]> Cypherpunks repositories - gostls13.git/log
gostls13.git
2 years agocmd/go/internal/modfetch: report error on failing to derive pseudo version from recen...
Zeke Lu [Sat, 27 Aug 2022 02:42:52 +0000 (02:42 +0000)]
cmd/go/internal/modfetch: report error on failing to derive pseudo version from recent tag

The current implementation ignores the error when it tries to get
the recent tag on revisions, which results in incorrect pseudo
version (v0.0.0-) is derived.

Fixes #53935

Change-Id: I153d851eb913fb7e40051e194c92b9ca5bf0e906
GitHub-Last-Rev: 6ba1d90df56cc0f30557863aa8418c54885abb14
GitHub-Pull-Request: golang/go#54701
Reviewed-on: https://go-review.googlesource.com/c/go/+/426079
Reviewed-by: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agotime: add fuzz test for Time.appendFormatRFC3339
Joe Tsai [Mon, 22 Aug 2022 20:33:56 +0000 (13:33 -0700)]
time: add fuzz test for Time.appendFormatRFC3339

Time.appendFormatRFC3339 is a specialized implementation of
Time.appendFormat. We expect the two to be identical.
Add a fuzz test to ensure this property.

Updates #54093

Change-Id: I0bc41ee6e016d3dac75d1ac372d8c9c7266d0299
Reviewed-on: https://go-review.googlesource.com/c/go/+/425100
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>

2 years agointernal/trace: fix typo in goroutines.go
Ikko Ashimine [Mon, 29 Aug 2022 15:48:47 +0000 (15:48 +0000)]
internal/trace: fix typo in goroutines.go

assosciated -> associated

Change-Id: Id1cbbdea12f0239a9e173ece934c18cc9ffcf0f4
GitHub-Last-Rev: 7fea9de6c27b95f91591db4b5bd9023eb80dc522
GitHub-Pull-Request: golang/go#54739
Reviewed-on: https://go-review.googlesource.com/c/go/+/425596
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Knyszek <mknyszek@google.com>

2 years agonet/http: FileServer method check + minimal OPTIONS implementation
Pascal S. de Kloe [Tue, 21 Jun 2022 16:50:35 +0000 (18:50 +0200)]
net/http: FileServer method check + minimal OPTIONS implementation

FileServer provides a read-only service. Methods other than GET or HEAD should
be denied with an Allow header.

Fixes #53501

Change-Id: I1d31b405eefd90565ecd474ac3f8d8d6e3b15072
Reviewed-on: https://go-review.googlesource.com/c/go/+/413554
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
2 years agocmd/compile/obj/arm64: fix encoding error of FMOVD/FMOVS $0|ZR
eric fang [Wed, 24 Aug 2022 03:28:16 +0000 (03:28 +0000)]
cmd/compile/obj/arm64: fix encoding error of FMOVD/FMOVS $0|ZR

Previously the first operand of FMOVD and FMOVS could be $0, which
would be converted to the ZR register. This is prohibited by CL 404316,
also it broken the encoding of "FMOVD/FMOVS ZR, Rn", this CL restores
this instruction format and fixes the encoding issue.

Fixes #54655.
Fixes #54729.

Change-Id: I9c42cd41296bed7ffd601609bd8ecaa27d11e659
Reviewed-on: https://go-review.googlesource.com/c/go/+/425188
Run-TryBot: Eric Fang <eric.fang@arm.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoreflect: simplify array value comparable check
Cuong Manh Le [Sat, 27 Aug 2022 14:38:45 +0000 (21:38 +0700)]
reflect: simplify array value comparable check

If array element is not interface, array or struct, we just need to
check whether the array element type is comparable.

Change-Id: I1ab94cfa17ae86feb6cd3fbdf878af5a776e7bec
Reviewed-on: https://go-review.googlesource.com/c/go/+/426194
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoimage/png: allow both PLTE and tRNS chunks for TrueColor
Nigel Tao [Fri, 19 Aug 2022 06:01:33 +0000 (16:01 +1000)]
image/png: allow both PLTE and tRNS chunks for TrueColor

Prior to this commit, png.Decode would allow TrueColor PNG images that
have one but not both of PLTE and tRNS chunks.

Fixes #54142

Change-Id: I259c1fff86a0aa5640dbadf7ad834e05fbd1430c
Reviewed-on: https://go-review.googlesource.com/c/go/+/424916
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Nigel Tao (INACTIVE; USE @golang.org INSTEAD) <nigeltao@google.com>
2 years agotesting: convert numFailed to atomic type
cuiweixie [Sat, 27 Aug 2022 02:24:49 +0000 (10:24 +0800)]
testing: convert numFailed to atomic type

Change-Id: Ic3464e95ad8901df5477d7717760b8c6d08ce97b
Reviewed-on: https://go-review.googlesource.com/c/go/+/426078
Run-TryBot: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agocmd/go/internal/modload: return error when duplicate module paths among modules in...
ianwoolf [Tue, 26 Jul 2022 13:55:26 +0000 (21:55 +0800)]
cmd/go/internal/modload: return error when duplicate module paths among modules in go.work

Fixes #54048

Change-Id: I27350af451ff50532856092f2d99b6cc6dc6743d
Reviewed-on: https://go-review.googlesource.com/c/go/+/419557
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agodebug/pe: check size in uint64 to avoid overflow
David Chase [Wed, 24 Aug 2022 19:53:40 +0000 (15:53 -0400)]
debug/pe: check size in uint64 to avoid overflow

uint32(sz) != n*uint32(ddSz) can go wrong if
the RHS overflows, so do it in wider precision.

Fixes #54640.

Change-Id: I776563330e46de6cdacd4055f6ff08e7de67797f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425364
Reviewed-by: Dan Kortschak <dan@kortschak.io>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/cgo: add and use runtime/cgo.Incomplete instead of //go:notinheap
Cuong Manh Le [Sun, 7 Aug 2022 11:41:42 +0000 (18:41 +0700)]
cmd/cgo: add and use runtime/cgo.Incomplete instead of //go:notinheap

Updates #46731

Change-Id: Ia83f27c177cc2f57e240cb5c6708d4552423f5be
Reviewed-on: https://go-review.googlesource.com/c/go/+/421879
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/compile: intrinsify Sub64 on riscv64
Wayne Zuo [Fri, 29 Jul 2022 14:14:53 +0000 (22:14 +0800)]
cmd/compile: intrinsify Sub64 on riscv64

After this CL, the performance difference in crypto/elliptic
benchmarks on linux/riscv64 are:

name                 old time/op    new time/op    delta
ScalarBaseMult/P256    1.64ms ± 1%    1.60ms ± 1%   -2.36%  (p=0.008 n=5+5)
ScalarBaseMult/P224    1.53ms ± 1%    1.47ms ± 2%   -4.24%  (p=0.008 n=5+5)
ScalarBaseMult/P384    5.12ms ± 2%    5.03ms ± 2%     ~     (p=0.095 n=5+5)
ScalarBaseMult/P521    22.3ms ± 2%    13.8ms ± 1%  -37.89%  (p=0.008 n=5+5)
ScalarMult/P256        4.49ms ± 2%    4.26ms ± 2%   -5.13%  (p=0.008 n=5+5)
ScalarMult/P224        4.33ms ± 1%    4.09ms ± 1%   -5.59%  (p=0.008 n=5+5)
ScalarMult/P384        16.3ms ± 1%    15.5ms ± 2%   -4.78%  (p=0.008 n=5+5)
ScalarMult/P521         101ms ± 0%      47ms ± 2%  -53.36%  (p=0.008 n=5+5)

Change-Id: I31cf0506e27f9d85f576af1813630a19c20dda8a
Reviewed-on: https://go-review.googlesource.com/c/go/+/420095
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/compile: intrinsify Add64 on riscv64
Wayne Zuo [Fri, 29 Jul 2022 06:24:26 +0000 (14:24 +0800)]
cmd/compile: intrinsify Add64 on riscv64

According to RISCV instruction set manual v2.2 Sec 2.4, we can
implement overflowing check for unsigned addition cheaply using
SLTU instructions.

After this CL, the performance difference in crypto/elliptic
benchmarks on linux/riscv64 are:

name                 old time/op    new time/op    delta
ScalarBaseMult/P256    1.93ms ± 1%    1.64ms ± 1%  -14.96%  (p=0.008 n=5+5)
ScalarBaseMult/P224    1.80ms ± 2%    1.53ms ± 1%  -14.89%  (p=0.008 n=5+5)
ScalarBaseMult/P384    6.15ms ± 2%    5.12ms ± 2%  -16.73%  (p=0.008 n=5+5)
ScalarBaseMult/P521    25.9ms ± 1%    22.3ms ± 2%  -13.78%  (p=0.008 n=5+5)
ScalarMult/P256        5.59ms ± 1%    4.49ms ± 2%  -19.79%  (p=0.008 n=5+5)
ScalarMult/P224        5.42ms ± 1%    4.33ms ± 1%  -20.01%  (p=0.008 n=5+5)
ScalarMult/P384        19.9ms ± 2%    16.3ms ± 1%  -18.15%  (p=0.008 n=5+5)
ScalarMult/P521        97.3ms ± 1%   100.7ms ± 0%   +3.48%  (p=0.008 n=5+5)

Change-Id: Ic4c82ced4b072a4a6575343fa9f29dd09b0cabc4
Reviewed-on: https://go-review.googlesource.com/c/go/+/420094
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
Reviewed-by: Joel Sing <joel@sing.id.au>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agonet: add FlagRunning to exactly reflect the states of an interface.
Jianwei Mao [Fri, 26 Aug 2022 08:45:23 +0000 (08:45 +0000)]
net: add FlagRunning to exactly reflect the states of an interface.

Correctly set this flag while parsing the syscall result.

The FlagUp flag can not distinguish the following situations:
1. interface is plugged, automatically up, and in running(UP) state
2. interface is not plugged, administratively or manually set to up,
but in DOWN state

So, We can't distinguish the state of a NIC by the FlagUp flag alone.

