This will allow the compiler to crunch Prog lists down to code as each
function is compiled, instead of waiting until the end, which should
reduce the working set of the compiler. But not until Go 1.7.
This also makes it easier to write some machine code output tests
for the assembler, which is why it's being done now.
For #13822.
Change-Id: I0811123bc6e5717cebb8948f9cea18e1b9baf6f7
Reviewed-on: https://go-review.googlesource.com/18311
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Consider this code:
func f(*int)
func g() {
p := new(int)
f(p)
}
where f is an assembly function.
In general liveness analysis assumes that during the call to f, p is dead
in this frame. If f has retained p, p will be found alive in f's frame and keep
the new(int) from being garbage collected. This is all correct and works.
We use the Go func declaration for f to give the assembly function
liveness information (the arguments are assumed live for the entire call).
Now consider this code:
func h1() {
p := new(int)
syscall.Syscall(1, 2, 3, uintptr(unsafe.Pointer(p)))
}
Here syscall.Syscall is taking the place of f, but because its arguments
are uintptr, the liveness analysis and the garbage collector ignore them.
Since p is no longer live in h once the call starts, if the garbage collector
scans the stack while the system call is blocked, it will find no reference
to the new(int) and reclaim it. If the kernel is going to write to *p once
the call finishes, reclaiming the memory is a mistake.
We can't change the arguments or the liveness information for
syscall.Syscall itself, both for compatibility and because sometimes the
arguments really are integers, and the garbage collector will get quite upset
if it finds an integer where it expects a pointer. The problem is that
these arguments are fundamentally untyped.
The solution we have taken in the syscall package's wrappers in past
releases is to insert a call to a dummy function named "use", to make
it look like the argument is live during the call to syscall.Syscall:
func h2() {
p := new(int)
syscall.Syscall(1, 2, 3, uintptr(unsafe.Pointer(p)))
use(unsafe.Pointer(p))
}
Keeping p alive during the call means that if the garbage collector
scans the stack during the system call now, it will find the reference to p.
Unfortunately, this approach is not available to users outside syscall,
because 'use' is unexported, and people also have to realize they need
to use it and do so. There is much existing code using syscall.Syscall
without a 'use'-like function. That code will fail very occasionally in
mysterious ways (see #13372).
This CL fixes all that existing code by making the compiler do the right
thing automatically, without any code modifications. That is, it takes h1
above, which is incorrect code today, and makes it correct code.
Specifically, if the compiler sees a foreign func definition (one
without a body) that has uintptr arguments, it marks those arguments
as "unsafe uintptrs". If it later sees the function being called
with uintptr(unsafe.Pointer(x)) as an argument, it arranges to mark x
as having escaped, and it makes sure to hold x in a live temporary
variable until the call returns, so that the garbage collector cannot
reclaim whatever heap memory x points to.
For now I am leaving the explicit calls to use in package syscall,
but they can be removed early in a future cycle (likely Go 1.7).
The rule has no effect on escape analysis, only on liveness analysis.
Fixes#13372.
Change-Id: I2addb83f70d08db08c64d394f9d06ff0a063c500
Reviewed-on: https://go-review.googlesource.com/18584
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This requires changing the tls access code to match the patterns documented in
the ABI documentation or the system linker will "optimize" it into ridiculousness.
With this change, -buildmode=pie works, although as it is tested in testshared,
the tests are not run yet.
Change-Id: I1efa6687af0a5b8db3385b10f6542a49056b2eb3
Reviewed-on: https://go-review.googlesource.com/15971
Reviewed-by: Russ Cox <rsc@golang.org>
The PowerPC ISA does not have a PC-relative load instruction, which poses
obvious challenges when generating position-independent code. The way the ELFv2
ABI addresses this is to specify that r2 points to a per "module" (shared
library or executable) TOC pointer. Maintaining this pointer requires
cooperation between codegen and the system linker:
* Non-leaf functions leave space on the stack at r1+24 to save the TOC pointer.
* A call to a function that *might* have to go via a PLT stub must be followed
by a nop instruction that the system linker can replace with "ld r1, 24(r1)"
to restore the TOC pointer (only when dynamically linking Go code).
* When calling a function via a function pointer, the address of the function
must be in r12, and the first couple of instructions (the "global entry
point") of the called function use this to derive the address of the TOC
for the module it is in.
* When calling a function that is implemented in the same module, the system
linker adjusts the call to skip over the instructions mentioned above (the
"local entry point"), assuming that r2 is already correctly set.
