]> Cypherpunks repositories - gostls13.git/log
gostls13.git
4 years agocmd/internal/obj/asm64: add support for moving BITCON to RSP
eric fang [Mon, 1 Feb 2021 07:47:06 +0000 (07:47 +0000)]
cmd/internal/obj/asm64: add support for moving BITCON to RSP

Constant of BITCON type can be moved into RSP by MOVD or MOVW instructions
directly, this CL enables this format of these two instructions.

For 32-bit ADDWop instructions with constant, rewrite the high 32-bit
to be a repetition of the low 32-bit, just as ANDWop instructions do,
so that we can optimize ADDW $bitcon, Rn, Rt as:
MOVW $bitcon, Rtmp
ADDW Rtmp, Rn, Rt
The original code is:
MOVZ $bitcon_low, Rtmp
MOVK $bitcon_high,Rtmp
ADDW Rtmp, Rn, Rt

Change-Id: I30e71972bcfd6470a8b6e6ffbacaee79d523805a
Reviewed-on: https://go-review.googlesource.com/c/go/+/289649
Trust: eric fang <eric.fang@arm.com>
Run-TryBot: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: eric fang <eric.fang@arm.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/asm: add arm64 instructions VUMAX and VUMIN
eric fang [Thu, 19 Nov 2020 07:18:41 +0000 (07:18 +0000)]
cmd/asm: add arm64 instructions VUMAX and VUMIN

This CL adds support for arm64 fp&simd instructions VUMAX and VUMIN.
Fixes #42326

Change-Id: I3757ba165dc31ce1ce70f3b06a9e5b94c14d2ab9
Reviewed-on: https://go-review.googlesource.com/c/go/+/271497
Trust: eric fang <eric.fang@arm.com>
Run-TryBot: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: fannie zhang <Fannie.Zhang@arm.com>
Reviewed-by: eric fang <eric.fang@arm.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/asm: add 128-bit FLDPQ and FSTPQ instructions for arm64
eric fang [Mon, 23 Nov 2020 10:59:33 +0000 (10:59 +0000)]
cmd/asm: add 128-bit FLDPQ and FSTPQ instructions for arm64

This CL adds assembly support for 128-bit FLDPQ and FSTPQ instructions.

This CL also deletes some wrong pre/post-indexed LDP and STP instructions,
such as {ALDP, C_UAUTO4K, C_NONE, C_NONE, C_PAIR, 74, 8, REGSP, 0, C_XPRE},
because when the offset type is C_UAUTO4K, pre and post don't work.

Change-Id: Ifd901d4440eb06eb9e86c9dd17518749fdf32848
Reviewed-on: https://go-review.googlesource.com/c/go/+/273668
Trust: eric fang <eric.fang@arm.com>
Run-TryBot: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: eric fang <eric.fang@arm.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agogo/types: use correct recv for parameterized embedded methods
Rob Findley [Wed, 3 Mar 2021 18:33:24 +0000 (13:33 -0500)]
go/types: use correct recv for parameterized embedded methods

This is a direct port of CL 298129 to go/types.

Fixes #44688

Change-Id: I950992ea7beea5b9c8bea0c296b5ce03b2aa9b12
Reviewed-on: https://go-review.googlesource.com/c/go/+/298349
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
4 years agocmd/compile/internal/types2: use correct recv for parameterized embedded methods
Robert Griesemer [Wed, 3 Mar 2021 05:52:09 +0000 (21:52 -0800)]
cmd/compile/internal/types2: use correct recv for parameterized embedded methods

Methods of generic types are instantiated lazily (upon use). Thus,
when we encounter a method of such a type, we need to instantiate
the method signature with the receiver type arguments. We infer
those type arguments from the method receiver. If the method is
embedded, we must use the actual embedded receiver type, otherwise
the receiver type declared with the method doesn't match up and
inference will fail.

(Note that there's no type inference in the source code here, it's
only the implementation which uses the existing inference mechanism
to easily identify the actual type arguments. If the implementation
is correct, the inference will always succeed.)

Updates #44688.

Change-Id: Ie35b62bebaeaf42037f2ca00cf8bd34fec2ddd9c
Reviewed-on: https://go-review.googlesource.com/c/go/+/298129
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
4 years agocmd/compile: remove races introduced in abiutils field update
David Chase [Wed, 3 Mar 2021 04:39:12 +0000 (23:39 -0500)]
cmd/compile: remove races introduced in abiutils field update

This fix uses mutex around the problematic store and subsequent access;
if this causes performance problems later a better fix is to do all the
ABI binding in gc/walk where it is single-threaded.

Change-Id: I488f28ab75beb8351c856fd50b0095cab463642e
Reviewed-on: https://go-review.googlesource.com/c/go/+/298109
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
4 years agocmd/compile: make modified Aux type for OpArgXXXX pass ssa/check
David Chase [Mon, 1 Mar 2021 16:02:48 +0000 (11:02 -0500)]
cmd/compile: make modified Aux type for OpArgXXXX pass ssa/check

For #40724.

Change-Id: I7d1e76139d187cd15a6e0df9d19542b7200589f6
Reviewed-on: https://go-review.googlesource.com/c/go/+/297911
Trust: David Chase <drchase@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: handle aggregate OpArg in registers
David Chase [Wed, 17 Feb 2021 15:38:03 +0000 (10:38 -0500)]
cmd/compile: handle aggregate OpArg in registers

Also handles case where OpArg does not escape but has its address
taken.

May have exposed a lurking bug in 1.16 expandCalls,
if e.g., loading len(someArrayOfstructThing[0].secondStringField)
from a local.  Maybe.

For #40724.

Change-Id: I0298c4ad5d652b5e3d7ed6a62095d59e2d8819c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/293396
Trust: David Chase <drchase@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: remove 8-byte alignment requirement of stack slot on s390x
eric fang [Tue, 2 Mar 2021 06:10:58 +0000 (06:10 +0000)]
cmd/compile: remove 8-byte alignment requirement of stack slot on s390x

This CL applies CL 267999 to s390x.

Updates #42385

Change-Id: Ie8e69ad1b3f7ddc2c8f05125f4af617aeac035ec
Reviewed-on: https://go-review.googlesource.com/c/go/+/297769
Reviewed-by: Keith Randall <khr@golang.org>
Trust: eric fang <eric.fang@arm.com>

4 years agocmd/compile: remove 8-byte alignment requirement of stack slot on mips
eric fang [Tue, 2 Mar 2021 05:53:49 +0000 (05:53 +0000)]
cmd/compile: remove 8-byte alignment requirement of stack slot on mips

This CL applies CL 267999 to mips.

Updates #42385

Change-Id: I8096e16c1b4def767b0c20c16add36fa2406bcab
Reviewed-on: https://go-review.googlesource.com/c/go/+/297772
Reviewed-by: Keith Randall <khr@golang.org>
Trust: eric fang <eric.fang@arm.com>

4 years agocmd/compile: intrinsify runtime/internal/atomic.{And,Or}{8,} on RISCV64
Joel Sing [Sat, 27 Feb 2021 08:07:32 +0000 (19:07 +1100)]
cmd/compile: intrinsify runtime/internal/atomic.{And,Or}{8,} on RISCV64

The 32 bit versions are easily implement with a single instruction, while the
8 bit versions require a bit more effort but use the same atomic instructions
via rewrite rules.

Change-Id: I42e8d457b239c8f75e39a8e282fc88c1bb292a99
Reviewed-on: https://go-review.googlesource.com/c/go/+/268098
Trust: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agosyscall: implement rawVforkSyscall for remaining linux platforms
Joel Sing [Tue, 7 May 2019 07:56:49 +0000 (17:56 +1000)]
syscall: implement rawVforkSyscall for remaining linux platforms

This allows the use of CLONE_VFORK and CLONE_VM for fork/exec, preventing
'fork/exec ...: cannot allocate memory' failures from occuring when attempting
to execute commands from a Go process that has a large memory footprint.
Additionally, this should reduce the latency of fork/exec on these platforms.

Fixes #31936

Change-Id: I4e28cf0763173145cacaa5340680dca9ff449305
Reviewed-on: https://go-review.googlesource.com/c/go/+/295849
Trust: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: retrieve Args from registers
David Chase [Sat, 13 Feb 2021 15:49:37 +0000 (10:49 -0500)]
cmd/compile: retrieve Args from registers

in progress; doesn't fully work until they are also passed on
register on the caller side.

For #40724.

