mirror of
https://github.com/golang/go.git
synced 2025-12-08 06:10:04 +00:00
[dev.typeparams] all: merge dev.regabi (5e4a0cd) into dev.typeparams
Merge List: + 2021-01-255e4a0cdde3[dev.regabi] all: merge master (bf0f7c9) into dev.regabi + 2021-01-25bf0f7c9d78doc/go1.16: mention os.DirFS in os section + 2021-01-25deaf29a8a8cmd/compile: fix order-of-assignment issue w/ defers + 2021-01-25ad2ca26a52doc/go1.16: mention os.DirEntry and types moved from os to io/fs + 2021-01-25a51921fa5bdoc/go1.16: mention new testing/iotest functions + 2021-01-25e6b6d107f7doc/go1.16: mention deprecation of io/ioutil + 2021-01-257eaaf28cae[dev.regabi] cmd/compile: disallow taking address of SSA'd values + 2021-01-2596a276363bdoc/go1.16: mention go/build changes + 2021-01-253d85c69a0bhtml/template: revert "avoid race when escaping updates template" + 2021-01-2554514c6b28cmd/go: fix TestScript/cgo_path, cgo_path_space when CC set + 2021-01-256f5e79f470[dev.regabi] cmd/compile/internal: specify memory layout + 2021-01-25cabffc199d[dev.regabi] cmd/compile/internal: add internal ABI specification + 2021-01-256de8443f3bdoc/asm: add a section on go_asm.h, clean up go_tls.h section + 2021-01-256a4739ccc5[dev.regabi] cmd/compile: enable rational constant arithmetic + 2021-01-25be9612a832[dev.regabi] os: disable TestDirFS until #42637 is fixed + 2021-01-258ee3d39838[dev.regabi] cmd/go: workaround -race issue on ppc64le + 2021-01-2554b251f542lib/time, time/tzdata: update tzdata to 2021a + 2021-01-255a76c3d548[dev.regabi] cmd/compile: modify abiutils for recently updated ABI + 2021-01-25ff82cc971aos: force consistent mtime before running fstest on directory on Windows + 2021-01-25044f937a73doc/go1.16: fix WalkDir and Walk links + 2021-01-23b634f5d97adoc/go1.16: add crypto/x509 memory optimization + 2021-01-239897655c61doc/go1.16: reword ambiguously parsable sentence + 2021-01-23cd99385ff4cmd/internal/obj/arm64: fix VMOVQ instruction encoding error + 2021-01-2366ee8b158fruntime: restore cgo_import_dynamic for libc.so on openbsd + 2021-01-2225c39e4fb5io/ioutil: fix example test for WriteFile to allow it to run in the playground + 2021-01-22eb21b31e48runtime: define dummy msanmove + 2021-01-223a778ff50fruntime: check for g0 stack last in signal handler + 2021-01-22a2cef9b544cmd/go: don't lookup the path for CC when invoking cgo Change-Id: Iede4f98ba5ddbee2e16075d20186f8a9c095e378
This commit is contained in:
Matthew Dempsky
commit
34704e374f
33 changed files with 8390 additions and 7221 deletions
72
doc/asm.html
72
doc/asm.html
|
|
@ -464,6 +464,57 @@ Function is the top of the call stack. Traceback should stop at this function.
|
||||||
</li>
|
</li>
|
||||||
</ul>
|
</ul>
|
||||||
|
|
||||||
|
<h3 id="data-offsets">Interacting with Go types and constants</h3>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
If a package has any .s files, then <code>go build</code> will direct
|
||||||
|
the compiler to emit a special header called <code>go_asm.h</code>,
|
||||||
|
which the .s files can then <code>#include</code>.
|
||||||
|
The file contains symbolic <code>#define</code> constants for the
|
||||||
|
offsets of Go struct fields, the sizes of Go struct types, and most
|
||||||
|
Go <code>const</code> declarations defined in the current package.
|
||||||
|
Go assembly should avoid making assumptions about the layout of Go
|
||||||
|
types and instead use these constants.
|
||||||
|
This improves the readability of assembly code, and keeps it robust to
|
||||||
|
changes in data layout either in the Go type definitions or in the
|
||||||
|
layout rules used by the Go compiler.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
Constants are of the form <code>const_<i>name</i></code>.
|
||||||
|
For example, given the Go declaration <code>const bufSize =
|
||||||
|
1024</code>, assembly code can refer to the value of this constant
|
||||||
|
as <code>const_bufSize</code>.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
Field offsets are of the form <code><i>type</i>_<i>field</i></code>.
|
||||||
|
Struct sizes are of the form <code><i>type</i>__size</code>.
|
||||||
|
For example, consider the following Go definition:
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<pre>
|
||||||
|
type reader struct {
|
||||||
|
buf [bufSize]byte
|
||||||
|
r int
|
||||||
|
}
|
||||||
|
</pre>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
Assembly can refer to the size of this struct
|
||||||
|
as <code>reader__size</code> and the offsets of the two fields
|
||||||
|
as <code>reader_buf</code> and <code>reader_r</code>.
|
||||||
|
Hence, if register <code>R1</code> contains a pointer to
|
||||||
|
a <code>reader</code>, assembly can reference the <code>r</code> field
|
||||||
|
as <code>reader_r(R1)</code>.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
If any of these <code>#define</code> names are ambiguous (for example,
|
||||||
|
a struct with a <code>_size</code> field), <code>#include
|
||||||
|
"go_asm.h"</code> will fail with a "redefinition of macro" error.
|
||||||
|
</p>
|
||||||
|
|
||||||
<h3 id="runtime">Runtime Coordination</h3>
|
<h3 id="runtime">Runtime Coordination</h3>
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
|
|
@ -615,21 +666,15 @@ Here follow some descriptions of key Go-specific details for the supported archi
|
||||||
<p>
|
<p>
|
||||||
The runtime pointer to the <code>g</code> structure is maintained
|
The runtime pointer to the <code>g</code> structure is maintained
|
||||||
through the value of an otherwise unused (as far as Go is concerned) register in the MMU.
|
through the value of an otherwise unused (as far as Go is concerned) register in the MMU.
|
||||||
An OS-dependent macro <code>get_tls</code> is defined for the assembler if the source is
|
In the runtime package, assembly code can include <code>go_tls.h</code>, which defines
|
||||||
in the <code>runtime</code> package and includes a special header, <code>go_tls.h</code>:
|
an OS- and architecture-dependent macro <code>get_tls</code> for accessing this register.
|
||||||
|
The <code>get_tls</code> macro takes one argument, which is the register to load the
|
||||||
|
<code>g</code> pointer into.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
<pre>
|
|
||||||
#include "go_tls.h"
|
|
||||||
</pre>
|
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
Within the runtime, the <code>get_tls</code> macro loads its argument register
|
For example, the sequence to load <code>g</code> and <code>m</code>
|
||||||
with a pointer to the <code>g</code> pointer, and the <code>g</code> struct
|
using <code>CX</code> looks like this:
|
||||||
contains the <code>m</code> pointer.
|
|
||||||
There's another special header containing the offsets for each
|
|
||||||
element of <code>g</code>, called <code>go_asm.h</code>.
|
|
||||||
The sequence to load <code>g</code> and <code>m</code> using <code>CX</code> looks like this:
|
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
<pre>
|
<pre>
|
||||||
|
|
@ -642,8 +687,7 @@ MOVL g_m(AX), BX // Move g.m into BX.
|
||||||
</pre>
|
</pre>
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
Note: The code above works only in the <code>runtime</code> package, while <code>go_tls.h</code> also
|
The <code>get_tls</code> macro is also defined on <a href="#amd64">amd64</a>.
|
||||||
applies to <a href="#arm">arm</a>, <a href="#amd64">amd64</a> and amd64p32, and <code>go_asm.h</code> applies to all architectures.
|
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
|
|
|
||||||
163
doc/go1.16.html
163
doc/go1.16.html
|
|
@ -461,10 +461,10 @@ func TestFoo(t *testing.T) {
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
The new <a href="/pkg/io/fs/"><code>io/fs</code></a> package
|
The new <a href="/pkg/io/fs/"><code>io/fs</code></a> package
|
||||||
defines an abstraction for read-only trees of files,
|
defines the <a href="/pkg/io/fs/#FS"><code>fs.FS</code></a> interface,
|
||||||
the <a href="/pkg/io/fs/#FS"><code>fs.FS</code></a> interface,
|
an abstraction for read-only trees of files.
|
||||||
and the standard library packages have
|
The standard library packages have been adapted to make use
|
||||||
been adapted to make use of the interface as appropriate.
|
of the interface as appropriate.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
<p>
|
<p>
|
||||||
|
|
@ -499,6 +499,44 @@ func TestFoo(t *testing.T) {
|
||||||
implementations.
|
implementations.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
|
<h3 id="ioutil">Deprecation of io/ioutil</h3>
|
||||||
|
|
||||||
|
<p>
|
||||||
|
The <a href="/pkg/io/ioutil/"><code>io/ioutil</code></a> package has
|
||||||
|
turned out to be a poorly defined and hard to understand collection
|
||||||
|
of things. All functionality provided by the package has been moved
|
||||||
|
to other packages. The <code>io/ioutil</code> package remains and
|
||||||
|
will continue to work as before, but we encourage new code to use
|
||||||
|
the new definitions in the <a href="/pkg/io/"><code>io</code></a> and
|
||||||
|
<a href="/pkg/os/"><code>os</code></a> packages.
|
||||||
|
|
||||||
|
Here is a list of the new locations of the names exported
|
||||||
|
by <code>io/ioutil</code>:
|
||||||
|
<ul>
|
||||||
|
<li><a href="/pkg/io/ioutil/#Discard"><code>Discard</code></a>
|
||||||
|
=> <a href="/pkg/io/#Discard"><code>io.Discard</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#NopCloser"><code>NopCloser</code></a>
|
||||||
|
=> <a href="/pkg/io/#NopCloser"><code>io.NopCloser</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#ReadAll"><code>ReadAll</code></a>
|
||||||
|
=> <a href="/pkg/io/#ReadAll"><code>io.ReadAll</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#ReadDir"><code>ReadDir</code></a>
|
||||||
|
=> <a href="/pkg/os/#ReadDir"><code>os.ReadDir</code></a>
|
||||||
|
(note: returns a slice of
|
||||||
|
<a href="/pkg/os/#DirEntry"><code>os.DirEntry</code></a>
|
||||||
|
rather than a slice of
|
||||||
|
<a href="/pkg/fs/#FileInfo"><code>fs.FileInfo</code></a>)
|
||||||
|
</li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#ReadFile"><code>ReadFile</code></a>
|
||||||
|
=> <a href="/pkg/os/#ReadFile"><code>os.ReadFile</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#TempDir"><code>TempDir</code></a>
|
||||||
|
=> <a href="/pkg/os/#MkdirTemp"><code>os.MkdirTemp</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#TempFile"><code>TempFile</code></a>
|
||||||
|
=> <a href="/pkg/os/#CreateTemp"><code>os.CreateTemp</code></a></li>
|
||||||
|
<li><a href="/pkg/io/ioutil/#WriteFile"><code>WriteFile</code></a>
|
||||||
|
=> <a href="/pkg/os/#WriteFile"><code>os.WriteFile</code></a></li>
|
||||||
|
</ul>
|
||||||
|
</p>
|
||||||
|
|
||||||
<!-- okay-after-beta1
|
<!-- okay-after-beta1
|
||||||
TODO: decide if any additional changes are worth factoring out from
|
TODO: decide if any additional changes are worth factoring out from
|
||||||
"Minor changes to the library" and highlighting in "Core library"
|
"Minor changes to the library" and highlighting in "Core library"
|
||||||
|
|
@ -623,6 +661,14 @@ func TestFoo(t *testing.T) {
|
||||||
method allows accessing the <a href="/pkg/crypto/x509/#SystemRootsError.Err"><code>Err</code></a>
|
method allows accessing the <a href="/pkg/crypto/x509/#SystemRootsError.Err"><code>Err</code></a>
|
||||||
field through the <a href="/pkg/errors"><code>errors</code></a> package functions.
|
field through the <a href="/pkg/errors"><code>errors</code></a> package functions.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 230025 -->
|
||||||
|
On Unix systems, the <code>crypto/x509</code> package is now more
|
||||||
|
efficient in how it stores its copy of the system cert pool.
|
||||||
|
Programs that use only a small number of roots will use around a
|
||||||
|
half megabyte less memory.
|
||||||
|
</p>
|
||||||
|
|
||||||
</dd>
|
</dd>
|
||||||
</dl><!-- crypto/x509 -->
|
</dl><!-- crypto/x509 -->
|
||||||
|
|
||||||
|
|
@ -685,6 +731,37 @@ func TestFoo(t *testing.T) {
|
||||||
</dd>
|
</dd>
|
||||||
</dl><!-- flag -->
|
</dl><!-- flag -->
|
||||||
|
|
||||||
|
<dl id="go/build"><dt><a href="/pkg/go/build/">go/build</a></dt>
|
||||||
|
<dd>
|
||||||
|
<p><!-- CL 243941, CL 283636 -->
|
||||||
|
The <a href="/pkg/go/build/#Package"><code>Package</code></a>
|
||||||
|
struct has new fields that report information
|
||||||
|
about <code>//go:embed</code> directives in the package:
|
||||||
|
<a href="/pkg/go/build/#Package.EmbedPatterns"><code>EmbedPatterns</code></a>,
|
||||||
|
<a href="/pkg/go/build/#Package.EmbedPatternPos"><code>EmbedPatternPos</code></a>,
|
||||||
|
<a href="/pkg/go/build/#Package.TestEmbedPatterns"><code>TestEmbedPatterns</code></a>,
|
||||||
|
<a href="/pkg/go/build/#Package.TestEmbedPatternPos"><code>TestEmbedPatternPos</code></a>,
|
||||||
|
<a href="/pkg/go/build/#Package.XTestEmbedPatterns"><code>XTestEmbedPatterns</code></a>,
|
||||||
|
<a href="/pkg/go/build/#Package.XTestEmbedPatternPos"><code>XTestEmbedPatternPos</code></a>.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 240551 -->
|
||||||
|
The <a href="/pkg/go/build/#Package"><code>Package</code></a> field
|
||||||
|
<a href="/pkg/go/build/#Package.IgnoredGoFiles"><code>IgnoredGoFiles</code></a>
|
||||||
|
will no longer include files that start with "_" or ".",
|
||||||
|
as those files are always ignored.
|
||||||
|
<code>IgnoredGoFiles</code> is for files ignored because of
|
||||||
|
build constraints.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 240551 -->
|
||||||
|
The new <a href="/pkg/go/build/#Package"><code>Package</code></a>
|
||||||
|
field <a href="/pkg/go/build/#Package.IgnoredOtherFiles"><code>IgnoredOtherFiles</code></a>
|
||||||
|
has a list of non-Go files ignored because of build constraints.
|
||||||
|
</p>
|
||||||
|
</dd>
|
||||||
|
</dl><!-- go/build -->
|
||||||
|
|
||||||
<dl id="html/template"><dt><a href="/pkg/html/template/">html/template</a></dt>
|
<dl id="html/template"><dt><a href="/pkg/html/template/">html/template</a></dt>
|
||||||
<dd>
|
<dd>
|
||||||
<p><!-- CL 243938 -->
|
<p><!-- CL 243938 -->
|
||||||
|
|
@ -703,6 +780,15 @@ func TestFoo(t *testing.T) {
|
||||||
The package now defines a
|
The package now defines a
|
||||||
<a href="/pkg/io/#ReadSeekCloser"><code>ReadSeekCloser</code></a> interface.
|
<a href="/pkg/io/#ReadSeekCloser"><code>ReadSeekCloser</code></a> interface.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 263141 -->
|
||||||
|
The package now defines
|
||||||
|
<a href="/pkg/io/#Discard"><code>Discard</code></a>,
|
||||||
|
<a href="/pkg/io/#NopCloser"><code>NopCloser</code></a>, and
|
||||||
|
<a href="/pkg/io/#ReadAll"><code>ReadAll</code></a>,
|
||||||
|
to be used instead of the same names in the
|
||||||
|
<a href="/pkg/io/ioutil/"><code>io/ioutil</code></a> package.
|
||||||
|
</p>
|
||||||
</dd>
|
</dd>
|
||||||
</dl><!-- io -->
|
</dl><!-- io -->
|
||||||
|
|
||||||
|
|
@ -857,6 +943,52 @@ func TestFoo(t *testing.T) {
|
||||||
instead of the unexported <code>errFinished</code> when the process has
|
instead of the unexported <code>errFinished</code> when the process has
|
||||||
already finished.
|
already finished.
|
||||||
</p>
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 261540 -->
|
||||||
|
The package defines a new type
|
||||||
|
<a href="/pkg/os/#DirEntry"><code>DirEntry</code></a>
|
||||||
|
as an alias for <a href="/pkg/io/fs/#DirEntry"><code>fs.DirEntry</code></a>.
|
||||||
|
The new <a href="/pkg/os/#ReadDir"><code>ReadDir</code></a>
|
||||||
|
function and the new
|
||||||
|
<a href="/pkg/os/#File.ReadDir"><code>File.ReadDir</code></a>
|
||||||
|
method can be used to read the contents of a directory into a
|
||||||
|
slice of <a href="/pkg/os/#DirEntry"><code>DirEntry</code></a>.
|
||||||
|
The <a href="/pkg/os/#File.Readdir"><code>File.Readdir</code></a>
|
||||||
|
method (note the lower case <code>d</code> in <code>dir</code>)
|
||||||
|
still exists, returning a slice of
|
||||||
|
<a href="/pkg/os/#FileInfo"><code>FileInfo</code></a>, but for
|
||||||
|
most programs it will be more efficient to switch to
|
||||||
|
<a href="/pkg/os/#File.ReadDir"><code>File.ReadDir</code></a>.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 263141 -->
|
||||||
|
The package now defines
|
||||||
|
<a href="/pkg/os/#CreateTemp"><code>CreateTemp</code></a>,
|
||||||
|
<a href="/pkg/os/#MkdirTemp"><code>MkdirTemp</code></a>,
|
||||||
|
<a href="/pkg/os/#ReadFile"><code>ReadFile</code></a>, and
|
||||||
|
<a href="/pkg/os/#WriteFile"><code>WriteFile</code></a>,
|
||||||
|
to be used instead of functions defined in the
|
||||||
|
<a href="/pkg/io/ioutil/"><code>io/ioutil</code></a> package.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 243906 -->
|
||||||
|
The types <a href="/pkg/os/#FileInfo"><code>FileInfo</code></a>,
|
||||||
|
<a href="/pkg/os/#FileMode"><code>FileMode</code></a>, and
|
||||||
|
<a href="/pkg/os/#PathError"><code>PathError</code></a>
|
||||||
|
are now aliases for types of the same name in the
|
||||||
|
<a href="/pkg/io/fs/"><code>io/fs</code></a> package.
|
||||||
|
Function signatures in the <a href="/pkg/os/"><code>os</code></a>
|
||||||
|
package have been updated to refer to the names in the
|
||||||
|
<a href="/pkg/io/fs/"><code>io/fs</code></a> package.
|
||||||
|
This should not affect any existing code.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 243911 -->
|
||||||
|
The new <a href="/pkg/os/#DirFS"><code>DirFS</code></a> function
|
||||||
|
provides an implementation of
|
||||||
|
<a href="/pkg/io/fs/#FS"><code>fs.FS</code></a> backed by a tree
|
||||||
|
of operating system files.