Fixes #53482

Change-Id: I43796bea1a7f72d1fddfef914efe603c81995e1b
GitHub-Last-Rev: 686b5d888e97e9b90cf36ac0c15943eb97e125d4
GitHub-Pull-Request: golang/go#53484
Reviewed-on: https://go-review.googlesource.com/c/go/+/413454
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ryan Schuster <shuey19831@gmail.com>
Reviewed-by: Jianwei Mao <maojianwei2020@gmail.com>
Reviewed-by: Damien Neil <dneil@google.com>
2 years agodebug/elf: fix reloc number of R_PPC64_SECTOFF_LO_DS
Paul E. Murphy [Mon, 13 Jun 2022 16:05:18 +0000 (11:05 -0500)]
debug/elf: fix reloc number of R_PPC64_SECTOFF_LO_DS

R_PPC64_SECTOFF_LO_DS is defined as reloc 62 on all PPC64 ELF ABIs.

Fixes #53356

Change-Id: I5fabf6be32f3310c5aed47d4d654e05fb7bc9de0
Reviewed-on: https://go-review.googlesource.com/c/go/+/411915
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Paul Murphy <murp@ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agonet: Resolver.LookupIP return error for empty string
ianwoolf [Wed, 27 Jul 2022 16:55:31 +0000 (00:55 +0800)]
net: Resolver.LookupIP return error for empty string

Fixes #53995

Change-Id: Ib0de237b57382feb6b8070f2310945aef6c7db01
Reviewed-on: https://go-review.googlesource.com/c/go/+/419734
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>

2 years agoreflect: add Value.{Comparable,Equal}
cuiweixie [Sun, 14 Aug 2022 08:45:33 +0000 (16:45 +0800)]
reflect: add Value.{Comparable,Equal}

For #46746

Change-Id: I879124974cdb55932cd9d07d3b384d49d5059857
Reviewed-on: https://go-review.googlesource.com/c/go/+/423794
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agocmd/compile: fix score for Select{0,1} with type flags
Lynn Boger [Tue, 16 Aug 2022 17:06:10 +0000 (12:06 -0500)]
cmd/compile: fix score for Select{0,1} with type flags

A recent change was made for ppc64x to treat ANDCCconst as
a tuple, allowing ANDconst to be removed from the list
of ops. Included in that change were some improvements to the
rules to avoid some extra code, mainly the elimination of a
cmp 0 following an andi. and in some cases the following
isel. While those changes worked for most cases, in a few
cases some extra unnecessary code was generated.

Currently the snippet appears in archive/zip.(*FileHeader).Mode:

        ANDCC R4,$1,R5                       // andi. r5,r4,1
        ANDCC R4,$16,R5                      // andi. r5,r4,16
        CMPW R5,R0                           // cmpw r5,r0
        ADDIS $0,$-32768,R5                  // lis r5,-32768
        OR R5,$511,R5                        // ori r5,r5,511
        MOVD $438,R6                         // li r6,438
        ISEL $2,R6,R5,R5                     // isel r5,r6,r5,eq
        MOVD $-147,R6                        // li r6,-147
        AND R6,R5,R6                         // and r6,r5,r6
        ANDCC R4,$1,R4                       // andi. r4,r4,1
        ISEL $2,R5,R6,R4                     // isel r4,r5,r6,eq

The first ANDCC is never used and should not be there.
From the ssa.html file, the scheduler is not putting the Select1
close to the ISEL, which results in the flag being clobbered
before it can be used. By changing the score for a Select0 or Select1
with type Flags, the extra ANDCC does not occur.

Change-Id: I82f4bc7c02afb1c2b1c048dc6995e0b3f9363fb3
Reviewed-on: https://go-review.googlesource.com/c/go/+/424294
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Paul Murphy <murp@ibm.com>
2 years agonet/http: don't panic on very large MaxBytesReaderLimit
cuiweixie [Fri, 12 Aug 2022 16:47:31 +0000 (16:47 +0000)]
net/http: don't panic on very large MaxBytesReaderLimit

Fixes #54408

Change-Id: I454199ae5bcd087b8fc4169b7622412105e71113
GitHub-Last-Rev: a33fe7e206d0c394440962acd360df3aa9b117c3
GitHub-Pull-Request: golang/go#54415
Reviewed-on: https://go-review.googlesource.com/c/go/+/423314
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: xie cui <523516579@qq.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: hopehook <hopehook@qq.com>
2 years agoruntime: add address sanitizer support for ppc64le
Archana R [Thu, 26 May 2022 10:24:38 +0000 (05:24 -0500)]
runtime: add address sanitizer support for ppc64le

updates #44853

Change-Id: I71905ee1bcb99ce7300bbed2daad3617d2643c53
Reviewed-on: https://go-review.googlesource.com/c/go/+/408814
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Benny Siegert <bsiegert@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Archana Ravindar <aravind5@in.ibm.com>

2 years agocmd/compile: remove branch in atomicCas{32,64} sequence on ppc64x.
Lynn Boger [Tue, 2 Aug 2022 13:26:23 +0000 (08:26 -0500)]
cmd/compile: remove branch in atomicCas{32,64} sequence on ppc64x.

This removes one of the branches in the atomicCas sequences for
ppc64/ppc64le.

Change-Id: Ibb46fbfdce052889c69e3da298f28caff540d99b
Reviewed-on: https://go-review.googlesource.com/c/go/+/422014
Reviewed-by: Archana Ravindar <aravind5@in.ibm.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Paul Murphy <murp@ibm.com>
2 years agonet/http: mention io.Seeker requirement in FS documentation
Alex Studer [Wed, 20 Oct 2021 19:18:03 +0000 (19:18 +0000)]
net/http: mention io.Seeker requirement in FS documentation

Both FileServer and NewFileTransport can try to seek a file, specifically
when MIME type sniffing is performed. This can be somewhat surprising to an
implementer of an fs.FS, as their filesystem will appear to work until a
user tries to access a file with an unrecognized extension (which requires
type sniffing and therefore seeking). With FileServer, this results in a
"seeker can't seek" message, which is not very clear for the developer.

The issue arises because fs.FS does not require Seek, while http.FileSystem
does. Therefore, this change adds a line to the documentation of net/http's
adapter function mentioning the requirement.

Change-Id: Ieb955b7a7f34e2be39dd696cb712513c70100b3a
GitHub-Last-Rev: fddccdae36e04fcb87d78b3b03fa4658dbb7d83d
GitHub-Pull-Request: golang/go#48781
Reviewed-on: https://go-review.googlesource.com/c/go/+/353874
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Damien Neil <dneil@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agonet/http: add errors.As support for x/net/http2.StreamError
Akshay Shah [Mon, 22 Aug 2022 23:08:13 +0000 (16:08 -0700)]
net/http: add errors.As support for x/net/http2.StreamError

To make it easier to extract the HTTP/2 error code (if any) from
net/http errors, implement an As method on the vendored copy of
golang.org/x/net/http2.StreamError. The new As method lets users work
with the vendored error type as though it were the x/net/http2
StreamError.

Fixes #53896.

Change-Id: Ib18eb428adc05a3c0e19a946ece936e2378e1c7c
Reviewed-on: https://go-review.googlesource.com/c/go/+/425104
Run-TryBot: Damien Neil <dneil@google.com>
Auto-Submit: Damien Neil <dneil@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: David Chase <drchase@google.com>
2 years agoruntime: drop padding alignment field for timeHistogram
Cuong Manh Le [Fri, 26 Aug 2022 02:16:47 +0000 (09:16 +0700)]
runtime: drop padding alignment field for timeHistogram

After CL 419449, timeHistogram always have 8-byte alignment.

Change-Id: I93145502bcafa1712b811b1a6d62da5d54d0db42
Reviewed-on: https://go-review.googlesource.com/c/go/+/425777
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agomime: allow duplicate media type parameters with equivalent values
Paschalis Tsilias [Wed, 10 Nov 2021 20:14:03 +0000 (22:14 +0200)]
mime: allow duplicate media type parameters with equivalent values

Fixes #48866

Change-Id: I2bd2c806e44eb4064b1fb9a6509d79cecbbef013
Reviewed-on: https://go-review.googlesource.com/c/go/+/363094
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Tim King <taking@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
Reviewed-by: David Chase <drchase@google.com>
2 years agoruntime: mark all methods of atomic types with go:nosplit
Andy Pan [Thu, 25 Aug 2022 17:29:57 +0000 (01:29 +0800)]
runtime: mark all methods of atomic types with go:nosplit

Atomic types may be used anywhere in the runtime,
so they must omit its usual stack overflow checks to avoid errors from stack splits.

Change-Id: Icb497334c860bcaa284a2d5a3edc47ee89844301
Reviewed-on: https://go-review.googlesource.com/c/go/+/425484
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agointernal/poll: optimize the maximum amount of bytes that can be transferred by one...
Andy Pan [Mon, 22 Aug 2022 08:08:00 +0000 (16:08 +0800)]
internal/poll: optimize the maximum amount of bytes that can be transferred by one splice(2) call