Change-Id: I29a6680e60bdbe9d132782530214f2a2b51fb8f6
Reviewed-on: https://go-review.googlesource.com/c/go/+/293394
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agoA+C: change email address for Baokun Lee
Baokun Lee [Mon, 14 Dec 2020 11:39:02 +0000 (19:39 +0800)]
A+C: change email address for Baokun Lee

Change-Id: Ib82e1c8c549c91e20c5aff5e50736617c6831f8c
Reviewed-on: https://go-review.googlesource.com/c/go/+/277792
Run-TryBot: Baokun Lee <bk@golangcn.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Alberto Donizetti <alb.donizetti@gmail.com>

4 years agocmd/compile: remove 8-byte alignment requirement of stack slot on mips64
eric fang [Tue, 2 Mar 2021 06:08:51 +0000 (06:08 +0000)]
cmd/compile: remove 8-byte alignment requirement of stack slot on mips64

This CL applies CL 267999 to mips64.

Updates #42385

Change-Id: Ideab21be0d8c1a36b3be7411b24adac70a3d16e0
Reviewed-on: https://go-review.googlesource.com/c/go/+/297771
Reviewed-by: eric fang <eric.fang@arm.com>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: eric fang <eric.fang@arm.com>
Run-TryBot: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agocmd/compile: use abiutils for all rcvr/in/out frame offsets.
David Chase [Tue, 9 Feb 2021 20:14:43 +0000 (15:14 -0500)]
cmd/compile: use abiutils for all rcvr/in/out frame offsets.

types thought it knew how to do this, but that's a lie, because types
doesn't know what the ABI is.

includes extra checking to help prevent things from accidentally working
if they need to be changed but aren't.

For #40724.

Change-Id: I166cd948f262344b7bebde6a2c25e7a7f878bbfb
Reviewed-on: https://go-review.googlesource.com/c/go/+/293393
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/link: disable flaky Darwin "symbols" test
David Chase [Mon, 1 Mar 2021 02:48:20 +0000 (21:48 -0500)]
cmd/link: disable flaky Darwin "symbols" test

About one run out of 3 it fails on my laptop, and
I am tired of having to be a nanny for my tests just
because of this one flaky test.  This has been a problem
for months.

Updates #32218.

Change-Id: I2871d4c6f47e9432d189ed7bdcda8f9c0871cfc5
Reviewed-on: https://go-review.googlesource.com/c/go/+/297469
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/go: remove -insecure flag on go get
witchard [Tue, 2 Mar 2021 21:16:24 +0000 (21:16 +0000)]
cmd/go: remove -insecure flag on go get

Resolves #37519

Change-Id: Iba675a180b0e61b12835cdb6ecd4c6dc61e0605c
GitHub-Last-Rev: aa018af6f8fc7f0b829820e831ad96734adcb8d0
GitHub-Pull-Request: golang/go#44724
Reviewed-on: https://go-review.googlesource.com/c/go/+/297709
Trust: Jay Conrod <jayconrod@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
4 years agocmd/go/internal/modload: do not resolve an arbitrary version for 'go list --versions'
Bryan C. Mills [Fri, 26 Feb 2021 22:40:18 +0000 (17:40 -0500)]
cmd/go/internal/modload: do not resolve an arbitrary version for 'go list --versions'

If we don't actually require the listed module, we previously
implicitly resolved "latest", but also (erroneously) forgot to apply
exclusions and retractions for it. But there is really no need to
resolve "latest" in this case at all — now we omit the version from
the reported module info entirely.

Fixes #44296

Change-Id: Id595f52f597c7213bd65b73bf066a678d9e1d694
Reviewed-on: https://go-review.googlesource.com/c/go/+/297150
Trust: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
4 years agocmd/go: add a test case that reproduces #44296
Bryan C. Mills [Fri, 26 Feb 2021 22:21:21 +0000 (17:21 -0500)]
cmd/go: add a test case that reproduces #44296

For #44296

Change-Id: I310f99ccd406622e39f3fbfa12f7a3bee39602db
Reviewed-on: https://go-review.googlesource.com/c/go/+/297149
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
4 years agocmd/go: resolve std-vendored dependencies as std packages except in 'go get' and...
Bryan C. Mills [Tue, 2 Mar 2021 14:52:55 +0000 (09:52 -0500)]
cmd/go: resolve std-vendored dependencies as std packages except in 'go get' and 'go mod'

In CL 251159, I removed a hard-coded special case changing the
rewriting behavior for std dependencies in GOROOT/src/vendor and
GOROOT/src/cmd/vendor. Unfortunately, that caused packages in 'std' to
be reported as stale when run within GOROOT/src.

This change restores the special-case behavior, but plumbs it through
the PackageOpts explicitly instead of comparing strings stored in
global variables.

Fixes #44725

Change-Id: If084fe74972ce1704715ca79b0b7e092dd90c88b
Reviewed-on: https://go-review.googlesource.com/c/go/+/297869
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
4 years agocmd/go/internal/mvs: prune spurious dependencies in Downgrade
Bryan C. Mills [Fri, 19 Feb 2021 19:03:45 +0000 (14:03 -0500)]
cmd/go/internal/mvs: prune spurious dependencies in Downgrade

Previously, mvs.Downgrade could introduce spurious dependencies if the
downgrade computed for one module lands on a “hidden” version (such as
a pseudo-version) due to a requirement introduced by the downgrade for
another module.

To eliminate those spurious dependencies, we can add one more call to
BuildList to recompute the “actual” downgraded versions, and then
including only those actual versions in the final call to BuildList.

For #36460

Change-Id: Icc6b54aa004907221b2bcbbae74598b0e4100776
Reviewed-on: https://go-review.googlesource.com/c/go/+/294294
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
4 years agodist: generate stub go.mod in workdir
Tamir Duberstein [Thu, 25 Feb 2021 21:44:46 +0000 (16:44 -0500)]
dist: generate stub go.mod in workdir

...and run commands from there.

This removes the requirement that bootstrap must not run inside a module
by ensuring that enclosing modules do not interfere with bootstrap.

Fixes #44209.

Change-Id: I700a81829226770b8160c8ff04127b855b6e26bf
Reviewed-on: https://go-review.googlesource.com/c/go/+/296610
Trust: Jay Conrod <jayconrod@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agocmd/compile: better version of check frame offsets against abi
David Chase [Mon, 1 Mar 2021 22:31:20 +0000 (17:31 -0500)]
cmd/compile: better version of check frame offsets against abi

improved to run on more architectures.

this is in preparation for turning off calculation of frame offsets
in types.CalcSize.

Replaces https://go-review.googlesource.com/c/go/+/293392 .
Updates #44675.
For #40724.

Change-Id: I40ba496172447cf09b86bc646148859363c11ad9
Reviewed-on: https://go-review.googlesource.com/c/go/+/297637
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: optimize single-precision floating point square root
fanzha02 [Mon, 7 Dec 2020 11:15:15 +0000 (19:15 +0800)]
cmd/compile: optimize single-precision floating point square root

Add generic rule to rewrite the single-precision square root expression
with one single-precision instruction. The optimization will reduce two
times of precision converting between double-precision and single-precision.

On arm64 flatform.

previous:
  FCVTSD F0, F0
  FSQRTD F0, F0
  FCVTDS F0, F0

optimized:
  FSQRTS S0, S0

And this patch adds the test case to check the correctness.

This patch refers to CL 241877, contributed by Alice Xu
(dianhong.xu@arm.com)

Change-Id: I6de5d02281c693017ac4bd4c10963dd55989bd7e
Reviewed-on: https://go-review.googlesource.com/c/go/+/276873
Trust: fannie zhang <Fannie.Zhang@arm.com>
Run-TryBot: fannie zhang <Fannie.Zhang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
4 years agointernal/poll, runtime: handle netpollopen error in poll_runtime_pollOpen
Tobias Klauser [Mon, 1 Mar 2021 18:55:22 +0000 (19:55 +0100)]
internal/poll, runtime: handle netpollopen error in poll_runtime_pollOpen

When netpollopen in poll_runtime_pollOpen returns an error, the work in
runtime_pollUnblock and runtime_pollClose can be avoided since the
underlying system call to set up the poller failed.

E.g. on linux, this avoids calling netpollclose and thus epoll_ctl(fd,
EPOLL_CTL_DEL, ...) in case the file does not support epoll, i.e.
epoll_ctl(fd, EPOLL_CTL_ADD, ...) in netpollopen failed.