|
||||||
|
</p>
|
||||||
</dd>
|
</dd>
|
||||||
</dl><!-- os -->
|
</dl><!-- os -->
|
||||||
|
|
||||||
|
|
@ -887,9 +1019,9 @@ func TestFoo(t *testing.T) {
|
||||||
<dd>
|
<dd>
|
||||||
<p><!-- CL 267887 -->
|
<p><!-- CL 267887 -->
|
||||||
The new function
|
The new function
|
||||||
<a href="/pkg/path/filepath/WalkDir"><code>WalkDir</code></a>
|
<a href="/pkg/path/filepath/#WalkDir"><code>WalkDir</code></a>
|
||||||
is similar to
|
is similar to
|
||||||
<a href="/pkg/path/filepath/Walk"><code>Walk</code></a>,
|
<a href="/pkg/path/filepath/#Walk"><code>Walk</code></a>,
|
||||||
but is typically more efficient.
|
but is typically more efficient.
|
||||||
The function passed to <code>WalkDir</code> receives a
|
The function passed to <code>WalkDir</code> receives a
|
||||||
<a href="/pkg/io/fs/#DirEntry"><code>fs.DirEntry</code></a>
|
<a href="/pkg/io/fs/#DirEntry"><code>fs.DirEntry</code></a>
|
||||||
|
|
@ -975,6 +1107,25 @@ func TestFoo(t *testing.T) {
|
||||||
</dd>
|
</dd>
|
||||||
</dl><!-- syscall -->
|
</dl><!-- syscall -->
|
||||||
|
|
||||||
|
<dl id="testing/iotest"><dt><a href="/pkg/testing/iotest/">testing/iotest</a></dt>
|
||||||
|
<dd>
|
||||||
|
<p><!-- CL 199501 -->
|
||||||
|
The new
|
||||||
|
<a href="/pkg/testing/iotest/#ErrReader"><code>ErrReader</code></a>
|
||||||
|
function returns an
|
||||||
|
<a href="/pkg/io/#Reader"><code>io.Reader</code></a> that always
|
||||||
|
returns an error.
|
||||||
|
</p>
|
||||||
|
|
||||||
|
<p><!-- CL 243909 -->
|
||||||
|
The new
|
||||||
|
<a href="/pkg/testing/iotest/#TestReader"><code>TestReader</code></a>
|
||||||
|
function tests that an <a href="/pkg/io/#Reader"><code>io.Reader</code></a>
|
||||||
|
behaves correctly.
|
||||||
|
</p>
|
||||||
|
</dd>
|
||||||
|
</dl><!-- testing/iotest -->
|
||||||
|
|
||||||
<dl id="text/template"><dt><a href="/pkg/text/template/">text/template</a></dt>
|
<dl id="text/template"><dt><a href="/pkg/text/template/">text/template</a></dt>
|
||||||
<dd>
|
<dd>
|
||||||
<p><!-- CL 254257, golang.org/issue/29770 -->
|
<p><!-- CL 254257, golang.org/issue/29770 -->
|
||||||
|
|
|
||||||
|
|
@ -8,8 +8,8 @@
|
||||||
# Consult https://www.iana.org/time-zones for the latest versions.
|
# Consult https://www.iana.org/time-zones for the latest versions.
|
||||||
|
|
||||||
# Versions to use.
|
# Versions to use.
|
||||||
CODE=2020f
|
CODE=2021a
|
||||||
DATA=2020f
|
DATA=2021a
|
||||||
|
|
||||||
set -e
|
set -e
|
||||||
rm -rf work
|
rm -rf work
|
||||||
|
|
|
||||||
Binary file not shown.
628
src/cmd/compile/internal-abi.md
Normal file
628
src/cmd/compile/internal-abi.md
Normal file
|
|
@ -0,0 +1,628 @@
|
||||||
|
# Go internal ABI specification
|
||||||
|
|
||||||
|
This document describes Go’s internal application binary interface
|
||||||
|
(ABI), known as ABIInternal.
|
||||||
|
Go's ABI defines the layout of data in memory and the conventions for
|
||||||
|
calling between Go functions.
|
||||||
|
This ABI is *unstable* and will change between Go versions.
|
||||||
|
If you’re writing assembly code, please instead refer to Go’s
|
||||||
|
[assembly documentation](/doc/asm.html), which describes Go’s stable
|
||||||
|
ABI, known as ABI0.
|
||||||
|
|
||||||
|
All functions defined in Go source follow ABIInternal.
|
||||||
|
However, ABIInternal and ABI0 functions are able to call each other
|
||||||
|
through transparent *ABI wrappers*, described in the [internal calling
|
||||||
|
convention proposal](https://golang.org/design/27539-internal-abi).
|
||||||
|
|
||||||
|
Go uses a common ABI design across all architectures.
|
||||||
|
We first describe the common ABI, and then cover per-architecture
|
||||||
|
specifics.
|
||||||
|
|
||||||
|
*Rationale*: For the reasoning behind using a common ABI across
|
||||||
|
architectures instead of the platform ABI, see the [register-based Go
|
||||||
|
calling convention proposal](https://golang.org/design/40724-register-calling).
|
||||||
|
|
||||||
|
## Memory layout
|
||||||
|
|
||||||
|
Go's built-in types have the following sizes and alignments.
|
||||||
|
Many, though not all, of these sizes are guaranteed by the [language
|
||||||
|
specification](/doc/go_spec.html#Size_and_alignment_guarantees).
|
||||||
|
Those that aren't guaranteed may change in future versions of Go (for
|
||||||
|
example, we've considered changing the alignment of int64 on 32-bit).
|
||||||
|
|
||||||
|
| Type | 64-bit | | 32-bit | |
|
||||||
|
| --- | --- | --- | --- | --- |
|
||||||
|
| | Size | Align | Size | Align |
|
||||||
|
| bool, uint8, int8 | 1 | 1 | 1 | 1 |
|
||||||
|
| uint16, int16 | 2 | 2 | 2 | 2 |
|
||||||
|
| uint32, int32 | 4 | 4 | 4 | 4 |
|
||||||
|
| uint64, int64 | 8 | 8 | 8 | 4 |
|
||||||
|
| int, uint | 8 | 8 | 4 | 4 |
|
||||||
|
| float32 | 4 | 4 | 4 | 4 |
|
||||||
|
| float64 | 8 | 8 | 8 | 4 |
|
||||||
|
| complex64 | 8 | 4 | 8 | 4 |
|
||||||
|
| complex128 | 16 | 8 | 16 | 4 |
|
||||||
|
| uintptr, *T, unsafe.Pointer | 8 | 8 | 4 | 4 |
|
||||||
|
|
||||||
|
The types `byte` and `rune` are aliases for `uint8` and `int32`,
|
||||||
|
respectively, and hence have the same size and alignment as these
|
||||||
|
types.
|
||||||
|
|
||||||
|
The layout of `map`, `chan`, and `func` types is equivalent to *T.
|
||||||
|
|
||||||
|
To describe the layout of the remaining composite types, we first
|
||||||
|
define the layout of a *sequence* S of N fields with types
|
||||||
|
t<sub>1</sub>, t<sub>2</sub>, ..., t<sub>N</sub>.
|
||||||
|
We define the byte offset at which each field begins relative to a
|
||||||
|
base address of 0, as well as the size and alignment of the sequence
|
||||||
|
as follows:
|
||||||
|
|
||||||
|
```
|
||||||
|
offset(S, i) = 0 if i = 1
|
||||||
|
= align(offset(S, i-1) + sizeof(t_(i-1)), alignof(t_i))
|
||||||
|
alignof(S) = 1 if N = 0
|
||||||
|
= max(alignof(t_i) | 1 <= i <= N)
|
||||||
|
sizeof(S) = 0 if N = 0
|
||||||
|
= align(offset(S, N) + sizeof(t_N), alignof(S))
|
||||||
|
```
|
||||||
|
|
||||||
|
Where sizeof(T) and alignof(T) are the size and alignment of type T,
|
||||||
|
respectively, and align(x, y) rounds x up to a multiple of y.
|
||||||
|
|
||||||
|
The `interface{}` type is a sequence of 1. a pointer to the runtime type
|
||||||
|
description for the interface's dynamic type and 2. an `unsafe.Pointer`
|
||||||
|
data field.
|
||||||
|
Any other interface type (besides the empty interface) is a sequence
|
||||||
|
of 1. a pointer to the runtime "itab" that gives the method pointers and
|
||||||
|
the type of the data field and 2. an `unsafe.Pointer` data field.
|
||||||
|
An interface can be "direct" or "indirect" depending on the dynamic
|
||||||
|
type: a direct interface stores the value directly in the data field,
|
||||||
|
and an indirect interface stores a pointer to the value in the data
|
||||||
|
field.
|
||||||
|
An interface can only be direct if the value consists of a single
|
||||||
|
pointer word.
|
||||||
|
|
||||||
|
An array type `[N]T` is a sequence of N fields of type T.
|
||||||
|
|
||||||
|
The slice type `[]T` is a sequence of a `*[cap]T` pointer to the slice
|
||||||
|
backing store, an `int` giving the `len` of the slice, and an `int`
|
||||||
|
giving the `cap` of the slice.
|
||||||
|
|
||||||
|
The `string` type is a sequence of a `*[len]byte` pointer to the
|
||||||
|
string backing store, and an `int` giving the `len` of the string.
|
||||||
|
|
||||||
|
A struct type `struct { f1 t1; ...; fM tM }` is laid out as the
|
||||||
|
sequence t1, ..., tM, tP, where tP is either:
|
||||||
|
|
||||||
|
- Type `byte` if sizeof(tM) = 0 and any of sizeof(t*i*) ≠ 0.
|
||||||
|
- Empty (size 0 and align 1) otherwise.
|
||||||
|
|
||||||
|
The padding byte prevents creating a past-the-end pointer by taking
|
||||||
|
the address of the final, empty fN field.
|
||||||
|
|
||||||
|
Note that user-written assembly code should generally not depend on Go
|
||||||
|
type layout and should instead use the constants defined in
|
||||||
|
[`go_asm.h`](/doc/asm.html#data-offsets).
|
||||||
|
|
||||||
|
## Function call argument and result passing
|
||||||
|
|
||||||
|
Function calls pass arguments and results using a combination of the
|
||||||
|
stack and machine registers.
|
||||||
|
Each argument or result is passed either entirely in registers or
|
||||||
|
entirely on the stack.
|
||||||
|
Because access to registers is generally faster than access to the
|
||||||
|
stack, arguments and results are preferentially passed in registers.
|
||||||
|
However, any argument or result that contains a non-trivial array or
|
||||||
|
does not fit entirely in the remaining available registers is passed
|
||||||
|
on the stack.
|
||||||
|
|
||||||
|
Each architecture defines a sequence of integer registers and a
|
||||||
|
sequence of floating-point registers.
|
||||||
|
At a high level, arguments and results are recursively broken down
|
||||||
|
into values of base types and these base values are assigned to
|
||||||
|
registers from these sequences.
|
||||||
|
|
||||||
|
Arguments and results can share the same registers, but do not share
|
||||||
|
the same stack space.
|
||||||
|
Beyond the arguments and results passed on the stack, the caller also
|
||||||
|
reserves spill space on the stack for all register-based arguments
|
||||||
|
(but does not populate this space).
|
||||||
|
|
||||||
|
The receiver, arguments, and results of function or method F are
|
||||||
|
assigned to registers or the stack using the following algorithm:
|
||||||
|
|
||||||
|
1. Let NI and NFP be the length of integer and floating-point register
|
||||||
|
sequences defined by the architecture.
|
||||||
|
Let I and FP be 0; these are the indexes of the next integer and
|
||||||
|
floating-pointer register.
|
||||||
|
Let S, the type sequence defining the stack frame, be empty.
|
||||||
|
1. If F is a method, assign F’s receiver.
|
||||||
|
1. For each argument A of F, assign A.
|
||||||
|
1. Add a pointer-alignment field to S. This has size 0 and the same
|
||||||
|
alignment as `uintptr`.
|
||||||
|
1. Reset I and FP to 0.
|
||||||
|
1. For each result R of F, assign R.
|
||||||
|
1. Add a pointer-alignment field to S.
|
||||||
|
1. For each register-assigned receiver and argument of F, let T be its
|
||||||
|
type and add T to the stack sequence S.
|
||||||
|
This is the argument's (or receiver's) spill space and will be
|
||||||
|
uninitialized at the call.
|
||||||
|
1. Add a pointer-alignment field to S.
|
||||||
|
|
||||||
|
Assigning a receiver, argument, or result V of underlying type T works
|
||||||
|
as follows:
|
||||||
|
|
||||||
|
1. Remember I and FP.
|
||||||
|
1. Try to register-assign V.
|
||||||
|
1. If step 2 failed, reset I and FP to the values from step 1, add T
|
||||||
|
to the stack sequence S, and assign V to this field in S.
|
||||||
|
|
||||||
|
Register-assignment of a value V of underlying type T works as follows:
|
||||||
|
|
||||||
|
1. If T is a boolean or integral type that fits in an integer
|
||||||
|
register, assign V to register I and increment I.
|
||||||
|
1. If T is an integral type that fits in two integer registers, assign
|
||||||
|
the least significant and most significant halves of V to registers
|
||||||
|
I and I+1, respectively, and increment I by 2
|
||||||
|
1. If T is a floating-point type and can be represented without loss
|
||||||
|
of precision in a floating-point register, assign V to register FP
|
||||||
|
and increment FP.
|
||||||
|
1. If T is a complex type, recursively register-assign its real and
|
||||||
|
imaginary parts.
|
||||||
|
1. If T is a pointer type, map type, channel type, or function type,
|
||||||
|
assign V to register I and increment I.
|
||||||
|
1. If T is a string type, interface type, or slice type, recursively
|
||||||
|
register-assign V’s components (2 for strings and interfaces, 3 for
|
||||||
|
slices).
|
||||||
|
1. If T is a struct type, recursively register-assign each field of V.
|
||||||
|
1. If T is an array type of length 0, do nothing.
|
||||||
|
1. If T is an array type of length 1, recursively register-assign its
|
||||||
|
one element.
|
||||||
|
1. If T is an array type of length > 1, fail.
|
||||||
|
1. If I > NI or FP > NFP, fail.
|
||||||
|
1. If any recursive assignment above fails, fail.
|
||||||
|
|
||||||
|
The above algorithm produces an assignment of each receiver, argument,
|
||||||
|
and result to registers or to a field in the stack sequence.
|
||||||
|
The final stack sequence looks like: stack-assigned receiver,
|
||||||
|
stack-assigned arguments, pointer-alignment, stack-assigned results,
|
||||||
|
pointer-alignment, spill space for each register-assigned argument,
|
||||||
|
pointer-alignment.
|
||||||
|
The following diagram shows what this stack frame looks like on the
|
||||||
|
stack, using the typical convention where address 0 is at the bottom:
|
||||||
|
|
||||||
|
+------------------------------+
|
||||||
|
| . . . |
|
||||||
|
| 2nd reg argument spill space |
|
||||||
|
| 1st reg argument spill space |
|
||||||
|
| <pointer-sized alignment> |
|
||||||
|
| . . . |
|
||||||
|
| 2nd stack-assigned result |
|
||||||
|
| 1st stack-assigned result |
|
||||||
|
| <pointer-sized alignment> |
|
||||||
|
| . . . |
|
||||||
|
| 2nd stack-assigned argument |
|
||||||
|
| 1st stack-assigned argument |
|
||||||
|
| stack-assigned receiver |
|
||||||
|
+------------------------------+ ↓ lower addresses
|
||||||
|
|
||||||
|
To perform a call, the caller reserves space starting at the lowest
|
||||||
|
address in its stack frame for the call stack frame, stores arguments
|
||||||
|
in the registers and argument stack fields determined by the above
|
||||||
|
algorithm, and performs the call.
|
||||||
|
At the time of a call, spill space, result stack fields, and result
|
||||||
|
registers are left uninitialized.
|
||||||
|
Upon return, the callee must have stored results to all result
|
||||||
|
registers and result stack fields determined by the above algorithm.
|
||||||
|
|
||||||
|
There are no callee-save registers, so a call may overwrite any
|
||||||
|
register that doesn’t have a fixed meaning, including argument
|
||||||
|
registers.
|
||||||
|
|
||||||
|
### Example
|
||||||
|
|
||||||
|
Consider the function `func f(a1 uint8, a2 [2]uintptr, a3 uint8) (r1
|
||||||
|
struct { x uintptr; y [2]uintptr }, r2 string)` on a 64-bit
|
||||||
|
architecture with hypothetical integer registers R0–R9.
|
||||||
|
|
||||||
|
On entry, `a1` is assigned to `R0`, `a3` is assigned to `R1` and the
|
||||||
|
stack frame is laid out in the following sequence:
|
||||||
|
|
||||||
|
a2 [2]uintptr
|
||||||
|
r1.x uintptr
|
||||||
|
r1.y [2]uintptr
|
||||||
|
a1Spill uint8
|
||||||
|
a2Spill uint8
|
||||||
|
_ [6]uint8 // alignment padding
|
||||||
|
|
||||||
|
In the stack frame, only the `a2` field is initialized on entry; the
|
||||||
|
rest of the frame is left uninitialized.
|
||||||
|
|
||||||
|
On exit, `r2.base` is assigned to `R0`, `r2.len` is assigned to `R1`,
|
||||||
|
and `r1.x` and `r1.y` are initialized in the stack frame.
|
||||||
|
|
||||||
|
There are several things to note in this example.
|
||||||
|
First, `a2` and `r1` are stack-assigned because they contain arrays.
|
||||||
|
The other arguments and results are register-assigned.
|
||||||
|
Result `r2` is decomposed into its components, which are individually
|
||||||
|
register-assigned.
|
||||||
|
On the stack, the stack-assigned arguments appear at lower addresses
|
||||||
|
than the stack-assigned results, which appear at lower addresses than
|
||||||
|
the argument spill area.
|
||||||
|
Only arguments, not results, are assigned a spill area on the stack.
|
||||||
|
|
||||||
|
### Rationale
|
||||||
|
|
||||||
|
Each base value is assigned to its own register to optimize
|
||||||
|
construction and access.
|
||||||
|
An alternative would be to pack multiple sub-word values into
|
||||||
|
registers, or to simply map an argument's in-memory layout to
|
||||||
|
registers (this is common in C ABIs), but this typically adds cost to
|
||||||
|
pack and unpack these values.
|
||||||
|
Modern architectures have more than enough registers to pass all
|
||||||
|
arguments and results this way for nearly all functions (see the
|
||||||
|
appendix), so there’s little downside to spreading base values across
|
||||||
|
registers.
|
||||||
|
|
||||||
|
Arguments that can’t be fully assigned to registers are passed
|
||||||
|
entirely on the stack in case the callee takes the address of that
|
||||||
|
argument.
|
||||||
|
If an argument could be split across the stack and registers and the
|
||||||
|
callee took its address, it would need to be reconstructed in memory,
|
||||||
|
a process that would be proportional to the size of the argument.
|
||||||
|
|
||||||
|
Non-trivial arrays are always passed on the stack because indexing
|
||||||
|
into an array typically requires a computed offset, which generally
|
||||||
|
isn’t possible with registers.
|
||||||
|
Arrays in general are rare in function signatures (only 0.7% of
|
||||||
|
functions in the Go 1.15 standard library and 0.2% in kubelet).
|
||||||
|
We considered allowing array fields to be passed on the stack while
|
||||||
|
the rest of an argument’s fields are passed in registers, but this
|
||||||
|
creates the same problems as other large structs if the callee takes
|
||||||
|
the address of an argument, and would benefit <0.1% of functions in
|
||||||
|
kubelet (and even these very little).
|
||||||
|
|
||||||
|
We make exceptions for 0 and 1-element arrays because these don’t
|
||||||
|
require computed offsets, and 1-element arrays are already decomposed
|
||||||
|
in the compiler’s SSA representation.
|
||||||
|
|
||||||
|
The ABI assignment algorithm above is equivalent to Go’s stack-based
|
||||||
|
ABI0 calling convention if there are zero architecture registers.
|
||||||
|
This is intended to ease the transition to the register-based internal
|
||||||
|
ABI and make it easy for the compiler to generate either calling
|
||||||
|
convention.
