg("Section contains executable code\n");
}
if (section->vsz < section->rsz) {
cli_dbgmsg("Section contains free space\n");
/*
cli_dbgmsg("Dumping %d bytes\n", section_hdr.SizeOfRawData - section_hdr.VirtualSize);
ddump(desc, section_hdr.PointerToRawData + section_hdr.VirtualSize, section_hdr.SizeOfRawData - section_hdr.VirtualSize, cli_gentemp(NULL));
*/
}
if (section->chr & 0x20000000)
cli_dbgmsg("Section's memory is executable\n");
if (section->chr & 0x80000000)
cli_dbgmsg("Section's memory is writeable\n");
cli_dbgmsg("------------------------------------\n");
2014-11-07 11:12:30 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!salign || (section->urva % salign)) { /* Bad section alignment */
cli_dbgmsg("cli_peheader: Broken PE - section's VirtualAddress is misaligned\n");
if (opts & CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO should we skip all of these checks if it's an empty
// section? Why the exception for uraw?
if (section->urva >> 31 || section->uvsz >> 31 || (section->rsz && section->uraw >> 31) || peinfo->sections[i].ursz >> 31) {
cli_dbgmsg("cli_peheader: Found PE values with sign bit set\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!i) {
if (section->urva != peinfo->hdr_size) { /* Bad first section RVA */
cli_dbgmsg("cli_peheader: First section doesn't start immediately after the header\n");
if (opts & CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->min = section->rva;
peinfo->max = section->rva + section->rsz;
} else {
if (section->urva - peinfo->sections[i - 1].urva != peinfo->sections[i - 1].vsz) { /* No holes, no overlapping, no virtual disorder */
cli_dbgmsg("cli_peheader: Virtually misplaced section (wrong order, overlapping, non contiguous)\n");
if (opts & CLI_PEHEADER_OPT_STRICT_ON_PE_ERRORS) {
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (section->rva < peinfo->min)
peinfo->min = section->rva;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (section->rva + section->rsz > peinfo->max) {
peinfo->max = section->rva + section->rsz;
peinfo->overlay_start = section->raw + section->rsz;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO This case might be possible, which would lead to us
// mislabelling the overlay
if (section->raw + section->rsz > peinfo->max) {
cli_dbgmsg("cli_peheader: Assumption Violated: Last section end RVA isn't tied to the last section\n");
}
2014-11-07 11:12:30 -05:00
}
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->overlay_size = fsize - peinfo->overlay_start;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// NOTE: For DLLs the entrypoint is likely to be zero
// TODO Should this offset include peinfo->offset?
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (!(peinfo->ep = cli_rawaddr(peinfo->vep, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size)) && err) {
cli_dbgmsg("cli_peheader: Broken PE file - Can't map EntryPoint to a file offset\n");
ret = CLI_PEHEADER_RET_BROKEN_PE;
goto done;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#if HAVE_JSON
if (opts & CLI_PEHEADER_OPT_COLLECT_JSON) {
cli_jsonint(pe_json, "EntryPointOffset", peinfo->ep);
if (cli_json_timeout_cycle_check(ctx, &toval) != CL_SUCCESS) {
ret = CLI_PEHEADER_RET_JSON_TIMEOUT;
goto done;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
}
2010-01-04 14:56:04 +01:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
#endif
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (opts & CLI_PEHEADER_OPT_DBG_PRINT_INFO) {
cli_dbgmsg("EntryPoint offset: 0x%x (%d)\n", peinfo->ep, peinfo->ep);
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
if (is_dll || peinfo->ndatadirs < 3 || !peinfo->dirs[2].Size)
2014-11-07 11:12:30 -05:00
peinfo->res_addr = 0;
else
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
peinfo->res_addr = EC32(peinfo->dirs[2].VirtualAddress);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
while (opts & CLI_PEHEADER_OPT_EXTRACT_VINFO &&
peinfo->ndatadirs >= 3 && peinfo->dirs[2].Size) {
2014-11-07 11:12:30 -05:00
struct vinfo_list vlist;
const uint8_t *vptr, *baseptr;
uint32_t rva, res_sz;
2010-01-04 14:56:04 +01:00
// TODO This code assumes peinfo->offset == 0, which might not always
// be the case.