Fixes #44552

Change-Id: I564d90340fd1ab3a6490526353616a447ae0cfb8
Reviewed-on: https://go-review.googlesource.com/c/go/+/297392
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/go: reject relative paths in GOMODCACHE environment
Baokun Lee [Mon, 18 Jan 2021 06:41:20 +0000 (14:41 +0800)]
cmd/go: reject relative paths in GOMODCACHE environment

Go already rejects relative paths in a couple environment variables,
It should reject relative paths in GOMODCACHE.

Fixes #43715

Change-Id: Id1ceff839c7ab21c00cf4ace45ce48324733a526
Reviewed-on: https://go-review.googlesource.com/c/go/+/284432
Run-TryBot: Baokun Lee <bk@golangcn.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Trust: Baokun Lee <bk@golangcn.org>

4 years agoall: fix spelling
John Bampton [Mon, 1 Mar 2021 09:47:09 +0000 (09:47 +0000)]
all: fix spelling

Change-Id: Iad14571c3e19b01740cd744f0b3025b3e2f1cb72
GitHub-Last-Rev: e8064019299f4e593116060ce2bbd14d62830af7
GitHub-Pull-Request: golang/go#44685
Reviewed-on: https://go-review.googlesource.com/c/go/+/297409
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agogo/parser,go/types: hide API changes related to type parameters
Rob Findley [Wed, 24 Feb 2021 15:31:11 +0000 (10:31 -0500)]
go/parser,go/types: hide API changes related to type parameters

While the dev.typeparams branch was merged, the type parameter API is
slated for go1.18. Hide these changes to the go/parser and go/types API.

This was done as follows:
 + For APIs that will probably not be needed for go1.18, simply unexport
   them.
 + For APIs that we expect to export in go1.18, prefix symbols with '_',
   so that the planned (upper-cased) symbol name is apparent.
 + For APIs that must be exported for testing, move both API and tests
   to files guarded by the go1.18 build constraint.
 + parser.ParseTypeParams is unexported and copied wherever it is
   needed.
 + The -G flag is removed from gofmt, replaced by enabling type
   parameters if built with the go1.18 build constraint.

Notably, changes related to type parameters in go/ast are currently left
exported. We're looking at the AST API separately.

The new API diff from 1.16 is:
+pkg go/ast, method (*FuncDecl) IsMethod() bool
+pkg go/ast, method (*ListExpr) End() token.Pos
+pkg go/ast, method (*ListExpr) Pos() token.Pos
+pkg go/ast, type FuncType struct, TParams *FieldList
+pkg go/ast, type ListExpr struct
+pkg go/ast, type ListExpr struct, ElemList []Expr
+pkg go/ast, type TypeSpec struct, TParams *FieldList
+pkg go/types, type Config struct, GoVersion string\f

Change-Id: I1baf67e26279b49092e774309a836c460979774a
Reviewed-on: https://go-review.googlesource.com/c/go/+/295929
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
4 years agocmd/link,debug/elf: mips32, add .gnu.attributes and .MIPS.abiflags sections
YunQiang Su [Sat, 20 Jun 2020 14:04:54 +0000 (14:04 +0000)]
cmd/link,debug/elf: mips32, add .gnu.attributes and .MIPS.abiflags sections

MIPS32 uses .gnu.attributes and .MIPS.abiflags sections to mark FP ABI
the object is using, and the kernel will set the correct FP mode for it.

Currrently Go doesn't generate these 2 sections. If we link object
without these 2 sections togather FPXX objects, the result will be FPXX,
which is wrong:
   FP32 + FPXX -> FP32
   FPXX + FP64 -> FP64
   FP32 + FP64 -> reject

These 2 sections is also needed to support FPXX and FP64 in future.

More details about FP32/FPXX/FP64 are explained in:
    https://web.archive.org/web/20180828210612/https://dmz-portal.mips.com/wiki/MIPS_O32_ABI_-_FR0_and_FR1_Interlinking

Fixes #39677

Change-Id: Ia34e7461dee38a4f575dd8ace609988744512ac1
Reviewed-on: https://go-review.googlesource.com/c/go/+/239217
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Meng Zhuo <mzh@golangcn.org>

4 years agoruntime: use entersyscall in syscall_syscallX on Darwin
Cherry Zhang [Mon, 25 Jan 2021 22:13:51 +0000 (17:13 -0500)]
runtime: use entersyscall in syscall_syscallX on Darwin

CL 197938 changed syscall* functions to call entersyscall, instead
of entersyscallblock. It missed syscall_syscallX, probably because
it was in sys_darwin_64.go, not sys_darwin.go like others. Change
that one as well.

Found during the review of CL 270380 (thanks Joel).

Change-Id: I0884fc766703f555a3895be332dccfa7d2431374
Reviewed-on: https://go-review.googlesource.com/c/go/+/286435
Trust: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agogo/types: review of expr.go
Rob Findley [Tue, 23 Feb 2021 21:29:33 +0000 (16:29 -0500)]
go/types: review of expr.go

The changes from the (reviewed) dev.regabi copy of expr.go can be seen
by comparing patchset 2 and 7. The actual change is some small
improvements to readability and consistency in untyped conversion,
adding some missing documentation, and removing the "// REVIEW
INCOMPLETE" marker.

Note that expr.go diverges from types2 in its handling of untyped
conversion.

Change-Id: I13a85f6e08f43343e249818245aa857b1f4bf29c
Reviewed-on: https://go-review.googlesource.com/c/go/+/295729
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
4 years agocmd/compile: import empty closure function correctly
Dan Scales [Mon, 1 Mar 2021 17:05:58 +0000 (09:05 -0800)]
cmd/compile:  import empty closure function correctly

On import, make sure that an empty closure is represented as a single
empty block statement. Otherwise, the closure is dropped. Block
statements are not exported explicitly, so must recreate on import.

Fixes #44330

Change-Id: I061598f0f859dd71d2d0cbd10c77cdd81525d1f2
Reviewed-on: https://go-review.googlesource.com/c/go/+/297569
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Trust: Dan Scales <danscales@google.com>

4 years agocmd: upgrade golang.org/x/mod to relax import path check
Jay Conrod [Mon, 1 Mar 2021 15:00:09 +0000 (10:00 -0500)]
cmd: upgrade golang.org/x/mod to relax import path check

This incorporates CL 297089, which allows leading dots in import path
elements but not module path elements. Also added a test.

Fixes #43985
Updates #34992

Change-Id: I2d5faabd8f7b23a7943d3f3ccb6707ab5dc2ce3c
Reviewed-on: https://go-review.googlesource.com/c/go/+/297530
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agocmd/compile/internal/ir: fix up stale comment
Than McIntosh [Fri, 26 Feb 2021 14:25:22 +0000 (09:25 -0500)]
cmd/compile/internal/ir: fix up stale comment

Fix a small stale comment in FinishCaptureNames (refers to old code
structure before the big refactoring).

Change-Id: I2dfb84ce238f919f6e17061439a8bd9b09459dae
Reviewed-on: https://go-review.googlesource.com/c/go/+/296829
Trust: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
4 years agocmd/go: add missing newline to retraction warning message
Jordan Liggitt [Sun, 28 Feb 2021 04:06:23 +0000 (23:06 -0500)]
cmd/go: add missing newline to retraction warning message

Fixes #44674

Change-Id: Icbdb79084bf7bd2f52cc0a53abcc1ec6f0c4a1bf
Reviewed-on: https://go-review.googlesource.com/c/go/+/297350
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Jay Conrod <jayconrod@google.com>

4 years agoRevert "cmd/compile: check frame offsets against abi"
Bryan C. Mills [Mon, 1 Mar 2021 14:43:24 +0000 (14:43 +0000)]
Revert "cmd/compile: check frame offsets against abi"

This reverts CL 293392.

Reason for revert: broke most non-x86 builders

Fixes #44675

Change-Id: I1e815c3ab27a02e83a2f0d221a617909dd021403
Reviewed-on: https://go-review.googlesource.com/c/go/+/297549
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
4 years agocmd/go/internal/modload: don't query when fixing canonical versions
Jay Conrod [Thu, 25 Feb 2021 23:38:43 +0000 (18:38 -0500)]
cmd/go/internal/modload: don't query when fixing canonical versions

If a canonical version is passed to fixVersion when loading the main
go.mod and that version don't match the module path's major version
suffix, don't call Query.

Query doesn't return a useful error in this case when the path is
malformed, for example, when it doens't have a dot in the first path
element. It's better to report the major version mismatch error.