|
||||||
|
An architecture may still define register meanings that aren’t
|
||||||
|
compatible with ABI0, but these differences should be easy to account
|
||||||
|
for in the compiler.
|
||||||
|
|
||||||
|
The algorithm reserves spill space for arguments in the caller’s frame
|
||||||
|
so that the compiler can generate a stack growth path that spills into
|
||||||
|
this reserved space.
|
||||||
|
If the callee has to grow the stack, it may not be able to reserve
|
||||||
|
enough additional stack space in its own frame to spill these, which
|
||||||
|
is why it’s important that the caller do so.
|
||||||
|
These slots also act as the home location if these arguments need to
|
||||||
|
be spilled for any other reason, which simplifies traceback printing.
|
||||||
|
|
||||||
|
There are several options for how to lay out the argument spill space.
|
||||||
|
We chose to lay out each argument according to its type's usual memory
|
||||||
|
layout but to separate the spill space from the regular argument
|
||||||
|
space.
|
||||||
|
Using the usual memory layout simplifies the compiler because it
|
||||||
|
already understands this layout.
|
||||||
|
Also, if a function takes the address of a register-assigned argument,
|
||||||
|
the compiler must spill that argument to memory in its usual memory
|
||||||
|
layout and it's more convenient to use the argument spill space for
|
||||||
|
this purpose.
|
||||||
|
|
||||||
|
Alternatively, the spill space could be structured around argument
|
||||||
|
registers.
|
||||||
|
In this approach, the stack growth spill path would spill each
|
||||||
|
argument register to a register-sized stack word.
|
||||||
|
However, if the function takes the address of a register-assigned
|
||||||
|
argument, the compiler would have to reconstruct it in memory layout
|
||||||
|
elsewhere on the stack.
|
||||||
|
|
||||||
|
The spill space could also be interleaved with the stack-assigned
|
||||||
|
arguments so the arguments appear in order whether they are register-
|
||||||
|
or stack-assigned.
|
||||||
|
This would be close to ABI0, except that register-assigned arguments
|
||||||
|
would be uninitialized on the stack and there's no need to reserve
|
||||||
|
stack space for register-assigned results.
|
||||||
|
We expect separating the spill space to perform better because of
|
||||||
|
memory locality.
|
||||||
|
Separating the space is also potentially simpler for `reflect` calls
|
||||||
|
because this allows `reflect` to summarize the spill space as a single
|
||||||
|
number.
|
||||||
|
Finally, the long-term intent is to remove reserved spill slots
|
||||||
|
entirely – allowing most functions to be called without any stack
|
||||||
|
setup and easing the introduction of callee-save registers – and
|
||||||
|
separating the spill space makes that transition easier.
|
||||||
|
|
||||||
|
## Closures
|
||||||
|
|
||||||
|
A func value (e.g., `var x func()`) is a pointer to a closure object.
|
||||||
|
A closure object begins with a pointer-sized program counter
|
||||||
|
representing the entry point of the function, followed by zero or more
|
||||||
|
bytes containing the closed-over environment.
|
||||||
|
|
||||||
|
Closure calls follow the same conventions as static function and
|
||||||
|
method calls, with one addition. Each architecture specifies a
|
||||||
|
*closure context pointer* register and calls to closures store the
|
||||||
|
address of the closure object in the closure context pointer register
|
||||||
|
prior to the call.
|
||||||
|
|
||||||
|
## Software floating-point mode
|
||||||
|
|
||||||
|
In "softfloat" mode, the ABI simply treats the hardware as having zero
|
||||||
|
floating-point registers.
|
||||||
|
As a result, any arguments containing floating-point values will be
|
||||||
|
passed on the stack.
|
||||||
|
|
||||||
|
*Rationale*: Softfloat mode is about compatibility over performance
|
||||||
|
and is not commonly used.
|
||||||
|
Hence, we keep the ABI as simple as possible in this case, rather than
|
||||||
|
adding additional rules for passing floating-point values in integer
|
||||||
|
registers.
|
||||||
|
|
||||||
|
## Architecture specifics
|
||||||
|
|
||||||
|
This section describes per-architecture register mappings, as well as
|
||||||
|
other per-architecture special cases.
|
||||||
|
|
||||||
|
### amd64 architecture
|
||||||
|
|
||||||
|
The amd64 architecture uses the following sequence of 9 registers for
|
||||||
|
integer arguments and results:
|
||||||
|
|
||||||
|
RAX, RBX, RCX, RDI, RSI, R8, R9, R10, R11
|
||||||
|
|
||||||
|
It uses X0 – X14 for floating-point arguments and results.
|
||||||
|
|
||||||
|
*Rationale*: These sequences are chosen from the available registers
|
||||||
|
to be relatively easy to remember.
|
||||||
|
|
||||||
|
Registers R12 and R13 are permanent scratch registers.
|
||||||
|
R15 is a scratch register except in dynamically linked binaries.
|
||||||
|
|
||||||
|
*Rationale*: Some operations such as stack growth and reflection calls
|
||||||
|
need dedicated scratch registers in order to manipulate call frames
|
||||||
|
without corrupting arguments or results.
|
||||||
|
|
||||||
|
Special-purpose registers are as follows:
|
||||||
|
|
||||||
|
| Register | Call meaning | Body meaning |
|
||||||
|
| --- | --- | --- |
|
||||||
|
| RSP | Stack pointer | Fixed |
|
||||||
|
| RBP | Frame pointer | Fixed |
|
||||||
|
| RDX | Closure context pointer | Scratch |
|
||||||
|
| R12 | None | Scratch |
|
||||||
|
| R13 | None | Scratch |
|
||||||
|
| R14 | Current goroutine | Scratch |
|
||||||
|
| R15 | GOT reference temporary | Fixed if dynlink |
|
||||||
|
| X15 | Zero value | Fixed |
|
||||||
|
|
||||||
|
TODO: We may start with the existing TLS-based g and move to R14
|
||||||
|
later.
|
||||||
|
|
||||||
|
*Rationale*: These register meanings are compatible with Go’s
|
||||||
|
stack-based calling convention except for R14 and X15, which will have
|
||||||
|
to be restored on transitions from ABI0 code to ABIInternal code.
|
||||||
|
In ABI0, these are undefined, so transitions from ABIInternal to ABI0
|
||||||
|
can ignore these registers.
|
||||||
|
|
||||||
|
*Rationale*: For the current goroutine pointer, we chose a register
|
||||||
|
that requires an additional REX byte.
|
||||||
|
While this adds one byte to every function prologue, it is hardly ever
|
||||||
|
accessed outside the function prologue and we expect making more
|
||||||
|
single-byte registers available to be a net win.
|
||||||
|
|
||||||
|
*Rationale*: We designate X15 as a fixed zero register because
|
||||||
|
functions often have to bulk zero their stack frames, and this is more
|
||||||
|
efficient with a designated zero register.
|
||||||
|
|
||||||
|
#### Stack layout
|
||||||
|
|
||||||
|
The stack pointer, RSP, grows down and is always aligned to 8 bytes.
|
||||||
|
|
||||||
|
The amd64 architecture does not use a link register.
|
||||||
|
|
||||||
|
A function's stack frame is laid out as follows:
|
||||||
|
|
||||||
|
+------------------------------+
|
||||||
|
| return PC |
|
||||||
|
| RBP on entry |
|
||||||
|
| ... locals ... |
|
||||||
|
| ... outgoing arguments ... |
|
||||||
|
+------------------------------+ ↓ lower addresses
|
||||||
|
|
||||||
|
The "return PC" is pushed as part of the standard amd64 `CALL`
|
||||||
|
operation.
|
||||||
|
On entry, a function subtracts from RSP to open its stack frame and
|
||||||
|
saves the value of RBP directly below the return PC.
|
||||||
|
A leaf function that does not require any stack space may omit the
|
||||||
|
saved RBP.
|
||||||
|
|
||||||
|
The Go ABI's use of RBP as a frame pointer register is compatible with
|
||||||
|
amd64 platform conventions so that Go can inter-operate with platform
|
||||||
|
debuggers and profilers.
|
||||||
|
|
||||||
|
#### Flags
|
||||||
|
|
||||||
|
The direction flag (D) is always cleared (set to the “forward”
|
||||||
|
direction) at a call.
|
||||||
|
The arithmetic status flags are treated like scratch registers and not
|
||||||
|
preserved across calls.
|
||||||
|
All other bits in RFLAGS are system flags.
|
||||||
|
|
||||||
|
The CPU is always in MMX technology state (not x87 mode).
|
||||||
|
|
||||||
|
*Rationale*: Go on amd64 uses the XMM registers and never uses the x87
|
||||||
|
registers, so it makes sense to assume the CPU is in MMX mode.
|
||||||
|
Otherwise, any function that used the XMM registers would have to
|
||||||
|
execute an EMMS instruction before calling another function or
|
||||||
|
returning (this is the case in the SysV ABI).
|
||||||
|
|
||||||
|
At calls, the MXCSR control bits are always set as follows:
|
||||||
|
|
||||||
|
| Flag | Bit | Value | Meaning |
|
||||||
|
| --- | --- | --- | --- |
|
||||||
|
| FZ | 15 | 0 | Do not flush to zero |
|
||||||
|
| RC | 14/13 | 0 (RN) | Round to nearest |
|
||||||
|
| PM | 12 | 1 | Precision masked |
|
||||||
|
| UM | 11 | 1 | Underflow masked |
|
||||||
|
| OM | 10 | 1 | Overflow masked |
|
||||||
|
| ZM | 9 | 1 | Divide-by-zero masked |
|
||||||
|
| DM | 8 | 1 | Denormal operations masked |
|
||||||
|
| IM | 7 | 1 | Invalid operations masked |
|
||||||
|
| DAZ | 6 | 0 | Do not zero de-normals |
|
||||||
|
|
||||||
|
The MXCSR status bits are callee-save.
|
||||||
|
|
||||||
|
*Rationale*: Having a fixed MXCSR control configuration allows Go
|
||||||
|
functions to use SSE operations without modifying or saving the MXCSR.
|
||||||
|
Functions are allowed to modify it between calls (as long as they
|
||||||
|
restore it), but as of this writing Go code never does.
|
||||||
|
The above fixed configuration matches the process initialization
|
||||||
|
control bits specified by the ELF AMD64 ABI.
|
||||||
|
|
||||||
|
The x87 floating-point control word is not used by Go on amd64.
|
||||||
|
|
||||||
|
## Future directions
|
||||||
|
|
||||||
|
### Spill path improvements
|
||||||
|
|
||||||
|
The ABI currently reserves spill space for argument registers so the
|
||||||
|
compiler can statically generate an argument spill path before calling
|
||||||
|
into `runtime.morestack` to grow the stack.
|
||||||
|
This ensures there will be sufficient spill space even when the stack
|
||||||
|
is nearly exhausted and keeps stack growth and stack scanning
|
||||||
|
essentially unchanged from ABI0.
|
||||||
|
|
||||||
|
However, this wastes stack space (the median wastage is 16 bytes per
|
||||||
|
call), resulting in larger stacks and increased cache footprint.
|
||||||
|
A better approach would be to reserve stack space only when spilling.
|
||||||
|
One way to ensure enough space is available to spill would be for
|
||||||
|
every function to ensure there is enough space for the function's own
|
||||||
|
frame *as well as* the spill space of all functions it calls.
|
||||||
|
For most functions, this would change the threshold for the prologue
|
||||||
|
stack growth check.
|
||||||
|
For `nosplit` functions, this would change the threshold used in the
|
||||||
|
linker's static stack size check.
|
||||||
|
|
||||||
|
Allocating spill space in the callee rather than the caller may also
|
||||||
|
allow for faster reflection calls in the common case where a function
|
||||||
|
takes only register arguments, since it would allow reflection to make
|
||||||
|
these calls directly without allocating any frame.
|
||||||
|
|
||||||
|
The statically-generated spill path also increases code size.
|
||||||
|
It is possible to instead have a generic spill path in the runtime, as
|
||||||
|
part of `morestack`.
|
||||||
|
However, this complicates reserving the spill space, since spilling
|
||||||
|
all possible register arguments would, in most cases, take
|
||||||
|
significantly more space than spilling only those used by a particular
|
||||||
|
function.
|
||||||
|
Some options are to spill to a temporary space and copy back only the
|
||||||
|
registers used by the function, or to grow the stack if necessary
|
||||||
|
before spilling to it (using a temporary space if necessary), or to
|
||||||
|
use a heap-allocated space if insufficient stack space is available.
|
||||||
|
These options all add enough complexity that we will have to make this
|
||||||
|
decision based on the actual code size growth caused by the static
|
||||||
|
spill paths.
|
||||||
|
|
||||||
|
### Clobber sets
|
||||||
|
|
||||||
|
As defined, the ABI does not use callee-save registers.
|
||||||
|
This significantly simplifies the garbage collector and the compiler's
|
||||||
|
register allocator, but at some performance cost.
|
||||||
|
A potentially better balance for Go code would be to use *clobber
|
||||||
|
sets*: for each function, the compiler records the set of registers it
|
||||||
|
clobbers (including those clobbered by functions it calls) and any
|
||||||
|
register not clobbered by function F can remain live across calls to
|
||||||
|
F.
|
||||||
|
|
||||||
|
This is generally a good fit for Go because Go's package DAG allows
|
||||||
|
function metadata like the clobber set to flow up the call graph, even
|
||||||
|
across package boundaries.
|
||||||
|
Clobber sets would require relatively little change to the garbage
|
||||||
|
collector, unlike general callee-save registers.
|
||||||
|
One disadvantage of clobber sets over callee-save registers is that
|
||||||
|
they don't help with indirect function calls or interface method
|
||||||
|
calls, since static information isn't available in these cases.
|
||||||
|
|
||||||
|
### Large aggregates
|
||||||
|
|
||||||
|
Go encourages passing composite values by value, and this simplifies
|
||||||
|
reasoning about mutation and races.
|
||||||
|
However, this comes at a performance cost for large composite values.
|
||||||
|
It may be possible to instead transparently pass large composite
|
||||||
|
values by reference and delay copying until it is actually necessary.
|
||||||
|
|
||||||
|
## Appendix: Register usage analysis
|
||||||
|
|
||||||
|
In order to understand the impacts of the above design on register
|
||||||
|
usage, we
|
||||||
|
[analyzed](https://github.com/aclements/go-misc/tree/master/abi) the
|
||||||
|
impact of the above ABI on a large code base: cmd/kubelet from
|
||||||
|
[Kubernetes](https://github.com/kubernetes/kubernetes) at tag v1.18.8.
|
||||||
|
|
||||||
|
The following table shows the impact of different numbers of available
|
||||||
|
integer and floating-point registers on argument assignment:
|
||||||
|
|
||||||
|
```
|
||||||
|
| | | | stack args | spills | stack total |
|
||||||
|
| ints | floats | % fit | p50 | p95 | p99 | p50 | p95 | p99 | p50 | p95 | p99 |
|
||||||
|
| 0 | 0 | 6.3% | 32 | 152 | 256 | 0 | 0 | 0 | 32 | 152 | 256 |
|
||||||
|
| 0 | 8 | 6.4% | 32 | 152 | 256 | 0 | 0 | 0 | 32 | 152 | 256 |
|
||||||
|
| 1 | 8 | 21.3% | 24 | 144 | 248 | 8 | 8 | 8 | 32 | 152 | 256 |
|
||||||
|
| 2 | 8 | 38.9% | 16 | 128 | 224 | 8 | 16 | 16 | 24 | 136 | 240 |
|
||||||
|
| 3 | 8 | 57.0% | 0 | 120 | 224 | 16 | 24 | 24 | 24 | 136 | 240 |
|
||||||
|
| 4 | 8 | 73.0% | 0 | 120 | 216 | 16 | 32 | 32 | 24 | 136 | 232 |
|
||||||
|
| 5 | 8 | 83.3% | 0 | 112 | 216 | 16 | 40 | 40 | 24 | 136 | 232 |
|
||||||
|
| 6 | 8 | 87.5% | 0 | 112 | 208 | 16 | 48 | 48 | 24 | 136 | 232 |
|
||||||
|
| 7 | 8 | 89.8% | 0 | 112 | 208 | 16 | 48 | 56 | 24 | 136 | 232 |
|
||||||
|
| 8 | 8 | 91.3% | 0 | 112 | 200 | 16 | 56 | 64 | 24 | 136 | 232 |
|
||||||
|
| 9 | 8 | 92.1% | 0 | 112 | 192 | 16 | 56 | 72 | 24 | 136 | 232 |
|
||||||
|
| 10 | 8 | 92.6% | 0 | 104 | 192 | 16 | 56 | 72 | 24 | 136 | 232 |
|
||||||
|
| 11 | 8 | 93.1% | 0 | 104 | 184 | 16 | 56 | 80 | 24 | 128 | 232 |
|
||||||
|
| 12 | 8 | 93.4% | 0 | 104 | 176 | 16 | 56 | 88 | 24 | 128 | 232 |
|
||||||
|
| 13 | 8 | 94.0% | 0 | 88 | 176 | 16 | 56 | 96 | 24 | 128 | 232 |
|
||||||
|
| 14 | 8 | 94.4% | 0 | 80 | 152 | 16 | 64 | 104 | 24 | 128 | 232 |
|
||||||
|
| 15 | 8 | 94.6% | 0 | 80 | 152 | 16 | 64 | 112 | 24 | 128 | 232 |
|
||||||
|
| 16 | 8 | 94.9% | 0 | 16 | 152 | 16 | 64 | 112 | 24 | 128 | 232 |
|
||||||
|
| ∞ | 8 | 99.8% | 0 | 0 | 0 | 24 | 112 | 216 | 24 | 120 | 216 |
|
||||||
|
```
|
||||||
|
|
||||||
|
The first two columns show the number of available integer and
|
||||||
|
floating-point registers.
|
||||||
|
The first row shows the results for 0 integer and 0 floating-point
|
||||||
|
registers, which is equivalent to ABI0.
|
||||||
|
We found that any reasonable number of floating-point registers has
|
||||||
|
the same effect, so we fixed it at 8 for all other rows.
|
||||||
|
|
||||||
|
The “% fit” column gives the fraction of functions where all arguments
|
||||||
|
and results are register-assigned and no arguments are passed on the
|
||||||
|
stack.
|
||||||
|
The three “stack args” columns give the median, 95th and 99th
|
||||||
|
percentile number of bytes of stack arguments.
|
||||||
|
The “spills” columns likewise summarize the number of bytes in
|
||||||
|
on-stack spill space.
|
||||||
|
And “stack total” summarizes the sum of stack arguments and on-stack
|
||||||
|
spill slots.
|
||||||
|
Note that these are three different distributions; for example,
|
||||||
|
there’s no single function that takes 0 stack argument bytes, 16 spill
|
||||||
|
bytes, and 24 total stack bytes.
|
||||||
|
|
||||||
|
From this, we can see that the fraction of functions that fit entirely
|
||||||
|
in registers grows very slowly once it reaches about 90%, though
|
||||||
|
curiously there is a small minority of functions that could benefit
|
||||||
|
from a huge number of registers.
|
||||||
|
Making 9 integer registers available on amd64 puts it in this realm.
|
||||||
|
We also see that the stack space required for most functions is fairly
|
||||||
|
small.
|
||||||
|
While the increasing space required for spills largely balances out
|
||||||
|
the decreasing space required for stack arguments as the number of
|
||||||
|
available registers increases, there is a general reduction in the
|
||||||
|
total stack space required with more available registers.
|
||||||
|
This does, however, suggest that eliminating spill slots in the future
|
||||||
|
would noticeably reduce stack requirements.
|
||||||
|
|
@ -25,9 +25,8 @@ import (
|
||||||
type ABIParamResultInfo struct {
|
type ABIParamResultInfo struct {
|
||||||
inparams []ABIParamAssignment // Includes receiver for method calls. Does NOT include hidden closure pointer.