Fixes #40222

name                          old time/op    new time/op    delta
Splice/tcp-to-tcp/1024-4        5.16µs ± 1%    5.17µs ± 4%     ~     (p=0.481 n=10+10)
Splice/tcp-to-tcp/2048-4        5.46µs ± 2%    5.48µs ± 2%     ~     (p=0.617 n=10+10)
Splice/tcp-to-tcp/4096-4        6.07µs ± 3%    6.09µs ± 5%     ~     (p=0.927 n=10+10)
Splice/tcp-to-tcp/8192-4        7.07µs ± 5%    7.16µs ± 3%     ~     (p=0.324 n=10+10)
Splice/tcp-to-tcp/16384-4       8.82µs ±10%    8.83µs ±11%     ~     (p=1.000 n=10+10)
Splice/tcp-to-tcp/32768-4       12.3µs ±12%    12.7µs ± 4%     ~     (p=0.268 n=10+9)
Splice/tcp-to-tcp/65536-4       22.1µs ± 3%    22.0µs ± 3%     ~     (p=0.912 n=10+10)
Splice/tcp-to-tcp/131072-4      40.0µs ± 4%    39.9µs ± 4%     ~     (p=1.000 n=9+8)
Splice/tcp-to-tcp/262144-4      69.1µs ± 3%    67.9µs ± 2%   -1.67%  (p=0.010 n=9+10)
Splice/tcp-to-tcp/524288-4       128µs ± 3%     124µs ± 2%   -2.67%  (p=0.001 n=9+10)
Splice/tcp-to-tcp/1048576-4      242µs ± 3%     242µs ± 2%     ~     (p=0.912 n=10+10)
Splice/unix-to-tcp/1024-4       1.97µs ± 3%    1.96µs ± 2%     ~     (p=0.342 n=10+10)
Splice/unix-to-tcp/2048-4       2.19µs ± 2%    2.21µs ± 2%     ~     (p=0.150 n=9+10)
Splice/unix-to-tcp/4096-4       2.52µs ± 5%    2.50µs ± 1%     ~     (p=0.393 n=10+10)
Splice/unix-to-tcp/8192-4       3.27µs ± 4%    3.24µs ± 7%     ~     (p=0.968 n=9+10)
Splice/unix-to-tcp/16384-4      4.68µs ± 9%    4.79µs ± 8%     ~     (p=0.739 n=10+10)
Splice/unix-to-tcp/32768-4      7.82µs ±12%    7.76µs ±12%     ~     (p=0.912 n=10+10)
Splice/unix-to-tcp/65536-4      17.9µs ± 3%    17.3µs ±10%     ~     (p=0.661 n=9+10)
Splice/unix-to-tcp/131072-4     35.9µs ± 1%    31.3µs ± 4%  -12.88%  (p=0.000 n=10+9)
Splice/unix-to-tcp/262144-4     72.4µs ± 3%    55.4µs ± 2%  -23.55%  (p=0.000 n=10+10)
Splice/unix-to-tcp/524288-4      146µs ± 3%     113µs ± 1%  -22.87%  (p=0.000 n=9+9)
Splice/unix-to-tcp/1048576-4     288µs ± 3%     223µs ± 3%  -22.49%  (p=0.000 n=10+9)

name                          old speed      new speed      delta
Splice/tcp-to-tcp/1024-4       198MB/s ± 1%   198MB/s ± 4%     ~     (p=0.481 n=10+10)
Splice/tcp-to-tcp/2048-4       375MB/s ± 2%   374MB/s ± 2%     ~     (p=0.631 n=10+10)
Splice/tcp-to-tcp/4096-4       674MB/s ± 3%   673MB/s ± 5%     ~     (p=0.912 n=10+10)
Splice/tcp-to-tcp/8192-4      1.16GB/s ± 6%  1.15GB/s ± 3%     ~     (p=0.315 n=10+10)
Splice/tcp-to-tcp/16384-4     1.86GB/s ±10%  1.84GB/s ±13%     ~     (p=0.720 n=10+9)
Splice/tcp-to-tcp/32768-4     2.68GB/s ±13%  2.57GB/s ± 7%     ~     (p=0.165 n=10+10)
Splice/tcp-to-tcp/65536-4     2.97GB/s ± 3%  2.98GB/s ± 3%     ~     (p=0.912 n=10+10)
Splice/tcp-to-tcp/131072-4    3.28GB/s ± 4%  3.29GB/s ± 4%     ~     (p=1.000 n=9+8)
Splice/tcp-to-tcp/262144-4    3.80GB/s ± 3%  3.86GB/s ± 2%   +1.70%  (p=0.010 n=9+10)
Splice/tcp-to-tcp/524288-4    4.10GB/s ± 3%  4.21GB/s ± 2%   +2.74%  (p=0.001 n=9+10)
Splice/tcp-to-tcp/1048576-4   4.34GB/s ± 3%  4.33GB/s ± 2%     ~     (p=0.912 n=10+10)
Splice/unix-to-tcp/1024-4      519MB/s ± 3%   523MB/s ± 2%     ~     (p=0.353 n=10+10)
Splice/unix-to-tcp/2048-4      934MB/s ± 2%   926MB/s ± 2%     ~     (p=0.156 n=9+10)
Splice/unix-to-tcp/4096-4     1.63GB/s ± 5%  1.64GB/s ± 1%     ~     (p=0.393 n=10+10)
Splice/unix-to-tcp/8192-4     2.51GB/s ± 3%  2.53GB/s ± 7%     ~     (p=0.968 n=9+10)
Splice/unix-to-tcp/16384-4    3.52GB/s ± 8%  3.43GB/s ± 8%     ~     (p=0.739 n=10+10)
Splice/unix-to-tcp/32768-4    4.21GB/s ±14%  4.24GB/s ±13%     ~     (p=0.912 n=10+10)
Splice/unix-to-tcp/65536-4    3.65GB/s ± 5%  3.81GB/s ±11%     ~     (p=0.436 n=10+10)
Splice/unix-to-tcp/131072-4   3.65GB/s ± 1%  4.19GB/s ± 4%  +14.81%  (p=0.000 n=10+9)
Splice/unix-to-tcp/262144-4   3.62GB/s ± 3%  4.73GB/s ± 2%  +30.78%  (p=0.000 n=10+10)
Splice/unix-to-tcp/524288-4   3.59GB/s ± 3%  4.66GB/s ± 1%  +29.64%  (p=0.000 n=9+9)
Splice/unix-to-tcp/1048576-4  3.65GB/s ± 3%  4.70GB/s ± 3%  +29.01%  (p=0.000 n=10+9)

name                          old alloc/op   new alloc/op   delta
Splice/tcp-to-tcp/1024-4         0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/2048-4         0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/4096-4         0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/8192-4         0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/16384-4        0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/32768-4        0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/65536-4        0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/131072-4       0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/262144-4       0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/524288-4       0.00B          0.00B          ~     (all equal)
Splice/tcp-to-tcp/1048576-4      0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/1024-4        0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/2048-4        0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/4096-4        0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/8192-4        0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/16384-4       0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/32768-4       0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/65536-4       0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/131072-4      0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/262144-4      0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/524288-4      0.00B          0.00B          ~     (all equal)
Splice/unix-to-tcp/1048576-4     0.00B          0.00B          ~     (all equal)

name                          old allocs/op  new allocs/op  delta
Splice/tcp-to-tcp/1024-4          0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/2048-4          0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/4096-4          0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/8192-4          0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/16384-4         0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/32768-4         0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/65536-4         0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/131072-4        0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/262144-4        0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/524288-4        0.00           0.00          ~     (all equal)
Splice/tcp-to-tcp/1048576-4       0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/1024-4         0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/2048-4         0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/4096-4         0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/8192-4         0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/16384-4        0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/32768-4        0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/65536-4        0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/131072-4       0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/262144-4       0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/524288-4       0.00           0.00          ~     (all equal)
Splice/unix-to-tcp/1048576-4      0.00           0.00          ~     (all equal)

Change-Id: I899c1fdcb3f7b9997faf2b0d2336016a3f1fd3ea
Reviewed-on: https://go-review.googlesource.com/c/go/+/425051
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

2 years agoruntime: remove gcCPULimiterState padding for atomic fields alignment
Cuong Manh Le [Thu, 25 Aug 2022 04:25:32 +0000 (11:25 +0700)]
runtime: remove gcCPULimiterState padding for atomic fields alignment

assistTimePool and lastUpdate are now atomic.Int64, so they are
guaranteed to have 64-bit alignment, even on 32-bit platforms.

Change-Id: Ib6062a47c3a92d46d43899e1ae9d119e7f5b8bb0
Reviewed-on: https://go-review.googlesource.com/c/go/+/425460
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agocmd/go/internal/modload: convert atomicLoadPkgFlags.bits to atomic type
hopehook [Thu, 25 Aug 2022 03:11:59 +0000 (11:11 +0800)]
cmd/go/internal/modload: convert atomicLoadPkgFlags.bits to atomic type

Change-Id: I9e59530953439dec6f4524c5a7adc75c98c12b8f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425456
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agocmd/go/internal/trace: convert traceStarted to atomic type
hopehook [Thu, 25 Aug 2022 03:17:46 +0000 (11:17 +0800)]
cmd/go/internal/trace: convert traceStarted to atomic type

Change-Id: Ia4214a29775f1178273b9b7dc84c0420bfa968de
Reviewed-on: https://go-review.googlesource.com/c/go/+/425457
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agodebug/macho: use saferio to read segment and section data
Dan Kortschak [Wed, 24 Aug 2022 12:30:11 +0000 (22:00 +0930)]
debug/macho: use saferio to read segment and section data

Avoid allocating large amounts of memory for corrupt input.

No test case because the problem can only happen for invalid data.
Let the fuzzer find cases like this.

Change-Id: Ib09d5fea54aabcb6941e541b42689222fba69632
Reviewed-on: https://go-review.googlesource.com/c/go/+/425303
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Dan Kortschak <dan@kortschak.io>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: hopehook <hopehook@qq.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/go/internal/imports: include ToolTags in the Tags map
Zeke Lu [Thu, 25 Aug 2022 04:17:00 +0000 (04:17 +0000)]
cmd/go/internal/imports: include ToolTags in the Tags map

This fixes a regression introduced when the "race" mode tag was moved to
the ToolTags field in CL 358539.

Fixes #54468

Change-Id: I107771948a4fe9d743cc13d1c15f324212b08e03
GitHub-Last-Rev: d211e351ef3331f2c38b16d327d992a32ebbfe30
GitHub-Pull-Request: golang/go#54618
Reviewed-on: https://go-review.googlesource.com/c/go/+/425154
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/go/internal/par: change cacheEntry.done type to atomic.Bool
Ludi Rehak [Tue, 9 Aug 2022 01:55:56 +0000 (18:55 -0700)]
cmd/go/internal/par: change cacheEntry.done type to atomic.Bool

Change-Id: I95c941f83f74d57dfdd2d6803c9059691fb649b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/422176
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: hopehook <hopehook@qq.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agocmd/test2json: add signal handler
ianwoolf [Sun, 24 Jul 2022 10:25:18 +0000 (18:25 +0800)]
cmd/test2json: add signal handler

Updates #53563

Change-Id: I35a3fd56718e198f68cbf73075a78b2fbc66bd7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/419295
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Bryan Mills <bcmills@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Ryan Schuster <shuey19831@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
2 years agoreflect: add Value.SetZero
Joe Tsai [Fri, 10 Jun 2022 01:20:47 +0000 (18:20 -0700)]
reflect: add Value.SetZero

The v.SetZero method is a faster equivalent of v.Set(Zero(v.Type())).