Fixes #44494

Change-Id: I97b1f64aee894fa0db6fb637aa03a51357ee782c
Reviewed-on: https://go-review.googlesource.com/c/go/+/296590
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agoos: mark pipes returned by os.Pipe() as inheritable by default
Jason A. Donenfeld [Sun, 31 Jan 2021 18:51:45 +0000 (19:51 +0100)]
os: mark pipes returned by os.Pipe() as inheritable by default

Now that we don't automatically pass all inheritable handles to new
processes, we can make pipes returned by os.Pipe() inheritable, just
like they are on Unix. This then allows them to be passed through the
SysProcAttr.AdditionalInheritedHandles parameter simply.

Updates #44011.
Fixes #21085.

Change-Id: I8eae329fbc74f9dc7962136fa9aae8fb66879751
Reviewed-on: https://go-review.googlesource.com/c/go/+/288299
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Alex Brainman <alex.brainman@gmail.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agocmd/compile: check frame offsets against abi
David Chase [Tue, 9 Feb 2021 23:09:47 +0000 (18:09 -0500)]
cmd/compile: check frame offsets against abi

this is in preparation for turning off calculation of frame offsets
in types.CalcSize

For #40724.

Change-Id: I2c29fd289c014674076e5ec5170055dbca5ea64b
Reviewed-on: https://go-review.googlesource.com/c/go/+/293392
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
4 years agocmd/compile: fix escape analysis of heap-allocated results
Matthew Dempsky [Sat, 27 Feb 2021 17:41:19 +0000 (09:41 -0800)]
cmd/compile: fix escape analysis of heap-allocated results

One of escape analysis's responsibilities is to summarize whether/how
each function parameter flows to the heap so we can correctly
incorporate those flows into callers' escape analysis data flow
graphs.

As an optimization, we separately record when parameters flow to
result parameters, so that we can more precisely analyze parameter
flows based on how the results are used at the call site. However, if
a named result parameter itself needs to be heap allocated, this
optimization isn't safe and the parameter needs to be recorded as
flowing to heap rather than flowing to result.

Escape analysis used to get this correct because it conservatively
rewalked the data-flow graph multiple times. So even though it would
incorrectly record the result parameter flow, it would separately find
a flow to the heap. However, CL 196811 (specifically, case 3)
optimized the walking logic to reduce unnecessary rewalks causing us
to stop finding the extra heap flow.

This CL fixes the issue by correcting location.leakTo to be sensitive
to sink.escapes and not record result-flows when the result parameter
escapes to the heap.

Fixes #44614.

Change-Id: I48742ed35a6cab591094e2d23a439e205bd65c50
Reviewed-on: https://go-review.googlesource.com/c/go/+/297289
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
4 years agocmd/compile: fixed which-result confusion in presence of 0-width types
David Chase [Fri, 26 Feb 2021 22:37:26 +0000 (17:37 -0500)]
cmd/compile: fixed which-result confusion in presence of 0-width types

A function returning multiple results, some of them zero-width,
will have more than one result present at an offset.  Be sure
that offset AND type match.

Includes test.

Change-Id: I3eb1f56116d989b4e73f533fefabb1bf554c901b
Reviewed-on: https://go-review.googlesource.com/c/go/+/297169
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
4 years agotime: correct unusual extension string cases
Ian Lance Taylor [Thu, 25 Feb 2021 18:01:56 +0000 (10:01 -0800)]
time: correct unusual extension string cases

This fixes two uncommon cases.

First, the tzdata code permits timezone offsets up to 24 * 7, although
the POSIX TZ parsing does not. The tzdata code uses this to specify a
day of week in some cases.

Second, we incorrectly rejected a negative time offset for when a time
zone change comes into effect.

Fixes #44385

Change-Id: I5f2efc1d385e9bfa974a0de3fa81e7a94b827602
Reviewed-on: https://go-review.googlesource.com/c/go/+/296392
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
4 years agoreflect: fix register ABI spill space calculation
Michael Anthony Knyszek [Wed, 17 Feb 2021 19:14:03 +0000 (19:14 +0000)]
reflect: fix register ABI spill space calculation

Currently this does things the old way by computing the number of
registers, but we're going to be using their ABI0 layout for the spill
space for now.

Change-Id: Ibcef1ee48fd834af7cbdaabe704bcabe066ed358
Reviewed-on: https://go-review.googlesource.com/c/go/+/293011
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Trust: Michael Knyszek <mknyszek@google.com>

4 years agocmd/compile: deal with closures in generic functions and instantiated function values
Dan Scales [Sun, 21 Feb 2021 18:54:38 +0000 (10:54 -0800)]
cmd/compile: deal with closures in generic functions and instantiated function values

 - Deal with closures in generic functions by fixing the stenciling code

 - Deal with instantiated function values (instantiated generic
   functions that are not immediately called) during stenciling. This
   requires changing the OFUNCINST node to an ONAME node for the
   appropriately instantiated function. We do this in a second pass,
   since this is uncommon, but requires editing the tree at multiple
   levels.

 - Check global assignments (as well as functions) for generic function
   instantiations.

 - Fix a bug in (*subst).typ where a generic type in a generic function
   may definitely not use all the type args of the function, so we need
   to translate the rparams of the type based on the tparams/targs of
   the function.

 - Added new test combine.go that tests out closures in generic
   functions and instantiated function values.

 - Added one new variant to the settable test.

 - Enabling inlining functions with closures for -G=3. (For now, set
   Ntype on closures in -G=3 mode to keep compatibility with later parts
   of compiler, and allow inlining of functions with closures.)

Change-Id: Iea63d5704c322e42e2f750a83adc8b44f911d4ec
Reviewed-on: https://go-review.googlesource.com/c/go/+/296269
Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>

4 years agosyscall: introduce SysProcAttr.ParentProcess on Windows
Jason A. Donenfeld [Sun, 31 Jan 2021 17:14:56 +0000 (18:14 +0100)]
syscall: introduce SysProcAttr.ParentProcess on Windows

This allows users to specify which process should be used as the parent
process when creating a new process.

Note that this doesn't just trivially pass the handle onward to
PROC_THREAD_ATTRIBUTE_PARENT_PROCESS, because inherited handles must be
valid in the parent process, so if we're changing the destination
process, then we must also change the origin of the parent handles. And,
the StartProcess function must clean up these handles successfully when
exiting, regardless of where the duplication happened. So, we take care
in this commit to use DuplicateHandle for both duplicating and for
closing the inherited handles.

The test was taken originally from CL 288272 and adjusted for use here.

Fixes #44011.

Change-Id: Ib3b132028dcab1aded3dc0e65126c8abebfa35eb
Reviewed-on: https://go-review.googlesource.com/c/go/+/288300
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Alex Brainman <alex.brainman@gmail.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agosyscall: introduce SysProcAttr.AdditionalInheritedHandles on Windows
Jason A. Donenfeld [Sun, 31 Jan 2021 17:07:43 +0000 (18:07 +0100)]
syscall: introduce SysProcAttr.AdditionalInheritedHandles on Windows

This allows users to specify handles that they explicitly want to be
inherited by the new process. These handles must already be marked as
inheritable.

Updates #44011.
Updates #21085.

Change-Id: Ib18322e7dc2909e68c4209e80385492804fa15d3
Reviewed-on: https://go-review.googlesource.com/c/go/+/288298
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agosyscall: restrict inherited handles on Windows
Jason A. Donenfeld [Sun, 31 Jan 2021 16:54:27 +0000 (17:54 +0100)]
syscall: restrict inherited handles on Windows

Windows does not have CLOEXEC, but rather handles are marked explicitly
for being inherited by new processes. This can cause problems when
different Windows functions create new processes from different threads.
syscall.StartProcess has traditionally used a mutex to prevent races
with itself, but this doesn't handle races with other win32 functions.

Fortunately there's a solution: PROC_THREAD_ATTRIBUTE_HANDLE_LIST allows
us to pass the entire list of handles that we want to be inherited. This
lets us get rid of the mutex and also makes process creation safe across
the Go runtime, no matter the context.

Updates #44011.

Change-Id: Ia3424cd2ec64868849cbd6cbb5b0d765224bf4ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/288297
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Alex Brainman <alex.brainman@gmail.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agosyscall: add support for proc thread attribute lists
Jason A. Donenfeld [Sun, 31 Jan 2021 16:37:20 +0000 (17:37 +0100)]
syscall: add support for proc thread attribute lists

This will allow us to pass additional attributes when starting
processes.