|
inparams []ABIParamAssignment // Includes receiver for method calls. Does NOT include hidden closure pointer.
|
||||||
outparams []ABIParamAssignment
|
outparams []ABIParamAssignment
|
||||||
intSpillSlots int
|
|
||||||
floatSpillSlots int
|
|
||||||
offsetToSpillArea int64
|
offsetToSpillArea int64
|
||||||
|
spillAreaSize int64
|
||||||
config *ABIConfig // to enable String() method
|
config *ABIConfig // to enable String() method
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -47,18 +46,14 @@ func (a *ABIParamResultInfo) OutParam(i int) ABIParamAssignment {
|
||||||
return a.outparams[i]
|
return a.outparams[i]
|
||||||
}
|
}
|
||||||
|
|
||||||
func (a *ABIParamResultInfo) IntSpillCount() int {
|
|
||||||
return a.intSpillSlots
|
|
||||||
}
|
|
||||||
|
|
||||||
func (a *ABIParamResultInfo) FloatSpillCount() int {
|
|
||||||
return a.floatSpillSlots
|
|
||||||
}
|
|
||||||
|
|
||||||
func (a *ABIParamResultInfo) SpillAreaOffset() int64 {
|
func (a *ABIParamResultInfo) SpillAreaOffset() int64 {
|
||||||
return a.offsetToSpillArea
|
return a.offsetToSpillArea
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func (a *ABIParamResultInfo) SpillAreaSize() int64 {
|
||||||
|
return a.spillAreaSize
|
||||||
|
}
|
||||||
|
|
||||||
// RegIndex stores the index into the set of machine registers used by
|
// RegIndex stores the index into the set of machine registers used by
|
||||||
// the ABI on a specific architecture for parameter passing. RegIndex
|
// the ABI on a specific architecture for parameter passing. RegIndex
|
||||||
// values 0 through N-1 (where N is the number of integer registers
|
// values 0 through N-1 (where N is the number of integer registers
|
||||||
|
|
@ -78,7 +73,27 @@ type RegIndex uint8
|
||||||
type ABIParamAssignment struct {
|
type ABIParamAssignment struct {
|
||||||
Type *types.Type
|
Type *types.Type
|
||||||
Registers []RegIndex
|
Registers []RegIndex
|
||||||
Offset int32
|
offset int32
|
||||||
|
}
|
||||||
|
|
||||||
|
// Offset returns the stack offset for addressing the parameter that "a" describes.
|
||||||
|
// This will panic if "a" describes a register-allocated parameter.
|
||||||
|
func (a *ABIParamAssignment) Offset() int32 {
|
||||||
|
if len(a.Registers) > 0 {
|
||||||
|
panic("Register allocated parameters have no offset")
|
||||||
|
}
|
||||||
|
return a.offset
|
||||||
|
}
|
||||||
|
|
||||||
|
// SpillOffset returns the offset *within the spill area* for the parameter that "a" describes.
|
||||||
|
// Registers will be spilled here; if a memory home is needed (for a pointer method e.g.)
|
||||||
|
// then that will be the address.
|
||||||
|
// This will panic if "a" describes a stack-allocated parameter.
|
||||||
|
func (a *ABIParamAssignment) SpillOffset() int32 {
|
||||||
|
if len(a.Registers) == 0 {
|
||||||
|
panic("Stack-allocated parameters have no spill offset")
|
||||||
|
}
|
||||||
|
return a.offset
|
||||||
}
|
}
|
||||||
|
|
||||||
// RegAmounts holds a specified number of integer/float registers.
|
// RegAmounts holds a specified number of integer/float registers.
|
||||||
|
|
@ -91,20 +106,58 @@ type RegAmounts struct {
|
||||||
// by the ABI rules for parameter passing and result returning.
|
// by the ABI rules for parameter passing and result returning.
|
||||||
type ABIConfig struct {
|
type ABIConfig struct {
|
||||||
// Do we need anything more than this?
|
// Do we need anything more than this?
|
||||||
regAmounts RegAmounts
|
regAmounts RegAmounts
|
||||||
|
regsForTypeCache map[*types.Type]int
|
||||||
}
|
}
|
||||||
|
|
||||||
// NewABIConfig returns a new ABI configuration for an architecture with
|
// NewABIConfig returns a new ABI configuration for an architecture with
|
||||||
// iRegsCount integer/pointer registers and fRegsCount floating point registers.
|
// iRegsCount integer/pointer registers and fRegsCount floating point registers.
|
||||||
func NewABIConfig(iRegsCount, fRegsCount int) *ABIConfig {
|
func NewABIConfig(iRegsCount, fRegsCount int) *ABIConfig {
|
||||||
return &ABIConfig{RegAmounts{iRegsCount, fRegsCount}}
|
return &ABIConfig{regAmounts: RegAmounts{iRegsCount, fRegsCount}, regsForTypeCache: make(map[*types.Type]int)}
|
||||||
|
}
|
||||||
|
|
||||||
|
// NumParamRegs returns the number of parameter registers used for a given type,
|
||||||
|
// without regard for the number available.
|
||||||
|
func (a *ABIConfig) NumParamRegs(t *types.Type) int {
|
||||||
|
if n, ok := a.regsForTypeCache[t]; ok {
|
||||||
|
return n
|
||||||
|
}
|
||||||
|
|
||||||
|
if t.IsScalar() || t.IsPtrShaped() {
|
||||||
|
var n int
|
||||||
|
if t.IsComplex() {
|
||||||
|
n = 2
|
||||||
|
} else {
|
||||||
|
n = (int(t.Size()) + types.RegSize - 1) / types.RegSize
|
||||||
|
}
|
||||||
|
a.regsForTypeCache[t] = n
|
||||||
|
return n
|
||||||
|
}
|
||||||
|
typ := t.Kind()
|
||||||
|
n := 0
|
||||||
|
switch typ {
|
||||||
|
case types.TARRAY:
|
||||||
|
n = a.NumParamRegs(t.Elem()) * int(t.NumElem())
|
||||||
|
case types.TSTRUCT:
|
||||||
|
for _, f := range t.FieldSlice() {
|
||||||
|
n += a.NumParamRegs(f.Type)
|
||||||
|
}
|
||||||
|
case types.TSLICE:
|
||||||
|
n = a.NumParamRegs(synthSlice)
|
||||||
|
case types.TSTRING:
|
||||||
|
n = a.NumParamRegs(synthString)
|
||||||
|
case types.TINTER:
|
||||||
|
n = a.NumParamRegs(synthIface)
|
||||||
|
}
|
||||||
|
a.regsForTypeCache[t] = n
|
||||||
|
return n
|
||||||
}
|
}
|
||||||
|
|
||||||
// ABIAnalyze takes a function type 't' and an ABI rules description
|
// ABIAnalyze takes a function type 't' and an ABI rules description
|
||||||
// 'config' and analyzes the function to determine how its parameters
|
// 'config' and analyzes the function to determine how its parameters
|
||||||
// and results will be passed (in registers or on the stack), returning
|
// and results will be passed (in registers or on the stack), returning
|
||||||
// an ABIParamResultInfo object that holds the results of the analysis.
|
// an ABIParamResultInfo object that holds the results of the analysis.
|
||||||
func ABIAnalyze(t *types.Type, config *ABIConfig) ABIParamResultInfo {
|
func (config *ABIConfig) ABIAnalyze(t *types.Type) ABIParamResultInfo {
|
||||||
setup()
|
setup()
|
||||||
s := assignState{
|
s := assignState{
|
||||||
rTotal: config.regAmounts,
|
rTotal: config.regAmounts,
|
||||||
|
|
@ -116,28 +169,27 @@ func ABIAnalyze(t *types.Type, config *ABIConfig) ABIParamResultInfo {
|
||||||
if t.NumRecvs() != 0 {
|
if t.NumRecvs() != 0 {
|
||||||
rfsl := ft.Receiver.FieldSlice()
|
rfsl := ft.Receiver.FieldSlice()
|
||||||
result.inparams = append(result.inparams,
|
result.inparams = append(result.inparams,
|
||||||
s.assignParamOrReturn(rfsl[0].Type))
|
s.assignParamOrReturn(rfsl[0].Type, false))
|
||||||
}
|
}
|
||||||
|
|
||||||
// Inputs
|
// Inputs
|
||||||
ifsl := ft.Params.FieldSlice()
|
ifsl := ft.Params.FieldSlice()
|
||||||
for _, f := range ifsl {
|
for _, f := range ifsl {
|
||||||
result.inparams = append(result.inparams,
|
result.inparams = append(result.inparams,
|
||||||
s.assignParamOrReturn(f.Type))
|
s.assignParamOrReturn(f.Type, false))
|
||||||
}
|
}
|
||||||
s.stackOffset = types.Rnd(s.stackOffset, int64(types.RegSize))
|
s.stackOffset = types.Rnd(s.stackOffset, int64(types.RegSize))
|
||||||
|
|
||||||
// Record number of spill slots needed.
|
|
||||||
result.intSpillSlots = s.rUsed.intRegs
|
|
||||||
result.floatSpillSlots = s.rUsed.floatRegs
|
|
||||||
|
|
||||||
// Outputs
|
// Outputs
|
||||||
s.rUsed = RegAmounts{}
|
s.rUsed = RegAmounts{}
|
||||||
ofsl := ft.Results.FieldSlice()
|
ofsl := ft.Results.FieldSlice()
|
||||||
for _, f := range ofsl {
|
for _, f := range ofsl {
|
||||||
result.outparams = append(result.outparams, s.assignParamOrReturn(f.Type))
|
result.outparams = append(result.outparams, s.assignParamOrReturn(f.Type, true))
|
||||||
}
|
}
|
||||||
result.offsetToSpillArea = s.stackOffset
|
// The spill area is at a register-aligned offset and its size is rounded up to a register alignment.
|
||||||
|
// TODO in theory could align offset only to minimum required by spilled data types.
|
||||||
|
result.offsetToSpillArea = alignTo(s.stackOffset, types.RegSize)
|
||||||
|
result.spillAreaSize = alignTo(s.spillOffset, types.RegSize)
|
||||||
|
|
||||||
return result
|
return result
|
||||||
}
|
}
|
||||||
|
|
@ -160,10 +212,14 @@ func (c *RegAmounts) regString(r RegIndex) string {
|
||||||
// form, suitable for debugging or unit testing.
|
// form, suitable for debugging or unit testing.
|
||||||
func (ri *ABIParamAssignment) toString(config *ABIConfig) string {
|
func (ri *ABIParamAssignment) toString(config *ABIConfig) string {
|
||||||
regs := "R{"
|
regs := "R{"
|
||||||
|
offname := "spilloffset" // offset is for spill for register(s)
|
||||||
|
if len(ri.Registers) == 0 {
|
||||||
|
offname = "offset" // offset is for memory arg
|
||||||
|
}
|
||||||
for _, r := range ri.Registers {
|
for _, r := range ri.Registers {
|
||||||
regs += " " + config.regAmounts.regString(r)
|
regs += " " + config.regAmounts.regString(r)
|
||||||
}
|
}
|
||||||
return fmt.Sprintf("%s } offset: %d typ: %v", regs, ri.Offset, ri.Type)
|
return fmt.Sprintf("%s } %s: %d typ: %v", regs, offname, ri.offset, ri.Type)
|
||||||
}
|
}
|
||||||
|
|
||||||
// toString method renders an ABIParamResultInfo in human-readable
|
// toString method renders an ABIParamResultInfo in human-readable
|
||||||
|
|
@ -176,8 +232,8 @@ func (ri *ABIParamResultInfo) String() string {
|
||||||
for k, r := range ri.outparams {
|
for k, r := range ri.outparams {
|
||||||
res += fmt.Sprintf("OUT %d: %s\n", k, r.toString(ri.config))
|
res += fmt.Sprintf("OUT %d: %s\n", k, r.toString(ri.config))
|
||||||
}
|
}
|
||||||
res += fmt.Sprintf("intspill: %d floatspill: %d offsetToSpillArea: %d",
|
res += fmt.Sprintf("offsetToSpillArea: %d spillAreaSize: %d",
|
||||||
ri.intSpillSlots, ri.floatSpillSlots, ri.offsetToSpillArea)
|
ri.offsetToSpillArea, ri.spillAreaSize)
|
||||||
return res
|
return res
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -188,16 +244,27 @@ type assignState struct {
|
||||||
rUsed RegAmounts // regs used by params completely assigned so far
|
rUsed RegAmounts // regs used by params completely assigned so far
|
||||||
pUsed RegAmounts // regs used by the current param (or pieces therein)
|
pUsed RegAmounts // regs used by the current param (or pieces therein)
|
||||||
stackOffset int64 // current stack offset
|
stackOffset int64 // current stack offset
|
||||||
|
spillOffset int64 // current spill offset
|
||||||
|
}
|
||||||
|
|
||||||
|
// align returns a rounded up to t's alignment
|
||||||
|
func align(a int64, t *types.Type) int64 {
|
||||||
|
return alignTo(a, int(t.Align))
|
||||||
|
}
|
||||||
|
|
||||||
|
// alignTo returns a rounded up to t, where t must be 0 or a power of 2.
|
||||||
|
func alignTo(a int64, t int) int64 {
|
||||||
|
if t == 0 {
|
||||||
|
return a
|
||||||
|
}
|
||||||
|
return types.Rnd(a, int64(t))
|
||||||
}
|
}
|
||||||
|
|
||||||
// stackSlot returns a stack offset for a param or result of the
|
// stackSlot returns a stack offset for a param or result of the
|
||||||
// specified type.
|
// specified type.
|
||||||
func (state *assignState) stackSlot(t *types.Type) int64 {
|
func (state *assignState) stackSlot(t *types.Type) int64 {
|
||||||
if t.Align > 0 {
|
rv := align(state.stackOffset, t)
|
||||||
state.stackOffset = types.Rnd(state.stackOffset, int64(t.Align))
|
state.stackOffset = rv + t.Width
|
||||||
}
|
|
||||||
rv := state.stackOffset
|
|
||||||
state.stackOffset += t.Width
|
|
||||||
return rv
|
return rv
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -225,11 +292,17 @@ func (state *assignState) allocateRegs() []RegIndex {
|
||||||
// regAllocate creates a register ABIParamAssignment object for a param
|
// regAllocate creates a register ABIParamAssignment object for a param
|
||||||
// or result with the specified type, as a final step (this assumes
|
// or result with the specified type, as a final step (this assumes
|
||||||
// that all of the safety/suitability analysis is complete).
|
// that all of the safety/suitability analysis is complete).
|
||||||
func (state *assignState) regAllocate(t *types.Type) ABIParamAssignment {
|
func (state *assignState) regAllocate(t *types.Type, isReturn bool) ABIParamAssignment {
|
||||||
|
spillLoc := int64(-1)
|
||||||
|
if !isReturn {
|
||||||
|
// Spill for register-resident t must be aligned for storage of a t.
|
||||||
|
spillLoc = align(state.spillOffset, t)
|
||||||
|
state.spillOffset = spillLoc + t.Size()
|
||||||
|
}
|
||||||
return ABIParamAssignment{
|
return ABIParamAssignment{
|
||||||
Type: t,
|
Type: t,
|
||||||
Registers: state.allocateRegs(),
|
Registers: state.allocateRegs(),
|
||||||
Offset: -1,
|
offset: int32(spillLoc),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -239,7 +312,7 @@ func (state *assignState) regAllocate(t *types.Type) ABIParamAssignment {
|
||||||
func (state *assignState) stackAllocate(t *types.Type) ABIParamAssignment {
|
func (state *assignState) stackAllocate(t *types.Type) ABIParamAssignment {
|
||||||
return ABIParamAssignment{
|
return ABIParamAssignment{
|
||||||
Type: t,
|
Type: t,
|
||||||
Offset: int32(state.stackSlot(t)),
|
offset: int32(state.stackSlot(t)),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -261,6 +334,9 @@ func (state *assignState) floatUsed() int {
|
||||||
// accordingly).
|
// accordingly).
|
||||||
func (state *assignState) regassignIntegral(t *types.Type) bool {
|
func (state *assignState) regassignIntegral(t *types.Type) bool {
|
||||||
regsNeeded := int(types.Rnd(t.Width, int64(types.PtrSize)) / int64(types.PtrSize))
|
regsNeeded := int(types.Rnd(t.Width, int64(types.PtrSize)) / int64(types.PtrSize))
|
||||||
|
if t.IsComplex() {
|
||||||
|
regsNeeded = 2
|
||||||
|
}
|
||||||
|
|
||||||
// Floating point and complex.
|
// Floating point and complex.
|
||||||
if t.IsFloat() || t.IsComplex() {
|
if t.IsFloat() || t.IsComplex() {
|
||||||
|
|
@ -371,14 +447,14 @@ func (state *assignState) regassign(pt *types.Type) bool {
|
||||||
// of type 'pt' to determine whether it can be register assigned.
|
// of type 'pt' to determine whether it can be register assigned.
|
||||||
// The result of the analysis is recorded in the result
|
// The result of the analysis is recorded in the result
|
||||||
// ABIParamResultInfo held in 'state'.
|
// ABIParamResultInfo held in 'state'.
|
||||||
func (state *assignState) assignParamOrReturn(pt *types.Type) ABIParamAssignment {
|
func (state *assignState) assignParamOrReturn(pt *types.Type, isReturn bool) ABIParamAssignment {
|
||||||
state.pUsed = RegAmounts{}
|
state.pUsed = RegAmounts{}
|
||||||
if pt.Width == types.BADWIDTH {
|
if pt.Width == types.BADWIDTH {
|
||||||
panic("should never happen")
|
panic("should never happen")
|
||||||
} else if pt.Width == 0 {
|
} else if pt.Width == 0 {
|
||||||
return state.stackAllocate(pt)
|
return state.stackAllocate(pt)
|
||||||
} else if state.regassign(pt) {
|
} else if state.regassign(pt) {
|
||||||
return state.regAllocate(pt)
|
return state.regAllocate(pt, isReturn)
|
||||||
} else {
|
} else {
|
||||||
return state.stackAllocate(pt)
|
return state.stackAllocate(pt)
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -1463,14 +1463,6 @@ func (p *noder) basicLit(lit *syntax.BasicLit) constant.Value {
|
||||||
p.errorAt(lit.Pos(), "malformed constant: %s", lit.Value)
|
p.errorAt(lit.Pos(), "malformed constant: %s", lit.Value)
|
||||||
}
|
}
|
||||||
|
|
||||||
// go/constant uses big.Rat by default, which is more precise, but
|
|
||||||
// causes toolstash -cmp and some tests to fail. For now, convert
|
|
||||||
// to big.Float to match cmd/compile's historical precision.
|
|
||||||
// TODO(mdempsky): Remove.
|
|
||||||
if v.Kind() == constant.Float {
|
|
||||||
v = constant.Make(ir.BigFloat(v))
|
|
||||||
}
|
|
||||||
|
|
||||||
return v
|
return v
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -434,6 +434,7 @@ func buildssa(fn *ir.Func, worker int) *ssa.Func {
|
||||||
// bitmask showing which of the open-coded defers in this function
|
// bitmask showing which of the open-coded defers in this function
|
||||||
// have been activated.
|
// have been activated.