Performance:

                     Direct         CachedZero     NewZero
SetZero/Bool         2.20ns ± 0%    8.97ns ± 5%    11.4ns ± 1%
SetZero/Int          3.08ns ± 1%    9.06ns ± 1%    11.5ns ± 0%
SetZero/Uint         2.88ns ± 1%    9.04ns ± 1%    11.7ns ± 5%
SetZero/Float        2.65ns ± 2%    9.05ns ± 1%    11.5ns ± 1%
SetZero/Complex      2.68ns ± 3%    9.31ns ± 1%    11.7ns ± 1%
SetZero/Array        6.69ns ± 4%    9.32ns ± 1%    11.8ns ± 1%
SetZero/Chan         3.31ns ± 1%    6.19ns ± 1%    8.20ns ± 1%
SetZero/Func         3.32ns ± 1%    6.20ns ± 0%    8.24ns ± 1%
SetZero/Interface    2.66ns ± 1%    9.31ns ± 1%    11.8ns ± 1%
SetZero/Map          3.31ns ± 1%    6.21ns ± 2%    8.19ns ± 1%
SetZero/Pointer      3.30ns ± 1%    6.22ns ± 1%    8.17ns ± 1%
SetZero/Slice        2.90ns ± 4%    9.13ns ± 1%    11.6ns ± 1%
SetZero/String       3.11ns ± 1%    9.30ns ± 1%    11.8ns ± 2%
SetZero/Struct       6.37ns ± 1%    9.18ns ± 4%    11.5ns ± 1%

where:

* Direct is measuring Value.SetZero
* CachedZero is measuring Value.Set with a cached Zero value
* NewZero is measuring Value.Set with a new Zero value

Fixes #52376

Change-Id: I793ca723aa97627824c5f5b356b2da30c8e46d71
Reviewed-on: https://go-review.googlesource.com/c/go/+/411476
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Dan Kortschak <dan@kortschak.io>
2 years agoruntime: convert forcegcstate.idle to internal atomic type
Andy Pan [Wed, 24 Aug 2022 19:03:35 +0000 (03:03 +0800)]
runtime: convert forcegcstate.idle to internal atomic type

Note that this changes a few unsynchronized operations of forcegcstate.idle to synchronized operations.

Updates #53821

Change-Id: I041654cc84a188fad45e2df7abce3a434f9a1f15
Reviewed-on: https://go-review.googlesource.com/c/go/+/425361
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agoruntime: convert g.selectDone to atomic type
hopehook [Thu, 25 Aug 2022 01:42:30 +0000 (09:42 +0800)]
runtime: convert g.selectDone to atomic type

On the write side, g.selectDone has been converted
from non-atomic to atomic access.

For #53821.

Change-Id: Iac46bc6acce7eed51dfd990285dd57f0d58b4ae2
Reviewed-on: https://go-review.googlesource.com/c/go/+/425414
Run-TryBot: hopehook <hopehook@qq.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: convert notifyList.wait to atomic type
hopehook [Fri, 26 Aug 2022 03:57:46 +0000 (11:57 +0800)]
runtime: convert notifyList.wait to atomic type

For #53821

Change-Id: Ib096332fe6111bbcd2f5c4cbb29c2fef7a808e7e
Reviewed-on: https://go-review.googlesource.com/c/go/+/425784
Run-TryBot: hopehook <hopehook@qq.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agoruntime: convert dlogger.owned to atomic type
hopehook [Fri, 26 Aug 2022 03:28:48 +0000 (11:28 +0800)]
runtime: convert dlogger.owned to atomic type

Note that this changes a non-atomic operation to atomic operation.

For #53821

Change-Id: I798914f505c8d7f85f9d7629fdc6493363a20aa1
Reviewed-on: https://go-review.googlesource.com/c/go/+/425782
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: hopehook <hopehook@qq.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agoruntime: convert p.statsSeq to internal atomic type
hopehook [Thu, 25 Aug 2022 02:13:03 +0000 (10:13 +0800)]
runtime: convert p.statsSeq to internal atomic type

For #53821.

Change-Id: I1cab3671a29c218b8a927aba9064e63b65900173
Reviewed-on: https://go-review.googlesource.com/c/go/+/425416
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: hopehook <hopehook@qq.com>

2 years agocmd/compile: align stack offset to alignment larger than PtrSize
Cherry Mui [Tue, 23 Aug 2022 22:19:06 +0000 (18:19 -0400)]
cmd/compile: align stack offset to alignment larger than PtrSize

In typebits.Set we check that the offset is a multiple of the
alignment, which makes perfect sense. But for values like
atomic.Int64, which has 8-byte alignment even on 32-bit platforms
(i.e. the alignment is larger than PtrSize), if it is on stack it
may be under-aligned, as the stack frame is only PtrSize aligned.

Normally we would prevent such values on stack, as the escape
analysis force values with higher alignment to heap. But for a
composite literal assignment like x = AlignedType{...}, the
compiler creates an autotmp for the RHS then copies it to the LHS.
The autotmp is on stack and may be under-aligned. Currently this
may cause an ICE in the typebits.Set check.

This CL makes it align the _offset_ of the autotmp to 8 bytes,
which satisfies the check. Note that this is actually lying: the
actual address at run time may not necessarily be 8-byte
aligned as we only align SP to 4 bytes.

The under-alignment is probably okay. The only purpose for the
autotmp is to copy the value to the LHS, and the copying code we
generate (at least currently) doesn't care the alignment beyond
stack alignment.

Fixes #54638.

Change-Id: I13c16afde2eea017479ff11dfc24092bcb8aba6a
Reviewed-on: https://go-review.googlesource.com/c/go/+/425256
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoreflect: FuncOf support more than 50 arguments
cuiweixie [Thu, 25 Aug 2022 06:41:23 +0000 (14:41 +0800)]
reflect: FuncOf support more than 50 arguments

Fixes #54669

Change-Id: I34cbe729d187437ddeafbaa910af6ed001b2603f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425461
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

2 years agogo/types: provide a better error message for [...] array types
Robert Griesemer [Thu, 25 Aug 2022 21:50:27 +0000 (14:50 -0700)]
go/types: provide a better error message for [...] array types

This matches types2 behavior.

For #54511.

Change-Id: Iea906e9fec7e334b7aa7f481de87373fa93d1c7c
Reviewed-on: https://go-review.googlesource.com/c/go/+/425715
Reviewed-by: Robert Griesemer <gri@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agogo/types: avoid declared but not used error for a couple of cases
Robert Griesemer [Thu, 25 Aug 2022 21:30:15 +0000 (14:30 -0700)]
go/types: avoid declared but not used error for a couple of cases

The change in typexpr.go matches types2 behavior.

For #54511.

Change-Id: I79c922a94f2ee0440c1814140935c321439c7d25
Reviewed-on: https://go-review.googlesource.com/c/go/+/425714
Reviewed-by: Robert Griesemer <gri@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
2 years agoruntime: convert netpollWaiters to internal atomic type
Andy Pan [Wed, 24 Aug 2022 18:26:08 +0000 (02:26 +0800)]
runtime: convert netpollWaiters to internal atomic type

Updates #53821

Change-Id: I8776382b3eb0b7752cfc0d9287b707039d3f05c6
Reviewed-on: https://go-review.googlesource.com/c/go/+/425358
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: hopehook <hopehook@qq.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agogo/types: remove support for "ERROR HERE" error markers in tests
Robert Griesemer [Thu, 25 Aug 2022 20:48:42 +0000 (13:48 -0700)]
go/types: remove support for "ERROR HERE" error markers in tests

There are only two tests that rely on the "ERROR HERE" markers;
yet those tests are trivialy adjustable (by adding an explicit
semicolon) such that they can just use the "ERROR" markers.

For #54511.

Change-Id: Idbb96ca8d35ae2584d195a4ac7c92640b8b492c5
Reviewed-on: https://go-review.googlesource.com/c/go/+/425674
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>

2 years agocmd/trace: display goroutines (PC=0) with clearer description
hitzhangjie [Thu, 25 Aug 2022 09:34:49 +0000 (09:34 +0000)]
cmd/trace: display goroutines (PC=0) with clearer description

This PR fixes: #54425 #49994

Change-Id: Id60a3ba6930f8e29b12b6d8f80945decd2ce31bc
GitHub-Last-Rev: 60a040aa2f111f64f571597799ce7ca317e1d281
GitHub-Pull-Request: golang/go#54575
Reviewed-on: https://go-review.googlesource.com/c/go/+/425042
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agoruntime: update gctrace docs and print lastStackScan instead of max
Michael Anthony Knyszek [Wed, 24 Aug 2022 20:52:51 +0000 (20:52 +0000)]
runtime: update gctrace docs and print lastStackScan instead of max

This change updates the gctrace docs to include stacks and globals in
the format line, and prints lastStackScan for "# MB stacks" instead of
maxStackScan, which is more accurate.

Fixes #54649.

Change-Id: Ibff2c390c9c9bf2b24b5b4e98ca346cc98d7cb2e
Reviewed-on: https://go-review.googlesource.com/c/go/+/425366
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agocrypto/x509: don't panic marshaling invalid ECDSA keys
Filippo Valsorda [Tue, 9 Aug 2022 20:02:57 +0000 (13:02 -0700)]
crypto/x509: don't panic marshaling invalid ECDSA keys

MarshalPKIXPublicKey, CreateCertificate, CreateCertificateRequest,
MarshalECPrivateKey, and MarshalPKCS8PrivateKey started raising a panic
when encoding an invalid ECDSA key in Go 1.19. Since they have an error
return value, they should return an error instead.

Fixes #54288

Change-Id: Iba132cd2f890ece36bb7d0396eb9a9a77bdb81df
Reviewed-on: https://go-review.googlesource.com/c/go/+/422298
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agogo/parser: match go/defer error message of syntax package
Robert Griesemer [Sat, 20 Aug 2022 02:01:05 +0000 (19:01 -0700)]
go/parser: match go/defer error message of syntax package

Adjust corresponding type checker tests accordingly.

For #54511.