So this changeset adds the global entry point instructions, sets the metadata so
the system linker knows where the local entry point is, inserts code to save the
TOC pointer at 24(r1), adds a nop after any call not known to be local and copes
with the odd non-local code transfer in the runtime (e.g. the stuff around
jmpdefer). It does not actually compile PIC yet.
Change-Id: I7522e22bdfd2f891745a900c60254fe9e372c854
Reviewed-on: https://go-review.googlesource.com/15967
Reviewed-by: Russ Cox <rsc@golang.org>
The larger stack frames causes the nosplit stack to overflow so the next change
increases the stackguard.
Change-Id: Ib2b4f24f0649eb1d13e3a58d265f13d1b6cc9bf9
Reviewed-on: https://go-review.googlesource.com/15964
Reviewed-by: Russ Cox <rsc@golang.org>
MIPS64 has 32 general purpose 64-bit integer registers (R0-R31), 32
64-bit floating point registers (F0-F31). Instructions are fixed-width,
and are 32-bit wide. Instructions are all in standard 1-, 2-, 3-operand
forms.
MIPS64-specific relocations are added. For this reason, test data of
cmd/newlink are regenerated.
No other changes are made to portable structures.
Branch delay slots are current filled with NOP instructions. The function
for instruction scheduling (try to fill the delay slot with a useful
instruction) is implemented but disabled for now.
Change-Id: Ic364999c7a33245260c1381fc26a2fa8972d38b3
Reviewed-on: https://go-review.googlesource.com/14442
Reviewed-by: Minux Ma <minux@golang.org>
On ppc64x, the thread pointer, held in R13, points 0x7000 bytes past where
thread-local storage begins (presumably to maximize the amount of storage that
can be accessed with a 16-bit signed displacement). The relocations used to
indicate thread-local storage to the platform linker account for this, so to be
able to support external linking we need to change things so the linker applies
this offset instead of the runtime assembly.
Change-Id: I2556c249ab2d802cae62c44b2b4c5b44787d7059
Reviewed-on: https://go-review.googlesource.com/14233
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Shared libraries on ppc64le will require a larger minimum stack frame (because
the ABI mandates that the TOC pointer is available at 24(R1)). Part 2a of
preparing for that is to have all bits of arch-independent and ppc64-specific
codegen that need to know call a function to find out.
Change-Id: I55899f73037e92227813c491049a3bd6f30bd41f
Reviewed-on: https://go-review.googlesource.com/15524
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Due to #9401, trailing empty fields will occupy at least 1 byte
of space.
Fixes#12884.
Change-Id: I838d3f1a73637e526f5a6dbc348981227d5bb2fd
Reviewed-on: https://go-review.googlesource.com/15660
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
obj.ProgInfo is a field inside obj.Prog, which is currently 320 bytes
on 64bit platforms. By moving the Flags field below the other fields
the size of obj.Prog drops into the 288 byte size class, a saving of
32 bytes per value allocated on the heap.
Change-Id: If8bb12f45328996d7df1d0bac9d1c019d2af73bd
Reviewed-on: https://go-review.googlesource.com/15522
Run-TryBot: Dave Cheney <dave@cheney.net>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
It's particularly nice to get rid of the android special cases in the linker.
Change-Id: I516363af7ce8a6b2f196fe49cb8887ac787a6dad
Reviewed-on: https://go-review.googlesource.com/14197
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Currently Go produces shared libraries that cannot be shared between processes
because they have relocations against the text segment (not text section). This
fixes this by moving some data to sections with magic names recognized by the
static linker.
The change in genasmsym to add STYPELINK to the switch should fix things on
darwin/arm64.
Fixes#10914
Updates #9210
Change-Id: Iab4a6678dd04cec6114e683caac5cf31b1063309
Reviewed-on: https://go-review.googlesource.com/14306
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This reverts commit 2c2cbb69c8.
Broke darwin/arm64
Change-Id: Ibd2dea475d6ce6a8b4b40e2da19a83fc0514025d
Reviewed-on: https://go-review.googlesource.com/14301
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Currently Go produces shared libraries that cannot be shared between processes
because they have relocations against the text segment (not text section). This
fixes this by moving some data to sections with magic names recognized by the
static linker.
Fixes#10914
Updates #9210
Change-Id: I7178daadc0ae87953d5a084aa3d580f4e3b46d47
Reviewed-on: https://go-review.googlesource.com/10300
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Nothing uses it any more.
Change-Id: I42ee7222b06b1a79b8b44894f3071752f9166d7a
Reviewed-on: https://go-review.googlesource.com/14193
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>
Same as we do for string symbols.
Fixes#11583.
Change-Id: Ia9264f6faf486697d987051b7f9851d37d8ad381
Reviewed-on: https://go-review.googlesource.com/12531
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
It looks like the test for whether symbols contain subsymbols is wrong.
In particular, symbols in C libraries are mistakenly considered container
symbols.
Fix the test so only symbols which actually have a subsymbol
are excluded from the symtab. When linking cgo programs the list
of containers is small, something like:
container _/home/khr/sandbox/symtab/misc/cgo/test(.text)<74>
container _/home/khr/sandbox/symtab/misc/cgo/test/issue8828(.text)<75>
container _/home/khr/sandbox/symtab/misc/cgo/test/issue9026(.text)<76>
container runtime/cgo(.text)<77>
I'm not sure this is the right fix. In particular I can't reproduce
the original problem. Anyone have a repro they can try and see if
this fix works?
Fixes#10747Fixes#11396
Change-Id: Id8b016389d33348b4a791fdcba0f9db8ae71ebf3
Reviewed-on: https://go-review.googlesource.com/11652
Reviewed-by: Russ Cox <rsc@golang.org>
They're each architecture-specific.
Let them share.
Reduces Prog size to 288, which is the
next smaller malloc class.
Reduces inuse_space while compiling the
rotate tests by ~3.2%.
Change-Id: Ica8ec90e466c97b569745fffff0e5acd364e55fa
Reviewed-on: https://go-review.googlesource.com/10514
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
That which cannot happen has not happened.
No immediate changes to Addr or Prog size.
Change-Id: I4cb9315f2c9f5f92eda340bfc4abb46395fa467f
Reviewed-on: https://go-review.googlesource.com/10513
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Printed and Width were unused.
Despite only removing two bytes, due to alignment, 8 bytes are saved
on 64-bit:
Before: unsafe.Sizeof(obj.Prog{}) == 304
After: unsafe.Sizeof(obj.Prog{}) == 296
The next size class below 320 (304=>19(320)) is 288. Still 8 bytes
away from that.
Change-Id: I8d1632dd40d387f7036c03c65ea4d64e9b6218c3
Reviewed-on: https://go-review.googlesource.com/10511
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
It is almost never set and Addr is large, so having the full struct
in the Prog wastes memory most of the time.
Before (on a 64-bit system):
$ sizeof -p cmd/internal/obj Addr Prog
Addr 80
Prog 376
$
After:
$ sizeof -p cmd/internal/obj Addr Prog
Addr 80
Prog 304
$
Change-Id: I491f201241f87543964a7d0f48b85830759be9d0
Reviewed-on: https://go-review.googlesource.com/10457
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
It shrinks Prog type from 448 bytes down to 376 bytes on amd64.
It also makes sense, because I don't know of any modern architecture
that have instructions which can write to two destinations, none of
which is a register (even x86 doesn't have such instructions).
Change-Id: I3061f1c9ac93d79ee2b92ecb9049641d0e0f6300
Reviewed-on: https://go-review.googlesource.com/10330
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
F3t was effectively a local variable.
Remove it.
This shrinks obj.Prog from 456 to 448 bytes,
which places it in a smaller malloc class.
This reduces the memory usage of the compiler
while compiling the rotate tests by ~2.75%.
Change-Id: I31cc9dd67269851a430b56bcc7d255c9349eb522
Reviewed-on: https://go-review.googlesource.com/10255
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This started out as trying to remove Bool2int calls, which it does a bit, but
mostly it ended up being removing the Link.Symmorestack array which seemed a
pointless bit of caching.
Change-Id: I91a51eb08cb4b08f3f9f093b575306499267b67a
Reviewed-on: https://go-review.googlesource.com/9239
Reviewed-by: Minux Ma <minux@golang.org>
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The long comment block in obj6.go:progedit talked about the two code sequences
for accessing g as "local exec" and "initial exec", but really they are both forms
of local exec. This stuff is confusing enough without using the wrong words for
things, so rewrite it to talk about 2-instruction and 1-instruction sequences.
Unfortunately the confusion has made it into code, with the R_TLS_IE relocation
now doing double duty as meaning actual initial exec when externally linking and
boring old local exec when linking internally (half of this is my fault). So this
stops using R_TLS_IE in the local exec case. There is a chance this might break
plan9 or windows, but I don't think so. Next step is working out what the heck is
going on on ARM...
Change-Id: I09da4388210cf49dbc99fd25f5172bbe517cee57
Reviewed-on: https://go-review.googlesource.com/9273
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
This lets us avoid loading string constants via the GOT and (together with
http://golang.org/cl/9102) results in the fannkuch benchmark having very similar
register usage with -dynlink as without.
Change-Id: Ic3892b399074982b76773c3e547cfbba5dabb6f9
Reviewed-on: https://go-review.googlesource.com/9103
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
They don't really make any sense on this side of the compiler/linker divide.
Some of the code touching these fields was the support for R_TLS when
thechar=='6' which turns out to be dead and so I just removed all of that.
Change-Id: I4e265613c4e7fcc30a965fffb7fd5f45017f06f3
Reviewed-on: https://go-review.googlesource.com/9107
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
http://golang.org/cl/7623 refactored how line history works and
introduced a new TrimPathPrefix field to replace the existing Trimpath
field, but never removed the latter or updated its users.
Fixes#10503.
Change-Id: Ief90a55b6cef2e8062b59856a4c7dcc0df01d3f2
Reviewed-on: https://go-review.googlesource.com/9113
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
This CL introduces R_ADDRARM64, which is similar to R_ADDRPOWER.
Fixes#10112.
Change-Id: I309e3df7608b9eef9339196fdc50dedf5f9439f1
Reviewed-on: https://go-review.googlesource.com/8438
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: David Crawshaw <crawshaw@golang.org>
A quick pass through link.go, mostly removing fields that are not
used on the "creating a single object file" side of the fence.
Change-Id: I35ba41378c2c418f7df2f2f88dce65bc64a1a45d
Reviewed-on: https://go-review.googlesource.com/7672
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The ProgInfo is loaded many times during each analysis pass.
Load it once at the beginning (in Flowstart if using that, or explicitly,
as in plive.go) and then refer to the cached copy.
Removes many calls to proginfo.
Makes Prog a little bigger, but the previous CL more than compensates.
Change-Id: If90a12fc6729878fdae10444f9c3bedc8d85026e
Reviewed-on: https://go-review.googlesource.com/7745
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
An interface{} is more in the spirit of the original union.
By my calculations, on 64-bit systems this reduces
Addr from 120 to 80 bytes, and Prog from 592 to 424 bytes.
Change-Id: I0d7b0981513c2a3c94c9ac76bb4f8816485b5a3c
Reviewed-on: https://go-review.googlesource.com/7744
Reviewed-by: Rob Pike <r@golang.org>
I think the file ended up in the order of the typedefs instead of the
order of the actual struct definitions. You can see where some of
the declarations were because some of the comments didn't move.
Put things back in the original order.
Change-Id: I0e3703008278b084b632c917cfb73bc81bdd4f23
Reviewed-on: https://go-review.googlesource.com/7743
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
In addition to possibly being clearer code,
this replaces an O(n) lookup with an O(log n) lookup.
Change-Id: I0a574c536a965a87f7ad6dcdcc30f737bc771cd5
Reviewed-on: https://go-review.googlesource.com/7623
Reviewed-by: Rob Pike <r@golang.org>
ARM64 (ARMv8) has 32 general purpose, 64-bit integer registers
(R0-R31), 32 64-bit scalar floating point registers (F0-F31), and
32 128-bit vector registers (unused, V0-V31).
R31 is either the stack pointer (RSP), or the zero register (ZR),
depending on the instruction. Note the distinction between the
hardware stack pointer, RSP, and the virtual stack pointer SP.
The (hardware) stack pointer must be 16-byte aligned at all times;
the RSP register itself must be aligned, offset(RSP) only has to
have natural alignment.
Instructions are fixed-width, and are 32-bit wide. ARM64 supports
ARMv7 too (32-bit ARM), but not in the same process. In general,
there is not much in common between 32-bit ARM and ARM64, it's a
new architecture.
All implementations have floating point instructions.
This change adds a Prog.To3 field analogous to Prog.To. It is used
by exclusive load/store instructions such as STLXR which read from
one register, and write to both a register and a memory address.
STLXRW R1, (R0), R3
This will store the word contained in R1 to the memory address
pointed by R0. R3 will be updated with the status result of the
store. It is used to implement atomic operations.
No other changes are made to the portable Prog and Addr structures.
Change-Id: Ie839029aa5265bbad35769d9689eca11e1c48c47
Reviewed-on: https://go-review.googlesource.com/7046
Reviewed-by: Russ Cox <rsc@golang.org>