|
||||||
deferBitsTemp := typecheck.TempAt(src.NoXPos, s.curfn, types.Types[types.TUINT8])
|
deferBitsTemp := typecheck.TempAt(src.NoXPos, s.curfn, types.Types[types.TUINT8])
|
||||||
|
deferBitsTemp.SetAddrtaken(true)
|
||||||
s.deferBitsTemp = deferBitsTemp
|
s.deferBitsTemp = deferBitsTemp
|
||||||
// For this value, AuxInt is initialized to zero by default
|
// For this value, AuxInt is initialized to zero by default
|
||||||
startDeferBits := s.entryNewValue0(ssa.OpConst8, types.Types[types.TUINT8])
|
startDeferBits := s.entryNewValue0(ssa.OpConst8, types.Types[types.TUINT8])
|
||||||
|
|
@ -5086,6 +5087,10 @@ func (s *state) addr(n ir.Node) *ssa.Value {
|
||||||
defer s.popLine()
|
defer s.popLine()
|
||||||
}
|
}
|
||||||
|
|
||||||
|
if s.canSSA(n) {
|
||||||
|
s.Fatalf("addr of canSSA expression: %+v", n)
|
||||||
|
}
|
||||||
|
|
||||||
t := types.NewPtr(n.Type())
|
t := types.NewPtr(n.Type())
|
||||||
linksymOffset := func(lsym *obj.LSym, offset int64) *ssa.Value {
|
linksymOffset := func(lsym *obj.LSym, offset int64) *ssa.Value {
|
||||||
v := s.entryNewValue1A(ssa.OpAddr, t, lsym, s.sb)
|
v := s.entryNewValue1A(ssa.OpAddr, t, lsym, s.sb)
|
||||||
|
|
|
||||||
|
|
@ -21,7 +21,7 @@ import (
|
||||||
// AMD64 registers available:
|
// AMD64 registers available:
|
||||||
// - integer: RAX, RBX, RCX, RDI, RSI, R8, R9, r10, R11
|
// - integer: RAX, RBX, RCX, RDI, RSI, R8, R9, r10, R11
|
||||||
// - floating point: X0 - X14
|
// - floating point: X0 - X14
|
||||||
var configAMD64 = abi.NewABIConfig(9,15)
|
var configAMD64 = abi.NewABIConfig(9, 15)
|
||||||
|
|
||||||
func TestMain(m *testing.M) {
|
func TestMain(m *testing.M) {
|
||||||
ssagen.Arch.LinkArch = &x86.Linkamd64
|
ssagen.Arch.LinkArch = &x86.Linkamd64
|
||||||
|
|
@ -46,9 +46,9 @@ func TestABIUtilsBasic1(t *testing.T) {
|
||||||
|
|
||||||
// expected results
|
// expected results
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 } offset: -1 typ: int32
|
IN 0: R{ I0 } spilloffset: 0 typ: int32
|
||||||
OUT 0: R{ I0 } offset: -1 typ: int32
|
OUT 0: R{ I0 } spilloffset: -1 typ: int32
|
||||||
intspill: 1 floatspill: 0 offsetToSpillArea: 0
|
offsetToSpillArea: 0 spillAreaSize: 8
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -75,39 +75,39 @@ func TestABIUtilsBasic2(t *testing.T) {
|
||||||
i8, i16, i32, i64},
|
i8, i16, i32, i64},
|
||||||
[]*types.Type{i32, f64, f64})
|
[]*types.Type{i32, f64, f64})
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 } offset: -1 typ: int8
|
IN 0: R{ I0 } spilloffset: 0 typ: int8
|
||||||
IN 1: R{ I1 } offset: -1 typ: int16
|
IN 1: R{ I1 } spilloffset: 2 typ: int16
|
||||||
IN 2: R{ I2 } offset: -1 typ: int32
|
IN 2: R{ I2 } spilloffset: 4 typ: int32
|
||||||
IN 3: R{ I3 } offset: -1 typ: int64
|
IN 3: R{ I3 } spilloffset: 8 typ: int64
|
||||||
IN 4: R{ F0 } offset: -1 typ: float32
|
IN 4: R{ F0 } spilloffset: 16 typ: float32
|
||||||
IN 5: R{ F1 } offset: -1 typ: float32
|
IN 5: R{ F1 } spilloffset: 20 typ: float32
|
||||||
IN 6: R{ F2 } offset: -1 typ: float64
|
IN 6: R{ F2 } spilloffset: 24 typ: float64
|
||||||
IN 7: R{ F3 } offset: -1 typ: float64
|
IN 7: R{ F3 } spilloffset: 32 typ: float64
|
||||||
IN 8: R{ I4 } offset: -1 typ: int8
|
IN 8: R{ I4 } spilloffset: 40 typ: int8
|
||||||
IN 9: R{ I5 } offset: -1 typ: int16
|
IN 9: R{ I5 } spilloffset: 42 typ: int16
|
||||||
IN 10: R{ I6 } offset: -1 typ: int32
|
IN 10: R{ I6 } spilloffset: 44 typ: int32
|
||||||
IN 11: R{ I7 } offset: -1 typ: int64
|
IN 11: R{ I7 } spilloffset: 48 typ: int64
|
||||||
IN 12: R{ F4 } offset: -1 typ: float32
|
IN 12: R{ F4 } spilloffset: 56 typ: float32
|
||||||
IN 13: R{ F5 } offset: -1 typ: float32
|
IN 13: R{ F5 } spilloffset: 60 typ: float32
|
||||||
IN 14: R{ F6 } offset: -1 typ: float64
|
IN 14: R{ F6 } spilloffset: 64 typ: float64
|
||||||
IN 15: R{ F7 } offset: -1 typ: float64
|
IN 15: R{ F7 } spilloffset: 72 typ: float64
|
||||||
IN 16: R{ F8 F9 } offset: -1 typ: complex128
|
IN 16: R{ F8 F9 } spilloffset: 80 typ: complex128
|
||||||
IN 17: R{ F10 F11 } offset: -1 typ: complex128
|
IN 17: R{ F10 F11 } spilloffset: 96 typ: complex128
|
||||||
IN 18: R{ F12 F13 } offset: -1 typ: complex128
|
IN 18: R{ F12 F13 } spilloffset: 112 typ: complex128
|
||||||
IN 19: R{ } offset: 0 typ: complex128
|
IN 19: R{ } offset: 0 typ: complex128
|
||||||
IN 20: R{ F14 } offset: -1 typ: complex64
|
IN 20: R{ } offset: 16 typ: complex64
|
||||||
IN 21: R{ I8 } offset: -1 typ: int8
|
IN 21: R{ I8 } spilloffset: 128 typ: int8
|
||||||
IN 22: R{ } offset: 16 typ: int16
|
IN 22: R{ } offset: 24 typ: int16
|
||||||
IN 23: R{ } offset: 20 typ: int32
|
IN 23: R{ } offset: 28 typ: int32
|
||||||
IN 24: R{ } offset: 24 typ: int64
|
IN 24: R{ } offset: 32 typ: int64
|
||||||
IN 25: R{ } offset: 32 typ: int8
|
IN 25: R{ } offset: 40 typ: int8
|
||||||
IN 26: R{ } offset: 34 typ: int16
|
IN 26: R{ } offset: 42 typ: int16
|
||||||
IN 27: R{ } offset: 36 typ: int32
|
IN 27: R{ } offset: 44 typ: int32
|
||||||
IN 28: R{ } offset: 40 typ: int64
|
IN 28: R{ } offset: 48 typ: int64
|
||||||
OUT 0: R{ I0 } offset: -1 typ: int32
|
OUT 0: R{ I0 } spilloffset: -1 typ: int32
|
||||||
OUT 1: R{ F0 } offset: -1 typ: float64
|
OUT 1: R{ F0 } spilloffset: -1 typ: float64
|
||||||
OUT 2: R{ F1 } offset: -1 typ: float64
|
OUT 2: R{ F1 } spilloffset: -1 typ: float64
|
||||||
intspill: 9 floatspill: 15 offsetToSpillArea: 48
|
offsetToSpillArea: 56 spillAreaSize: 136
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -123,15 +123,15 @@ func TestABIUtilsArrays(t *testing.T) {
|
||||||
[]*types.Type{a2, a1, ae, aa1})
|
[]*types.Type{a2, a1, ae, aa1})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 } offset: -1 typ: [1]int32
|
IN 0: R{ I0 } spilloffset: 0 typ: [1]int32
|
||||||
IN 1: R{ } offset: 0 typ: [0]int32
|
IN 1: R{ } offset: 0 typ: [0]int32
|
||||||
IN 2: R{ I1 } offset: -1 typ: [1][1]int32
|
IN 2: R{ I1 } spilloffset: 4 typ: [1][1]int32
|
||||||
IN 3: R{ } offset: 0 typ: [2]int32
|
IN 3: R{ } offset: 0 typ: [2]int32
|
||||||
OUT 0: R{ } offset: 8 typ: [2]int32
|
OUT 0: R{ } offset: 8 typ: [2]int32
|
||||||
OUT 1: R{ I0 } offset: -1 typ: [1]int32
|
OUT 1: R{ I0 } spilloffset: -1 typ: [1]int32
|
||||||
OUT 2: R{ } offset: 16 typ: [0]int32
|
OUT 2: R{ } offset: 16 typ: [0]int32
|
||||||
OUT 3: R{ I1 } offset: -1 typ: [1][1]int32
|
OUT 3: R{ I1 } spilloffset: -1 typ: [1][1]int32
|
||||||
intspill: 2 floatspill: 0 offsetToSpillArea: 16
|
offsetToSpillArea: 16 spillAreaSize: 8
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -147,13 +147,13 @@ func TestABIUtilsStruct1(t *testing.T) {
|
||||||
[]*types.Type{s, i8, i32})
|
[]*types.Type{s, i8, i32})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 } offset: -1 typ: int8
|
IN 0: R{ I0 } spilloffset: 0 typ: int8
|
||||||
IN 1: R{ I1 I2 I3 I4 } offset: -1 typ: struct { int8; int8; struct {}; int8; int16 }
|
IN 1: R{ I1 I2 I3 I4 } spilloffset: 2 typ: struct { int8; int8; struct {}; int8; int16 }
|
||||||
IN 2: R{ I5 } offset: -1 typ: int64
|
IN 2: R{ I5 } spilloffset: 8 typ: int64
|
||||||
OUT 0: R{ I0 I1 I2 I3 } offset: -1 typ: struct { int8; int8; struct {}; int8; int16 }
|
OUT 0: R{ I0 I1 I2 I3 } spilloffset: -1 typ: struct { int8; int8; struct {}; int8; int16 }
|
||||||
OUT 1: R{ I4 } offset: -1 typ: int8
|
OUT 1: R{ I4 } spilloffset: -1 typ: int8
|
||||||
OUT 2: R{ I5 } offset: -1 typ: int32
|
OUT 2: R{ I5 } spilloffset: -1 typ: int32
|
||||||
intspill: 6 floatspill: 0 offsetToSpillArea: 0
|
offsetToSpillArea: 0 spillAreaSize: 16
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -168,12 +168,12 @@ func TestABIUtilsStruct2(t *testing.T) {
|
||||||
[]*types.Type{fs, fs})
|
[]*types.Type{fs, fs})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 } offset: -1 typ: struct { int64; struct {} }
|
IN 0: R{ I0 } spilloffset: 0 typ: struct { int64; struct {} }
|
||||||
IN 1: R{ I1 } offset: -1 typ: struct { int64; struct {} }
|
IN 1: R{ I1 } spilloffset: 16 typ: struct { int64; struct {} }
|
||||||
IN 2: R{ I2 F0 } offset: -1 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
IN 2: R{ I2 F0 } spilloffset: 32 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
||||||
OUT 0: R{ I0 F0 } offset: -1 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
OUT 0: R{ I0 F0 } spilloffset: -1 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
||||||
OUT 1: R{ I1 F1 } offset: -1 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
OUT 1: R{ I1 F1 } spilloffset: -1 typ: struct { float64; struct { int64; struct {} }; struct {} }
|
||||||
intspill: 3 floatspill: 1 offsetToSpillArea: 0
|
offsetToSpillArea: 0 spillAreaSize: 64
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -189,19 +189,19 @@ func TestABIUtilsSliceString(t *testing.T) {
|
||||||
[]*types.Type{str, i64, str, sli32})
|
[]*types.Type{str, i64, str, sli32})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 I1 I2 } offset: -1 typ: []int32
|
IN 0: R{ I0 I1 I2 } spilloffset: 0 typ: []int32
|
||||||
IN 1: R{ I3 } offset: -1 typ: int8
|
IN 1: R{ I3 } spilloffset: 24 typ: int8
|
||||||
IN 2: R{ I4 I5 I6 } offset: -1 typ: []int32
|
IN 2: R{ I4 I5 I6 } spilloffset: 32 typ: []int32
|
||||||
IN 3: R{ I7 } offset: -1 typ: int8
|
IN 3: R{ I7 } spilloffset: 56 typ: int8
|
||||||
IN 4: R{ } offset: 0 typ: string
|
IN 4: R{ } offset: 0 typ: string
|
||||||
IN 5: R{ I8 } offset: -1 typ: int8
|
IN 5: R{ I8 } spilloffset: 57 typ: int8
|
||||||
IN 6: R{ } offset: 16 typ: int64
|
IN 6: R{ } offset: 16 typ: int64
|
||||||
IN 7: R{ } offset: 24 typ: []int32
|
IN 7: R{ } offset: 24 typ: []int32
|
||||||
OUT 0: R{ I0 I1 } offset: -1 typ: string
|
OUT 0: R{ I0 I1 } spilloffset: -1 typ: string
|
||||||
OUT 1: R{ I2 } offset: -1 typ: int64
|
OUT 1: R{ I2 } spilloffset: -1 typ: int64
|
||||||
OUT 2: R{ I3 I4 } offset: -1 typ: string
|
OUT 2: R{ I3 I4 } spilloffset: -1 typ: string
|
||||||
OUT 3: R{ I5 I6 I7 } offset: -1 typ: []int32
|
OUT 3: R{ I5 I6 I7 } spilloffset: -1 typ: []int32
|
||||||
intspill: 9 floatspill: 0 offsetToSpillArea: 48
|
offsetToSpillArea: 48 spillAreaSize: 64
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -219,17 +219,17 @@ func TestABIUtilsMethod(t *testing.T) {
|
||||||
[]*types.Type{a7, f64, i64})
|
[]*types.Type{a7, f64, i64})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 I1 I2 } offset: -1 typ: struct { int16; int16; int16 }
|
IN 0: R{ I0 I1 I2 } spilloffset: 0 typ: struct { int16; int16; int16 }
|
||||||
IN 1: R{ I3 } offset: -1 typ: *struct { int16; int16; int16 }
|
IN 1: R{ I3 } spilloffset: 8 typ: *struct { int16; int16; int16 }
|
||||||
IN 2: R{ } offset: 0 typ: [7]*struct { int16; int16; int16 }
|
IN 2: R{ } offset: 0 typ: [7]*struct { int16; int16; int16 }
|
||||||
IN 3: R{ F0 } offset: -1 typ: float64
|
IN 3: R{ F0 } spilloffset: 16 typ: float64
|
||||||
IN 4: R{ I4 } offset: -1 typ: int16
|
IN 4: R{ I4 } spilloffset: 24 typ: int16
|
||||||
IN 5: R{ I5 } offset: -1 typ: int16
|
IN 5: R{ I5 } spilloffset: 26 typ: int16
|
||||||
IN 6: R{ I6 } offset: -1 typ: int16
|
IN 6: R{ I6 } spilloffset: 28 typ: int16
|
||||||
OUT 0: R{ } offset: 56 typ: [7]*struct { int16; int16; int16 }
|
OUT 0: R{ } offset: 56 typ: [7]*struct { int16; int16; int16 }
|
||||||
OUT 1: R{ F0 } offset: -1 typ: float64
|
OUT 1: R{ F0 } spilloffset: -1 typ: float64
|
||||||
OUT 2: R{ I0 } offset: -1 typ: int64
|
OUT 2: R{ I0 } spilloffset: -1 typ: int64
|
||||||
intspill: 7 floatspill: 1 offsetToSpillArea: 112
|
offsetToSpillArea: 112 spillAreaSize: 32
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
|
|
@ -252,18 +252,44 @@ func TestABIUtilsInterfaces(t *testing.T) {
|
||||||
[]*types.Type{ei, nei, pei})
|
[]*types.Type{ei, nei, pei})
|
||||||
|
|
||||||
exp := makeExpectedDump(`
|
exp := makeExpectedDump(`
|
||||||
IN 0: R{ I0 I1 I2 } offset: -1 typ: struct { int16; int16; bool }
|
IN 0: R{ I0 I1 I2 } spilloffset: 0 typ: struct { int16; int16; bool }
|
||||||
IN 1: R{ I3 I4 } offset: -1 typ: interface {}
|
IN 1: R{ I3 I4 } spilloffset: 8 typ: interface {}
|
||||||
IN 2: R{ I5 I6 } offset: -1 typ: interface {}
|
IN 2: R{ I5 I6 } spilloffset: 24 typ: interface {}
|
||||||
IN 3: R{ I7 I8 } offset: -1 typ: interface { () untyped string }
|
IN 3: R{ I7 I8 } spilloffset: 40 typ: interface { () untyped string }
|
||||||
IN 4: R{ } offset: 0 typ: *interface {}
|
IN 4: R{ } offset: 0 typ: *interface {}
|
||||||
IN 5: R{ } offset: 8 typ: interface { () untyped string }
|
IN 5: R{ } offset: 8 typ: interface { () untyped string }
|
||||||
IN 6: R{ } offset: 24 typ: int16
|
IN 6: R{ } offset: 24 typ: int16
|
||||||
OUT 0: R{ I0 I1 } offset: -1 typ: interface {}
|
OUT 0: R{ I0 I1 } spilloffset: -1 typ: interface {}
|
||||||
OUT 1: R{ I2 I3 } offset: -1 typ: interface { () untyped string }
|
OUT 1: R{ I2 I3 } spilloffset: -1 typ: interface { () untyped string }
|
||||||
OUT 2: R{ I4 } offset: -1 typ: *interface {}
|
OUT 2: R{ I4 } spilloffset: -1 typ: *interface {}
|
||||||
intspill: 9 floatspill: 0 offsetToSpillArea: 32
|
offsetToSpillArea: 32 spillAreaSize: 56
|
||||||
`)
|
`)
|
||||||
|
|
||||||
abitest(t, ft, exp)
|
abitest(t, ft, exp)
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func TestABINumParamRegs(t *testing.T) {
|
||||||
|
i8 := types.Types[types.TINT8]
|
||||||
|
i16 := types.Types[types.TINT16]
|
||||||
|
i32 := types.Types[types.TINT32]
|
||||||
|
i64 := types.Types[types.TINT64]
|
||||||
|
f32 := types.Types[types.TFLOAT32]
|
||||||
|
f64 := types.Types[types.TFLOAT64]
|
||||||
|
c64 := types.Types[types.TCOMPLEX64]
|
||||||
|
c128 := types.Types[types.TCOMPLEX128]
|
||||||
|
|
||||||
|
s := mkstruct([]*types.Type{i8, i8, mkstruct([]*types.Type{}), i8, i16})
|
||||||
|
a := types.NewArray(s, 3)
|
||||||
|
|
||||||
|
nrtest(t, i8, 1)
|
||||||
|
nrtest(t, i16, 1)
|
||||||
|
nrtest(t, i32, 1)
|
||||||
|
nrtest(t, i64, 1)
|
||||||
|
nrtest(t, f32, 1)
|
||||||
|
nrtest(t, f64, 1)
|
||||||
|
nrtest(t, c64, 2)
|
||||||
|
nrtest(t, c128, 2)
|
||||||
|
nrtest(t, s, 4)
|
||||||
|
nrtest(t, a, 12)
|
||||||
|
|
||||||
|
}
|
||||||
|
|
@ -78,9 +78,9 @@ func tokenize(src string) []string {
|
||||||
|
|
||||||
func verifyParamResultOffset(t *testing.T, f *types.Field, r abi.ABIParamAssignment, which string, idx int) int {
|
func verifyParamResultOffset(t *testing.T, f *types.Field, r abi.ABIParamAssignment, which string, idx int) int {
|
||||||
n := ir.AsNode(f.Nname).(*ir.Name)
|
n := ir.AsNode(f.Nname).(*ir.Name)
|
||||||
if n.FrameOffset() != int64(r.Offset) {
|
if n.FrameOffset() != int64(r.Offset()) {
|
||||||
t.Errorf("%s %d: got offset %d wanted %d t=%v",
|
t.Errorf("%s %d: got offset %d wanted %d t=%v",
|
||||||
which, idx, r.Offset, n.Offset_, f.Type)
|
which, idx, r.Offset(), n.Offset_, f.Type)
|
||||||
return 1
|
return 1
|
||||||
}
|
}
|
||||||
return 0
|
return 0
|
||||||
|
|
@ -106,12 +106,20 @@ func difftokens(atoks []string, etoks []string) string {
|
||||||
return ""
|
return ""
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func nrtest(t *testing.T, ft *types.Type, expected int) {
|
||||||
|
types.CalcSize(ft)
|
||||||
|
got := configAMD64.NumParamRegs(ft)
|
||||||
|
if got != expected {
|
||||||
|
t.Errorf("]\nexpected num regs = %d, got %d, type %v", expected, got, ft)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
func abitest(t *testing.T, ft *types.Type, exp expectedDump) {
|
func abitest(t *testing.T, ft *types.Type, exp expectedDump) {
|
||||||
|
|
||||||
types.CalcSize(ft)
|
types.CalcSize(ft)
|
||||||
|
|
||||||
// Analyze with full set of registers.