Change-Id: Ieaf29f26c0877973fc0acbde35292cd69a4b709c
Reviewed-on: https://go-review.googlesource.com/c/go/+/425007
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agopath/filepath, io/fs: add SkipAll
Paschalis Tsilias [Sat, 13 Nov 2021 00:18:30 +0000 (02:18 +0200)]
path/filepath, io/fs: add SkipAll

Fixes #47209

Change-Id: If75b0dd38f2c30a23517205d80c7a6683a5c921c
Reviewed-on: https://go-review.googlesource.com/c/go/+/363814
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Bryan Mills <bcmills@google.com>
2 years agocmd/compile/internal/noder: fix inlined function literal positions
Matthew Dempsky [Wed, 24 Aug 2022 22:03:50 +0000 (15:03 -0700)]
cmd/compile/internal/noder: fix inlined function literal positions

When inlining function calls, we rewrite the position information on
all of the nodes to keep track of the inlining context. This is
necessary so that at runtime, we can synthesize additional stack
frames so that the inlining is transparent to the user.

However, for function literals, we *don't* want to apply this
rewriting to the underlying function. Because within the function
literal (when it's not itself inlined), the inlining context (if any)
will have already be available at the caller PC instead.

Unified IR was already getting this right in the case of user-written
statements within the function literal, which is what the unit test
for #46234 tested. However, it was still using inline-adjusted
positions for the function declaration and its parameters, which
occasionally end up getting used for generated code (e.g., loading
captured values from the closure record).

I've manually verified that this fixes the hang in
https://go.dev/play/p/avQ0qgRzOgt, and spot-checked the
-d=pctab=pctoinline output for kube-apiserver and kubelet and they
seem better.

However, I'm still working on a more robust test for this (hence
"Updates" not "Fixes") and internal assertions to verify that we're
emitting correct inline trees. In particular, there are still other
cases (even in the non-unified frontend) where we're producing
corrupt (but at least acyclic) inline trees.

Updates #54625.

Change-Id: Iacfd2e1eb06ae8dc299c0679f377461d3d46c15a
Reviewed-on: https://go-review.googlesource.com/c/go/+/425395
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agotesting: add Elapsed method to testing.B
hopehook [Sat, 30 Jul 2022 04:47:51 +0000 (12:47 +0800)]
testing: add Elapsed method to testing.B

Elapsed returns the measured elapsed time of the benchmark,
but does not change the running state of the timer.

Fixes #43620.

Change-Id: Idd9f64c4632518eec759d2ffccbf0050d84fcc03
Reviewed-on: https://go-review.googlesource.com/c/go/+/420254
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: hopehook <hopehook@qq.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
2 years agogo/types: match types2 string when printing composite literals
Robert Griesemer [Sat, 20 Aug 2022 01:31:56 +0000 (18:31 -0700)]
go/types: match types2 string when printing composite literals

Given a composite literal type S, rather than always printing
(S literal) for a composite literals, print S{} if the literal
has no elements, and print S{…} as a short form (suitable for
error messages) if there are elements. This matches types2 and
also Go1.17 compiler behavior (except that the original compiler
would print ... rather than …). Using … rather than ... makes
it clearer that we don't have real Go syntax, and it's also more
compact.

For #54511.

Change-Id: I5991e8060232f16ecbf4a1fe4ae091598fc76b68
Reviewed-on: https://go-review.googlesource.com/c/go/+/425006
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
2 years agocrypto/tls: support ECDHE when ec_point_formats is missing
Filippo Valsorda [Wed, 24 Aug 2022 10:39:20 +0000 (12:39 +0200)]
crypto/tls: support ECDHE when ec_point_formats is missing

Fixes #49126

Change-Id: I9d6f6392b1a6748bdac1d2c6371b22d75829a2b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/425295
Run-TryBot: Filippo Valsorda <filippo@golang.org>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Alex Scheel <alex.scheel@hashicorp.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agoruntime: mark morestack_noctxt SPWRITE on LR architectures
Cherry Mui [Wed, 24 Aug 2022 22:33:21 +0000 (18:33 -0400)]
runtime: mark morestack_noctxt SPWRITE on LR architectures

On LR architectures, morestack (and morestack_noctxt) are called
with a special calling convention, where the caller doesn't save
LR on stack but passes it as a register, which morestack will save
to g.sched.lr. The stack unwinder currently doesn't understand it,
and would fail to unwind from it. morestack already writes SP (as
it switches stack), but morestack_noctxt (which tailcalls
morestack) doesn't. If a profiling signal lands right in
morestack_noctxt, the unwinder will try to unwind the stack and
go off, and possibly crash.

Marking morestack_noctxt SPWRITE stops the unwinding.

Ideally we could teach the unwinder about the special calling
convention, or change the calling convention to be less special
(so the unwinder doesn't need to fetch a register from the signal
context). This is a stop-gap solution, to stop the unwinder from
crashing.

Fixes #54332.

Change-Id: I75295f2e27ddcf05f1ea0b541aedcb9000ae7576
Reviewed-on: https://go-review.googlesource.com/c/go/+/425396
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agomisc/cgo/testsanitizers: determine compiler version for tsan tests on ppc64le
Lynn Boger [Wed, 24 Aug 2022 17:14:20 +0000 (12:14 -0500)]
misc/cgo/testsanitizers: determine compiler version for tsan tests on ppc64le

Some tests in misc/cgo/testsanitizers had been disabled on ppc64le
until recently, due to an intermittent error in the tsan tests,
with the goal of trying to understand the failure.

After further investigation, I found that the code for tsan within
gcc does not work consistently when ASLR is enabled on ppc64le. A
fix for that problem was integrated in gcc 9.

This adds a check to testsanitizers to determine the gcc compiler
version on ppc64le and skip the test if the version is too old.

A similar check is needed for asan too.

Updates #54645

Change-Id: I70717d1aa9e967cf1e871566e72b3862b91fea3f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425355
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Archana Ravindar <aravind5@in.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>

2 years agogo/types, types2: add more tests for unsafe.Slice/SliceData/String/StringData
Robert Griesemer [Wed, 24 Aug 2022 23:59:22 +0000 (16:59 -0700)]
go/types, types2: add more tests for unsafe.Slice/SliceData/String/StringData

Also:
- fine-tune the implementation for some of the new builtin functions
- make sure the go/types code is an exact as possible copy of the
  types2 code
- fix the description and examples for errorcodes.go

Follow-up on CL 423754.

For #53003.

Change-Id: I5c70b74e90c724cf6c842cedc6f8ace26fde372b
Reviewed-on: https://go-review.googlesource.com/c/go/+/425454
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agogo/types,types2: add support for unsafe.{String,StringData,SliceData}
cuiweixie [Sun, 14 Aug 2022 06:41:19 +0000 (14:41 +0800)]
go/types,types2: add support for unsafe.{String,StringData,SliceData}

For #53003
Change-Id: Id3125268523fed855ffac20cde6128010e3513f0
Reviewed-on: https://go-review.googlesource.com/c/go/+/423754
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agogo/internal/gcimporter: call Complete on cloned Interfaces too
Matthew Dempsky [Wed, 24 Aug 2022 20:47:48 +0000 (13:47 -0700)]
go/internal/gcimporter: call Complete on cloned Interfaces too

For "type T interface{ M() }", go/types users expect T's underlying
interface type to specify T as the receiver parameter type (#49906).
The unified importer handles this by cloning the interface to rewrite
the receiver parameters before calling SetUnderlying.

I missed in CL 425360 that these interfaces would need to have
Complete called too.

Manually tested to confirm that this actually fixes "go test -race
golang.org/x/tools/go/analysis/internal/checker" now (when both CLs
are ported to the x/tools importer).

Updates #54653.

Change-Id: I51e6db925db56947cd39dbe880230f14734ca01c
Reviewed-on: https://go-review.googlesource.com/c/go/+/425365
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agogo/internal/gcimporter: call Interface.Complete in unified importer
Matthew Dempsky [Wed, 24 Aug 2022 18:53:42 +0000 (11:53 -0700)]
go/internal/gcimporter: call Interface.Complete in unified importer

To support concurrent use of the go/types API, importers need to call
Interface.Complete on constructed interfaces before returning.

There's an issue that the interfaces may contain embedded defined
types, whose underlying type isn't known yet. This issue will
eventually go away once CL 424876 lands, but that CL needs to wait for
CL 424854 to re-land, which needs to wait for CL 421879 to land...

In the mean time, this CL implements the same solution used by the
indexed importer: maintaining a list of constructed interfaces, and
calling Interface.Complete on them after the SetUnderlying loop and
just before returning the imported package.

Updates #54653.

Change-Id: I0f42c915a4b7d28c628bbab7ac2eab2415c7858f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425360
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agodoc/go1.20: populate initial TODOs via relnote
Dmitri Shuralyov [Wed, 24 Aug 2022 17:04:42 +0000 (13:04 -0400)]
doc/go1.20: populate initial TODOs via relnote

Add content generated by 'relnote -html'. This covers all of known
TODOs available by 2022-08-24 such that relnote produces no output
when -exclude-from=doc/go1.20.html flag is used.

For #54202.

Change-Id: Icd9be5364bc73c3de56d5d3039e6f313908af38e
Reviewed-on: https://go-review.googlesource.com/c/go/+/425354
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>

2 years agocmd/compile: deadcode for LoweredMuluhilo on riscv64
Wayne Zuo [Wed, 24 Aug 2022 14:17:51 +0000 (22:17 +0800)]
cmd/compile: deadcode for LoweredMuluhilo on riscv64

This is a follow up of CL 425101 on RISCV64.

According to RISCV Volume 1, Unprivileged Spec v. 20191213 Chapter 7.1:
If both the high and low bits of the same product are required, then the
recommended code sequence is: MULH[[S]U] rdh, rs1, rs2; MUL rdl, rs1, rs2
(source register specifiers must be in same order and rdh cannot be the
same as rs1 or rs2). Microarchitectures can then fuse these into a single
multiply operation instead of performing two separate multiplies.

So we should not split Muluhilo to separate instructions.

Updates #54607

Change-Id: If47461f3aaaf00e27cd583a9990e144fb8bcdb17
Reviewed-on: https://go-review.googlesource.com/c/go/+/425203
Auto-Submit: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agodoc: start draft Go 1.20 release notes
Dmitri Shuralyov [Wed, 24 Aug 2022 15:21:55 +0000 (11:21 -0400)]
doc: start draft Go 1.20 release notes

This initial release notes template is based on previous releases.
CL 425354 adds initial content generated by 'relnote -html'.

For #54202.

Change-Id: I36e15723edc9610d61986ba226ee84c2b30a33af
Reviewed-on: https://go-review.googlesource.com/c/go/+/425334
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agonet/http: remove unused doneChan
Alexander Yastrebov [Wed, 1 Jun 2022 00:30:31 +0000 (00:30 +0000)]
net/http: remove unused doneChan

The https://golang.org/cl/43230 removed use of `getDoneChan`.