if (0 != peinfo->offset) {
cli_dbgmsg("cli_peheader: Assumption Violated: Looking for version info when peinfo->offset != 0\n");
}
2014-11-07 11:12:30 -05:00
memset(&vlist, 0, sizeof(vlist));
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
findres(0x10, 0xffffffff, map, peinfo, versioninfo_cb, &vlist);
if (!vlist.count)
2014-11-07 11:12:30 -05:00
break; /* No version_information */
2010-01-04 14:56:04 +01:00
if (cli_hashset_init(&peinfo->vinfo, 32, 80)) {
2014-11-07 11:12:30 -05:00
cli_errmsg("cli_peheader: Unable to init vinfo hashset\n");
goto done;
2014-11-07 11:12:30 -05:00
}
2010-01-04 14:56:04 +01:00
2014-11-07 11:12:30 -05:00
err = 0;
for (i = 0; i < vlist.count; i++) { /* enum all version_information res - RESUMABLE */
cli_dbgmsg("cli_peheader: parsing version info @ rva %x (%zu/%u)\n", vlist.rvas[i], i + 1, vlist.count);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rva = cli_rawaddr(vlist.rvas[i], peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err)
2014-11-07 11:12:30 -05:00
continue;
if (!(vptr = fmap_need_off_once(map, rva, 16)))
2014-11-07 11:12:30 -05:00
continue;
baseptr = vptr - rva;
/* parse resource */
rva = cli_readint32(vptr); /* ptr to version_info */
res_sz = cli_readint32(vptr + 4); /* sizeof(resource) */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
rva = cli_rawaddr(rva, peinfo->sections, peinfo->nsections, &err, fsize, peinfo->hdr_size);
if (err)
2014-11-07 11:12:30 -05:00
continue;
if (!(vptr = fmap_need_off_once(map, rva, res_sz)))
2014-11-07 11:12:30 -05:00
continue;
while (res_sz > 4) { /* look for version_info - NOT RESUMABLE (expecting exactly one versioninfo) */
2014-11-07 11:12:30 -05:00
uint32_t vinfo_sz, vinfo_val_sz, got_varfileinfo = 0;
vinfo_sz = vinfo_val_sz = cli_readint32(vptr);
vinfo_sz &= 0xffff;
if (vinfo_sz > res_sz)
2014-11-07 11:12:30 -05:00
break; /* the content is larger than the container */
vinfo_val_sz >>= 16;
if (vinfo_sz <= 6 + 0x20 + 2 + 0x34 ||
vinfo_val_sz != 0x34 ||
memcmp(vptr + 6, "V\0S\0_\0V\0E\0R\0S\0I\0O\0N\0_\0I\0N\0F\0O\0\0\0", 0x20) ||
(unsigned int)cli_readint32(vptr + 0x28) != 0xfeef04bd) {
2014-11-07 11:12:30 -05:00
/* - there should be enough room for the header(6), the key "VS_VERSION_INFO"(20), the padding(2) and the value(34)
* - the value should be sizeof(fixedfileinfo)
* - the key should match
* - there should be some proper magic for fixedfileinfo */
break; /* there's no point in looking further */
}
2010-01-04 14:56:04 +01:00
2014-11-07 11:12:30 -05:00
/* move to the end of fixedfileinfo where the child elements are located */
vptr += 6 + 0x20 + 2 + 0x34;
vinfo_sz -= 6 + 0x20 + 2 + 0x34;
2010-01-05 01:01:48 +01:00
while (vinfo_sz > 6) { /* look for stringfileinfo - NOT RESUMABLE (expecting at most one stringfileinfo) */
2014-11-07 11:12:30 -05:00
uint32_t sfi_sz = cli_readint32(vptr) & 0xffff;
2010-01-04 14:56:04 +01:00
if (sfi_sz > vinfo_sz)
2014-11-07 11:12:30 -05:00
break; /* the content is larger than the container */
2010-01-04 14:56:04 +01:00
if (!got_varfileinfo && sfi_sz > 6 + 0x18 && !memcmp(vptr + 6, "V\0a\0r\0F\0i\0l\0e\0I\0n\0f\0o\0\0\0", 0x18)) {
2014-11-07 11:12:30 -05:00
/* skip varfileinfo as it sometimes appear before stringtableinfo */
vptr += sfi_sz;
vinfo_sz -= sfi_sz;
got_varfileinfo = 1;
continue;
}
2011-11-04 00:55:54 +01:00
if (sfi_sz <= 6 + 0x1e || memcmp(vptr + 6, "S\0t\0r\0i\0n\0g\0F\0i\0l\0e\0I\0n\0f\0o\0\0\0", 0x1e)) {
2014-11-07 11:12:30 -05:00
/* - there should be enough room for the header(6) and the key "StringFileInfo"(1e)
* - the key should match */
break; /* this is an implicit hard fail: parent is not resumable */
}
2010-01-04 14:56:04 +01:00
2014-11-07 11:12:30 -05:00
/* move to the end of stringfileinfo where the child elements are located */
vptr += 6 + 0x1e;
sfi_sz -= 6 + 0x1e;
2010-01-04 14:56:04 +01:00
while (sfi_sz > 6) { /* enum all stringtables - RESUMABLE */
uint32_t st_sz = cli_readint32(vptr) & 0xffff;