|
// Analyze with full set of registers.
|
||||||
regRes := abi.ABIAnalyze(ft, configAMD64)
|
regRes := configAMD64.ABIAnalyze(ft)
|
||||||
regResString := strings.TrimSpace(regRes.String())
|
regResString := strings.TrimSpace(regRes.String())
|
||||||
|
|
||||||
// Check results.
|
// Check results.
|
||||||
|
|
@ -122,8 +130,8 @@ func abitest(t *testing.T, ft *types.Type, exp expectedDump) {
|
||||||
}
|
}
|
||||||
|
|
||||||
// Analyze again with empty register set.
|
// Analyze again with empty register set.
|
||||||
empty := &abi.ABIConfig{}
|
empty := abi.NewABIConfig(0, 0)
|
||||||
emptyRes := abi.ABIAnalyze(ft, empty)
|
emptyRes := empty.ABIAnalyze(ft)
|
||||||
emptyResString := emptyRes.String()
|
emptyResString := emptyRes.String()
|
||||||
|
|
||||||
// Walk the results and make sure the offsets assigned match
|
// Walk the results and make sure the offsets assigned match
|
||||||
|
|
|
||||||
|
|
@ -462,12 +462,16 @@ func (p *iexporter) doDecl(n *ir.Name) {
|
||||||
}
|
}
|
||||||
|
|
||||||
case ir.OLITERAL:
|
case ir.OLITERAL:
|
||||||
|
// TODO(mdempsky): Extend check to all declarations.
|
||||||
|
if n.Typecheck() == 0 {
|
||||||
|
base.FatalfAt(n.Pos(), "missed typecheck: %v", n)
|
||||||
|
}
|
||||||
|
|
||||||
// Constant.
|
// Constant.
|
||||||
// TODO(mdempsky): Do we still need this typecheck? If so, why?
|
|
||||||
n = Expr(n).(*ir.Name)
|
|
||||||
w.tag('C')
|
w.tag('C')
|
||||||
w.pos(n.Pos())
|
w.pos(n.Pos())
|
||||||
w.value(n.Type(), n.Val())
|
w.value(n.Type(), n.Val())
|
||||||
|
w.constExt(n)
|
||||||
|
|
||||||
case ir.OTYPE:
|
case ir.OTYPE:
|
||||||
if types.IsDotAlias(n.Sym()) {
|
if types.IsDotAlias(n.Sym()) {
|
||||||
|
|
@ -956,6 +960,17 @@ func (w *exportWriter) mpfloat(v constant.Value, typ *types.Type) {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func (w *exportWriter) mprat(v constant.Value) {
|
||||||
|
r, ok := constant.Val(v).(*big.Rat)
|
||||||
|
if !w.bool(ok) {
|
||||||
|
return
|
||||||
|
}
|
||||||
|
// TODO(mdempsky): Come up with a more efficient binary
|
||||||
|
// encoding before bumping iexportVersion to expose to
|
||||||
|
// gcimporter.
|
||||||
|
w.string(r.String())
|
||||||
|
}
|
||||||
|
|
||||||
func (w *exportWriter) bool(b bool) bool {
|
func (w *exportWriter) bool(b bool) bool {
|
||||||
var x uint64
|
var x uint64
|
||||||
if b {
|
if b {
|
||||||
|
|
@ -971,7 +986,37 @@ func (w *exportWriter) string(s string) { w.uint64(w.p.stringOff(s)) }
|
||||||
|
|
||||||
// Compiler-specific extensions.
|
// Compiler-specific extensions.
|
||||||
|
|
||||||
func (w *exportWriter) varExt(n ir.Node) {
|
func (w *exportWriter) constExt(n *ir.Name) {
|
||||||
|
// Internally, we now represent untyped float and complex
|
||||||
|
// constants with infinite-precision rational numbers using
|
||||||
|
// go/constant, but the "public" export data format known to
|
||||||
|
// gcimporter only supports 512-bit floating point constants.
|
||||||
|
// In case rationals turn out to be a bad idea and we want to
|
||||||
|
// switch back to fixed-precision constants, for now we
|
||||||
|
// continue writing out the 512-bit truncation in the public
|
||||||
|
// data section, and write the exact, rational constant in the
|
||||||
|
// compiler's extension data. Also, we only need to worry
|
||||||
|
// about exporting rationals for declared constants, because
|
||||||
|
// constants that appear in an expression will already have
|
||||||
|
// been coerced to a concrete, fixed-precision type.
|
||||||
|
//
|
||||||
|
// Eventually, assuming we stick with using rationals, we
|
||||||
|
// should bump iexportVersion to support rationals, and do the
|
||||||
|
// whole gcimporter update song-and-dance.
|
||||||
|
//
|
||||||
|
// TODO(mdempsky): Prepare vocals for that.
|
||||||
|
|
||||||
|
switch n.Type() {
|
||||||
|
case types.UntypedFloat:
|
||||||
|
w.mprat(n.Val())
|
||||||
|
case types.UntypedComplex:
|
||||||
|
v := n.Val()
|
||||||
|
w.mprat(constant.Real(v))
|
||||||
|
w.mprat(constant.Imag(v))
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
func (w *exportWriter) varExt(n *ir.Name) {
|
||||||
w.linkname(n.Sym())
|
w.linkname(n.Sym())
|
||||||
w.symIdx(n.Sym())
|
w.symIdx(n.Sym())
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -303,7 +303,9 @@ func (r *importReader) doDecl(sym *types.Sym) *ir.Name {
|
||||||
typ := r.typ()
|
typ := r.typ()
|
||||||
val := r.value(typ)
|
val := r.value(typ)
|
||||||
|
|
||||||
return importconst(r.p.ipkg, pos, sym, typ, val)
|
n := importconst(r.p.ipkg, pos, sym, typ, val)
|
||||||
|
r.constExt(n)
|
||||||
|
return n
|
||||||
|
|
||||||
case 'F':
|
case 'F':
|
||||||
typ := r.signature(nil)
|
typ := r.signature(nil)
|
||||||
|
|
@ -440,6 +442,15 @@ func (p *importReader) float(typ *types.Type) constant.Value {
|
||||||
return constant.Make(&f)
|
return constant.Make(&f)
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func (p *importReader) mprat(orig constant.Value) constant.Value {
|
||||||
|
if !p.bool() {
|
||||||
|
return orig
|
||||||
|
}
|
||||||
|
var rat big.Rat
|
||||||
|
rat.SetString(p.string())
|
||||||
|
return constant.Make(&rat)
|
||||||
|
}
|
||||||
|
|
||||||
func (r *importReader) ident(selector bool) *types.Sym {
|
func (r *importReader) ident(selector bool) *types.Sym {
|
||||||
name := r.string()
|
name := r.string()
|
||||||
if name == "" {
|
if name == "" {
|
||||||
|
|
@ -641,7 +652,19 @@ func (r *importReader) byte() byte {
|
||||||
|
|
||||||
// Compiler-specific extensions.
|
// Compiler-specific extensions.
|
||||||
|
|
||||||
func (r *importReader) varExt(n ir.Node) {
|
func (r *importReader) constExt(n *ir.Name) {
|
||||||
|
switch n.Type() {
|
||||||
|
case types.UntypedFloat:
|
||||||
|
n.SetVal(r.mprat(n.Val()))
|
||||||
|
case types.UntypedComplex:
|
||||||
|
v := n.Val()
|
||||||
|
re := r.mprat(constant.Real(v))
|
||||||
|
im := r.mprat(constant.Imag(v))
|
||||||
|
n.SetVal(makeComplex(re, im))
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
func (r *importReader) varExt(n *ir.Name) {
|
||||||
r.linkname(n.Sym())
|
r.linkname(n.Sym())
|
||||||
r.symIdx(n.Sym())
|
r.symIdx(n.Sym())
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -289,11 +289,14 @@ func ascompatee(op ir.Op, nl, nr []ir.Node) []ir.Node {
|
||||||
}
|
}
|
||||||
|
|
||||||
var assigned ir.NameSet
|
var assigned ir.NameSet
|
||||||
var memWrite bool
|
var memWrite, deferResultWrite bool
|
||||||
|
|
||||||
// affected reports whether expression n could be affected by
|
// affected reports whether expression n could be affected by
|
||||||
// the assignments applied so far.
|
// the assignments applied so far.
|
||||||
affected := func(n ir.Node) bool {
|
affected := func(n ir.Node) bool {
|
||||||
|
if deferResultWrite {
|
||||||
|
return true
|
||||||
|
}
|
||||||
return ir.Any(n, func(n ir.Node) bool {
|
return ir.Any(n, func(n ir.Node) bool {
|
||||||
if n.Op() == ir.ONAME && assigned.Has(n.(*ir.Name)) {
|
if n.Op() == ir.ONAME && assigned.Has(n.(*ir.Name)) {
|
||||||
return true
|
return true
|
||||||
|
|
@ -369,21 +372,40 @@ func ascompatee(op ir.Op, nl, nr []ir.Node) []ir.Node {
|
||||||
|
|
||||||
appendWalkStmt(&late, convas(ir.NewAssignStmt(base.Pos, lorig, r), &late))
|
appendWalkStmt(&late, convas(ir.NewAssignStmt(base.Pos, lorig, r), &late))
|
||||||
|
|
||||||
if name != nil && ir.IsBlank(name) {
|
// Check for reasons why we may need to compute later expressions
|
||||||
// We can ignore assignments to blank.
|
// before this assignment happens.
|
||||||
continue
|
|
||||||
}
|
if name == nil {
|
||||||
if op == ir.ORETURN && types.OrigSym(name.Sym()) == nil {
|
// Not a direct assignment to a declared variable.
|
||||||
// We can also ignore assignments to anonymous result
|
// Conservatively assume any memory access might alias.
|
||||||
// parameters. These can't appear in expressions anyway.
|
memWrite = true
|
||||||
continue
|
continue
|
||||||
}
|
}
|
||||||
|
|
||||||
if name != nil && name.OnStack() && !name.Addrtaken() {
|
if name.Class == ir.PPARAMOUT && ir.CurFunc.HasDefer() {
|
||||||
assigned.Add(name)
|
// Assignments to a result parameter in a function with defers
|
||||||
} else {
|
// becomes visible early if evaluation of any later expression
|
||||||
memWrite = true
|
// panics (#43835).
|
||||||
|
deferResultWrite = true
|
||||||
|
continue
|
||||||
}
|
}
|
||||||
|
|
||||||
|
if sym := types.OrigSym(name.Sym()); sym == nil || sym.IsBlank() {
|
||||||
|
// We can ignore assignments to blank or anonymous result parameters.
|
||||||
|
// These can't appear in expressions anyway.
|
||||||
|
continue
|
||||||
|
}
|
||||||
|
|
||||||
|
if name.Addrtaken() || !name.OnStack() {
|
||||||
|
// Global variable, heap escaped, or just addrtaken.
|
||||||
|
// Conservatively assume any memory access might alias.
|
||||||
|
memWrite = true
|
||||||
|
continue
|
||||||
|
}
|
||||||
|
|
||||||
|
// Local, non-addrtaken variable.
|
||||||
|
// Assignments can only alias with direct uses of this variable.
|
||||||
|
assigned.Add(name)
|
||||||
}
|
}
|
||||||
|
|
||||||
early.Append(late.Take()...)
|
early.Append(late.Take()...)
|
||||||
|
|
|
||||||
|
|
@ -57,9 +57,6 @@ type Builder struct {
|
||||||
id sync.Mutex
|
id sync.Mutex
|
||||||
toolIDCache map[string]string // tool name -> tool ID
|
toolIDCache map[string]string // tool name -> tool ID
|
||||||
buildIDCache map[string]string // file name -> build ID
|
buildIDCache map[string]string // file name -> build ID
|
||||||
|
|
||||||
cgoEnvOnce sync.Once
|
|
||||||
cgoEnvCache []string
|
|
||||||
}
|
}
|
||||||
|
|
||||||
// NOTE: Much of Action would not need to be exported if not for test.
|
// NOTE: Much of Action would not need to be exported if not for test.
|
||||||
|
|
|
||||||
|
|
@ -1165,7 +1165,10 @@ func (b *Builder) vet(ctx context.Context, a *Action) error {
|
||||||
}
|
}
|
||||||
|
|
||||||
// TODO(rsc): Why do we pass $GCCGO to go vet?
|
// TODO(rsc): Why do we pass $GCCGO to go vet?
|
||||||
env := b.cgoEnv()
|
env := b.cCompilerEnv()
|
||||||
|
if cfg.BuildToolchainName == "gccgo" {
|
||||||
|
env = append(env, "GCCGO="+BuildToolchain.compiler())
|
||||||
|
}
|
||||||
|
|
||||||
p := a.Package
|
p := a.Package
|
||||||
tool := VetTool
|
tool := VetTool
|
||||||
|
|
@ -2111,24 +2114,6 @@ func (b *Builder) cCompilerEnv() []string {
|
||||||
return []string{"TERM=dumb"}
|
return []string{"TERM=dumb"}
|
||||||
}
|
}
|
||||||
|
|
||||||
// cgoEnv returns environment variables to set when running cgo.
|
|
||||||
// Some of these pass through to cgo running the C compiler,
|
|
||||||
// so it includes cCompilerEnv.
|
|
||||||
func (b *Builder) cgoEnv() []string {
|
|
||||||
b.cgoEnvOnce.Do(func() {
|
|
||||||
cc, err := exec.LookPath(b.ccExe()[0])
|
|
||||||
if err != nil || filepath.Base(cc) == cc { // reject relative path
|
|
||||||
cc = "/missing-cc"
|
|
||||||
}
|
|
||||||
gccgo := GccgoBin
|
|
||||||
if filepath.Base(gccgo) == gccgo { // reject relative path
|
|
||||||
gccgo = "/missing-gccgo"
|
|
||||||
}
|
|
||||||
b.cgoEnvCache = append(b.cCompilerEnv(), "CC="+cc, "GCCGO="+gccgo)
|
|
||||||
})
|
|
||||||
return b.cgoEnvCache
|
|
||||||
}
|
|
||||||
|
|
||||||
// mkdir makes the named directory.
|
// mkdir makes the named directory.
|
||||||
func (b *Builder) Mkdir(dir string) error {
|
func (b *Builder) Mkdir(dir string) error {
|
||||||
// Make Mkdir(a.Objdir) a no-op instead of an error when a.Objdir == "".
|
// Make Mkdir(a.Objdir) a no-op instead of an error when a.Objdir == "".
|
||||||
|
|
@ -2729,7 +2714,7 @@ func (b *Builder) cgo(a *Action, cgoExe, objdir string, pcCFLAGS, pcLDFLAGS, cgo
|
||||||
// along to the host linker. At this point in the code, cgoLDFLAGS
|
// along to the host linker. At this point in the code, cgoLDFLAGS
|
||||||
// consists of the original $CGO_LDFLAGS (unchecked) and all the
|
// consists of the original $CGO_LDFLAGS (unchecked) and all the
|
||||||
// flags put together from source code (checked).
|
// flags put together from source code (checked).
|
||||||
cgoenv := b.cgoEnv()
|
cgoenv := b.cCompilerEnv()
|
||||||
if len(cgoLDFLAGS) > 0 {
|
if len(cgoLDFLAGS) > 0 {
|
||||||
flags := make([]string, len(cgoLDFLAGS))
|
flags := make([]string, len(cgoLDFLAGS))
|
||||||
for i, f := range cgoLDFLAGS {
|
for i, f := range cgoLDFLAGS {
|
||||||
|
|
@ -2966,7 +2951,7 @@ func (b *Builder) dynimport(a *Action, p *load.Package, objdir, importGo, cgoExe
|
||||||
if p.Standard && p.ImportPath == "runtime/cgo" {
|
if p.Standard && p.ImportPath == "runtime/cgo" {
|
||||||
cgoflags = []string{"-dynlinker"} // record path to dynamic linker
|
cgoflags = []string{"-dynlinker"} // record path to dynamic linker
|
||||||
}
|
}
|
||||||
return b.run(a, base.Cwd, p.ImportPath, b.cgoEnv(), cfg.BuildToolexec, cgoExe, "-dynpackage", p.Name, "-dynimport", dynobj, "-dynout", importGo, cgoflags)
|
return b.run(a, base.Cwd, p.ImportPath, b.cCompilerEnv(), cfg.BuildToolexec, cgoExe, "-dynpackage", p.Name, "-dynimport", dynobj, "-dynout", importGo, cgoflags)
|
||||||
}
|
}
|
||||||
|
|
||||||
// Run SWIG on all SWIG input files.
|
// Run SWIG on all SWIG input files.
|
||||||
|
|
|
||||||
|
|
@ -129,7 +129,11 @@ func (gcToolchain) gc(b *Builder, a *Action, archive string, importcfg, embedcfg
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
args := []interface{}{cfg.BuildToolexec, base.Tool("compile"), "-o", ofile, "-trimpath", a.trimpath(), gcflags, gcargs, "-D", p.Internal.LocalPrefix}
|
args := []interface{}{cfg.BuildToolexec, base.Tool("compile"), "-o", ofile, "-trimpath", a.trimpath(), gcflags, gcargs}
|
||||||
|
if p.Internal.LocalPrefix != "" {
|
||||||
|
// Workaround #43883.
|
||||||
|
args = append(args, "-D", p.Internal.LocalPrefix)
|
||||||
|
}
|
||||||
if importcfg != nil {
|
if importcfg != nil {
|
||||||
if err := b.writeFile(objdir+"importcfg", importcfg); err != nil {
|
if err := b.writeFile(objdir+"importcfg", importcfg); err != nil {
|
||||||
return "", nil, err
|
return "", nil, err
|
||||||
|
|
|
||||||
12
src/cmd/go/testdata/script/cgo_path.txt
vendored
12
src/cmd/go/testdata/script/cgo_path.txt
vendored
|
|
@ -1,12 +1,20 @@
|
||||||
[!cgo] skip
|
[!cgo] skip
|
||||||
|
|
||||||
|
# Set CC explicitly to something that requires a PATH lookup.
|
||||||
|
# Normally, the default is gcc or clang, but if CC was set during make.bash,
|
||||||
|
# that becomes the default.
|
||||||
|
[exec:clang] env CC=clang
|
||||||
|
[exec:gcc] env CC=gcc
|
||||||
|
[!exec:clang] [!exec:gcc] skip 'Unknown C compiler'
|
||||||
|
|
||||||
env GOCACHE=$WORK/gocache # Looking for compile flags, so need a clean cache.
|
env GOCACHE=$WORK/gocache # Looking for compile flags, so need a clean cache.