Updates #44011.

Change-Id: I4af365c5544a6d421830f247593ec970200e5e03
Reviewed-on: https://go-review.googlesource.com/c/go/+/288296
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Alex Brainman <alex.brainman@gmail.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agocmd/link: recognize ARM64 PE files and relocations
Jason A. Donenfeld [Wed, 3 Feb 2021 23:11:12 +0000 (00:11 +0100)]
cmd/link: recognize ARM64 PE files and relocations

For now, this only add a single relocation type, which is sufficient for
Windows resources. Later we'll see if we need more for cgo.

In order to ensure these code paths are actually tested, this expands
the rsrc tests to include all the architectures of PE objects that we
need to be recognizing, and splits things more clearly between binutils
and llvm objects, which have a slightly different layout, so that we
test both.

This CL is part of a stack adding windows/arm64
support (#36439), intended to land in the Go 1.17 cycle.

Change-Id: Ia1ee840265e9d12c0b12dd1c5d0810f8b300e557
Reviewed-on: https://go-review.googlesource.com/c/go/+/289429
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
4 years agocmd/link: handle types as converted to interface when dynlink
Cherry Zhang [Fri, 26 Feb 2021 01:01:53 +0000 (20:01 -0500)]
cmd/link: handle types as converted to interface when dynlink

When using plugins, a type (whose value) may be pass to a plugin
and get converted to interface there, or vice versa. We need to
treat the type as potentially converted to interface, and retain
its methods.

Should fix #44586.

Change-Id: I80dd35e68baedaa852a317543ccd78d94628d13b
Reviewed-on: https://go-review.googlesource.com/c/go/+/296709
Trust: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
4 years agocmd/compile: fix mishandling of unsafe-uintptr arguments with call method in go/defer
Cuong Manh Le [Fri, 26 Feb 2021 03:17:09 +0000 (10:17 +0700)]
cmd/compile: fix mishandling of unsafe-uintptr arguments with call method in go/defer

In CL 253457, we did the same fix for direct function calls. But for
method calls, the receiver argument also need to be passed through the
wrapper function, which we are not doing so the compiler crashes with
the code in #44415.

It will be nicer if we can rewrite OCALLMETHOD to normal OCALLFUNC, but
that will be for future CL. The passing receiver argument to wrapper
function is easier for backporting to go1.16 branch.

Fixes #44415

Change-Id: I03607a64429042c6066ce673931db9769deb3124
Reviewed-on: https://go-review.googlesource.com/c/go/+/296490
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agocmd/compile: change StaticCall to return a "Results"
David Chase [Thu, 4 Feb 2021 21:42:35 +0000 (16:42 -0500)]
cmd/compile: change StaticCall to return a "Results"

StaticLECall (multiple value in +mem, multiple value result +mem) ->
StaticCall (multiple ergister value in +mem,
   multiple register-sized-value result +mem) ->
ARCH CallStatic (multiple ergister value in +mem,
   multiple register-sized-value result +mem)

But the architecture-dependent stuff is indifferent to whether
it is mem->mem or (mem)->(mem) until Prog generation.

Deal with OpSelectN -> Prog in ssagen/ssa.go, others, as they
appear.

For #40724.

Change-Id: I1d0436f6371054f1881862641d8e7e418e4a6a16
Reviewed-on: https://go-review.googlesource.com/c/go/+/293391
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: fix missing descend in Addrtaken for closures.
Dan Scales [Thu, 25 Feb 2021 20:13:23 +0000 (12:13 -0800)]
cmd/compile:  fix missing descend in Addrtaken for closures.

ComputeAddrtaken needs to descend into closures, now that imported
bodies can include closures. The bug was that we weren't properly
setting Addrtaken for a variable inside a closure inside a function that
we were importing.

For now, still disable inlining of functions with closures for -G mode.
I'll enable it in a later change -- there are just a few fixes related
to the fact that we don't need to set Ntype for closure functions.

Added a test derived from the cilium repro in the issue.

Fixes #44370

Change-Id: Ida2a403636bf8740b471b3ad68b5474951811e19
Reviewed-on: https://go-review.googlesource.com/c/go/+/296649
Run-TryBot: Dan Scales <danscales@google.com>
Trust: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
4 years agocmd/internal/obj: add Prog.SetFrom3{Reg,Const}
Josh Bleecher Snyder [Thu, 7 Jan 2021 22:01:29 +0000 (14:01 -0800)]
cmd/internal/obj: add Prog.SetFrom3{Reg,Const}

These are the the most common uses, and they reduce line noise.

I don't love adding new deprecated APIs,
but since they're trivial wrappers,
it'll be very easy to update them along with the rest.
No functional changes; passes toolstash-check.

Change-Id: I691a8175cfef9081180e463c63f326376af3f3a6
Reviewed-on: https://go-review.googlesource.com/c/go/+/296009
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
4 years agosyscall: comment on fields omitted from the win32finddata1 struct
Bryan C. Mills [Fri, 15 Jan 2021 14:48:39 +0000 (09:48 -0500)]
syscall: comment on fields omitted from the win32finddata1 struct

Updates #42637

Change-Id: I4c7d38034b60c2c04efdeb530a97d96deddfd6fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/284152
Trust: Bryan C. Mills <bcmills@google.com>
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Alex Brainman <alex.brainman@gmail.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
4 years agoRevert "cmd/compile: fix mishandling of unsafe-uintptr arguments with call method...
Matthew Dempsky [Thu, 25 Feb 2021 21:32:29 +0000 (21:32 +0000)]
Revert "cmd/compile: fix mishandling of unsafe-uintptr arguments with call method in go/defer"

This reverts commit CL 294849.

Reason for revert: this doesn't actually fix the issue, as revealed
by the noopt builder's failures.

Change-Id: Ib4ea9ceb4d75e46b3b91ec348b365fd8c83316ad
Reviewed-on: https://go-review.googlesource.com/c/go/+/296629
Trust: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agoreflect: add Method.IsExported and StructField.IsExported methods
Joe Tsai [Thu, 15 Oct 2020 01:41:16 +0000 (18:41 -0700)]
reflect: add Method.IsExported and StructField.IsExported methods

The IsExported method is a more intuitive helper for checking whether
the method or field is exported than checking whether PkgPath is empty.

In the same CL, modify the standard library to make use of this helper.

Fixes #41563

Change-Id: Iaacfb3b74449501f98e2707aa32095a32bd3c3c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/266197
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/internal/obj: fix typo in docs
Josh Bleecher Snyder [Thu, 25 Feb 2021 17:58:05 +0000 (09:58 -0800)]
cmd/internal/obj: fix typo in docs

Change-Id: I5a3d26a4cc59b327d46ca24bcb01ef594758c230
Reviewed-on: https://go-review.googlesource.com/c/go/+/296531
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/compile: declare inlined result params early for empty returns
Matthew Dempsky [Thu, 18 Feb 2021 01:49:40 +0000 (17:49 -0800)]
cmd/compile: declare inlined result params early for empty returns

The code for delayed declaration of inlined result parameters only
handles non-empty return statements. This is generally okay, because
we already early declare if there are any (non-blank) named result
parameters.

But if a user writes a function with only blank result parameters and
with exactly one return statement, which is empty, then they could end
up hitting the dreaded "Value live at entry" ICE.

This CL fixes the issue by ensuring we always early declare inlined
result parameters if there are any empty return statements.

Fixes #44355.

Change-Id: I315f3853be436452883b1ce31da1bdffdf24d506
Reviewed-on: https://go-review.googlesource.com/c/go/+/293293
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
4 years agoos: avoid allocation in File.WriteString
Josh Bleecher Snyder [Sat, 7 Nov 2020 01:37:03 +0000 (17:37 -0800)]
os: avoid allocation in File.WriteString

Instead of alloc+copy to convert the string
to a byte slice, do an unsafe conversion.

Rely on the kernel not to scribble on the
buffer during the write.

Fixes #42406

Change-Id: I66f4838b43a11bcc3d67bbfa1706726318d55343
Reviewed-on: https://go-review.googlesource.com/c/go/+/268020
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
4 years agodatabase/sql: close driver.Connector if it implements io.Closer
Ivan Trubach [Thu, 25 Feb 2021 19:15:04 +0000 (19:15 +0000)]
database/sql: close driver.Connector if it implements io.Closer

This change allows driver implementations to manage resources in
driver.Connector, e.g. to share the same underlying database handle
between multiple connections. That is, it allows embedded databases
with in-memory backends like SQLite and Genji to safely release the
resources once the sql.DB is closed.

This makes it possible to address oddities with in-memory stores in
SQLite and Genji drivers without introducing too much complexity in
the driver implementations.

See also:
- https://github.com/mattn/go-sqlite3/issues/204
- https://github.com/mattn/go-sqlite3/issues/511
- https://github.com/genjidb/genji/issues/210

Fixes #41790

Change-Id: Idbd19763134438ed38288b9d44f16608e4e97fd7
GitHub-Last-Rev: 962c785dfb3bb6ad98b2216bcedd84ba383fe872
GitHub-Pull-Request: golang/go#41710
Reviewed-on: https://go-review.googlesource.com/c/go/+/258360
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Trust: Emmanuel Odeke <emmanuel@orijtech.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agocmd/compile: automate resultInArg0 register checks
Josh Bleecher Snyder [Fri, 8 Jan 2021 03:08:37 +0000 (19:08 -0800)]
cmd/compile: automate resultInArg0 register checks

No functional changes; passes toolstash-check.
No measureable performance changes.

Change-Id: I2629f73d4a3cc56d80f512f33cf57cf41d8f15d3
Reviewed-on: https://go-review.googlesource.com/c/go/+/296010
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
4 years agocmd/compile: fix mishandling of unsafe-uintptr arguments with call method in go/defer
Cuong Manh Le [Sun, 21 Feb 2021 15:09:03 +0000 (22:09 +0700)]
cmd/compile: fix mishandling of unsafe-uintptr arguments with call method in go/defer

In CL 253457, we did the same fix for direct function calls. But for
method calls, the receiver argument also need to be passed through the
wrapper function, which we are not doing so the compiler crashes with
the code in #44415.

As we already rewrite t.M(...) into T.M(t, ...) during walkCall1, to fix
this, we can do the same trick in wrapCall, so the receiver argument
will be treated as others.

Fixes #44415

Change-Id: I396182983c85d9c5e4494657da79d25636e8a079
Reviewed-on: https://go-review.googlesource.com/c/go/+/294849
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agoruntime: use pipe2 for nonblockingPipe on dragonfly
Tobias Klauser [Wed, 24 Feb 2021 08:43:59 +0000 (09:43 +0100)]
runtime: use pipe2 for nonblockingPipe on dragonfly

The pipe2 syscall is available since DragonflyBSD 4.2, see
https://www.dragonflybsd.org/release42/

Change-Id: Ifc67c4935cc59bae29be459167e2fa765843ac03
Reviewed-on: https://go-review.googlesource.com/c/go/+/295471
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agoruntime: batch moving gFree list between local p and global schedt
Andy Pan [Mon, 23 Nov 2020 07:42:48 +0000 (15:42 +0800)]
runtime: batch moving gFree list between local p and global schedt

Change-Id: I0ca1fcee6d3f08bdfcfa51f0dc774118d7355636
Reviewed-on: https://go-review.googlesource.com/c/go/+/271914
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
4 years agocmd: upgrade golang.org/x/mod to fix go.mod parser
Jay Conrod [Wed, 24 Feb 2021 23:24:45 +0000 (18:24 -0500)]
cmd: upgrade golang.org/x/mod to fix go.mod parser

modfile.Parse passed an empty string to the VersionFixer for the
module path. This caused errors for v2+ versions.

Fixes #44494

Change-Id: I13b86b6ecf6815c4bc9a96ec0668284c9228c205
Reviewed-on: https://go-review.googlesource.com/c/go/+/296131
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agocmd/go: add a script test corresponding to the downhiddenartifact MVS test
Bryan C. Mills [Fri, 19 Feb 2021 19:41:02 +0000 (14:41 -0500)]
cmd/go: add a script test corresponding to the downhiddenartifact MVS test

For #36460

Change-Id: I95abff45bb325732a19eb8b9c0d3fc34df08b4d4
Reviewed-on: https://go-review.googlesource.com/c/go/+/294293
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
4 years agocmd/go/internal/mvs: add test cases for downgrade interaction with hidden versions
Bryan C. Mills [Fri, 19 Feb 2021 01:20:12 +0000 (20:20 -0500)]
cmd/go/internal/mvs: add test cases for downgrade interaction with hidden versions

For #36460

Change-Id: I889c4ece0d2caff7528cf437f89f7446dbd83955
Reviewed-on: https://go-review.googlesource.com/c/go/+/294292
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
4 years agotest: add test case that failed with gccgo
Ian Lance Taylor [Mon, 15 Feb 2021 01:59:48 +0000 (17:59 -0800)]
test: add test case that failed with gccgo

bug511.dir/b.go:10:14: error: reference to undefined field or method 'M'

Change-Id: I9f96dc5c7254b310bc3e15b0bc588d62718cb4b2
Reviewed-on: https://go-review.googlesource.com/c/go/+/292009
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agotesting/fstest: treat dash specially when building glob
Ian Lance Taylor [Sun, 21 Feb 2021 23:28:41 +0000 (15:28 -0800)]
testing/fstest: treat dash specially when building glob

"[-]" is not a valid path.Match pattern.

Fixes #44474

Change-Id: I0932bbf08ffb8ad0c5337d69d0893f53c1ba89ad
Reviewed-on: https://go-review.googlesource.com/c/go/+/294869
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
4 years agosyscall: return error if GetQueuedCompletionStatus truncates key
Jason A. Donenfeld [Thu, 25 Feb 2021 01:58:04 +0000 (02:58 +0100)]
syscall: return error if GetQueuedCompletionStatus truncates key

This function has the wrong signature, so return an error when that
actually might lead to unexpected results. Users should switch to
x/sys/windows for the real version of this function.

Updates #44538.

Change-Id: I4d1f3d1e380815733ecfea683f939b1d25dcc32a
Reviewed-on: https://go-review.googlesource.com/c/go/+/296154
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agosyscall, os: use pipe2 syscall on DragonflyBSD instead of pipe
Tobias Klauser [Tue, 23 Feb 2021 13:58:32 +0000 (14:58 +0100)]
syscall, os: use pipe2 syscall on DragonflyBSD instead of pipe

Follow the implementation used by the other BSDs and account for the
intricacy of having to pass the fds array even though the file
descriptors are returned.

Re-submit of CL 130996 with corrected pipe2 wrapper.

Change-Id: Ie36d8214cba60c4fdb579f18bfc1c1ab3ead3ddc
Reviewed-on: https://go-review.googlesource.com/c/go/+/295372
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/compile: fix typo in rewrite_test.go
Ikko Ashimine [Thu, 25 Feb 2021 03:03:45 +0000 (03:03 +0000)]
cmd/compile: fix typo in rewrite_test.go

insted -> instead

Change-Id: Ib8a0423cf99f615976f058468873fb576dd96db6
GitHub-Last-Rev: 8e1a1d08807a35c55d65a2e3f8bb28418a43b3a8
GitHub-Pull-Request: golang/go#44598
Reviewed-on: https://go-review.googlesource.com/c/go/+/296309
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agotest: fix inline.go test for linux-amd64-noopt
Egon Elbre [Wed, 24 Feb 2021 19:08:52 +0000 (21:08 +0200)]
test: fix inline.go test for linux-amd64-noopt

math.Float32bits was not being inlined across package boundaries.
Create a private func that can be inlined with -l.

Change-Id: Ic50bf4727dd8ade09d011eb204006b7ee88db34a
Reviewed-on: https://go-review.googlesource.com/c/go/+/295989
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agoruntime: subtract one from ip when determining abort
Jason A. Donenfeld [Thu, 25 Feb 2021 00:29:58 +0000 (01:29 +0100)]
runtime: subtract one from ip when determining abort

On windows/arm, the abort is given from one byte off of the function
address, perhaps because Windows wants to simulate x86/amd64 modes, or
because it's jumping from thumb mode. This is not the case with
windows/arm64, though.

This prevents a failure in the builders with the TestAbort test:

    crash_test.go:727: output contains BAD:
        panic: runtime error: invalid memory address or nil pointer dereference [recovered]
                panic: BAD: recovered from abort
        [signal 0xc0000005 code=0x0 addr=0x0 pc=0x6a5721]

Change-Id: I8939c60611863cc0c325e179a772601acea9fd4a
Reviewed-on: https://go-review.googlesource.com/c/go/+/296153
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agosyscall: restore broken GetQueuedCompletionStatus signature but make it not crash
Jason A. Donenfeld [Wed, 24 Feb 2021 21:59:59 +0000 (22:59 +0100)]
syscall: restore broken GetQueuedCompletionStatus signature but make it not crash

This reverts commit dc4698f52b5ad3f0251e0cc25bc7ffbd10e23f2c, and then
fixes the memory corruption issue. It also suggests users switch to
x/sys/windows for the proper function.

This requires CL 295174 to be submitted first.

Updates #44538.

Change-Id: I0ee602f1c1d6a89cc585aff426833a4cb3f2be50
Reviewed-on: https://go-review.googlesource.com/c/go/+/296149
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/compile: plumb abi info into expandCalls
David Chase [Mon, 1 Feb 2021 18:26:47 +0000 (13:26 -0500)]
cmd/compile: plumb abi info into expandCalls

Work in progress.

TODO:
- insert debugging output for all the steps listed below
- emit modified call instructions w/ multiple register inputs
  and Result-typed outputs (next CL)
  - initially just change output from "mem" to "Result{mem}"
  = most places this hits will be future work.
- change OpArg to use registerized variants
  - (done) match abi paramresultinfo with particular arg, use Name
  - (this CL) push register offsets for "loads" and "stores" into
    recursive decomposition.
- hand registerized Result to exit block

For #40724.

Change-Id: Ie5de9d71f8fd4e092f5ee9260b54de35abf91016
Reviewed-on: https://go-review.googlesource.com/c/go/+/293390
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
4 years agocmd/compile: disable inlining functions with closures for now
Dan Scales [Wed, 24 Feb 2021 21:03:17 +0000 (13:03 -0800)]
cmd/compile: disable inlining functions with closures for now

Added a flag '-d=inlfuncswithclosures=1' to allow inlining functions with
closures, and change the default to off for now, until #44370 is fixed.

Updates #44370.

Change-Id: Ic17723aa5c091d91f5f5004d8b63ec7125257acf
Reviewed-on: https://go-review.googlesource.com/c/go/+/296049
Run-TryBot: Dan Scales <danscales@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>

4 years agocmd/link: use ctxt.Logf instead of package log
Josh Bleecher Snyder [Wed, 24 Feb 2021 19:48:09 +0000 (11:48 -0800)]
cmd/link: use ctxt.Logf instead of package log

Fixes #43601

Change-Id: I28b745cb92932d875a66f64c63355650a092f096
Reviewed-on: https://go-review.googlesource.com/c/go/+/296029
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agocmd/compile: ARM64 optimize []float64 and []float32 access
Egon Elbre [Fri, 27 Nov 2020 15:10:33 +0000 (17:10 +0200)]
cmd/compile: ARM64 optimize []float64 and []float32 access

Optimize load and store to []float64 and []float32.
Previously it used LSL instead of shifted register indexed load/store.

Before:

    LSL   $3, R0, R0
    FMOVD F0, (R1)(R0)

After:

    FMOVD F0, (R1)(R0<<3)

Fixes #42798

Change-Id: I0c0912140c3dce5aa6abc27097c0eb93833cc589
Reviewed-on: https://go-review.googlesource.com/c/go/+/273706
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Giovanni Bajo <rasky@develer.com>

4 years agocmd/compile/internal-abi: fix ABI0-equivalence for zero-sized values
Austin Clements [Wed, 24 Feb 2021 01:48:22 +0000 (20:48 -0500)]
cmd/compile/internal-abi: fix ABI0-equivalence for zero-sized values

This fixes a bug in the internal ABI specification that made it not
equivalent to ABI0 even with zero architectural argument registers in
the case of a zero-sized argument with alignment > 1.

In ABI0, even zero-sized arguments cause alignment padding in the
stack frame.

Currently, in the internal ABI, zero-sized arguments get
register-assigned even if there are no registers because they don't
consume any registers. Hence, they don't create alignment padding in
the stack frame.

Fix this by stack-assigning zero-sized arguments.

For #40724.

Change-Id: I1f5a95a94fed8b5313a360e5e76875701ba5f562
Reviewed-on: https://go-review.googlesource.com/c/go/+/295791
Trust: Austin Clements <austin@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
4 years agocmd/compile/internal-abi: update internal ABI spec for g register
Austin Clements [Wed, 24 Feb 2021 02:22:20 +0000 (21:22 -0500)]
cmd/compile/internal-abi: update internal ABI spec for g register

We've already implemented dedicating R14 as the G register on amd64,
so remove the TODO saying we might want to hold off on this.

For #40724.

Change-Id: I45b24ced03cac862127b53f5e9a4b4bcf6b1f86c
Reviewed-on: https://go-review.googlesource.com/c/go/+/295790
Trust: Austin Clements <austin@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
4 years agocmd/compile/internal-abi: use x87 mode, not MMX mode
Austin Clements [Wed, 24 Feb 2021 03:03:29 +0000 (22:03 -0500)]
cmd/compile/internal-abi: use x87 mode, not MMX mode

Florian Weimer pointed out that my justification for using MMX mode
was nonsense and that staying in x87 mode simplifies transitions to
and from C. Hence, switch the spec to say we're always in x87 mode.

For #40724.

Change-Id: Iad916b2c376db41f95614aa6897f6b1184576bb7
Reviewed-on: https://go-review.googlesource.com/c/go/+/295789
Trust: Austin Clements <austin@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
4 years agocmd/compile: add rule to coalesce writes
Alejandro García Montoro [Wed, 30 Dec 2020 17:41:36 +0000 (18:41 +0100)]
cmd/compile: add rule to coalesce writes

The code generated when storing eight bytes loaded from memory created a
series of small writes instead of a single, large one. The specific
pattern of instructions generated stored 1 byte, then 2 bytes, then 4
bytes, and finally 1 byte.

The new rules match this specific pattern both for amd64 and for s390x,
and convert it into a single instruction to store the 8 bytes. arm64 and
ppc64le already generated the right code, but the new codegen test
covers also those architectures.

Fixes #41663

Change-Id: Ifb9b464be2d59c2ed5034acf7b9c3e473f344030
Reviewed-on: https://go-review.googlesource.com/c/go/+/280456
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agocmd/internal/goobj: add test case for object file reader
Than McIntosh [Tue, 15 Dec 2020 21:01:34 +0000 (16:01 -0500)]
cmd/internal/goobj: add test case for object file reader

Add test in which a input Go object file contains a very large number
of relocations (more than 1<<20).

Updates #41621.

Change-Id: If1ebf3c4fefbf55ddec4e05c5299e7c48fc697d8
Reviewed-on: https://go-review.googlesource.com/c/go/+/278493
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Trust: Than McIntosh <thanm@google.com>

4 years agocontext: avoid importing context package twice
Kevin Burke [Wed, 24 Feb 2021 17:12:34 +0000 (09:12 -0800)]
context: avoid importing context package twice

Change-Id: Id0a127e080dda8ee62738922c6de8caf3719dd68
Reviewed-on: https://go-review.googlesource.com/c/go/+/295949
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Kevin Burke <kev@inburke.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>

4 years agosyscall: do not overflow key memory in GetQueuedCompletionStatus
Jason A. Donenfeld [Tue, 23 Feb 2021 12:29:40 +0000 (13:29 +0100)]
syscall: do not overflow key memory in GetQueuedCompletionStatus

The third argument to GetQueuedCompletionStatus is a pointer to a
uintptr, not a uint32. Users of this functions have therefore been
corrupting their memory every time they used it. Either that memory
corruption was silent (dangerous), or their programs didn't work so they
chose a different API to use.

Fixes #44538.

RELNOTES=yes

Change-Id: Idf48d4c712d13da29791e9a460159255f963105b
Reviewed-on: https://go-review.googlesource.com/c/go/+/295371
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocmd/compile: plumb abi info into ssagen/ssa
David Chase [Thu, 21 Jan 2021 17:04:46 +0000 (12:04 -0500)]
cmd/compile: plumb abi info into ssagen/ssa

Plumb abi information into ssa/ssagen for plain calls
and plain functions (not methods).  Does not extend all the
way through the compiler (yet).

One test disabled because it extends far enough to break the test.

Normalized all the compiler's register args TODOs to
// TODO(register args) ...

For #40724.

Change-Id: I0173a4579f032ac3c9db3aef1749d40da5ea01ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/293389
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
4 years agocmd/compile: reduce inline cost of OCONVOP
Egon Elbre [Sat, 2 Jan 2021 14:28:11 +0000 (16:28 +0200)]
cmd/compile: reduce inline cost of OCONVOP

OCONVOP doesn't have effect in the compiled code so, it can be safely
excluded from inline cost calculation.

Also make sequence ODEREF OCONVNOP* OADDR cost 1. This is rather common
conversion, such as *(*uint32)(unsafe.Pointer(&x)).

Fixes #42788

Change-Id: I5001f7e89d985c198b6405694cdd5b819cf3f47a
Reviewed-on: https://go-review.googlesource.com/c/go/+/281232
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Elias Naur <mail@eliasnaur.com>

4 years agotesting: print late arriving log line in panic
Josh Bleecher Snyder [Thu, 14 Jan 2021 22:38:55 +0000 (14:38 -0800)]
testing: print late arriving log line in panic

When you log after a test has completed,
the testing package panics.

Print the logged line as part of that panic,
to aid in debugging.

Change-Id: I3d6689d1eed57c03e300afe37db0c15b2f4acda4
Reviewed-on: https://go-review.googlesource.com/c/go/+/283972
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
4 years agocontext: reduce contention in cancelCtx.Done
Josh Bleecher Snyder [Fri, 29 Jan 2021 23:41:18 +0000 (15:41 -0800)]
context: reduce contention in cancelCtx.Done

Use an atomic.Value to hold the done channel.
Conveniently, we have a mutex handy to coordinate writes to it.

name                 old time/op  new time/op  delta
ContextCancelDone-8  67.5ns ±10%   2.2ns ±11%  -96.74%  (p=0.000 n=30+28)

Fixes #42564

Change-Id: I5d72e0e87fb221d4e230209e5fb4698bea4053c6
Reviewed-on: https://go-review.googlesource.com/c/go/+/288193
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Sameer Ajmani <sameer@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
4 years agocmd/go: fix version validation in 'go mod edit -exclude'
Bryan C. Mills [Mon, 22 Feb 2021 20:05:20 +0000 (15:05 -0500)]
cmd/go: fix version validation in 'go mod edit -exclude'

The fix is to pull in CL 295089 from the x/mod repo.

Fixes #44497

Change-Id: I008b58d0f4bb48c09d4f1e6ed31d11a714f87dc0
Reviewed-on: https://go-review.googlesource.com/c/go/+/295150
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
4 years agocmd/go, cmd/cgo: pass -mfp32 and -mhard/soft-float to MIPS GCC
YunQiang Su [Tue, 9 Jun 2020 04:09:58 +0000 (04:09 +0000)]
cmd/go, cmd/cgo: pass -mfp32 and -mhard/soft-float to MIPS GCC

For mips32 currently, we are using FP32, while the gcc may be FPXX,
which may generate .MIPS.abiflags and .gnu.attributes section with
value as FPXX. So the kernel will treat the exe as FPXX, and may
choose to use FR=1 FPU mode for it.
Currently, in Go, we use 2 lwc1 to load both half of a double value
to a pair of even-odd FPR. This behavior can only work with FR=0 mode.

In FR=1 mode, all of 32 FPR are 64bit. If we lwc1 the high-half of a double
value to an odd FPR, and try to use the previous even FPR to compute, the
real high-half of even FPR will be unpredicatable.
We set -mfp32 to force the gcc generate FP32 code and section value.

More details about FP32/FPXX/FP64 are explained in:
https://web.archive.org/web/20180828210612/https://dmz-portal.mips.com/wiki/MIPS_O32_ABI_-_FR0_and_FR1_Interlinking

When GOMIPS/GOMIPS64 is set as softfloat, we should also pass
 -msoft-float to gcc.

Here we also add -mno-odd-spreg option, since Loongson's CPU cannot use
odd-number FR in FR=0 mode.

Fixes #39435

Change-Id: I54026ad416a815fe43a9261ebf6d02e5519c3930
Reviewed-on: https://go-review.googlesource.com/c/go/+/237058
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Meng Zhuo <mzh@golangcn.org>

4 years agoio/ioutil: forward TempFile and TempDir to os package
Ian Lance Taylor [Fri, 22 Jan 2021 00:59:29 +0000 (16:59 -0800)]
io/ioutil: forward TempFile and TempDir to os package

For #42026
Fixes #44311

Change-Id: I3dabcf902d155f95800b4adf1d7578906a194ce6
Reviewed-on: https://go-review.googlesource.com/c/go/+/285378
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
4 years agoencoding/json: reduce allocated space in Unmarshal
Josh Bleecher Snyder [Wed, 18 Nov 2020 20:50:29 +0000 (12:50 -0800)]
encoding/json: reduce allocated space in Unmarshal

The decodeState type is a large part of the allocated space during Unmarshal.
The errorContext field is infrequently used, and only on error.
Extract it into a pointer and allocate it separate when necessary.

name                old time/op    new time/op    delta
UnmarshalString-8      115ns ± 5%     114ns ± 3%     ~     (p=0.170 n=15+15)
UnmarshalFloat64-8     113ns ± 2%     106ns ± 1%   -6.42%  (p=0.000 n=15+14)
UnmarshalInt64-8      93.3ns ± 1%    86.9ns ± 4%   -6.89%  (p=0.000 n=14+15)

name                old alloc/op   new alloc/op   delta
UnmarshalString-8       192B ± 0%      160B ± 0%  -16.67%  (p=0.000 n=15+15)
UnmarshalFloat64-8      180B ± 0%      148B ± 0%  -17.78%  (p=0.000 n=15+15)
UnmarshalInt64-8        176B ± 0%      144B ± 0%  -18.18%  (p=0.000 n=15+15)

name                old allocs/op  new allocs/op  delta
UnmarshalString-8       2.00 ± 0%      2.00 ± 0%     ~     (all equal)
UnmarshalFloat64-8      2.00 ± 0%      2.00 ± 0%     ~     (all equal)
UnmarshalInt64-8        1.00 ± 0%      1.00 ± 0%     ~     (all equal)

Change-Id: I53f3f468e6c65f77a12e5138a2626455b197012d
Reviewed-on: https://go-review.googlesource.com/c/go/+/271338
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
4 years agodatabase: remove race in TestTxContextWait
Josh Bleecher Snyder [Mon, 1 Feb 2021 16:35:36 +0000 (08:35 -0800)]
database: remove race in TestTxContextWait

This test contained a data race.
On line 437, db.BeginTx starts a goroutine that runs tx.awaitDone,
which reads tx.keepConnOnRollback.
On line 445, the test writes to tx.keepConnOnRollback.
tx.awaitDone waits on ctx, but because ctx is timeout-based,
there's no ordering guarantee between the write and the read.

The race detector never caught this before
because the context package implementation of Done
contained enough synchronization to make it safe.
That synchronization is not package of the context API or guarantees,
and the first several releases it was not present.
Another commit soon will remove that synchronization,
exposing the latent data race.

To fix the race, emulate a time-based context
using an explicit cancellation-based context.
This gives us enough control to avoid the race.

Change-Id: I103fe9b987b1d4c02e7a20ac3c22a682652128b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/288493
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
4 years agogo/types: review of call.go
Rob Findley [Tue, 23 Feb 2021 17:01:26 +0000 (12:01 -0500)]
go/types: review of call.go

The changes from the (reviewed) dev.regabi copy of call.go can be seen
by comparing patchset 1 and 4. The actual changes are removing the
"// REVIEW INCOMPLETE" marker, deleting some leftover handling of type
instantiation in Checker.call, and adding a comment that exprOrTypeList
should be refactored.

I started to refactor exprOrTypeList, but thought it best to mark this
code as reviewed before diverging from types2.

Change-Id: Icf7fbff5a8def49c5f1781472fd7ba7b73dd9a9c
Reviewed-on: https://go-review.googlesource.com/c/go/+/295531
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
4 years agohash/maphash: remove duplicate from Hash documentation
Alberto Donizetti [Sun, 14 Feb 2021 10:19:39 +0000 (11:19 +0100)]
hash/maphash: remove duplicate from Hash documentation

Fixes #44255

Change-Id: I14d2edbee0a0c39e04111414a57d70ee2fdfb6af
Reviewed-on: https://go-review.googlesource.com/c/go/+/291631
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
4 years agoruntime: remove unused const stackSystem on dragonfly
Tobias Klauser [Tue, 23 Feb 2021 14:35:43 +0000 (15:35 +0100)]
runtime: remove unused const stackSystem on dragonfly

Change-Id: I778c2bd7cf0b12275bae344cb2130a7959500481
Reviewed-on: https://go-review.googlesource.com/c/go/+/295470
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>