Change-Id: I33390c0e3aea6d98367363773ebe39d9c1f64ae9
GitHub-Last-Rev: fe1e4154eaad4e485acfe4272f27d63367144e47
GitHub-Pull-Request: golang/go#53172
Reviewed-on: https://go-review.googlesource.com/c/go/+/409538
Run-TryBot: Damien Neil <dneil@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: hopehook <hopehook@qq.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/go: skip link_syso tests in short mode
Bryan C. Mills [Wed, 24 Aug 2022 14:10:12 +0000 (10:10 -0400)]
cmd/go: skip link_syso tests in short mode

These tests invoke the system C compiler and linker.
Skipping them saves a little over half a second of time in short mode.

Updates #54423.

Change-Id: I3e8aa7b53c0c91f7d1e001ec2cd5f7b4954de52d
Reviewed-on: https://go-review.googlesource.com/c/go/+/425206
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agocmd/go/internal/work: make NewBuilder safe for concurrent and repeated use
Bryan C. Mills [Tue, 23 Aug 2022 20:33:01 +0000 (16:33 -0400)]
cmd/go/internal/work: make NewBuilder safe for concurrent and repeated use

Ever since 'go build' was added (in CL 5483069), it has used an atexit
handler to clean up working directories. At some point (prior to CL
95900044), Init was called multiple times per builder, registering
potentially many atexit handlers that execute asynchronously and make
debugging more difficult.

The use of an AtExit handler also makes the Builder (and anything that
uses it) prone to races: the base.AtExit API is not designed for
concurrent use, but cmd/go is becoming increasingly concurrent over
time. The AtExit handler also makes the Builder inappropriate to use
within a unit-test, since the handlers do not run during the test
function and accumulate over time.

This change makes NewBuilder safe for concurrent use by registering
the AtExit handler only once (during BuildInit, which was already not
safe for concurrent use), and using a sync.Map to store the set of
builders that need cleanup in case of an unclean exit. In addition, it
causes the test variant of cmd/go to fail if any Builder instance
leaks from a clean exit, helping to ensure that functions that create
Builders do not leak them indefinitely, especially in tests.

Updates #54423.

Change-Id: Ia227b15b8fa53c33177c71271d756ac0858feebe
Reviewed-on: https://go-review.googlesource.com/c/go/+/425254
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agocmd/go: avoid overwriting cached Origin metadata
Bryan C. Mills [Tue, 23 Aug 2022 22:00:34 +0000 (18:00 -0400)]
cmd/go: avoid overwriting cached Origin metadata

Fixes #54631.

Change-Id: I17d2fa282642aeb1ae2a6e29a0756b8960bea34b
Reviewed-on: https://go-review.googlesource.com/c/go/+/425255
Run-TryBot: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
2 years agocmd/go: avoid registering AtExit handlers in tests
Bryan C. Mills [Wed, 24 Aug 2022 13:45:18 +0000 (09:45 -0400)]
cmd/go: avoid registering AtExit handlers in tests

Ever since 'go build' was added (in CL 5483069), it has used an atexit
handler to clean up working directories.

CL 154109 introduced 'cc' command to the script test framework that
called Init on a builder once per invocation. Unfortunately, since
base.AtExit is unsynchronized, the Init added there caused any script
that invokes that command to be unsafe for concurrent use.

This change fixes the race by having the 'cc' command pass in its
working directory instead of allowing the Builder to allocate one.
Following modern Go best practices, it also replaces the in-place Init
method (which is prone to typestate and aliasing bugs) with a
NewBuilder constructor function.

Fixes #54423.

Change-Id: I8fc2127a7d877bb39a1174e398736bb51d03d4d2
Reviewed-on: https://go-review.googlesource.com/c/go/+/425205
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
2 years agocmd/compile: defer transitive inlining until after AST is edited
Matthew Dempsky [Tue, 23 Aug 2022 23:34:26 +0000 (16:34 -0700)]
cmd/compile: defer transitive inlining until after AST is edited

This CL changes the inliner to process transitive inlining iteratively
after the AST has actually been edited, rather than recursively and
immediately. This is important for handling indirect function calls
correctly, because ir.reassigned walks the function body looking for
reassignments; whereas previously the inlined reassignments might not
have been actually added to the AST yet.

Fixes #54632.

Change-Id: I0dd69813c8a70b965174e0072335bc00afedf286
Reviewed-on: https://go-review.googlesource.com/c/go/+/425257
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
2 years agoimage/png: have DecodeConfig read tRNS chunks
Nigel Tao [Fri, 19 Aug 2022 06:39:38 +0000 (16:39 +1000)]
image/png: have DecodeConfig read tRNS chunks

Fixes #54325

Change-Id: Ie468180c4d6f21db7672dd71bd2a40f3a5881b7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/424917
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Nigel Tao (INACTIVE; USE @golang.org INSTEAD) <nigeltao@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
2 years agoreflect: optimize Value.IsZero
Joe Tsai [Fri, 10 Jun 2022 01:32:53 +0000 (18:32 -0700)]
reflect: optimize Value.IsZero

If a struct or array is comparable, then we can leverage rtype.equal,
which is almost always faster than what Go reflection can achieve.

As a secondary optimization, pre-compute Value.Len and Value.NumField
outside of the loop conditional.

Performance:

name                       old time/op  new time/op  delta
IsZero/ArrayComparable      136ns ± 4%    16ns ± 1%  -88.28%  (p=0.008 n=5+5)
IsZero/ArrayIncomparable    197ns ±10%   123ns ± 1%  -37.74%  (p=0.008 n=5+5)
IsZero/StructComparable    26.4ns ± 0%   9.6ns ± 1%  -63.68%  (p=0.016 n=4+5)
IsZero/StructIncomparable  43.5ns ± 1%  27.8ns ± 1%  -36.21%  (p=0.008 n=5+5)

The incomparable types gain a performance boost since
they are generally constructed from nested comparable types.

Change-Id: If2c1929f8bb1b5b19306ef0c69f3c95a27d4b60d
Reviewed-on: https://go-review.googlesource.com/c/go/+/411478
Reviewed-by: Dan Kortschak <dan@kortschak.io>
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoencoding/json: move some misplaced benchmark tests to bench_test.go
Andy Pan [Sat, 20 Aug 2022 11:21:39 +0000 (19:21 +0800)]
encoding/json: move some misplaced benchmark tests to bench_test.go

Change-Id: I5987eed00ee825421abe62699a06e9b66499f35f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425016
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agointernal/singleflight: make DoChan only return Result channel
Cuong Manh Le [Sat, 13 Aug 2022 17:35:24 +0000 (00:35 +0700)]
internal/singleflight: make DoChan only return Result channel

So next CL can delete "internal/singleflight" and vendor
"golang.org/x/sync/singleflight" instead.

For #31697

Change-Id: I020da1e5a48d484637b538c010029218f5a4a744
Reviewed-on: https://go-review.googlesource.com/c/go/+/423655
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agonet: avoid relying on singleflight.Group.DoChan to detect hook called
Cuong Manh Le [Sat, 13 Aug 2022 17:29:20 +0000 (00:29 +0700)]
net: avoid relying on singleflight.Group.DoChan to detect hook called

So next CLs can revert changes to "internal/singleflight" in CL #82795,
then replace it with "golang.org/x/sync/singleflight" instead.

For #31697

Change-Id: I873ce30d7e051539aa6dc5d4f21e558869a6d132
Reviewed-on: https://go-review.googlesource.com/c/go/+/423654
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>

2 years agocmd/internal/obj/loong64: add MASKEQZ and MASKNEZ instructions support
Wayne Zuo [Sat, 9 Jul 2022 03:28:50 +0000 (11:28 +0800)]
cmd/internal/obj/loong64: add MASKEQZ and MASKNEZ instructions support

Change-Id: Ied16c3be47c863a94d46bd568191057ded4b7d0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/416734
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Wayne Zuo <wdvxdr@golangcn.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: xiaodong liu <teaofmoli@gmail.com>
2 years agoruntime/pprof: remove round-to-file-start adjustment
Keith Randall [Tue, 16 Aug 2022 19:35:36 +0000 (12:35 -0700)]
runtime/pprof: remove round-to-file-start adjustment

This causes a problem in the test sometimes. With a mapping like:

00400000-00411000 r--p 00000000 fe:01 4459044                            /tmp/go-build1710804385/b001/pprof.test
00411000-00645000 r-xp 00011000 fe:01 4459044                            /tmp/go-build1710804385/b001/pprof.test

The removed code would make the first mapping 0x400000-0x645000. Tests
then grab the first few addresses to use as PCs, thinking they are in
an executable range. But those addresses are really not in an
executable range, causing the tests to fail.

Change-Id: I5a69d0259d1fd70ff9745df1cbad4d54c5898e7b
Reviewed-on: https://go-review.googlesource.com/c/go/+/424295
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Keith Randall <khr@golang.org>

2 years agocmd/compile: generic constant folding: Floor Ceil Trunc RoundToEven
Jorropo [Wed, 8 Jun 2022 21:27:05 +0000 (23:27 +0200)]
cmd/compile: generic constant folding: Floor Ceil Trunc RoundToEven

Change-Id: I553a6d0bd3ae45e5bf62191411e71102b3f44cd8
Reviewed-on: https://go-review.googlesource.com/c/go/+/411215
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Joedian Reid <joedian@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
2 years agonet/url: consistently remove ../ elements in JoinPath
Damien Neil [Fri, 12 Aug 2022 23:21:09 +0000 (16:21 -0700)]
net/url: consistently remove ../ elements in JoinPath

JoinPath would fail to remove relative elements from the start of
the path when the first path element is "".

In addition, JoinPath would return the original path unmodified
when provided with no elements to join, violating the documented
behavior of always cleaning the resulting path.

Correct both these cases.

    JoinPath("http://go.dev", "../go")
    // before: http://go.dev/../go
    // after:  http://go.dev/go

    JoinPath("http://go.dev/../go")
    // before: http://go.dev/../go
    // after:  http://go.dev/go

Fixes #54385.