|
||||||
[!windows] env PATH=.:$PATH
|
[!windows] env PATH=.:$PATH
|
||||||
[!windows] chmod 0777 p/gcc p/clang
|
[!windows] chmod 0755 p/gcc p/clang
|
||||||
[!windows] exists -exec p/gcc p/clang
|
[!windows] exists -exec p/gcc p/clang
|
||||||
[windows] exists -exec p/gcc.bat p/clang.bat
|
[windows] exists -exec p/gcc.bat p/clang.bat
|
||||||
! exists p/bug.txt
|
! exists p/bug.txt
|
||||||
go build -x
|
! go build -x
|
||||||
|
stderr '^cgo: exec (clang|gcc): (clang|gcc) resolves to executable relative to current directory \(.[/\\](clang|gcc)(.bat)?\)$'
|
||||||
! exists p/bug.txt
|
! exists p/bug.txt
|
||||||
|
|
||||||
-- go.mod --
|
-- go.mod --
|
||||||
|
|
|
||||||
56
src/cmd/go/testdata/script/cgo_path_space.txt
vendored
Normal file
56
src/cmd/go/testdata/script/cgo_path_space.txt
vendored
Normal file
|
|
@ -0,0 +1,56 @@
|
||||||
|
# Check that if the PATH directory containing the C compiler has a space,
|
||||||
|
# we can still use that compiler with cgo.
|
||||||
|
# Verifies #43808.
|
||||||
|
[!cgo] skip
|
||||||
|
|
||||||
|
# Set CC explicitly to something that requires a PATH lookup.
|
||||||
|
# Normally, the default is gcc or clang, but if CC was set during make.bash,
|
||||||
|
# that becomes the default.
|
||||||
|
[exec:clang] env CC=clang
|
||||||
|
[exec:gcc] env CC=gcc
|
||||||
|
[!exec:clang] [!exec:gcc] skip 'Unknown C compiler'
|
||||||
|
|
||||||
|
[!windows] chmod 0755 $WORK/'program files'/clang
|
||||||
|
[!windows] chmod 0755 $WORK/'program files'/gcc
|
||||||
|
[!windows] exists -exec $WORK/'program files'/clang
|
||||||
|
[!windows] exists -exec $WORK/'program files'/gcc
|
||||||
|
[!windows] env PATH=$WORK/'program files':$PATH
|
||||||
|
[windows] exists -exec $WORK/'program files'/gcc.bat
|
||||||
|
[windows] exists -exec $WORK/'program files'/clang.bat
|
||||||
|
[windows] env PATH=$WORK\'program files';%PATH%
|
||||||
|
|
||||||
|
! exists $WORK/log.txt
|
||||||
|
? go build -x
|
||||||
|
exists $WORK/log.txt
|
||||||
|
rm $WORK/log.txt
|
||||||
|
|
||||||
|
# TODO(#41400, #43078): when CC is set explicitly, it should be allowed to
|
||||||
|
# contain spaces separating arguments, and it should be possible to quote
|
||||||
|
# arguments with spaces (including the path), as in CGO_CFLAGS and other
|
||||||
|
# variables. For now, this doesn't work.
|
||||||
|
[!windows] env CC=$WORK/'program files'/gcc
|
||||||
|
[windows] env CC=$WORK\'program files'\gcc.bat
|
||||||
|
! go build -x
|
||||||
|
! exists $WORK/log.txt
|
||||||
|
|
||||||
|
-- go.mod --
|
||||||
|
module m
|
||||||
|
|
||||||
|
-- m.go --
|
||||||
|
package m
|
||||||
|
|
||||||
|
// #define X 1
|
||||||
|
import "C"
|
||||||
|
|
||||||
|
-- $WORK/program files/gcc --
|
||||||
|
#!/bin/sh
|
||||||
|
|
||||||
|
echo ok >$WORK/log.txt
|
||||||
|
-- $WORK/program files/clang --
|
||||||
|
#!/bin/sh
|
||||||
|
|
||||||
|
echo ok >$WORK/log.txt
|
||||||
|
-- $WORK/program files/gcc.bat --
|
||||||
|
echo ok >%WORK%\log.txt
|
||||||
|
-- $WORK/program files/clang.bat --
|
||||||
|
echo ok >%WORK%\log.txt
|
||||||
|
|
@ -280,7 +280,7 @@ func MOVCONST(d int64, s int, rt int) uint32 {
|
||||||
const (
|
const (
|
||||||
// Optab.flag
|
// Optab.flag
|
||||||
LFROM = 1 << 0 // p.From uses constant pool
|
LFROM = 1 << 0 // p.From uses constant pool
|
||||||
LFROM3 = 1 << 1 // p.From3 uses constant pool
|
LFROM128 = 1 << 1 // p.From3<<64+p.From forms a 128-bit constant in literal pool
|
||||||
LTO = 1 << 2 // p.To uses constant pool
|
LTO = 1 << 2 // p.To uses constant pool
|
||||||
NOTUSETMP = 1 << 3 // p expands to multiple instructions, but does NOT use REGTMP
|
NOTUSETMP = 1 << 3 // p expands to multiple instructions, but does NOT use REGTMP
|
||||||
)
|
)
|
||||||
|
|
@ -419,7 +419,7 @@ var optab = []Optab{
|
||||||
{AMOVD, C_LACON, C_NONE, C_NONE, C_RSP, 34, 8, REGSP, LFROM, 0},
|
{AMOVD, C_LACON, C_NONE, C_NONE, C_RSP, 34, 8, REGSP, LFROM, 0},
|
||||||
|
|
||||||
// Move a large constant to a vector register.
|
// Move a large constant to a vector register.
|
||||||
{AVMOVQ, C_VCON, C_NONE, C_VCON, C_VREG, 101, 4, 0, LFROM | LFROM3, 0},
|
{AVMOVQ, C_VCON, C_NONE, C_VCON, C_VREG, 101, 4, 0, LFROM128, 0},
|
||||||
{AVMOVD, C_VCON, C_NONE, C_NONE, C_VREG, 101, 4, 0, LFROM, 0},
|
{AVMOVD, C_VCON, C_NONE, C_NONE, C_VREG, 101, 4, 0, LFROM, 0},
|
||||||
{AVMOVS, C_LCON, C_NONE, C_NONE, C_VREG, 101, 4, 0, LFROM, 0},
|
{AVMOVS, C_LCON, C_NONE, C_NONE, C_VREG, 101, 4, 0, LFROM, 0},
|
||||||
|
|
||||||
|
|
@ -995,8 +995,8 @@ func span7(ctxt *obj.Link, cursym *obj.LSym, newprog obj.ProgAlloc) {
|
||||||
if o.flag&LFROM != 0 {
|
if o.flag&LFROM != 0 {
|
||||||
c.addpool(p, &p.From)
|
c.addpool(p, &p.From)
|
||||||
}
|
}
|
||||||
if o.flag&LFROM3 != 0 {
|
if o.flag&LFROM128 != 0 {
|
||||||
c.addpool(p, p.GetFrom3())
|
c.addpool128(p, &p.From, p.GetFrom3())
|
||||||
}
|
}
|
||||||
if o.flag<O != 0 {
|
if o.flag<O != 0 {
|
||||||
c.addpool(p, &p.To)
|
c.addpool(p, &p.To)
|
||||||
|
|
@ -1201,6 +1201,36 @@ func (c *ctxt7) flushpool(p *obj.Prog, skip int) {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// addpool128 adds a 128-bit constant to literal pool by two consecutive DWORD
|
||||||
|
// instructions, the 128-bit constant is formed by ah.Offset<<64+al.Offset.
|
||||||
|
func (c *ctxt7) addpool128(p *obj.Prog, al, ah *obj.Addr) {
|
||||||
|
lit := al.Offset
|
||||||
|
q := c.newprog()
|
||||||
|
q.As = ADWORD
|
||||||
|
q.To.Type = obj.TYPE_CONST
|
||||||
|
q.To.Offset = lit
|
||||||
|
q.Pc = int64(c.pool.size)
|
||||||
|
|
||||||
|
lit = ah.Offset
|
||||||
|
t := c.newprog()
|
||||||
|
t.As = ADWORD
|
||||||
|
t.To.Type = obj.TYPE_CONST
|
||||||
|
t.To.Offset = lit
|
||||||
|
t.Pc = int64(c.pool.size + 8)
|
||||||
|
q.Link = t
|
||||||
|
|
||||||
|
if c.blitrl == nil {
|
||||||
|
c.blitrl = q
|
||||||
|
c.pool.start = uint32(p.Pc)
|
||||||
|
} else {
|
||||||
|
c.elitrl.Link = q
|
||||||
|
}
|
||||||
|
|
||||||
|
c.elitrl = t
|
||||||
|
c.pool.size += 16
|
||||||
|
p.Pool = q
|
||||||
|
}
|
||||||
|
|
||||||
/*
|
/*
|
||||||
* MOVD foo(SB), R is actually
|
* MOVD foo(SB), R is actually
|
||||||
* MOVD addr, REGTMP
|
* MOVD addr, REGTMP
|
||||||
|
|
|
||||||
|
|
@ -47,7 +47,7 @@ func TestLarge(t *testing.T) {
|
||||||
|
|
||||||
// assemble generated file
|
// assemble generated file
|
||||||
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-S", "-o", filepath.Join(dir, "test.o"), tmpfile)
|
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-S", "-o", filepath.Join(dir, "test.o"), tmpfile)
|
||||||
cmd.Env = append(os.Environ(), "GOARCH=arm64", "GOOS=linux")
|
cmd.Env = append(os.Environ(), "GOOS=linux")
|
||||||
out, err := cmd.CombinedOutput()
|
out, err := cmd.CombinedOutput()
|
||||||
if err != nil {
|
if err != nil {
|
||||||
t.Errorf("Assemble failed: %v, output: %s", err, out)
|
t.Errorf("Assemble failed: %v, output: %s", err, out)
|
||||||
|
|
@ -62,7 +62,7 @@ func TestLarge(t *testing.T) {
|
||||||
|
|
||||||
// build generated file
|
// build generated file
|
||||||
cmd = exec.Command(testenv.GoToolPath(t), "tool", "asm", "-o", filepath.Join(dir, "x.o"), tmpfile)
|
cmd = exec.Command(testenv.GoToolPath(t), "tool", "asm", "-o", filepath.Join(dir, "x.o"), tmpfile)
|
||||||
cmd.Env = append(os.Environ(), "GOARCH=arm64", "GOOS=linux")
|
cmd.Env = append(os.Environ(), "GOOS=linux")
|
||||||
out, err = cmd.CombinedOutput()
|
out, err = cmd.CombinedOutput()
|
||||||
if err != nil {
|
if err != nil {
|
||||||
t.Errorf("Build failed: %v, output: %s", err, out)
|
t.Errorf("Build failed: %v, output: %s", err, out)
|
||||||
|
|
@ -96,7 +96,7 @@ func TestNoRet(t *testing.T) {
|
||||||
t.Fatal(err)
|
t.Fatal(err)
|
||||||
}
|
}
|
||||||
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-o", filepath.Join(dir, "x.o"), tmpfile)
|
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-o", filepath.Join(dir, "x.o"), tmpfile)
|
||||||
cmd.Env = append(os.Environ(), "GOARCH=arm64", "GOOS=linux")
|
cmd.Env = append(os.Environ(), "GOOS=linux")
|
||||||
if out, err := cmd.CombinedOutput(); err != nil {
|
if out, err := cmd.CombinedOutput(); err != nil {
|
||||||
t.Errorf("%v\n%s", err, out)
|
t.Errorf("%v\n%s", err, out)
|
||||||
}
|
}
|
||||||
|
|
@ -134,7 +134,7 @@ func TestPCALIGN(t *testing.T) {
|
||||||
t.Fatal(err)
|
t.Fatal(err)
|
||||||
}
|
}
|
||||||
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-S", "-o", tmpout, tmpfile)
|
cmd := exec.Command(testenv.GoToolPath(t), "tool", "asm", "-S", "-o", tmpout, tmpfile)
|
||||||
cmd.Env = append(os.Environ(), "GOARCH=arm64", "GOOS=linux")
|
cmd.Env = append(os.Environ(), "GOOS=linux")
|
||||||
out, err := cmd.CombinedOutput()
|
out, err := cmd.CombinedOutput()
|
||||||
if err != nil {
|
if err != nil {
|
||||||
t.Errorf("The %s build failed: %v, output: %s", test.name, err, out)
|
t.Errorf("The %s build failed: %v, output: %s", test.name, err, out)
|
||||||
|
|
@ -150,3 +150,13 @@ func TestPCALIGN(t *testing.T) {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
func testvmovq() (r1, r2 uint64)
|
||||||
|
|
||||||
|
// TestVMOVQ checks if the arm64 VMOVQ instruction is working properly.
|
||||||
|
func TestVMOVQ(t *testing.T) {
|
||||||
|
a, b := testvmovq()
|
||||||
|
if a != 0x7040201008040201 || b != 0x3040201008040201 {
|
||||||
|
t.Errorf("TestVMOVQ got: a=0x%x, b=0x%x, want: a=0x7040201008040201, b=0x3040201008040201", a, b)
|
||||||
|
}
|
||||||
|
}
|
||||||
14
src/cmd/internal/obj/arm64/asm_arm64_test.s
Normal file
14
src/cmd/internal/obj/arm64/asm_arm64_test.s
Normal file
|
|
@ -0,0 +1,14 @@
|
||||||
|
// Copyright 2021 The Go Authors. All rights reserved.
|
||||||
|
// Use of this source code is governed by a BSD-style
|
||||||
|
// license that can be found in the LICENSE file.
|
||||||
|
|
||||||
|
#include "textflag.h"
|
||||||
|
|
||||||
|
// testvmovq() (r1, r2 uint64)
|
||||||
|
TEXT ·testvmovq(SB), NOSPLIT, $0-16
|
||||||
|
VMOVQ $0x7040201008040201, $0x3040201008040201, V1
|
||||||
|
VMOV V1.D[0], R0
|
||||||
|
VMOV V1.D[1], R1
|
||||||
|
MOVD R0, r1+0(FP)
|
||||||
|
MOVD R1, r2+8(FP)
|
||||||
|
RET
|
||||||
|
|
@ -1720,6 +1720,8 @@ var v = "v";
|
||||||
`
|
`
|
||||||
|
|
||||||
func TestEscapeRace(t *testing.T) {
|
func TestEscapeRace(t *testing.T) {
|
||||||
|
t.Skip("this test currently fails with -race; see issue #39807")
|
||||||
|
|
||||||
tmpl := New("")
|
tmpl := New("")
|
||||||
_, err := tmpl.New("templ.html").Parse(raceText)
|
_, err := tmpl.New("templ.html").Parse(raceText)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
|
|
@ -1777,6 +1779,39 @@ func TestRecursiveExecute(t *testing.T) {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// recursiveInvoker is for TestRecursiveExecuteViaMethod.
|
||||||
|
type recursiveInvoker struct {
|
||||||
|
t *testing.T
|
||||||
|
tmpl *Template
|
||||||
|
}
|
||||||
|
|
||||||
|
func (r *recursiveInvoker) Recur() (string, error) {
|
||||||
|
var sb strings.Builder
|
||||||
|
if err := r.tmpl.ExecuteTemplate(&sb, "subroutine", nil); err != nil {
|
||||||
|
r.t.Fatal(err)
|
||||||
|
}
|
||||||
|
return sb.String(), nil
|
||||||
|
}
|
||||||
|
|
||||||
|
func TestRecursiveExecuteViaMethod(t *testing.T) {
|
||||||
|
tmpl := New("")
|
||||||
|
top, err := tmpl.New("x.html").Parse(`{{.Recur}}`)
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
_, err = tmpl.New("subroutine").Parse(`<a href="/x?p={{"'a<b'"}}">`)
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
r := &recursiveInvoker{
|
||||||
|
t: t,
|
||||||
|
tmpl: tmpl,
|
||||||
|
}
|
||||||
|
if err := top.Execute(io.Discard, r); err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
// Issue 43295.
|
// Issue 43295.
|
||||||
func TestTemplateFuncsAfterClone(t *testing.T) {
|
func TestTemplateFuncsAfterClone(t *testing.T) {
|
||||||
s := `{{ f . }}`
|
s := `{{ f . }}`
|
||||||
|
|
|
||||||
|
|
@ -11,7 +11,6 @@ import (
|
||||||
"os"
|
"os"
|
||||||
"path"
|
"path"
|
||||||
"path/filepath"
|
"path/filepath"
|
||||||
"reflect"
|
|
||||||
"sync"
|
"sync"
|
||||||
"text/template"
|
"text/template"
|
||||||
"text/template/parse"
|
"text/template/parse"
|
||||||
|
|
@ -36,20 +35,18 @@ var escapeOK = fmt.Errorf("template escaped correctly")
|
||||||
|
|
||||||
// nameSpace is the data structure shared by all templates in an association.
|
// nameSpace is the data structure shared by all templates in an association.
|
||||||
type nameSpace struct {
|
type nameSpace struct {
|
||||||
mu sync.RWMutex
|
mu sync.Mutex
|
||||||
set map[string]*Template
|
set map[string]*Template
|
||||||
escaped bool
|
escaped bool
|
||||||
esc escaper
|
esc escaper
|
||||||
// The original functions, before wrapping.
|
|
||||||
funcMap FuncMap
|
|
||||||
}
|
}
|
||||||
|
|
||||||
// Templates returns a slice of the templates associated with t, including t
|
// Templates returns a slice of the templates associated with t, including t
|
||||||
// itself.
|
// itself.
|
||||||
func (t *Template) Templates() []*Template {
|
func (t *Template) Templates() []*Template {
|
||||||
ns := t.nameSpace
|
ns := t.nameSpace
|
||||||
ns.mu.RLock()
|
ns.mu.Lock()
|
||||||
defer ns.mu.RUnlock()
|
defer ns.mu.Unlock()
|
||||||
// Return a slice so we don't expose the map.
|
// Return a slice so we don't expose the map.
|
||||||
m := make([]*Template, 0, len(ns.set))
|
m := make([]*Template, 0, len(ns.set))
|
||||||
for _, v := range ns.set {
|
for _, v := range ns.set {
|
||||||
|
|
@ -87,8 +84,8 @@ func (t *Template) checkCanParse() error {
|
||||||
if t == nil {
|
if t == nil {
|
||||||
return nil
|
return nil
|
||||||
}
|
}
|
||||||
t.nameSpace.mu.RLock()
|
t.nameSpace.mu.Lock()
|
||||||
defer t.nameSpace.mu.RUnlock()
|
defer t.nameSpace.mu.Unlock()
|
||||||
if t.nameSpace.escaped {
|
if t.nameSpace.escaped {
|
||||||
return fmt.Errorf("html/template: cannot Parse after Execute")
|
return fmt.Errorf("html/template: cannot Parse after Execute")
|
||||||
}
|
}
|
||||||
|
|
@ -97,16 +94,6 @@ func (t *Template) checkCanParse() error {
|
||||||
|
|
||||||
// escape escapes all associated templates.
|
// escape escapes all associated templates.
|
||||||
func (t *Template) escape() error {
|
func (t *Template) escape() error {
|
||||||
t.nameSpace.mu.RLock()
|
|
||||||
escapeErr := t.escapeErr
|
|
||||||
t.nameSpace.mu.RUnlock()
|
|
||||||
if escapeErr != nil {
|
|
||||||
if escapeErr == escapeOK {
|
|
||||||
return nil
|
|
||||||
}
|
|
||||||
return escapeErr
|
|
||||||
}
|
|
||||||
|
|
||||||
t.nameSpace.mu.Lock()
|
t.nameSpace.mu.Lock()
|
||||||
defer t.nameSpace.mu.Unlock()
|
defer t.nameSpace.mu.Unlock()
|
||||||
t.nameSpace.escaped = true
|
t.nameSpace.escaped = true
|
||||||
|
|
@ -134,8 +121,6 @@ func (t *Template) Execute(wr io.Writer, data interface{}) error {
|
||||||
if err := t.escape(); err != nil {
|
if err := t.escape(); err != nil {
|
||||||
return err
|
return err
|
||||||
}
|
}
|
||||||
t.nameSpace.mu.RLock()
|
|
||||||
defer t.nameSpace.mu.RUnlock()
|
|
||||||
return t.text.Execute(wr, data)
|
return t.text.Execute(wr, data)
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -151,8 +136,6 @@ func (t *Template) ExecuteTemplate(wr io.Writer, name string, data interface{})
|
||||||
if err != nil {
|
if err != nil {
|
||||||
return err
|
return err
|
||||||
}
|
}
|
||||||
t.nameSpace.mu.RLock()
|
|
||||||
defer t.nameSpace.mu.RUnlock()
|
|
||||||
return tmpl.text.Execute(wr, data)
|
return tmpl.text.Execute(wr, data)
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -160,27 +143,13 @@ func (t *Template) ExecuteTemplate(wr io.Writer, name string, data interface{})
|
||||||
// is escaped, or returns an error if it cannot be. It returns the named
|
// is escaped, or returns an error if it cannot be. It returns the named
|
||||||
// template.