2014-11-07 11:12:30 -05:00
const uint8_t *next_vptr = vptr + st_sz;
uint32_t next_sfi_sz = sfi_sz - st_sz;
2010-01-04 14:56:04 +01:00
if (st_sz > sfi_sz || st_sz <= 24) {
2014-11-07 11:12:30 -05:00
/* - the content is larger than the container
- there's no room for a stringtables (headers(6) + key(16) + padding(2)) */
break; /* this is an implicit hard fail: parent is not resumable */
}
2010-01-04 14:56:04 +01:00
2014-11-07 11:12:30 -05:00
/* move to the end of stringtable where the child elements are located */
vptr += 24;
st_sz -= 24;
while (st_sz > 6) { /* enum all strings - RESUMABLE */
2014-11-07 11:12:30 -05:00
uint32_t s_sz, s_key_sz, s_val_sz;
s_sz = (cli_readint32(vptr) & 0xffff) + 3;
s_sz &= ~3;
if (s_sz > st_sz || s_sz <= 6 + 2 + 8) {
2014-11-07 11:12:30 -05:00
/* - the content is larger than the container
* - there's no room for a minimal string
* - there's no room for the value */
st_sz = 0;
2014-11-07 11:12:30 -05:00
sfi_sz = 0;
break; /* force a hard fail */
}
/* ~wcstrlen(key) */
for (s_key_sz = 6; s_key_sz + 1 < s_sz; s_key_sz += 2) {
if (vptr[s_key_sz] || vptr[s_key_sz + 1])
2014-11-07 11:12:30 -05:00
continue;
s_key_sz += 2;
break;
}
s_key_sz += 3;
s_key_sz &= ~3;
if (s_key_sz >= s_sz) {
2014-11-07 11:12:30 -05:00
/* key overflow */
vptr += s_sz;
st_sz -= s_sz;
continue;
}
s_val_sz = s_sz - s_key_sz;
s_key_sz -= 6;
if (s_val_sz <= 2) {
2014-11-07 11:12:30 -05:00
/* skip unset value */
vptr += s_sz;
st_sz -= s_sz;
continue;
}
if (cli_hashset_addkey(&peinfo->vinfo, (uint32_t)(vptr - baseptr + 6))) {
2014-11-07 11:12:30 -05:00
cli_errmsg("cli_peheader: Unable to add rva to vinfo hashset\n");
goto done;
2014-11-07 11:12:30 -05:00
}
if (cli_debug_flag) {
2014-11-07 11:12:30 -05:00
char *k, *v, *s;
/* FIXME: skip too long strings */
k = cli_utf16toascii((const char *)vptr + 6, s_key_sz);
if (k) {
v = cli_utf16toascii((const char *)vptr + s_key_sz + 6, s_val_sz);
if (v) {
s = cli_str2hex((const char *)vptr + 6, s_key_sz + s_val_sz);
if (s) {
2014-11-07 11:12:30 -05:00
cli_dbgmsg("VersionInfo (%x): '%s'='%s' - VI:%s\n", (uint32_t)(vptr - baseptr + 6), k, v, s);
free(s);
}
free(v);
}
free(k);
}
}
vptr += s_sz;
st_sz -= s_sz;
} /* enum all strings - RESUMABLE */
vptr = next_vptr;
2014-11-07 11:12:30 -05:00
sfi_sz = next_sfi_sz * (sfi_sz != 0);
} /* enum all stringtables - RESUMABLE */
break;
} /* look for stringfileinfo - NOT RESUMABLE */
break;
} /* look for version_info - NOT RESUMABLE */
} /* enum all version_information res - RESUMABLE */
2014-11-07 11:12:30 -05:00
break;
2010-01-05 01:01:48 +01:00
} /* while(dirs[2].Size) */
2010-01-04 14:56:04 +01:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// Do final preperations for peinfo to be passed back
peinfo->is_dll = is_dll;
ret = CLI_PEHEADER_RET_SUCCESS;
done:
/* In the fail case, peinfo will get destroyed by the caller */
if (NULL != section_hdrs) {
free(section_hdrs);
}
return ret;
}
2012-01-08 03:19:29 +01:00
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// TODO We should sort based on VirtualAddress instead, since PointerToRawData
// will be zero for sections where SizeOfRawData is zero. This also aligns
// with what tools like pefile do.
static int sort_sects(const void *first, const void *second)
{
2012-01-08 03:19:29 +01:00
const struct cli_exe_section *a = first, *b = second;
return (a->raw - b->raw);
}
/* Check the given PE file for an authenticode signature and return whether
* the signature is valid. There are two cases that this function should
* handle:
* - A PE file has an embedded Authenticode section
* - The PE file has no embedded Authenticode section but is covered by a
* catalog file that was loaded in via a -d
*
* If peinfo is NULL, one will be created internally and used
*
* CL_VERIFIED will be returned if the file was trusted based on its
* signature. CL_VIRUS will be returned if the file was blocked based on
* its signature. Otherwise, a cl_error_t error value will be returned.
*
* If CL_VIRUS is returned, cli_append_virus will get called, adding the
* name associated with the block list CRB rules to the list of found viruses.*/
cl_error_t cli_check_auth_header(cli_ctx *ctx, struct cli_exe_info *peinfo)
{
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
size_t at;
unsigned int i, j, hlen;
2012-01-08 03:19:29 +01:00
size_t fsize;
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
fmap_t *map = ctx->fmap;
void *hashctx = NULL;
struct pe_certificate_hdr cert_hdr;
struct cli_mapped_region *regions = NULL;
unsigned int nregions;
cl_error_t ret = CL_EVERIFY;
uint8_t authsha[SHA256_HASH_SIZE];
uint32_t sec_dir_offset;
uint32_t sec_dir_size;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
struct cli_exe_info _peinfo;
// If Authenticode parsing has been disabled via DCONF or an engine
// option, then don't continue on.
if (!(DCONF & PE_CONF_CERTS))
return CL_EVERIFY;
if (ctx->engine->engine_options & ENGINE_OPTIONS_DISABLE_PE_CERTS)
return CL_EVERIFY;
// If peinfo is NULL, initialize one. This makes it so that this function
// can be used easily by sigtool
if (NULL == peinfo) {
peinfo = &_peinfo;
cli_exe_info_init(peinfo, 0);
2012-01-08 03:19:29 +01:00
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
if (cli_peheader(ctx->fmap, peinfo, CLI_PEHEADER_OPT_NONE, NULL) != CLI_PEHEADER_RET_SUCCESS) {
cli_exe_info_destroy(peinfo);
return CL_EFORMAT;
}
2012-01-08 03:19:29 +01:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
sec_dir_offset = EC32(peinfo->dirs[4].VirtualAddress);
sec_dir_size = EC32(peinfo->dirs[4].Size);
// As an optimization, check the security DataDirectory here and if
// it's less than 8-bytes (and we aren't relying on this code to compute
// the section hashes), bail out if we don't have any Authenticode hashes
// loaded from .cat files. The value 2 in these calls is the sentinel value
// for the 'PE' .cat Authenticode hash file type.
2021-07-16 19:11:53 -07:00
if (sec_dir_size < 8 &&
!cli_hm_have_size(ctx->engine->hm_fp, CLI_HASH_SHA1, 2) &&
!cli_hm_have_size(ctx->engine->hm_fp, CLI_HASH_SHA256, 2)) {
ret = CL_BREAK;
goto finish;
}
2012-01-08 03:19:29 +01:00
fsize = map->len;
// We'll build a list of the regions that need to be hashed and pass it to
// asn1_check_mscat to do hash verification there (the hash algorithm is
// specified in the PKCS7 structure). We need to hash up to 4 regions
regions = (struct cli_mapped_region *)cli_calloc(4, sizeof(struct cli_mapped_region));
if (!regions) {
ret = CL_EMEM;
goto finish;
}
nregions = 0;
#define add_chunk_to_hash_list(_offset, _size) \
do { \
regions[nregions].offset = (_offset); \
regions[nregions].size = (_size); \
nregions++; \
} while (0)
2012-01-08 03:19:29 +01:00
// Pretty much every case below should return CL_EFORMAT
ret = CL_EFORMAT;
/* MZ to checksum */
at = 0;
hlen = peinfo->e_lfanew + sizeof(struct pe_image_file_hdr) + (peinfo->is_pe32plus ? offsetof(struct pe_image_optional_hdr64, CheckSum) : offsetof(struct pe_image_optional_hdr32, CheckSum));
add_chunk_to_hash_list(0, hlen);
at = hlen + 4;
/* Checksum to security */
if (peinfo->is_pe32plus)
hlen = sizeof(struct pe_image_optional_hdr64) - offsetof(struct pe_image_optional_hdr64, CheckSum) - 4;
else
hlen = sizeof(struct pe_image_optional_hdr32) - offsetof(struct pe_image_optional_hdr32, CheckSum) - 4;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
hlen += sizeof(struct pe_image_data_dir) * 4;
add_chunk_to_hash_list(at, hlen);
at += hlen + 8;
2012-01-08 03:19:29 +01:00
if (at > peinfo->hdr_size) {
goto finish;
}
2012-01-08 03:19:29 +01:00
if (sec_dir_offset) {
// Verify that we have all the bytes we expect in the authenticode sig
// and that the certificate table is the last thing in the file
// (according to the MS13-098 bulletin, this is a requirement)
if (fsize != sec_dir_size + sec_dir_offset) {
cli_dbgmsg("cli_check_auth_header: expected authenticode data at the end of the file\n");
goto finish;
}
// Hash everything else up to the start of the security section. Allow
// the case where at == sec_dir_offset without adding another region
// to hash, since this could technically be valid (although I haven't
// verified this).
if (at < sec_dir_offset) {
hlen = sec_dir_offset - at;
add_chunk_to_hash_list(at, hlen);
2020-01-03 15:53:29 -05:00
} else if (at > sec_dir_offset) {
cli_dbgmsg("cli_check_auth_header: security directory offset appears to overlap with the PE header\n");
goto finish;
}
2012-01-08 03:19:29 +01:00
// Parse the security directory header
if (fmap_readn(map, &cert_hdr, sec_dir_offset, sizeof(cert_hdr)) != sizeof(cert_hdr)) {
goto finish;
}
if (EC16(cert_hdr.revision) != WIN_CERT_REV_2) {
cli_dbgmsg("cli_check_auth_header: unsupported authenticode data revision\n");
goto finish;
}
if (EC16(cert_hdr.type) != WIN_CERT_TYPE_PKCS7) {
cli_dbgmsg("cli_check_auth_header: unsupported authenticode data type\n");
goto finish;
}
hlen = sec_dir_size;
if (EC32(cert_hdr.length) != hlen) {
/* This is the case that MS13-098 aimed to address, but it got
* pushback to where the fix (not allowing additional, non-zero
* bytes in the security directory) is now opt-in via a registry
* key. Given that most machines will treat these binaries as
* valid, we'll still parse the signature and just trust that
* our trust signatures are tailored enough to where any
* instances of this are reasonable (for instance, I saw one
* binary that appeared to use this to embed a license key.) */
cli_dbgmsg("cli_check_auth_header: MS13-098 violation detected, but continuing on to verify certificate\n");
}
at = sec_dir_offset + sizeof(cert_hdr);
hlen -= sizeof(cert_hdr);
ret = asn1_check_mscat((struct cl_engine *)(ctx->engine), map, at, hlen, regions, nregions, ctx);
if (CL_VERIFIED == ret) {
// We validated the embedded signature. Hooray!
goto finish;
} else if (CL_VIRUS == ret) {
// A block list rule hit - don't continue on to check hm_fp for a match
goto finish;
}
// Otherwise, we still need to check to see whether this file is
// covered by a .cat file (it's common these days for driver files
// to have .cat files covering PEs with embedded signatures)
} else {
// Hash everything else
if (at < fsize) {
hlen = fsize - at;
add_chunk_to_hash_list(at, hlen);
}
}
// At this point we should compute the SHA1 authenticode hash to see
// whether we've had any hashes added from external catalog files
static const struct supported_hashes {
const cli_hash_type_t hashtype;
const char *hashctx_name;
} supported_hashes[] = {
{CLI_HASH_SHA1, "sha1"},
{CLI_HASH_SHA256, "sha256"},
};
2021-07-16 19:11:53 -07:00
for (i = 0; i < (sizeof(supported_hashes) / sizeof(supported_hashes[0])); i++) {
const cli_hash_type_t hashtype = supported_hashes[i].hashtype;
const char *hashctx_name = supported_hashes[i].hashctx_name;
2021-07-16 19:11:53 -07:00
if (!cli_hm_have_size(ctx->engine->hm_fp, hashtype, 2)) {
continue;
}
hashctx = cl_hash_init(hashctx_name);
if (NULL == hashctx) {
ret = CL_EMEM;
goto finish;
}
2014-01-20 14:15:51 -05:00
for (j = 0; j < nregions; j++) {
const uint8_t *hptr;
if (0 == regions[j].size) {
continue;
}
if (!(hptr = fmap_need_off_once(map, regions[j].offset, regions[j].size))) {
break;
}
2014-01-20 14:15:51 -05:00
cl_update_hash(hashctx, hptr, regions[j].size);
}
if (j != nregions) {
goto finish;
}
cl_finish_hash(hashctx, authsha);
hashctx = NULL;
if (cli_hm_scan(authsha, 2, NULL, ctx->engine->hm_fp, hashtype) == CL_VIRUS) {
cli_dbgmsg("cli_check_auth_header: PE file trusted by catalog file (%s)\n", hashctx_name);
ret = CL_VERIFIED;
goto finish;
}
}
ret = CL_EVERIFY;
finish:
if (NULL != hashctx) {
cl_hash_destroy(hashctx);
}
if (NULL != regions) {
free(regions);
2014-01-21 14:32:09 -05:00
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
// If we created the peinfo, then destroy it. Otherwise we don't own it
if (&_peinfo == peinfo) {
cli_exe_info_destroy(peinfo);
}
return ret;
2012-01-08 03:19:29 +01:00
}
/* Print out either the MD5, SHA1, or SHA256 associated with the imphash or
* the individual sections. Also, this function computes the hashes of each
* section (sorted based on the RVAs of the sections) if hashes is non-NULL.
*
* If the section hashes are to be computed and returned, this function
* allocates memory for the section hashes, and it's up to the caller to free
* it. hashes->sections will be initialized to NULL at the beginning of the
* function, and if after the call it's value is non-NULL, the memory should be
* freed. Furthermore, if hashes->sections is non-NULL, the hashes can assume
* to be valid regardless of the return code.
*
* Also, a few other notes:
* - If a section has a virtual size of zero, it's corresponding hash value
* will not be computed and the hash contents will be all zeroes.
* - If a section extends beyond the end of the file, the section data and
* length will be truncated, and the hash generated accordingly
* - If a section exists completely outside of the file, it won't be included
* in the list of sections, and nsections will be adjusted accordingly.
*/
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
cl_error_t cli_genhash_pe(cli_ctx *ctx, unsigned int class, int type, stats_section_t *hashes)
{
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
unsigned int i;
struct cli_exe_info _peinfo;
struct cli_exe_info *peinfo = &_peinfo;
unsigned char *hash, *hashset[CLI_HASH_AVAIL_TYPES];
int genhash[CLI_HASH_AVAIL_TYPES];
int hlen = 0;
if (hashes) {
hashes->sections = NULL;
if (class != CL_GENHASH_PE_CLASS_SECTION || type != 1) {
cli_dbgmsg("`hashes` can only be populated with MD5 PE section data\n");
return CL_EARG;
}
}
if (class >= CL_GENHASH_PE_CLASS_LAST)
return CL_EARG;
// TODO see if peinfo can be passed in (or lives in ctx or something) and
// if so, use that to avoid having to re-parse the header
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_init(peinfo, 0);
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
if (cli_peheader(ctx->fmap, peinfo, CLI_PEHEADER_OPT_NONE, NULL) != CLI_PEHEADER_RET_SUCCESS) {
cli_exe_info_destroy(peinfo);
return CL_EFORMAT;
}
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_qsort(peinfo->sections, peinfo->nsections, sizeof(*(peinfo->sections)), sort_sects);
/* pick hashtypes to generate */
memset(genhash, 0, sizeof(genhash));
2016-07-12 11:53:59 -04:00
memset(hashset, 0, sizeof(hashset));
switch (type) {
case 1:
genhash[CLI_HASH_MD5] = 1;
hlen = hashlen[CLI_HASH_MD5];
hash = hashset[CLI_HASH_MD5] = cli_calloc(hlen, sizeof(char));
break;
case 2:
genhash[CLI_HASH_SHA1] = 1;
hlen = hashlen[CLI_HASH_SHA1];
hash = hashset[CLI_HASH_SHA1] = cli_calloc(hlen, sizeof(char));
break;
default:
genhash[CLI_HASH_SHA256] = 1;
hlen = hashlen[CLI_HASH_SHA256];
hash = hashset[CLI_HASH_SHA256] = cli_calloc(hlen, sizeof(char));
break;
}
if (!hash) {
cli_errmsg("cli_genhash_pe: cli_malloc failed!\n");
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return CL_EMEM;
}
if (hashes) {
hashes->nsections = peinfo->nsections;
hashes->sections = cli_calloc(peinfo->nsections, sizeof(struct cli_section_hash));
if (!(hashes->sections)) {
cli_exe_info_destroy(peinfo);
free(hash);
return CL_EMEM;
}
}
if (class == CL_GENHASH_PE_CLASS_SECTION) {
char *dstr;
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
for (i = 0; i < peinfo->nsections; i++) {
/* Generate hashes */
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
if (cli_hashsect(ctx->fmap, &peinfo->sections[i], hashset, genhash, genhash) == 1) {
if (cli_debug_flag) {
dstr = cli_str2hex((char *)hash, hlen);
cli_dbgmsg("Section{%u}: %u:%s\n", i, peinfo->sections[i].rsz, dstr ? (char *)dstr : "(NULL)");
if (dstr != NULL) {
free(dstr);
}
2016-07-12 11:53:59 -04:00
}
if (hashes) {
memcpy(hashes->sections[i].md5, hash, sizeof(hashes->sections[i].md5));
hashes->sections[i].len = peinfo->sections[i].rsz;
}
} else if (peinfo->sections[i].rsz) {
2016-07-12 11:53:59 -04:00
cli_dbgmsg("Section{%u}: failed to generate hash for section\n", i);
} else {
cli_dbgmsg("Section{%u}: section contains no data\n", i);
2016-07-12 11:53:59 -04:00
}
}
} else if (class == CL_GENHASH_PE_CLASS_IMPTBL) {
char *dstr;
uint32_t impsz = 0;
libclamav: Fix scan recursion tracking Scan recursion is the process of identifying files embedded in other files and then scanning them, recursively. Internally this process is more complex than it may sound because a file may have multiple layers of types before finding a new "file". At present we treat the recursion count in the scanning context as an index into both our fmap list AND our container list. These two lists are conceptually a part of the same thing and should be unified. But what's concerning is that the "recursion level" isn't actually incremented or decremented at the same time that we add a layer to the fmap or container lists but instead is more touchy-feely, increasing when we find a new "file". To account for this shadiness, the size of the fmap and container lists has always been a little longer than our "max scan recursion" limit so we don't accidentally overflow the fmap or container arrays (!). I've implemented a single recursion-stack as an array, similar to before, which includes a pointer to each fmap at each layer, along with the size and type. Push and pop functions add and remove layers whenever a new fmap is added. A boolean argument when pushing indicates if the new layer represents a new buffer or new file (descriptor). A new buffer will reset the "nested fmap level" (described below). This commit also provides a solution for an issue where we detect embedded files more than once during scan recursion. For illustration, imagine a tarball named foo.tar.gz with this structure: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | But suppose baz.exe embeds a ZIP archive and a 7Z archive, like this: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | baz.exe | PE | 0 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | │   └── hello.txt | ASCII | 2 | 0 | | └── sfx.7z | 7Z | 1 | 1 | |    └── world.txt | ASCII | 2 | 0 | (A) If we scan for embedded files at any layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | ├── foo.tar | TAR | 1 | 0 | | │ ├── bar.zip | ZIP | 2 | 1 | | │ │   └── hola.txt | ASCII | 3 | 0 | | │ ├── baz.exe | PE | 2 | 1 | | │ │ ├── sfx.zip | ZIP | 3 | 1 | | │ │ │   └── hello.txt | ASCII | 4 | 0 | | │ │ └── sfx.7z | 7Z | 3 | 1 | | │ │    └── world.txt | ASCII | 4 | 0 | | │ ├── sfx.zip | ZIP | 2 | 1 | | │ │   └── hello.txt | ASCII | 3 | 0 | | │ └── sfx.7z | 7Z | 2 | 1 | | │   └── world.txt | ASCII | 3 | 0 | | ├── sfx.zip | ZIP | 1 | 1 | | └── sfx.7z | 7Z | 1 | 1 | (A) is bad because it scans content more than once. Note that for the GZ layer, it may detect the ZIP and 7Z if the signature hits on the compressed data, which it might, though extracting the ZIP and 7Z will likely fail. The reason the above doesn't happen now is that we restrict embedded type scans for a bunch of archive formats to include GZ and TAR. (B) If we scan for embedded files at the foo.tar layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | ├── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 2 | 1 | | │   └── hello.txt | ASCII | 3 | 0 | | └── sfx.7z | 7Z | 2 | 1 | |    └── world.txt | ASCII | 3 | 0 | (B) is almost right. But we can achieve it easily enough only scanning for embedded content in the current fmap when the "nested fmap level" is 0. The upside is that it should safely detect all embedded content, even if it may think the sfz.zip and sfx.7z are in foo.tar instead of in baz.exe. The biggest risk I can think of affects ZIPs. SFXZIP detection is identical to ZIP detection, which is why we don't allow SFXZIP to be detected if insize of a ZIP. If we only allow embedded type scanning at fmap-layer 0 in each buffer, this will fail to detect the embedded ZIP if the bar.exe was not compressed in foo.zip and if non-compressed files extracted from ZIPs aren't extracted as new buffers: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.zip | ZIP | 0 | 0 | | └── bar.exe | PE | 1 | 1 | | └── sfx.zip | ZIP | 2 | 2 | Provided that we ensure all files extracted from zips are scanned in new buffers, option (B) should be safe. (C) If we scan for embedded files at the baz.exe layer, we may detect: | description | type | rec level | nested fmap level | | ------------------------- | ----- | --------- | ----------------- | | foo.tar.gz | GZ | 0 | 0 | | └── foo.tar | TAR | 1 | 0 | | ├── bar.zip | ZIP | 2 | 1 | | │   └── hola.txt | ASCII | 3 | 0 | | └── baz.exe | PE | 2 | 1 | | ├── sfx.zip | ZIP | 3 | 1 | | │   └── hello.txt | ASCII | 4 | 0 | | └── sfx.7z | 7Z | 3 | 1 | |    └── world.txt | ASCII | 4 | 0 | (C) is right. But it's harder to achieve. For this example we can get it by restricting 7ZSFX and ZIPSFX detection only when scanning an executable. But that may mean losing detection of archives embedded elsewhere. And we'd have to identify allowable container types for each possible embedded type, which would be very difficult. So this commit aims to solve the issue the (B)-way. Note that in all situations, we still have to scan with file typing enabled to determine if we need to reassign the current file type, such as re-identifying a Bzip2 archive as a DMG that happens to be Bzip2- compressed. Detection of DMG and a handful of other types rely on finding data partway through or near the ned of a file before reassigning the entire file as the new type. Other fixes and considerations in this commit: - The utf16 HTML parser has weak error handling, particularly with respect to creating a nested fmap for scanning the ascii decoded file. This commit cleans up the error handling and wraps the nested scan with the recursion-stack push()/pop() for correct recursion tracking. Before this commit, each container layer had a flag to indicate if the container layer is valid. We need something similar so that the cli_recursion_stack_get_*() functions ignore normalized layers. Details... Imagine an LDB signature for HTML content that specifies a ZIP container. If the signature actually alerts on the normalized HTML and you don't ignore normalized layers for the container check, it will appear as though the alert is in an HTML container rather than a ZIP container. This commit accomplishes this with a boolean you set in the scan context before scanning a new layer. Then when the new fmap is created, it will use that flag to set similar flag for the layer. The context flag is reset those that anything after this doesn't have that flag. The flag allows the new recursion_stack_get() function to ignore normalized layers when iterating the stack to return a layer at a requested index, negative or positive. Scanning normalized extracted/normalized javascript and VBA should also use the 'layer is normalized' flag. - This commit also fixes Heuristic.Broken.Executable alert for ELF files to make sure that: A) these only alert if cli_append_virus() returns CL_VIRUS (aka it respects the FP check). B) all broken-executable alerts for ELF only happen if the SCAN_HEURISTIC_BROKEN option is enabled. - This commit also cleans up the error handling in cli_magic_scan_dir(). This was needed so we could correctly apply the layer-is-normalized-flag to all VBA macros extracted to a directory when scanning the directory. - Also fix an issue where exceeding scan maximums wouldn't cause embedded file detection scans to abort. Granted we don't actually want to abort if max filesize or max recursion depth are exceeded... only if max scansize, max files, and max scantime are exceeded. Add 'abort_scan' flag to scan context, to protect against depending on correct error propagation for fatal conditions. Instead, setting this flag in the scan context should guarantee that a fatal condition deep in scan recursion isn't lost which result in more stuff being scanned instead of aborting. This shouldn't be necessary, but some status codes like CL_ETIMEOUT never used to be fatal and it's easier to do this than to verify every parser only returns CL_ETIMEOUT and other "fatal status codes" in fatal conditions. - Remove duplicate is_tar() prototype from filestypes.c and include is_tar.h instead. - Presently we create the fmap hash when creating the fmap. This wastes a bit of CPU if the hash is never needed. Now that we're creating fmap's for all embedded files discovered with file type recognition scans, this is a much more frequent occurence and really slows things down. This commit fixes the issue by only creating fmap hashes as needed. This should not only resolve the perfomance impact of creating fmap's for all embedded files, but also should improve performance in general. - Add allmatch check to the zip parser after the central-header meta match. That way we don't multiple alerts with the same match except in allmatch mode. Clean up error handling in the zip parser a tiny bit. - Fixes to ensure that the scan limits such as scansize, filesize, recursion depth, # of embedded files, and scantime are always reported if AlertExceedsMax (--alert-exceeds-max) is enabled. - Fixed an issue where non-fatal alerts for exceeding scan maximums may mask signature matches later on. I changed it so these alerts use the "possibly unwanted" alert-type and thus only alert if no other alerts were found or if all-match or heuristic-precedence are enabled. - Added the "Heuristics.Limits.Exceeded.*" events to the JSON metadata when the --gen-json feature is enabled. These will show up once under "ParseErrors" the first time a limit is exceeded. In the present implementation, only one limits-exceeded events will be added, so as to prevent a malicious or malformed sample from filling the JSON buffer with millions of events and using a tonne of RAM.
2021-09-11 14:15:21 -07:00
cl_error_t ret;
/* Generate hash */
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
ret = hash_imptbl(ctx, hashset, &impsz, genhash, peinfo);
2016-07-12 11:53:59 -04:00
if (ret == CL_SUCCESS) {
if (cli_debug_flag) {
dstr = cli_str2hex((char *)hash, hlen);
cli_dbgmsg("Imphash: %s:%u\n", dstr ? (char *)dstr : "(NULL)", impsz);
if (dstr != NULL) {
free(dstr);
}
2016-07-12 11:53:59 -04:00
}
} else {
cli_dbgmsg("Imphash: failed to generate hash for import table (%d)\n", ret);
}
} else {
cli_dbgmsg("cli_genhash_pe: unknown pe genhash class: %u\n", class);
}
free(hash);
PE parsing code improvements, db loading bug fixes Consolidate the PE parsing code into one function. I tried to preserve all existing functionality from the previous, distinct implementations to a large extent (with the exceptions mentioned below). If I noticed potential bugs/improvements, I added a TODO statement about those so that they can be fixed in a smaller commit later. Also, there are more TODOs in places where I'm not entirely sure why certain actions are performed - more research is needed for these. I'm submitting a pull request now so that regression testing can be done, and because merging what I have thus far now will likely have fewer conflicts than if I try to merge later PE parsing code improvements: - PEs without all 16 data directories are parsed more appropriately now - Added lots more debug statements Also: - Allow MAX_BC and MAX_TRACKED_PCRE to be specified via CFLAGS When doing performance testing with the latest CVD, MAX_BC and MAX_TRACKED_PCRE need to be raised to track all the events. Allow these to be specified via CFLAGS by not redefining them if they are already defined - Fix an issue preventing wildcard sizes in .MDB/.MSB rules I'm not sure what the original intent of the check I removed was, but it prevents using wildcard sizes in .MDB/.MSB rules. AFAICT these wildcard sizes should be handled appropriately by the MD5 section hash computation code, so I don't think a check on that is needed. - Fix several issues related to db loading - .imp files will now get loaded if they exist in a directory passed via clamscan's '-d' flag - .pwdb files will now get loaded if they exist in a directory passed via clamscan's '-d' flag even when compiling without yara support - Changes to .imp, .ign, and .ign2 files will now be reflected in calls to cl_statinidir and cl_statchkdir (and also .pwdb files, even when compiling without yara support) - The contents of .sfp files won't be included in some of the signature counts, and the contents of .cud files will be - Any local.gdb files will no longer be loaded twice - For .imp files, you are no longer required to specify a minimum flevel for wildcard rules, since this isn't needed
2019-01-08 00:09:08 -05:00
cli_exe_info_destroy(peinfo);
return CL_SUCCESS;
}