Change-Id: I6d22cd160d097c50703dd96e4f453c6c118fd5d9
Reviewed-on: https://go-review.googlesource.com/c/go/+/423514
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
2 years agocmd/compile: move SSA rotate instruction detection to arch-independent rules
Keith Randall [Thu, 4 Aug 2022 05:58:30 +0000 (22:58 -0700)]
cmd/compile: move SSA rotate instruction detection to arch-independent rules

Detect rotate instructions while still in architecture-independent form.
It's easier to do here, and we don't need to repeat it in each
architecture file.

Change-Id: I9396954b3f3b3bfb96c160d064a02002309935bb
Reviewed-on: https://go-review.googlesource.com/c/go/+/421195
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Eric Fang <eric.fang@arm.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Joedian Reid <joedian@golang.org>
Reviewed-by: Ruinan Sun <Ruinan.Sun@arm.com>
Run-TryBot: Keith Randall <khr@golang.org>

2 years agocmd/compile: fix error message formatting in decomposeBuiltInPhi
Jorropo [Thu, 9 Jun 2022 04:50:54 +0000 (06:50 +0200)]
cmd/compile: fix error message formatting in decomposeBuiltInPhi

Change-Id: Ib2b2eb6bebb59134f671a79739ee1acaf32ed443
Reviewed-on: https://go-review.googlesource.com/c/go/+/411217
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Joedian Reid <joedian@golang.org>
2 years agoall: append(bytes, str...) works out of the box
Daniel Martí [Mon, 22 Aug 2022 21:11:03 +0000 (22:11 +0100)]
all: append(bytes, str...) works out of the box

From the append docs in the builtin package:

As a special case, it is legal to append a string to a byte slice, like this:

slice = append([]byte("hello "), "world"...)

Change-Id: Ib14039a7476873b12a3aefccd8863e8d628b9249
Reviewed-on: https://go-review.googlesource.com/c/go/+/425102
Reviewed-by: hopehook <hopehook@qq.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>

2 years agonet/http/httputil: do not duplicate 'Connection: close' headers
Miguel Ángel Jimeno [Tue, 23 Aug 2022 12:06:19 +0000 (14:06 +0200)]
net/http/httputil: do not duplicate 'Connection: close' headers

Fixes #54616.

Change-Id: I46f41bbccff7b49f26f60c1edf65acfd1701aed5
Reviewed-on: https://go-review.googlesource.com/c/go/+/425155
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>

2 years agosyscall: rely on utf16.AppendRune
qmuntal [Mon, 22 Aug 2022 16:25:29 +0000 (18:25 +0200)]
syscall: rely on utf16.AppendRune

Using utf16.AppendRune instead of utf16.Encode safe a bunch
of allocations across the board, as many higher level functions
use it to call Windows syscalls, for example to `os` package:

name                old alloc/op   new alloc/op   delta
Readdirname-12        15.6kB ± 0%    15.6kB ± 0%   +0.26%  (p=0.008 n=5+5)
Readdir-12            29.4kB ± 0%    29.4kB ± 0%   +0.14%  (p=0.008 n=5+5)
ReadDir-12            29.4kB ± 0%    29.4kB ± 0%   +0.14%  (p=0.016 n=4+5)
StatDot-12              552B ± 0%      560B ± 0%   +1.45%  (p=0.008 n=5+5)
StatFile-12             512B ± 0%      336B ± 0%  -34.38%  (p=0.008 n=5+5)
StatDir-12              432B ± 0%      288B ± 0%  -33.33%  (p=0.008 n=5+5)
LstatDot-12             552B ± 0%      560B ± 0%   +1.45%  (p=0.008 n=5+5)
LstatFile-12            512B ± 0%      336B ± 0%  -34.38%  (p=0.008 n=5+5)
LstatDir-12             432B ± 0%      288B ± 0%  -33.33%  (p=0.008 n=5+5)
StatFile-12             4.00 ± 0%      2.00 ± 0%  -50.00%  (p=0.008 n=5+5)
StatDir-12              4.00 ± 0%      2.00 ± 0%  -50.00%  (p=0.008 n=5+5)
LstatFile-12            4.00 ± 0%      2.00 ± 0%  -50.00%  (p=0.008 n=5+5)
LstatDir-12             4.00 ± 0%      2.00 ± 0%  -50.00%  (p=0.008 n=5+5)

Updates #51786

Change-Id: I0a088cf1a96e9c304da9311bb3895b70443c1637
Reviewed-on: https://go-review.googlesource.com/c/go/+/425054
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

2 years agodebug/plan9obj: don't crash on EOF before symbol type
Ian Lance Taylor [Tue, 23 Aug 2022 00:42:54 +0000 (17:42 -0700)]
debug/plan9obj: don't crash on EOF before symbol type

No debug/plan9obj test case because the problem can only happen for
invalid data. Let the fuzzer find cases like this.

For #47653
Fixes #54585

Change-Id: I8d3e15725b9bc09dd0e6f2750769987021f5e982
Reviewed-on: https://go-review.googlesource.com/c/go/+/425115
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
2 years agodebug/macho: don't use narch for seenArches map size
Ian Lance Taylor [Tue, 23 Aug 2022 00:37:07 +0000 (17:37 -0700)]
debug/macho: don't use narch for seenArches map size

If narch is very large we would allocate a lot of memory for seenArches.
In practice we aren't going to see many different architectures so
don't bother to specify a size for the seenArches map.

No debug/macho test case because the problem can only happen for
invalid data. Let the fuzzer find cases like this.

For #47653
For #52523

Change-Id: I5a3b0e3aa6172ddffd6f44d9ae513c39a00d8764
Reviewed-on: https://go-review.googlesource.com/c/go/+/425114
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
2 years agostrconv: optimize Parse for []byte arguments
Joe Tsai [Thu, 26 Aug 2021 19:02:47 +0000 (12:02 -0700)]
strconv: optimize Parse for []byte arguments

When one has a []byte on hand, but desires to call the Parse functions,
the conversion from []byte to string would allocate.

    var b []byte = ...
    v, err := strconv.ParseXXX(string(b), ...)

This changes it such that the input string never escapes from
any of the Parse functions. Together with the compiler optimization
where the compiler stack allocates any string smaller than 32B
this makes most valid inputs for strconv.ParseXXX(string(b), ...)
not require an allocation for the input string.
For example, the longest int64 or uint64 encoded in decimal is 20B.
Also, the longest decimal formatting of a float64 in appendix B
of RFC 8785 is 25B.

Previously, this was not possible since the input leaked to the error,
which causes the prover to give up and instead heap copy the []byte.
We fix this by copying the input string in the error case.
The advantage of this change is that you can now call strconv.ParseXXX
with a []byte without allocations (most times) in the non-error case.
The detriment is that the error-case now has an extra allocation.
We should optimize for the non-error path, rather than the error path.

The effects of this change is transitively seen through packages
that must use strconv.ParseXXX on a []byte such as "encoding/json":

    name              old time/op    new time/op    delta
    UnmarshalFloat64  186ns          157ns          -15.89%  (p=0.000 n=10+10)

    name              old alloc/op   new alloc/op   delta
    UnmarshalFloat64  148B           144B            -2.70%  (p=0.000 n=10+10)

    name              old allocs/op  new allocs/op  delta
    UnmarshalFloat64  2.00           1.00           -50.00%  (p=0.000 n=10+10)

In order for "encoding/json" to benefit, there needs to be a
small change made to how "encoding/json" calls strconv.ParseXXX.
That will be a future change.

Credit goes to Jeff Wendling for a similar patch.

Fixes #42429

Change-Id: I512d6927f965f82e95bd7ec14a28a587f23b7203
Reviewed-on: https://go-review.googlesource.com/c/go/+/345488
Reviewed-by: Martin Möhrmann <martin@golang.org>
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agobytes: rely on utf8.AppendRune
Joe Tsai [Thu, 16 Jun 2022 00:03:27 +0000 (17:03 -0700)]
bytes: rely on utf8.AppendRune

This is both simpler and more performant.

WriteRune                                 23.9µs ± 1%   22.8µs ± 8%   -4.43%  (p=0.006 n=8+10)
ToUpper/longɐstringɐwithɐnonasciiⱯchars    573ns ± 4%    474ns ± 6%  -17.27%  (p=0.000 n=10+10)
ToUpper/ɐɐɐɐɐ                              236ns ± 6%    202ns ± 5%  -14.12%  (p=0.000 n=10+10)
ToUpper/a\u0080\U0010ffff                 98.8ns ± 6%   91.2ns ± 3%   -7.67%  (p=0.000 n=10+10)
ToLower/LONGⱯSTRINGⱯWITHⱯNONASCIIⱯCHARS    511ns ± 3%    409ns ± 4%  -20.02%  (p=0.000 n=10+10)
ToLower/ⱭⱭⱭⱭⱭ                              178ns ± 4%    173ns ± 2%   -2.76%  (p=0.005 n=10+10)
ToLower/A\u0080\U0010ffff                  100ns ± 3%     91ns ± 2%   -8.55%  (p=0.000 n=10+10)

Change-Id: I968ef194da2c115cbdcac2d3575c34c65e6a8a56
Reviewed-on: https://go-review.googlesource.com/c/go/+/412337
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>

2 years agoruntime: name nil semaphore of pollDesc
Andy Pan [Fri, 19 Aug 2022 10:45:51 +0000 (18:45 +0800)]
runtime: name nil semaphore of pollDesc

Use explicit name pdNil for nil semaphore of a pollDesc to make it self-explanatory like pdReady and pdWait.

Change-Id: Ibfb246e14419d366edadbccac4d3717f0c135cb0
Reviewed-on: https://go-review.googlesource.com/c/go/+/424923
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>

2 years agostrings: rely on utf8.AppendRune
Joe Tsai [Thu, 16 Jun 2022 00:50:04 +0000 (17:50 -0700)]
strings: rely on utf8.AppendRune

This is both simpler and more performant.

Change-Id: I66ef8e49c059a722932392ee3ecfb951d9b8e121
Reviewed-on: https://go-review.googlesource.com/c/go/+/412339
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Joseph Tsai <joetsai@digital-static.net>
Auto-Submit: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agoencoding/json: give it a chance to put encodeState back in pool when error occurs
Andy Pan [Sat, 13 Aug 2022 18:29:33 +0000 (02:29 +0800)]
encoding/json: give it a chance to put encodeState back in pool when error occurs

name                       old time/op    new time/op    delta
CodeEncoderError-10           688µs ± 8%     496µs ±15%   -27.92%  (p=0.000 n=10+9)
CodeMarshalError-10           747µs ± 6%     546µs ± 4%   -26.86%  (p=0.000 n=10+10)
MarshalBytesError/32-10       284µs ± 2%     273µs ± 1%    -3.84%  (p=0.000 n=10+10)
MarshalBytesError/256-10      281µs ± 2%     278µs ± 4%      ~     (p=0.053 n=9+10)
MarshalBytesError/4096-10     290µs ± 1%     279µs ± 3%    -3.52%  (p=0.000 n=10+10)

name                       old speed      new speed      delta
CodeEncoderError-10        2.83GB/s ± 8%  3.84GB/s ±20%   +36.03%  (p=0.000 n=10+10)
CodeMarshalError-10        2.60GB/s ± 5%  3.56GB/s ± 4%   +36.61%  (p=0.000 n=10+10)

name                       old alloc/op   new alloc/op   delta
CodeEncoderError-10          4.05MB ± 1%    0.00MB ± 1%  -100.00%  (p=0.000 n=10+9)
CodeMarshalError-10          6.05MB ± 0%    1.99MB ± 1%   -67.13%  (p=0.000 n=10+10)
MarshalBytesError/32-10      66.0kB ± 0%     0.2kB ± 0%   -99.67%  (p=0.000 n=9+8)
MarshalBytesError/256-10     50.1kB ± 0%     0.9kB ± 0%   -98.23%  (p=0.000 n=9+9)
MarshalBytesError/4096-10    87.4kB ± 0%     7.5kB ± 0%   -91.47%  (p=0.000 n=8+10)

name                       old allocs/op  new allocs/op  delta
CodeEncoderError-10            25.0 ± 0%       4.0 ± 0%   -84.00%  (p=0.000 n=9+10)
CodeMarshalError-10            27.0 ± 0%       6.0 ± 0%   -77.78%  (p=0.000 n=10+10)
MarshalBytesError/32-10        18.0 ± 0%       5.0 ± 0%   -72.22%  (p=0.000 n=10+10)
MarshalBytesError/256-10       17.0 ± 0%       6.0 ± 0%   -64.71%  (p=0.000 n=10+10)
MarshalBytesError/4096-10      16.0 ± 0%       6.0 ± 0%   -62.50%  (p=0.000 n=10+10)

Change-Id: I48070bb05f55707251c694e40d2570403bbf61f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/423694
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: initialize pointer bits of noscan spans
Keith Randall [Tue, 23 Aug 2022 17:40:31 +0000 (10:40 -0700)]
runtime: initialize pointer bits of noscan spans

Some code paths in the runtime (cgo, heapdump) request heap bits
without first checking that the span is !noscan. Instead of trying
to find and work around all those cases, just set the pointer bits
of noscan spans correctly. It's somewhat safer than ensuring we
caught all the possible cases.

Fixes #54557
Fixes #54558

Change-Id: Ibd476e6cdea77c962e4d15aad26f29df66fd94e8
Reviewed-on: https://go-review.googlesource.com/c/go/+/425194
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
2 years agoruntime: convert ticksType.val to atomic type
Cuong Manh Le [Fri, 19 Aug 2022 11:46:18 +0000 (18:46 +0700)]
runtime: convert ticksType.val to atomic type

Updates #53821

Change-Id: Ia0c58d7e7e11a1b52bbb7c19ebbb131e3eea5314
Reviewed-on: https://go-review.googlesource.com/c/go/+/424926
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agocmd/compile: handle partially overlapping assignments
Keith Randall [Mon, 22 Aug 2022 17:26:50 +0000 (10:26 -0700)]
cmd/compile: handle partially overlapping assignments

Normally, when moving Go values of type T from one location to another,
we don't need to worry about partial overlaps. The two Ts must either be
in disjoint (nonoverlapping) memory or in exactly the same location.
There are 2 cases where this isn't true:
 1) Using unsafe you can arrange partial overlaps.
 2) Since Go 1.17, you can use a cast from a slice to a ptr-to-array.
    https://go.dev/ref/spec#Conversions_from_slice_to_array_pointer
    This feature can be used to construct partial overlaps of array types.
      var a [3]int
      p := (*[2]int)(a[:])
      q := (*[2]int)(a[1:])
      *p = *q
We don't care about solving 1. Or at least, we haven't historically
and no one has complained.
For 2, we need to ensure that if there might be partial overlap,
then we can't use OpMove; we must use memmove instead.
(memmove handles partial overlap by copying in the correct
direction. OpMove does not.)

Note that we have to be careful here not to introduce a call when
we're marshaling arguments to a call or unmarshaling results from a call.

Fixes #54467

Change-Id: I1ca6aba8041576849c1d85f1fa33ae61b80a373d
Reviewed-on: https://go-review.googlesource.com/c/go/+/425076
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>

2 years agoRevert "cmd/compile: restore test/nested.go test cases"
Matthew Dempsky [Tue, 23 Aug 2022 18:55:12 +0000 (18:55 +0000)]
Revert "cmd/compile: restore test/nested.go test cases"

This reverts CL 424854.

Reason for revert: broke misc/cgo/stdio.TestTestRun on several builders.

Will re-land after CL 421879 is submitted.

Change-Id: I2548c70d33d7c178cc71c1d491cd81c22660348f
Reviewed-on: https://go-review.googlesource.com/c/go/+/425214
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Auto-Submit: Matthew Dempsky <mdempsky@google.com>

2 years agoruntime: fix ppc64 startup on newer linux kernels
Paul E. Murphy [Fri, 19 Aug 2022 14:20:02 +0000 (09:20 -0500)]
runtime: fix ppc64 startup on newer linux kernels

R0 needs to be cleared at startup as it may not always be cleared
by the kernel on newer kernels.

Fixes #54547

Change-Id: Id7055699aaa8d8b193b7e3e784f075ce29ac3f1d
Reviewed-on: https://go-review.googlesource.com/c/go/+/424927
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Paul Murphy <murp@ibm.com>

2 years agoall: use go/parser.SkipObjectResolution in more places
Daniel Martí [Thu, 21 Apr 2022 13:17:40 +0000 (14:17 +0100)]
all: use go/parser.SkipObjectResolution in more places

None of cgo, "go test", nor srcimporter make use of go/ast's object
resolution via go/ast.Object. As such, we can skip that work during
parse time, which should save some CPU time.

We don't have any benchmark numbers, as none of the three packages have
any usable benchmarks, but we measured gofmt to be about 5% faster
thanks to this tweak in https://go.dev/cl/401454.
These three packages are quite different to gofmt, but one can expect
similar speed-ups in the 1-5% range.

Two notable exceptions, which do make use of go/ast.Object, are cmd/fix
and cmd/doc - we do not modify those here.

See #46485.

Change-Id: Ie3e65600d4790641c4e4d6f1c379be477fa02cee
Reviewed-on: https://go-review.googlesource.com/c/go/+/401455
Run-TryBot: Bryan Mills <bcmills@google.com>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoencoding/xml: add (*Encoder).Close
Axel Wagner [Thu, 18 Aug 2022 14:18:34 +0000 (16:18 +0200)]
encoding/xml: add (*Encoder).Close

Flush can not check for unclosed elements, as more data might be encoded
after Flush is called. Close implicitly calls Flush and also checks that
all opened elements are closed as well.

Fixes #53346

Change-Id: I889b9f5ae54e5dfabb9e6948d96c5ed7bc1110f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/424777
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: David Chase <drchase@google.com>
2 years agocmd/go/internal/modload: remove ImportMap and PackageDir
Bryan C. Mills [Fri, 2 Jul 2021 17:15:37 +0000 (13:15 -0400)]
cmd/go/internal/modload: remove ImportMap and PackageDir

These two functions together duplicated much of the functionality of
modload.Lookup. Use that instead in modcmd.vendorPkg, and reduce the
modload surface area.

Updates #42504
Updates #40775
For #26904

Change-Id: Ib8aaac495d090178dd56971aef9e5aa44ffa818b
Reviewed-on: https://go-review.googlesource.com/c/go/+/332571
Reviewed-by: Michael Matloob <matloob@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>

2 years agocmd/compile: mark booleans as registerizable
Jakub Ciolek [Wed, 25 May 2022 10:22:22 +0000 (12:22 +0200)]
cmd/compile: mark booleans as registerizable

Boolean values fit in registers, mark them accordingly. Improves codegen a bit.

compilecmp for darwin/amd64:

compress/gzip
compress/gzip.(*Reader).Reset 1017 -> 985  (-3.15%)

net
net.newRequest 1002 -> 970  (-3.19%)

crypto/tls
crypto/tls.(*sessionState).unmarshal 1054 -> 968  (-8.16%)

cmd/compile/internal/syntax
cmd/compile/internal/syntax.Fprint 518 -> 453  (-12.55%)

cmd/vendor/github.com/ianlancetaylor/demangle
cmd/vendor/github.com/ianlancetaylor/demangle.ASTToString 389 -> 325  (-16.45%)

cmd/go/internal/load
cmd/go/internal/load.PackagesAndErrors 3453 -> 3381  (-2.09%)

cmd/compile/internal/ssa
cmd/compile/internal/ssa.registerizable 249 -> 255  (+2.41%)

cmd/compile/internal/ssagen
cmd/compile/internal/ssagen.buildssa 9388 -> 9356  (-0.34%)

file                                            before   after    Δ       %
compress/gzip.s                                 8247     8215     -32     -0.388%
net.s                                           266667   266635   -32     -0.012%
crypto/tls.s                                    290324   290238   -86     -0.030%
cmd/compile/internal/syntax.s                   156422   156357   -65     -0.042%
cmd/vendor/github.com/ianlancetaylor/demangle.s 268313   268249   -64     -0.024%
cmd/go/internal/load.s                          122946   122874   -72     -0.059%
cmd/compile/internal/ssa.s                      3551201  3551207  +6      +0.000%
cmd/compile/internal/ssagen.s                   362299   362267   -32     -0.009%
total                                           19725872 19725495 -377    -0.002%

Change-Id: I4cd40b54d8b2da6d1f946e51f16689315a369dca
Reviewed-on: https://go-review.googlesource.com/c/go/+/408474
Run-TryBot: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@google.com>