|
// template.
|
||||||
func (t *Template) lookupAndEscapeTemplate(name string) (tmpl *Template, err error) {
|
func (t *Template) lookupAndEscapeTemplate(name string) (tmpl *Template, err error) {
|
||||||
t.nameSpace.mu.RLock()
|
|
||||||
tmpl = t.set[name]
|
|
||||||
var escapeErr error
|
|
||||||
if tmpl != nil {
|
|
||||||
escapeErr = tmpl.escapeErr
|
|
||||||
}
|
|
||||||
t.nameSpace.mu.RUnlock()
|
|
||||||
|
|
||||||
if tmpl == nil {
|
|
||||||
return nil, fmt.Errorf("html/template: %q is undefined", name)
|
|
||||||
}
|
|
||||||
if escapeErr != nil {
|
|
||||||
if escapeErr != escapeOK {
|
|
||||||
return nil, escapeErr
|
|
||||||
}
|
|
||||||
return tmpl, nil
|
|
||||||
}
|
|
||||||
|
|
||||||
t.nameSpace.mu.Lock()
|
t.nameSpace.mu.Lock()
|
||||||
defer t.nameSpace.mu.Unlock()
|
defer t.nameSpace.mu.Unlock()
|
||||||
t.nameSpace.escaped = true
|
t.nameSpace.escaped = true
|
||||||
|
tmpl = t.set[name]
|
||||||
|
if tmpl == nil {
|
||||||
|
return nil, fmt.Errorf("html/template: %q is undefined", name)
|
||||||
|
}
|
||||||
if tmpl.escapeErr != nil && tmpl.escapeErr != escapeOK {
|
if tmpl.escapeErr != nil && tmpl.escapeErr != escapeOK {
|
||||||
return nil, tmpl.escapeErr
|
return nil, tmpl.escapeErr
|
||||||
}
|
}
|
||||||
|
|
@ -286,13 +255,6 @@ func (t *Template) Clone() (*Template, error) {
|
||||||
}
|
}
|
||||||
ns := &nameSpace{set: make(map[string]*Template)}
|
ns := &nameSpace{set: make(map[string]*Template)}
|
||||||
ns.esc = makeEscaper(ns)
|
ns.esc = makeEscaper(ns)
|
||||||
if t.nameSpace.funcMap != nil {
|
|
||||||
ns.funcMap = make(FuncMap, len(t.nameSpace.funcMap))
|
|
||||||
for name, fn := range t.nameSpace.funcMap {
|
|
||||||
ns.funcMap[name] = fn
|
|
||||||
}
|
|
||||||
}
|
|
||||||
wrapFuncs(ns, textClone, ns.funcMap)
|
|
||||||
ret := &Template{
|
ret := &Template{
|
||||||
nil,
|
nil,
|
||||||
textClone,
|
textClone,
|
||||||
|
|
@ -307,13 +269,12 @@ func (t *Template) Clone() (*Template, error) {
|
||||||
return nil, fmt.Errorf("html/template: cannot Clone %q after it has executed", t.Name())
|
return nil, fmt.Errorf("html/template: cannot Clone %q after it has executed", t.Name())
|
||||||
}
|
}
|
||||||
x.Tree = x.Tree.Copy()
|
x.Tree = x.Tree.Copy()
|
||||||
tc := &Template{
|
ret.set[name] = &Template{
|
||||||
nil,
|
nil,
|
||||||
x,
|
x,
|
||||||
x.Tree,
|
x.Tree,
|
||||||
ret.nameSpace,
|
ret.nameSpace,
|
||||||
}
|
}
|
||||||
ret.set[name] = tc
|
|
||||||
}
|
}
|
||||||
// Return the template associated with the name of this template.
|
// Return the template associated with the name of this template.
|
||||||
return ret.set[ret.Name()], nil
|
return ret.set[ret.Name()], nil
|
||||||
|
|
@ -382,43 +343,10 @@ type FuncMap map[string]interface{}
|
||||||
// type. However, it is legal to overwrite elements of the map. The return
|
// type. However, it is legal to overwrite elements of the map. The return
|
||||||
// value is the template, so calls can be chained.
|
// value is the template, so calls can be chained.
|
||||||
func (t *Template) Funcs(funcMap FuncMap) *Template {
|
func (t *Template) Funcs(funcMap FuncMap) *Template {
|
||||||
t.nameSpace.mu.Lock()
|
t.text.Funcs(template.FuncMap(funcMap))
|
||||||
if t.nameSpace.funcMap == nil {
|
|
||||||
t.nameSpace.funcMap = make(FuncMap, len(funcMap))
|
|
||||||
}
|
|
||||||
for name, fn := range funcMap {
|
|
||||||
t.nameSpace.funcMap[name] = fn
|
|
||||||
}
|
|
||||||
t.nameSpace.mu.Unlock()
|
|
||||||
|
|
||||||
wrapFuncs(t.nameSpace, t.text, funcMap)
|
|
||||||
return t
|
return t
|
||||||
}
|
}
|
||||||
|
|
||||||
// wrapFuncs records the functions with text/template. We wrap them to
|
|
||||||
// unlock the nameSpace. See TestRecursiveExecute for a test case.
|
|
||||||
func wrapFuncs(ns *nameSpace, textTemplate *template.Template, funcMap FuncMap) {
|
|
||||||
if len(funcMap) == 0 {
|
|
||||||
return
|
|
||||||
}
|
|
||||||
tfuncs := make(template.FuncMap, len(funcMap))
|
|
||||||
for name, fn := range funcMap {
|
|
||||||
fnv := reflect.ValueOf(fn)
|
|
||||||
wrapper := func(args []reflect.Value) []reflect.Value {
|
|
||||||
ns.mu.RUnlock()
|
|
||||||
defer ns.mu.RLock()
|
|
||||||
if fnv.Type().IsVariadic() {
|
|
||||||
return fnv.CallSlice(args)
|
|
||||||
} else {
|
|
||||||
return fnv.Call(args)
|
|
||||||
}
|
|
||||||
}
|
|
||||||
wrapped := reflect.MakeFunc(fnv.Type(), wrapper)
|
|
||||||
tfuncs[name] = wrapped.Interface()
|
|
||||||
}
|
|
||||||
textTemplate.Funcs(tfuncs)
|
|
||||||
}
|
|
||||||
|
|
||||||
// Delims sets the action delimiters to the specified strings, to be used in
|
// Delims sets the action delimiters to the specified strings, to be used in
|
||||||
// subsequent calls to Parse, ParseFiles, or ParseGlob. Nested template
|
// subsequent calls to Parse, ParseFiles, or ParseGlob. Nested template
|
||||||
// definitions will inherit the settings. An empty delimiter stands for the
|
// definitions will inherit the settings. An empty delimiter stands for the
|
||||||
|
|
@ -432,8 +360,8 @@ func (t *Template) Delims(left, right string) *Template {
|
||||||
// Lookup returns the template with the given name that is associated with t,
|
// Lookup returns the template with the given name that is associated with t,
|
||||||
// or nil if there is no such template.
|
// or nil if there is no such template.
|
||||||
func (t *Template) Lookup(name string) *Template {
|
func (t *Template) Lookup(name string) *Template {
|
||||||
t.nameSpace.mu.RLock()
|
t.nameSpace.mu.Lock()
|
||||||
defer t.nameSpace.mu.RUnlock()
|
defer t.nameSpace.mu.Unlock()
|
||||||
return t.set[name]
|
return t.set[name]
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -125,7 +125,7 @@ func ExampleReadFile() {
|
||||||
|
|
||||||
func ExampleWriteFile() {
|
func ExampleWriteFile() {
|
||||||
message := []byte("Hello, Gophers!")
|
message := []byte("Hello, Gophers!")
|
||||||
err := ioutil.WriteFile("testdata/hello", message, 0644)
|
err := ioutil.WriteFile("hello", message, 0644)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
log.Fatal(err)
|
log.Fatal(err)
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -11,6 +11,7 @@ import (
|
||||||
"fmt"
|
"fmt"
|
||||||
"internal/testenv"
|
"internal/testenv"
|
||||||
"io"
|
"io"
|
||||||
|
"io/fs"
|
||||||
"os"
|
"os"
|
||||||
. "os"
|
. "os"
|
||||||
osexec "os/exec"
|
osexec "os/exec"
|
||||||
|
|
@ -2689,6 +2690,32 @@ func TestOpenFileKeepsPermissions(t *testing.T) {
|
||||||
}
|
}
|
||||||
|
|
||||||
func TestDirFS(t *testing.T) {
|
func TestDirFS(t *testing.T) {
|
||||||
|
// On Windows, we force the MFT to update by reading the actual metadata from GetFileInformationByHandle and then
|
||||||
|
// explicitly setting that. Otherwise it might get out of sync with FindFirstFile. See golang.org/issues/42637.
|
||||||
|
if runtime.GOOS == "windows" {
|
||||||
|
if err := filepath.WalkDir("./testdata/dirfs", func(path string, d fs.DirEntry, err error) error {
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
info, err := d.Info()
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
stat, err := Stat(path) // This uses GetFileInformationByHandle internally.
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
if stat.ModTime() == info.ModTime() {
|
||||||
|
return nil
|
||||||
|
}
|
||||||
|
if err := Chtimes(path, stat.ModTime(), stat.ModTime()); err != nil {
|
||||||
|
t.Log(err) // We only log, not die, in case the test directory is not writable.
|
||||||
|
}
|
||||||
|
return nil
|
||||||
|
}); err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
}
|
||||||
if err := fstest.TestFS(DirFS("./testdata/dirfs"), "a", "b", "dir/x"); err != nil {
|
if err := fstest.TestFS(DirFS("./testdata/dirfs"), "a", "b", "dir/x"); err != nil {
|
||||||
t.Fatal(err)
|
t.Fatal(err)
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -16,7 +16,8 @@ const msanenabled = false
|
||||||
|
|
||||||
// Because msanenabled is false, none of these functions should be called.
|
// Because msanenabled is false, none of these functions should be called.
|
||||||
|
|
||||||
func msanread(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
func msanread(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
||||||
func msanwrite(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
func msanwrite(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
||||||
func msanmalloc(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
func msanmalloc(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
||||||
func msanfree(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
func msanfree(addr unsafe.Pointer, sz uintptr) { throw("msan") }
|
||||||
|
func msanmove(dst, src unsafe.Pointer, sz uintptr) { throw("msan") }
|
||||||
|
|
|
||||||
|
|
@ -1251,6 +1251,11 @@ func mstart() {
|
||||||
// Initialize stack bounds from system stack.
|
// Initialize stack bounds from system stack.
|
||||||
// Cgo may have left stack size in stack.hi.
|
// Cgo may have left stack size in stack.hi.
|
||||||
// minit may update the stack bounds.
|
// minit may update the stack bounds.
|
||||||
|
//
|
||||||
|
// Note: these bounds may not be very accurate.
|
||||||
|
// We set hi to &size, but there are things above
|
||||||
|
// it. The 1024 is supposed to compensate this,
|
||||||
|
// but is somewhat arbitrary.
|
||||||
size := _g_.stack.hi
|
size := _g_.stack.hi
|
||||||
if size == 0 {
|
if size == 0 {
|
||||||
size = 8192 * sys.StackGuardMultiplier
|
size = 8192 * sys.StackGuardMultiplier
|
||||||
|
|
|
||||||
|
|
@ -475,6 +475,14 @@ func adjustSignalStack(sig uint32, mp *m, gsigStack *gsignalStack) bool {
|
||||||
return false
|
return false
|
||||||
}
|
}
|
||||||
|
|
||||||
|
var st stackt
|
||||||
|
sigaltstack(nil, &st)
|
||||||
|
stsp := uintptr(unsafe.Pointer(st.ss_sp))
|
||||||
|
if st.ss_flags&_SS_DISABLE == 0 && sp >= stsp && sp < stsp+st.ss_size {
|
||||||
|
setGsignalStack(&st, gsigStack)
|
||||||
|
return true
|
||||||
|
}
|
||||||
|
|
||||||
if sp >= mp.g0.stack.lo && sp < mp.g0.stack.hi {
|
if sp >= mp.g0.stack.lo && sp < mp.g0.stack.hi {
|
||||||
// The signal was delivered on the g0 stack.
|
// The signal was delivered on the g0 stack.
|
||||||
// This can happen when linked with C code
|
// This can happen when linked with C code
|
||||||
|
|
@ -483,29 +491,25 @@ func adjustSignalStack(sig uint32, mp *m, gsigStack *gsignalStack) bool {
|
||||||
// the signal handler directly when C code,
|
// the signal handler directly when C code,
|
||||||
// including C code called via cgo, calls a
|
// including C code called via cgo, calls a
|
||||||
// TSAN-intercepted function such as malloc.
|
// TSAN-intercepted function such as malloc.
|
||||||
|
//
|
||||||
|
// We check this condition last as g0.stack.lo
|
||||||
|
// may be not very accurate (see mstart).
|
||||||
st := stackt{ss_size: mp.g0.stack.hi - mp.g0.stack.lo}
|
st := stackt{ss_size: mp.g0.stack.hi - mp.g0.stack.lo}
|
||||||
setSignalstackSP(&st, mp.g0.stack.lo)
|
setSignalstackSP(&st, mp.g0.stack.lo)
|
||||||
setGsignalStack(&st, gsigStack)
|
setGsignalStack(&st, gsigStack)
|
||||||
return true
|
return true
|
||||||
}
|
}
|
||||||
|
|
||||||
var st stackt
|
// sp is not within gsignal stack, g0 stack, or sigaltstack. Bad.
|
||||||
sigaltstack(nil, &st)
|
setg(nil)
|
||||||
|
needm()
|
||||||
if st.ss_flags&_SS_DISABLE != 0 {
|
if st.ss_flags&_SS_DISABLE != 0 {
|
||||||
setg(nil)
|
|
||||||
needm()
|
|
||||||
noSignalStack(sig)
|
noSignalStack(sig)
|
||||||
dropm()
|
} else {
|
||||||
}
|
|
||||||
stsp := uintptr(unsafe.Pointer(st.ss_sp))
|
|
||||||
if sp < stsp || sp >= stsp+st.ss_size {
|
|
||||||
setg(nil)
|
|
||||||
needm()
|
|
||||||
sigNotOnStack(sig)
|
sigNotOnStack(sig)
|
||||||
dropm()
|
|
||||||
}
|
}
|
||||||
setGsignalStack(&st, gsigStack)
|
dropm()
|
||||||
return true
|
return false
|
||||||
}
|
}
|
||||||
|
|
||||||
// crashing is the number of m's we have waited for when implementing
|
// crashing is the number of m's we have waited for when implementing
|
||||||
|
|
|
||||||
|
|
@ -57,3 +57,4 @@ func pthread_create_trampoline()
|
||||||
//go:cgo_import_dynamic libc_pthread_sigmask pthread_sigmask "libpthread.so"
|
//go:cgo_import_dynamic libc_pthread_sigmask pthread_sigmask "libpthread.so"
|
||||||
|
|
||||||
//go:cgo_import_dynamic _ _ "libpthread.so"
|
//go:cgo_import_dynamic _ _ "libpthread.so"
|
||||||
|
//go:cgo_import_dynamic _ _ "libc.so"
|
||||||
|
|
|
||||||
File diff suppressed because it is too large
Load diff
45
test/fixedbugs/issue43835.go
Normal file
45
test/fixedbugs/issue43835.go
Normal file
|
|
@ -0,0 +1,45 @@
|
||||||
|
// run
|
||||||
|
|
||||||
|
// Copyright 2021 The Go Authors. All rights reserved.
|
||||||
|
// Use of this source code is governed by a BSD-style
|
||||||
|
// license that can be found in the LICENSE file.
|
||||||
|
|
||||||
|
package main
|
||||||
|
|
||||||
|
func main() {
|
||||||
|
if f() {
|
||||||
|
panic("FAIL")
|
||||||
|
}
|
||||||
|
if bad, _ := g(); bad {
|
||||||
|
panic("FAIL")
|
||||||
|
}
|
||||||
|
if bad, _ := h(); bad {
|
||||||
|
panic("FAIL")
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
func f() (bad bool) {
|
||||||
|
defer func() {
|
||||||
|
recover()
|
||||||
|
}()
|
||||||
|
var p *int
|
||||||
|
bad, _ = true, *p
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
func g() (bool, int) {
|
||||||
|
defer func() {
|
||||||
|
recover()
|
||||||
|
}()
|
||||||
|
var p *int
|
||||||
|
return true, *p
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
func h() (_ bool, _ int) {
|
||||||
|
defer func() {
|
||||||
|
recover()
|
||||||
|
}()
|
||||||
|
var p *int
|
||||||
|
return true, *p
|
||||||
|
}
|
||||||
|
|
@ -21,7 +21,7 @@ func main() {
|
||||||
var prec float64
|
var prec float64
|
||||||
switch runtime.Compiler {
|
switch runtime.Compiler {
|
||||||
case "gc":
|
case "gc":
|
||||||
prec = 512
|
prec = math.Inf(1) // exact precision using rational arithmetic
|
||||||
case "gccgo":
|
case "gccgo":
|
||||||
prec = 256
|
prec = 256
|
||||||
default:
|
default:
|
||||||
|
|
|
||||||
|
|
@ -37,12 +37,11 @@ var x = []interface{}{
|
||||||
|
|
||||||
// If the compiler's internal floating point representation
|
// If the compiler's internal floating point representation
|
||||||
// is shorter than 1024 bits, it cannot distinguish max64+ulp64/2-1 and max64+ulp64/2.
|
// is shorter than 1024 bits, it cannot distinguish max64+ulp64/2-1 and max64+ulp64/2.
|
||||||
// gc uses fewer than 1024 bits, so allow it to print the overflow error for the -1 case.
|
|
||||||
float64(max64 + ulp64/2 - two1024/two256), // ok
|
float64(max64 + ulp64/2 - two1024/two256), // ok
|
||||||
float64(max64 + ulp64/2 - 1), // GC_ERROR "constant 1\.79769e\+308 overflows float64"
|
float64(max64 + ulp64/2 - 1), // ok
|
||||||
float64(max64 + ulp64/2), // ERROR "constant 1\.79769e\+308 overflows float64"
|
float64(max64 + ulp64/2), // ERROR "constant 1\.79769e\+308 overflows float64"
|
||||||
|
|
||||||
float64(-max64 - ulp64/2 + two1024/two256), // ok
|
float64(-max64 - ulp64/2 + two1024/two256), // ok
|
||||||
float64(-max64 - ulp64/2 + 1), // GC_ERROR "constant -1\.79769e\+308 overflows float64"
|
float64(-max64 - ulp64/2 + 1), // ok
|
||||||
float64(-max64 - ulp64/2), // ERROR "constant -1\.79769e\+308 overflows float64"
|
float64(-max64 - ulp64/2), // ERROR "constant -1\.79769e\+308 overflows float64"
|
||||||
}
|
}
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue