]> Cypherpunks repositories - gostls13.git/log
gostls13.git
3 years agodoc/go1.17: mention enabling frame pointer on all ARM64
Cherry Mui [Wed, 19 May 2021 20:34:58 +0000 (16:34 -0400)]
doc/go1.17: mention enabling frame pointer on all ARM64

Updates #44513.

Change-Id: I43e95de0423779b3311d96c56f7c8c1cc5be27b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/321311
Trust: Cherry Mui <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
3 years agogo/types: ensure that Named.check is nilled out once it is expanded
Rob Findley [Fri, 7 May 2021 02:28:37 +0000 (22:28 -0400)]
go/types: ensure that Named.check is nilled out once it is expanded

To support lazy expansion of defined types, *Named holds on to a
*Checker field, which can pin the *Checker in memory. This can have
meaningful memory implications for applications that keep type
information around.

Ensure that the Checker field is nilled out for any Named types that are
instantiated during the type checking pass, by deferring a clean up to
'later' boundaries.

In testing this almost exactly offset the ~6% memory footprint increase
I observed with 1.17.

Fixes #45580

Change-Id: I8aa5bb777573a924afe36e79fa65f8729336bceb
Reviewed-on: https://go-review.googlesource.com/c/go/+/318849
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoruntime,cmd/link/internal/ld: fix typos
tyltr [Wed, 26 May 2021 15:41:27 +0000 (15:41 +0000)]
runtime,cmd/link/internal/ld: fix typos

Change-Id: I558590cef7e2311aadbdcb4088033e350d3aae32
GitHub-Last-Rev: 513944a6238e0e32e2a2c266b70f7d50c9db508d
GitHub-Pull-Request: golang/go#46389
Reviewed-on: https://go-review.googlesource.com/c/go/+/322809
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agomath/big: move division into natdiv.go
Russ Cox [Thu, 13 May 2021 03:04:25 +0000 (23:04 -0400)]
math/big: move division into natdiv.go

Code moved and functions reordered to be in a consistent
top-down dependency order, but otherwise unchanged.

First step toward commenting division algorithms.

Change-Id: Ib5e604fb5b2867edff3a228ba4e57b5cb32c4137
Reviewed-on: https://go-review.googlesource.com/c/go/+/321077
Trust: Russ Cox <rsc@golang.org>
Trust: Katie Hockman <katie@golang.org>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agodoc/go1.17: fix formatting for time changes
Tobias Klauser [Tue, 25 May 2021 08:33:02 +0000 (10:33 +0200)]
doc/go1.17: fix formatting for time changes

Also add a link to the time.Time type and adjust the wording a bit.

Change-Id: I2f4210ada6d253eb5804e6327b2432487beb8a05
Reviewed-on: https://go-review.googlesource.com/c/go/+/321811
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoarchive/zip: only preallocate File slice if reasonably sized
Roland Shoemaker [Tue, 11 May 2021 18:31:31 +0000 (11:31 -0700)]
archive/zip: only preallocate File slice if reasonably sized

Since the number of files in the EOCD record isn't validated, it isn't
safe to preallocate Reader.Files using that field. A malformed archive
can indicate it contains up to 1 << 128 - 1 files. We can still safely
preallocate the slice by checking if the specified number of files in
the archive is reasonable, given the size of the archive.

Thanks to the OSS-Fuzz project for discovering this issue and to
Emmanuel Odeke for reporting it.

Fixes #46242
Fixes CVE-2021-33196

Change-Id: I3c76d8eec178468b380d87fdb4a3f2cb06f0ee76
Reviewed-on: https://go-review.googlesource.com/c/go/+/318909
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Trust: Joe Tsai <thebrokentoaster@gmail.com>
Run-TryBot: Roland Shoemaker <roland@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
3 years agodoc: add Go 1.17 release note about inlining functions with closures
Dan Scales [Sat, 22 May 2021 20:40:20 +0000 (13:40 -0700)]
doc: add Go 1.17 release note about inlining functions with closures

Fixes #45781

Change-Id: Ia5bc2845f7f94aff4f3f0ff15533feb148223adb
Reviewed-on: https://go-review.googlesource.com/c/go/+/322089
Trust: Dan Scales <danscales@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/go: add a -compat flag to 'go mod tidy'
Bryan C. Mills [Fri, 14 May 2021 20:53:06 +0000 (16:53 -0400)]
cmd/go: add a -compat flag to 'go mod tidy'

Fixes #46141

Change-Id: I9d4032e75252ade9eaa937389ea97ef3fb287499
Reviewed-on: https://go-review.googlesource.com/c/go/+/321071
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agonet: verify results from Lookup* are valid domain names
Roland Shoemaker [Mon, 17 May 2021 17:06:51 +0000 (10:06 -0700)]
net: verify results from Lookup* are valid domain names

For the methods LookupCNAME, LookupSRV, LookupMX, LookupNS, and
LookupAddr check that the returned domain names are in fact valid DNS
names using the existing isDomainName function.

Thanks to Philipp Jeitner and Haya Shulman from Fraunhofer SIT for
reporting this issue.

Fixes #46241
Fixes CVE-2021-33195

Change-Id: Icf231acd93178a3b6aec3f178cff7e693f74ef8c
Reviewed-on: https://go-review.googlesource.com/c/go/+/320949
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
3 years agomisc/wasm: ensure correct stack pointer in catch clauses
Richard Musiol [Sun, 23 May 2021 21:06:43 +0000 (23:06 +0200)]
misc/wasm: ensure correct stack pointer in catch clauses

The stack pointer may have changed after a call from JavaScript into Go
code because of stack growth. The normal case already updated the
sp variable accordingly, but the catch case did not yet.

Fixes #45433

Change-Id: I3e0a33381929626f6b21902948935eb5ffb26c96
Reviewed-on: https://go-review.googlesource.com/c/go/+/321936
Trust: Richard Musiol <neelance@gmail.com>
Run-TryBot: Richard Musiol <neelance@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/go: align checks of module path during initialization.
Constantin Konstantinidis [Sun, 16 May 2021 09:03:34 +0000 (11:03 +0200)]
cmd/go: align checks of module path during initialization.

Fixes #45025.

Change-Id: I70c2b745f764484e4b3a2824adc470f168fb2c50
Reviewed-on: https://go-review.googlesource.com/c/go/+/320310
Trust: Jay Conrod <jayconrod@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: add tests illustrating what happens when Go 1.16 is used in a Go 1.17 main...
Bryan C. Mills [Fri, 14 May 2021 20:01:17 +0000 (16:01 -0400)]
cmd/go: add tests illustrating what happens when Go 1.16 is used in a Go 1.17 main module

For #46141
Updates #46160

Change-Id: Ib22435b8051aaf3fa74d43d3b7f2d091e67f05e2
Reviewed-on: https://go-review.googlesource.com/c/go/+/320172
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/compile: ensure equal functions don't do unaligned loads
Keith Randall [Sun, 23 May 2021 19:38:59 +0000 (12:38 -0700)]
cmd/compile: ensure equal functions don't do unaligned loads

On architectures which don't support unaligned loads, make sure we
don't generate code that requires them.

Generated hash functions also matter in this respect, but they all look ok.

Update #37716
Fixes #46283

Change-Id: I6197fdfe04da4428092c99bd871d93738789e16b
Reviewed-on: https://go-review.googlesource.com/c/go/+/322151
Trust: Keith Randall <khr@golang.org>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: record regabi status in DW_AT_producer
Alessandro Arzilli [Sun, 9 May 2021 07:05:45 +0000 (09:05 +0200)]
cmd/compile: record regabi status in DW_AT_producer

Records if regabi was enabled during compilation in the DW_AT_producer
attribute of each compile unit.
This is useful to debuggers that support the debugCall protocol.

Change-Id: I5ad2c48ebf126aeb8bfb459b53a1a5304550036a
Reviewed-on: https://go-review.googlesource.com/c/go/+/318050
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: always include underlying type for map types
Keith Randall [Sun, 23 May 2021 04:59:00 +0000 (21:59 -0700)]
cmd/compile: always include underlying type for map types

This is a different fix for #37716.

Should help make the fix for #46283 easier, since we will no longer
need to keep compiler-generated hash functions and the runtime
hash function in sync.

Change-Id: I84cb93144e425dcd03afc552b5fbd0f2d2cc6d39
Reviewed-on: https://go-review.googlesource.com/c/go/+/322150
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime: account for spill slots in Windows callback compilation
Michael Anthony Knyszek [Fri, 21 May 2021 20:30:02 +0000 (20:30 +0000)]
runtime: account for spill slots in Windows callback compilation

The Go ABI, as it stands, requires spill space to be reserved for
register arguments. syscall.NewCallback (because of compileCallback)
does not actually reserve this space, leading to issues if the Go code
it invokes actually makes use of it.

Fixes #46301.

Change-Id: Idbc3578accaaaa29e4ba32291ef08d464da0b7b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/322029
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Egon Elbre <egonelbre@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/go/internal/modload: set the default GoVersion in a single location
Bryan C. Mills [Fri, 14 May 2021 20:51:57 +0000 (16:51 -0400)]
cmd/go/internal/modload: set the default GoVersion in a single location

For #46141
Updates #36460

Change-Id: Ie4c13c73a451650d1e8abb8e5cebfc30d0a71a70
Reviewed-on: https://go-review.googlesource.com/c/go/+/321070
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/go/internal/modcmd: factor out a type for flags whose arguments are Go versions
Bryan C. Mills [Sat, 15 May 2021 03:14:22 +0000 (23:14 -0400)]
cmd/go/internal/modcmd: factor out a type for flags whose arguments are Go versions

For #46141
Updates #45094

Change-Id: I6553600c69273762a81795ef021c66f4e0872b6b
Reviewed-on: https://go-review.googlesource.com/c/go/+/321069
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/compile: revert CL/316890
Than McIntosh [Fri, 21 May 2021 14:53:10 +0000 (10:53 -0400)]
cmd/compile: revert CL/316890

This is a revert of https://go-review.googlesource.com/c/go/+/316890,
which has positive effects on debugging + DWARF variable locations
for register parameters when the reg abi is in effect, but also
turns out to interact badly with the register allocator.

Fixes #46304.

Change-Id: I624bd980493411a9cde45d44fcd3c46cad796909
Reviewed-on: https://go-review.googlesource.com/c/go/+/321830
Trust: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agodoc/go1.17: document changes to net/http package
ian woolf [Fri, 7 May 2021 08:31:00 +0000 (16:31 +0800)]
doc/go1.17: document changes to net/http package

Changes include:
* ReadRequest function now returns an error when a request has multiple
  Host headers.

For #44513.
Updates #46015.

Change-Id: I48ea7c5cee3f1d1a247035fd37191362a53d1f04
Reviewed-on: https://go-review.googlesource.com/c/go/+/317914
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Trust: Heschi Kreinick <heschi@google.com>

3 years agodoc: document additional atomic.Value methods
Keith Randall [Wed, 12 May 2021 20:55:59 +0000 (13:55 -0700)]
doc: document additional atomic.Value methods

For #44513. Fixes #46022.

Change-Id: Id1d87fbd4034461953760ce77201f87ed723ff88
Reviewed-on: https://go-review.googlesource.com/c/go/+/319549
Trust: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/go: in TestScript/mod_replace, download an explicit module path
Bryan C. Mills [Fri, 21 May 2021 20:33:42 +0000 (16:33 -0400)]
cmd/go: in TestScript/mod_replace, download an explicit module path

As of CL 318629, 'go mod download' without arguments does not save
checksums for module source code. Without a checksum, 'go list' will
not report the location of the source code even if it is present, in
order to prevent accidental access of mismatched code.

Downloading an explicit module here also more clearly expresses the
intent of the test (“download this module and see where it is”), and
may be somewhat more efficient (since the test doesn't need source
code for the other modules in the build list).

Updates #45332

Change-Id: Ic589b22478e3ed140b95365bb6729101dd598ccc
Reviewed-on: https://go-review.googlesource.com/c/go/+/321956
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoos: document that StartProcess puts files into blocking mode
Ian Lance Taylor [Fri, 21 May 2021 19:50:52 +0000 (12:50 -0700)]
os: document that StartProcess puts files into blocking mode

Fixes #43894

Change-Id: I2add7b8a4f6ae69a5ef1c48703fde21a4b74307c
Reviewed-on: https://go-review.googlesource.com/c/go/+/321852
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
3 years agocmd/dist: display first class port status in json output
Constantin Konstantinidis [Fri, 14 May 2021 13:00:46 +0000 (15:00 +0200)]
cmd/dist: display first class port status in json output

Fixes #38874

Change-Id: I819dd008fd6869d335888b4aa03dcf739da9a9a4
Reviewed-on: https://go-review.googlesource.com/c/go/+/320069
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Trust: Daniel Martí <mvdan@mvdan.cc>

3 years agocmd/go: in 'go mod download' without args, don't save module zip sums
Jay Conrod [Mon, 10 May 2021 22:10:18 +0000 (18:10 -0400)]
cmd/go: in 'go mod download' without args, don't save module zip sums

'go mod download' without arguments is frequently used to populate the
module cache. It tends to fetch a lot of extra files (for modules in
the build list that aren't needed to build packages in the main
module). It's annoying when sums are written for these extra files.

'go mod download mod@version' will still write sums for specific
modules in the build list. 'go mod download all' still has the
previous behavior.

For now, all invocations of 'go mod download' still update go.mod and
go.sum with changes needed to load the build list (1.15 behavior).

Fixes #45332

Change-Id: I9e17d18a7466ac7271a0e1a2b663f6b3cb168c97
Reviewed-on: https://go-review.googlesource.com/c/go/+/318629
Trust: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agodoc/go1.17: document database/sql changes for Go 1.17
sryoya [Sun, 16 May 2021 14:36:05 +0000 (23:36 +0900)]
doc/go1.17: document database/sql changes for Go 1.17

For #44513
Fixes #46008

Change-Id: If80d484f73a0eb6946abdc654eb2c0d3dd6db416
Reviewed-on: https://go-review.googlesource.com/c/go/+/320251
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Robert Findley <rfindley@google.com>

3 years agodoc/go1.17: document io/fs changes for Go 1.17
Adam Mitha [Fri, 14 May 2021 17:58:01 +0000 (10:58 -0700)]
doc/go1.17: document io/fs changes for Go 1.17

For #44513
Fixes #46011

Change-Id: I862ef9a4314cd34fb8c828a8cd7d0a7b36c6f683
Reviewed-on: https://go-review.googlesource.com/c/go/+/320151
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Heschi Kreinick <heschi@google.com>
Trust: Robert Findley <rfindley@google.com>
Trust: Dmitri Shuralyov <dmitshur@golang.org>

3 years agodoc/go1.17: document archive/zip changes for Go 1.17
Aaron Sheah [Tue, 18 May 2021 14:54:02 +0000 (14:54 +0000)]
doc/go1.17: document archive/zip changes for Go 1.17

For #44513. Fixes #46000

Change-Id: I299d0b5657f1f96174d6e35d60daac8b36e59d29
GitHub-Last-Rev: e63461bff042a8abe79e0ec3515eefbf56ba1d82
GitHub-Pull-Request: golang/go#46235
Reviewed-on: https://go-review.googlesource.com/c/go/+/320809
Reviewed-by: Heschi Kreinick <heschi@google.com>
Trust: Robert Findley <rfindley@google.com>

3 years agocmd/go: remove warning from module deprecation notice printing
Tim Heckman [Thu, 20 May 2021 23:51:28 +0000 (16:51 -0700)]
cmd/go: remove warning from module deprecation notice printing

Fixes #46294

Change-Id: I50023006dab83dee455f98a702ca1c72e61764ea
Reviewed-on: https://go-review.googlesource.com/c/go/+/321649
Trust: Jay Conrod <jayconrod@google.com>
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agoruntime: wait for Go runtime to initialize in Windows signal test
Jason A. Donenfeld [Fri, 21 May 2021 10:01:39 +0000 (12:01 +0200)]
runtime: wait for Go runtime to initialize in Windows signal test

The test harness waits for "ready" as a sign that the Go runtime has
installed its signal handler and is ready to be tested. But actually,
while LoadLibrary starts the loading of the Go runtime, it does so
asynchronously, so the "ready" sign is potentially premature and
certainly racy. However, all exported cgo entry points make a call to
_cgo_wait_runtime_init_done which waits for that asynchronous
initialization to complete. Therefore, this commit fixes the test to
call into the exported "Dummy" cgo function before emitting the "ready"
sign, so that we're sure the Go runtime is actually loaded.

Updates #45638.

Change-Id: I9b12b172d45bdcc09d54dd301de3a3e499544834
Reviewed-on: https://go-review.googlesource.com/c/go/+/321769
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
3 years agoio/fs: added an example for io/fs.WalkDir
Adam Mitha [Fri, 14 May 2021 03:52:41 +0000 (20:52 -0700)]
io/fs: added an example for io/fs.WalkDir

The documentation currently does not show how to get an `FS` instance for the operating system's filesystem. This example demonstrates how to accomplish this using the `os` package.

Fixes #46083

Change-Id: I053111c12ab09ef13f0d04fcdff8a6ea0dccf379
Reviewed-on: https://go-review.googlesource.com/c/go/+/319989
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Emmanuel Odeke <emmanuel@orijtech.com>

3 years agocmd: go get golang.org/x/tools/analysis@49064d23 && go mod vendor
Tim King [Thu, 20 May 2021 01:45:53 +0000 (18:45 -0700)]
cmd: go get golang.org/x/tools/analysis@49064d23 && go mod vendor

This brings in CLs 312829, 317431, 319211.

Fixes #40356.
Fixes #46129.

Change-Id: I2ee1f858b2a41ffa60d88b0c17511ccad57f1816
Reviewed-on: https://go-review.googlesource.com/c/go/+/321389
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Trust: Tim King <taking@google.com>
Run-TryBot: Tim King <taking@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agodoc/go1.17: document changes to os package
Vishal Dalwadi [Fri, 7 May 2021 04:35:36 +0000 (10:05 +0530)]
doc/go1.17: document changes to os package

Documents the changes to File.WriteString method.

For #44513.
Fixes #46018.

Change-Id: I3a8ef9df9f84662614d54802710bd705d626b995
Reviewed-on: https://go-review.googlesource.com/c/go/+/317910
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Trust: Cherry Mui <cherryyz@google.com>

3 years agocrypto/x509: add new FreeBSD 12.2+ trusted certificate folder
Lapo Luchini [Thu, 20 May 2021 15:41:02 +0000 (15:41 +0000)]
crypto/x509: add new FreeBSD 12.2+ trusted certificate folder

Up to FreeBSD 12.1 the package ca_root_nss was needed in order to have
certificates under /usr/local/share/certs as the base system didn't have
a system trusted certificate store.

This has been fixed in FreeBSD 12.2 using /etc/ssl/certs:
https://svnweb.freebsd.org/base?view=revision&revision=357082

Fixes #46284

Change-Id: I912b1bacc30cdf20d19e3ef9d09b69bb8055ff49
GitHub-Last-Rev: 0fa5542ea3c70ecb03e621381d7c34fbadf7ea47
GitHub-Pull-Request: golang/go#46276
Reviewed-on: https://go-review.googlesource.com/c/go/+/321190
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>

3 years agotest: re-enable test on riscv64 now that it supports external linking
Joel Sing [Wed, 19 May 2021 16:23:04 +0000 (02:23 +1000)]
test: re-enable test on riscv64 now that it supports external linking

Update #36739

Change-Id: I14ab2cd0e29966b9a2f992e8c3bcb415203e63e6
Reviewed-on: https://go-review.googlesource.com/c/go/+/321449
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agodoc/go1.17: add release notes for OpenBSD ports
Cherry Mui [Wed, 19 May 2021 20:22:34 +0000 (16:22 -0400)]
doc/go1.17: add release notes for OpenBSD ports

Updates #44513.

Change-Id: I8758768f6231fd7fcbaa7109eb49ee56cd60d93d
Reviewed-on: https://go-review.googlesource.com/c/go/+/321310
Trust: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
3 years agodoc/go1.17: add release note for windows/arm64 port
Cherry Mui [Wed, 19 May 2021 20:07:52 +0000 (16:07 -0400)]
doc/go1.17: add release note for windows/arm64 port

Updates #44513, #42604.

Change-Id: I8200e8087c219a0042ab2a6770a7275c3b17942a
Reviewed-on: https://go-review.googlesource.com/c/go/+/321309
Trust: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
3 years agodoc/go1.17: document new math constants
Vishal Dalwadi [Fri, 7 May 2021 04:55:27 +0000 (10:25 +0530)]
doc/go1.17: document new math constants

Documents the newly introduced:
* MaxInt
* MinInt
* MaxUint

Updates #28538.
For #44513.
Fixes #46012.

Change-Id: Iab6bbcf8f76ebe105b973d5fd39b86b8cd078348
Reviewed-on: https://go-review.googlesource.com/c/go/+/317911
Trust: Heschi Kreinick <heschi@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
3 years agosyscall: document NewCallback and NewCallbackCDecl limitations
Michael Anthony Knyszek [Wed, 19 May 2021 15:46:44 +0000 (15:46 +0000)]
syscall: document NewCallback and NewCallbackCDecl limitations

Currently NewCallback and NewCallbackCDecl may only be called a limited
number of times in a single Go process, but this property of the API is
not documented. This change fixes that, but does not document the
precise limit to avoid making that limit part of the API, leaving us
open to increasing or decreasing the limit in the future as needed.

Although the API avoids documenting a limit, it does guarantee a minimum
callback count so users can rely on at least some amount of callbacks
working.

Updates #46184.

Change-Id: I5129bf5fe301efff73ac112ba1f207ab32058833
Reviewed-on: https://go-review.googlesource.com/c/go/+/321133
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agomisc/cgo/testplugin: skip TestIssue25756pie on darwin/arm64 builder
Dmitri Shuralyov [Wed, 19 May 2021 22:51:11 +0000 (22:51 +0000)]
misc/cgo/testplugin: skip TestIssue25756pie on darwin/arm64 builder

This test is known to be broken on the darwin/arm64 builder.
Skip it while it's being investigated so it doesn't mask other failures.

For #46239.
Updates #43228.

Change-Id: I8fe57a0636bba84c3100337146dcb96cc264e524
Reviewed-on: https://go-review.googlesource.com/c/go/+/321349
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/internal/moddeps: use filepath.SkipDir only on directories
Dmitri Shuralyov [Wed, 19 May 2021 03:29:14 +0000 (23:29 -0400)]
cmd/internal/moddeps: use filepath.SkipDir only on directories

If a filepath.WalkFunc returns filepath.SkipDir when invoked on a
non-directory file, it skips the remaining files in the containing
directory.¹

CL 276272 accidentally added a code path that triggers this behavior
whenever filepath.Walk reaches a non-directory file that begins with
a dot, such as .gitattributes or .DS_Store, causing findGorootModules
to return early without finding any modules in GOROOT. Tests that use
it ceased to provide test coverage that the tree is tidy.

Add an explicit check for info.IsDir in the 5 places that intend to
use filepath.SkipDir to skip traversing that directory. Even paths
like GOROOT/bin and GOROOT/pkg which are unlikely to be anything but
a directory are worth checking, since the goal of moddeps is to take
a possibly problematic GOROOT tree as input and detect problems.

While the goal of findGorootModules is to find all modules in GOROOT
programmatically (in case new modules are added or modified), there
are 4 modules now that are quite likely to exist, so check for their
presence to avoid similar regressions. (It's not hard to update this
test if a well-known GOROOT module is removed or otherwise modified;
but if it becomes hard we can simplify it to check for a reasonable
number of modules instead.)

Also fix the minor skew that has crept in since the test got disabled.

¹ This wasn't necessarily an intentional design decision, but it was
  found only when Go 1.4 was already out. See CL 11690 for details.

Fixes #46254.

Change-Id: Id55ed926f8c0094b1af923070de72bacca05996f
Reviewed-on: https://go-review.googlesource.com/c/go/+/320991
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
3 years agonet: return nil UDPAddr from ReadFromUDP
Michael Pratt [Tue, 18 May 2021 16:23:56 +0000 (12:23 -0400)]
net: return nil UDPAddr from ReadFromUDP

In cases where the socket operation has no underlying address,
golang.org/cl/291509 unintentionally changed ReadFromUDP from return a
nil *UDPAddr to a non-nil (but zero value) *UDPAddr.

This may break callers that assume "no address" is always addr == nil,
so change it back to remain nil.

Fixes #46238

Change-Id: I8531e8fa16b853ed7560088eabda0b9e3e53f5be
Reviewed-on: https://go-review.googlesource.com/c/go/+/320909
Trust: Michael Pratt <mpratt@google.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
3 years agotest: check portable error message on issue46234.go
Cherry Mui [Tue, 18 May 2021 22:25:44 +0000 (18:25 -0400)]
test: check portable error message on issue46234.go

issue46234.go expects an error output "segmentation violation",
which is UNIX-specific. Check for "nil pointer dereference"
instead, which is emitted by the Go runtime and should work on all
platforms.

Should fix Windows builders.

Change-Id: I3f5a66a687d43cae5eaf6a9e942b877e5a248900
Reviewed-on: https://go-review.googlesource.com/c/go/+/321072
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
3 years agogo/build/constraint: fix parsing of "// +build" (with no args)
Russ Cox [Tue, 18 May 2021 15:16:38 +0000 (11:16 -0400)]
go/build/constraint: fix parsing of "// +build" (with no args)

"// +build" by itself was like "// +build !" - unsatisfiable.
Make it so again (right now it panics).

Fixes #44487.

Change-Id: Iacbc1398af6f988ef011f9f438e792eb62f8f434
Reviewed-on: https://go-review.googlesource.com/c/go/+/320829
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/compile: don't emit inltree for closure within body of inlined func
Than McIntosh [Tue, 18 May 2021 16:58:02 +0000 (12:58 -0400)]
cmd/compile: don't emit inltree for closure within body of inlined func

When inlining functions with closures, ensure that we don't mark the
body of the closure with a src.Pos marker that reflects the inline,
since this will result in the generation of an inltree table for the
closure itself (as opposed to the routine that the func-with-closure
was inlined into).

Fixes #46234.

Change-Id: I348296de6504fc4745d99adab436640f50be299a
Reviewed-on: https://go-review.googlesource.com/c/go/+/320913
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Than McIntosh <thanm@google.com>

3 years agocrypto/x509: remove duplicate import
Roland Shoemaker [Mon, 10 May 2021 16:59:07 +0000 (09:59 -0700)]
crypto/x509: remove duplicate import

Change-Id: I86742ae7aa4ff49a38f8e3bc1d64fb223feae73e
Reviewed-on: https://go-review.googlesource.com/c/go/+/318409
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
3 years agomake.bash: fix misuse of continue
Cherry Mui [Thu, 13 May 2021 23:22:57 +0000 (19:22 -0400)]
make.bash: fix misuse of continue

Apparently, in bash, the "continue" keyword can only be used
inside of a loop, not in an if block. If readelf exists but $CC
does not, make.bash emits a warning:

./make.bash: line 135: continue: only meaningful in a `for', `while', or `until' loop

Change it to a conditional.

Change-Id: I00a0940ed99bc0c565094e506705961b6b3d362e
Reviewed-on: https://go-review.googlesource.com/c/go/+/320170
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
3 years agodoc/go1.17: fix typo "avoding" -> "avoiding"
Jeff Widman [Tue, 18 May 2021 06:09:20 +0000 (06:09 +0000)]
doc/go1.17: fix typo "avoding" -> "avoiding"

Change-Id: Ice4a6e7ec8175caf3f049ac1ca39929059f90e9c
GitHub-Last-Rev: a2d59d55511ac7825cbeca6651a79b1060d90de3
GitHub-Pull-Request: golang/go#46227
Reviewed-on: https://go-review.googlesource.com/c/go/+/320729
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Alberto Donizetti <alb.donizetti@gmail.com>
Trust: Alberto Donizetti <alb.donizetti@gmail.com>

3 years agotime: rewrite the documentation for layout strings
Rob Pike [Mon, 17 May 2021 00:30:01 +0000 (10:30 +1000)]
time: rewrite the documentation for layout strings

People continue to be confused by how these work. Address that by some
rejiggering.

Introduce a constant called Layout that both defines the time and
provides a reference point for Parse and Format to refer to. We can
then delete much redundancy, especially for Format's comments, but
Parse tightens a bit too.

Then change the way the concept of the layout string is introduced,
and provide a clearer catalog of what its elements are.

Fixes #38871

Change-Id: Ib967ae70c7d5798a97b865cdda1fda4daed8a99a
Reviewed-on: https://go-review.googlesource.com/c/go/+/320252
Trust: Rob Pike <r@golang.org>
Run-TryBot: Rob Pike <r@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
3 years agogo/token: correct the interval notation used in some panic messages
Rob Findley [Mon, 17 May 2021 20:11:51 +0000 (16:11 -0400)]
go/token: correct the interval notation used in some panic messages

Fix an apparent typo for the right-hand bound in a couple panic
messages, where '[' was used instead of ']'.

Fixes #46215

Change-Id: Ie419c404ca72ed085a83a2c38ea1a5d6ed326cca
Reviewed-on: https://go-review.googlesource.com/c/go/+/320510
Trust: Robert Findley <rfindley@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoall: update golang.org/x/net to latest
Tobias Klauser [Mon, 17 May 2021 10:18:12 +0000 (12:18 +0200)]
all: update golang.org/x/net to latest

To pull in CL 318309.

For #41184

Change-Id: I99adb0478e71dbd72e13551a87ed09eae2a0ef2c
Reviewed-on: https://go-review.googlesource.com/c/go/+/320312
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agosyscall: some containers may fail syscall.TestSetuidEtc
Andrew G. Morgan [Thu, 13 May 2021 03:41:26 +0000 (20:41 -0700)]
syscall: some containers may fail syscall.TestSetuidEtc

The test previously had the hardcoded assumption that /proc/self/status
files had "Groups:" lines containing numerical IDs in ascending order.
Because of the possibility of non-monotonic ordering of GIDs in user
namespaces, this assumption was not universally true for all
/proc/self/gid_map setups.

To ensure this test can pass in those setups, sanity check failed
"Groups:" line matches with a string sorted version of the expected
values. (For the test cases here, numerical and string sorted order
are guaranteed to match.)

Fixes #46145

Change-Id: Ia060e80b123604bc394a15c02582fc406f944d36
Reviewed-on: https://go-review.googlesource.com/c/go/+/319591
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>

3 years agocmd/go: don't print 'go get' deprecation notices in the main module
Jay Conrod [Mon, 10 May 2021 21:19:50 +0000 (17:19 -0400)]
cmd/go: don't print 'go get' deprecation notices in the main module

If a user runs 'go get example.com/cmd' for a package in the main
module, it's more likely they intend to fill in missing dependencies
for that package (especially with -u). If the intent were only to
build and install, 'go install example.com/cmd' would be a better
choice.

For #43684

Resolving a comment on CL 305670.

Change-Id: I5c80ffdcdb3425b448f2f49cc20b07a18cb2bbe9
Reviewed-on: https://go-review.googlesource.com/c/go/+/318570
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agoruntime/cgo: fix crosscall2 on ppc64x
Lynn Boger [Wed, 12 May 2021 17:11:03 +0000 (12:11 -0500)]
runtime/cgo: fix crosscall2 on ppc64x

Some uses of crosscall2 did not work on ppc64le and probably
aix-ppc64. In particular, if there was a main program compiled
with -buildmode=pie and used a plugin which invoked crosscall2,
then failures could occur due to R2 getting set incorrectly along the
way. The problem was due to R2 being saved on the caller's
stack; it is now saved on the crosscall2 stack. More details can be
found in the issue.

This adds a testcase where the main program is built with pie
and the plugin invokes crosscall2.

This also changes the save of the CR bits from MOVD to MOVW as
it should be.

Fixes #43228

Change-Id: Ib5673e25a2ec5ee46bf9a1ffb0cb1f3ef5449086
Reviewed-on: https://go-review.googlesource.com/c/go/+/319489
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: Heschi Kreinick <heschi@google.com>

3 years agocmd/go: error out of 'go mod tidy' if the go version is newer than supported
Bryan C. Mills [Thu, 13 May 2021 13:48:40 +0000 (09:48 -0400)]
cmd/go: error out of 'go mod tidy' if the go version is newer than supported

Fixes #46142

Change-Id: Ib7a0a159e53cbe476be6aa9a050add10cc750dec
Reviewed-on: https://go-review.googlesource.com/c/go/+/319669
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agoruntime: mark osyield nosplit on OpenBSD
Cherry Mui [Thu, 13 May 2021 22:19:42 +0000 (18:19 -0400)]
runtime: mark osyield nosplit on OpenBSD

osyield is called in code paths that are not allowed to split
stack, e.g. casgstatus called from entersyscall/exitsyscall.
It is nosplit on all other platforms. Mark it nosplit on OpenBSD
as well.

Change-Id: I3fed5d7f58b3d50610beca6eed2c7e902b8ec52c
Reviewed-on: https://go-review.googlesource.com/c/go/+/319969
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Joel Sing <joel@sing.id.au>
3 years agonet/http: prevent infinite wait during TestMissingStatusNoPanic
Michael Fraenkel [Thu, 13 May 2021 15:41:45 +0000 (09:41 -0600)]
net/http: prevent infinite wait during TestMissingStatusNoPanic

If the client request never makes it to the server, the outstanding
accept is never broken. Change the test to always close the listening
socket when the client request completes.

Updates #45358

Change-Id: I744a91dfa11704e7e528163d7669c394e90456dc
Reviewed-on: https://go-review.googlesource.com/c/go/+/319275
Trust: Heschi Kreinick <heschi@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/go/internal/load: override Package.Root in module mode
Manlio Perillo [Wed, 12 May 2021 10:09:18 +0000 (12:09 +0200)]
cmd/go/internal/load: override Package.Root in module mode

The Context.ImportDir method in the go/build package sets Package.Root
to $GOPATH, if a package is inside a GOPATH workspace.  The
loadPackageData function keeps this value even when modules are enabled.

Override Package.Root when modules are enabled, instead of just set its
value when Context.ImportDir was unable to set it.

Add a regression test.

Fixes #46119

Change-Id: I900a33fe13a445cb771e2952d0d830f1b4a5921f
Reviewed-on: https://go-review.googlesource.com/c/go/+/319209
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: fix a portability issue in the cd script command
Manlio Perillo [Thu, 13 May 2021 14:08:59 +0000 (16:08 +0200)]
cmd/go: fix a portability issue in the cd script command

Currently all script tests use the UNIX path separator with the cd
command, causing the PWD environment variable to have the incorrect path
separator on Windows.

Call filepath.FromSlash on the cd command argument.
Update the testdata/script/README to document that the cd argument must
use slashes.

Add a regression test.

To reproduce this issue, a test must use the cd command followed by a
stdout or stderr command containing the pattern $PWD.

Change-Id: Ib2dc5f185cc2476451402787996d14df91f7dddb
Reviewed-on: https://go-review.googlesource.com/c/go/+/319311
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: fix spacing in help text of -overlay flag
Ben Hoyt [Thu, 13 May 2021 21:59:20 +0000 (09:59 +1200)]
cmd/go: fix spacing in help text of -overlay flag

There was a space missing in the first line shown below, and an extra
space in the second line shown. Thanks Peter Bourgon for noting this.

BEFORE:
$ go help build | grep -A1 'has some limitations'
has some limitations:importantly, cgo files included from outside the
include path must be  in the same directory as the Go package they are

AFTER:
$ go help build | grep -A1 'has some limitations'
has some limitations: importantly, cgo files included from outside the
include path must be in the same directory as the Go package they are

Note that I edited alldocs.go by hand here, as the mkalldocs.sh script
produces a lot more changes, for example adding the -insecure flag
documentation in. Not sure what's wrong there.

Change-Id: I303f6d6b42b0e24cec0748a949dc23beec64b917
Reviewed-on: https://go-review.googlesource.com/c/go/+/319949
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/internal/obj/arm64: disable AL and NV for some condition operation instructions
eric fang [Fri, 7 May 2021 05:48:18 +0000 (05:48 +0000)]
cmd/internal/obj/arm64: disable AL and NV for some condition operation instructions

According to the armv8-a reference manual, conditions AL and NV are not allowed
for instructions CINC, CINV, CNEG, CSET and CSETM. This CL adds this check and
the corresponding test cases.

Change-Id: Icb496b7b13a353f41491f2de4d939a5cd88abb04
Reviewed-on: https://go-review.googlesource.com/c/go/+/317912
Reviewed-by: eric fang <eric.fang@arm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: eric fang <eric.fang@arm.com>
Run-TryBot: eric fang <eric.fang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agodebug/macho: fix a typo in macho.go
itchyny [Thu, 13 May 2021 04:03:18 +0000 (04:03 +0000)]
debug/macho: fix a typo in macho.go

Change-Id: Ica47b53decf6690fbd37e666e9de5098117b82de
GitHub-Last-Rev: 6aabb208a6499e29fe32a2f0d928c4e027d556b2
GitHub-Pull-Request: golang/go#46147
Reviewed-on: https://go-review.googlesource.com/c/go/+/319592
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
Trust: Heschi Kreinick <heschi@google.com>

3 years agoall: fix spelling
John Bampton [Fri, 7 May 2021 10:47:28 +0000 (10:47 +0000)]
all: fix spelling

Change-Id: Id991d8e81e04835cabfb02d07a2199bfb553726c
GitHub-Last-Rev: 5d0b55d49127a2e2ba3148d20a58c52debcae12b
GitHub-Pull-Request: golang/go#44802
Reviewed-on: https://go-review.googlesource.com/c/go/+/299069
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Heschi Kreinick <heschi@google.com>

3 years agocmd/link: always mark runtime.unreachableMethod symbol
Cherry Mui [Thu, 13 May 2021 20:48:50 +0000 (16:48 -0400)]
cmd/link: always mark runtime.unreachableMethod symbol

In the deadcode path we mark runtime.unreachableMethod symbol,
which is a special symbol used for redirecting unreachable
methods. Currently this code is conditioned on not -linkshared.
This is wrong. It should be marked with -linkshared mode as well.

In fact, -linkshared should only affect the entry symbol. Change
the code accordingly.

Change-Id: I252abf850212a930f275589ef0035a43e52cb9cc
Reviewed-on: https://go-review.googlesource.com/c/go/+/319893
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
3 years agocmd/link: resolve ABI alias for runtime.unreachableMethod
Cherry Mui [Thu, 13 May 2021 20:36:32 +0000 (16:36 -0400)]
cmd/link: resolve ABI alias for runtime.unreachableMethod

We redirect references to unreachable methods to
runtime.unreachableMethod. We choose to use ABIInternal symbol
for this, because runtime.unreachableMethod is a defined Go
function.

When linking against shared libraries, and ABI wrappers are not
enabled, the imported function symbols are all ABI0 and aliased
to ABIInternal. We need to resolve ABI alias in this case.

Change-Id: Idd64ef46ce0b5f54882ea0069ce0d59dc9b7a599
Reviewed-on: https://go-review.googlesource.com/c/go/+/319891
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: permit .tbd files as a linker flag
Kevin Albertson [Tue, 16 Feb 2021 14:40:49 +0000 (14:40 +0000)]
cmd/go: permit .tbd files as a linker flag

A .tbd file is a macOS text-based stub library and is a valid input to
the macOS linker. This change adds .tbd to the allow-list for acceptable
linker flags.

Fixes golang/go#44263

Change-Id: Ie5439a13325dbc908e42f95ec70aca518bb549f9
GitHub-Last-Rev: 6055c3b5fa8c8146c0c5c976aa8e1d219e01e414
GitHub-Pull-Request: golang/go#44276
Reviewed-on: https://go-review.googlesource.com/c/go/+/292269
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Ian Lance Taylor <iant@golang.org>

3 years agocmd/go: be less strict about go version syntax in dependency go.mod files
Russ Cox [Thu, 6 May 2021 18:01:20 +0000 (14:01 -0400)]
cmd/go: be less strict about go version syntax in dependency go.mod files

It is unclear what the future holds for the go line in go.mod files.
Perhaps at some point we will switch to semver numbering.
Perhaps at some point we will allow specifying minor versions
or even betas and release candidates.
Those kinds of changes are difficult today because the go line
is parsed in dependency modules, meaning that older
versions of the Go toolchain need to understand newer go lines.

This CL makes that case - parsing a go line in a dependency's
go.mod file - a bit more lax about how to find the version.
It allows a leading v and any trailing non-digit-prefixed string
after the MAJOR.MINOR section.

There are no concrete plans to make use of any of these changes,
but if in the future we want to make them, having a few Go releases
under out belt that will accept the syntax in dependencies will
make any changes significantly easier.

Change-Id: I79bb84bba4b769048ac4b14d5c275eb9a3f270c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/317690
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agoregexp: fix repeat of preferred empty match
Russ Cox [Tue, 11 May 2021 14:52:44 +0000 (10:52 -0400)]
regexp: fix repeat of preferred empty match

In Perl mode, (|a)* should match an empty string at the start of the
input. Instead it matches as many a's as possible.
Because (|a)+ is handled correctly, matching only an empty string,
this leads to the paradox that e* can match more text than e+
(for e = (|a)) and that e+ is sometimes different from ee*.

This is a very old bug that ultimately derives from the picture I drew
for e* in https://swtch.com/~rsc/regexp/regexp1.html. The picture is
correct for longest-match (POSIX) regexps but subtly wrong for
preferred-match (Perl) regexps in the case where e has a preferred
empty match. Pointed out by Andrew Gallant in private mail.

The current code treats e* and e+ as the same structure, with
different entry points. In the case of e* the preference list ends up
not quite in the right order, in part because the “before e” and
“after e” states are the same state. Splitting them apart fixes the
preference list, and that can be done by compiling e* as if it were
(e+)?.

Like with any bug fix, there is a very low chance of breaking a
program that accidentally depends on the buggy behavior.

RE2, Go, and Rust all have this bug, and we've all agreed to fix it,
to keep the implementations in sync.

Fixes #46123.

Change-Id: I70e742e71e0a23b626593b16ddef3c1e73b413b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/318750
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile/internal/dwarfgen: fix DWARF param DIE ordering
Than McIntosh [Tue, 11 May 2021 15:29:27 +0000 (11:29 -0400)]
cmd/compile/internal/dwarfgen: fix DWARF param DIE ordering

The DWARF standard requires that the DIEs in a subprogram
corresponding to input and output parameters appear in declaration
order; this patch adds some new code in dwarfgen to enforce this
ordering (relying on the existing fn.Dcl ordering is not sufficient).

Prior to the register ABI, it was easy to keep vars/decls sorted
during DWARF generation since you could always rely on frame offset;
with the ABI sorting by frame offset no longer gives you the original
declaration order in all cases.

Fixes #46055.

Change-Id: I0e070cb781d6453caba896e5d3bee7cd5388050d
Reviewed-on: https://go-review.googlesource.com/c/go/+/318829
Trust: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agodebug/dwarf: delay array type fixup to handle type cycles
Than McIntosh [Wed, 12 May 2021 15:03:48 +0000 (11:03 -0400)]
debug/dwarf: delay array type fixup to handle type cycles

A user encountered a debug/dwarf crash when running the dwarf2json
tool (https://github.com/volatilityfoundation/dwarf2json) on a
debug-built copy of the linux kernel. In this crash, the DWARF type
reader was trying to examine the contents of an array type while that
array type was still in the process of being constructed (due to
cycles in the type graph).

To avoid such situations, this patch extends the mechanism introduced
in https://go-review.googlesource.com/18459 (which handles typedef
types) to delay fixup of array types as well.

Change-Id: I303f6ce5db1ca4bd79da3581957dfc2bfc17cc01
Reviewed-on: https://go-review.googlesource.com/c/go/+/319329
Trust: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Yi Chou <yich@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/vendor: update golang.org/x/sys to latest
Tobias Klauser [Wed, 12 May 2021 16:07:43 +0000 (18:07 +0200)]
cmd/vendor: update golang.org/x/sys to latest

To pull in CL 318212.

For #41184

Change-Id: Iec0d7bee2d3a5874c24a55ec20efd7746bf70902
Reviewed-on: https://go-review.googlesource.com/c/go/+/319410
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoall: add //go:build lines to assembly files
Tobias Klauser [Wed, 12 May 2021 15:55:42 +0000 (17:55 +0200)]
all: add //go:build lines to assembly files

Don't add them to files in vendor and cmd/vendor though. These will be
pulled in by updating the respective dependencies.

For #41184

Change-Id: Icc57458c9b3033c347124323f33084c85b224c70
Reviewed-on: https://go-review.googlesource.com/c/go/+/319389
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
3 years agocmd/go/internal/modload: in updateLazyRoots, do not require the main module explicitly
Bryan C. Mills [Tue, 11 May 2021 14:15:57 +0000 (10:15 -0400)]
cmd/go/internal/modload: in updateLazyRoots, do not require the main module explicitly

Fixes #46078

Change-Id: I8044dac717459f1eeae1d8381a6503f22f9f51ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/319009
Trust: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/compile/abi-internal.md: fix table format
Cherry Mui [Wed, 12 May 2021 17:08:16 +0000 (13:08 -0400)]
cmd/compile/abi-internal.md: fix table format

The table was not rendered correctly because one line missed a
column.

Change-Id: I1373e4e9fb8b8f2dcd9fd0db339083362cce9b71
Reviewed-on: https://go-review.googlesource.com/c/go/+/319291
Trust: Cherry Mui <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
3 years agocmd/compile: add arch-specific inlining for runtime.memmove
Ruslan Andreev [Tue, 19 Jan 2021 14:30:10 +0000 (22:30 +0800)]
cmd/compile: add arch-specific inlining for runtime.memmove

This CL add runtime.memmove inlining for AMD64 and ARM64.
According to ssa dump from testcases generic rules can't inline
memmomve properly due to one of the arguments is Phi operation. But this
Phi op will be optimized out by later optimization stages. As a result
memmove can be inlined during arch-specific rules.
The commit add new optimization rules to arch-specific rules that can
inline runtime.memmove if it possible during lowering stage.
Optimization fires 5 times in Go source-code using regabi.

Fixes #41662

Change-Id: Iaffaf4c482d068b5f0683d141863892202cc8824
Reviewed-on: https://go-review.googlesource.com/c/go/+/289151
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: David Chase <drchase@google.com>

3 years agoruntime/internal/atomic: add LSE atomics instructions to arm64
Jonathan Swinney [Sat, 17 Apr 2021 01:12:28 +0000 (01:12 +0000)]
runtime/internal/atomic: add LSE atomics instructions to arm64

As a follow up to an earlier change[1] to add ARMv8+LSE instructions in
the compiler generated atomic intrinsics, make the same change in the
runtime library. Since not all ARMv8 systems support LSE instructions,
they are protected by a feature-flag branch.

[1]: golang.org/cl/234217 commit: ecc3f5112eba

Change-Id: I0e2fb22e78d5eddb6547863667a8865946679a00
Reviewed-on: https://go-review.googlesource.com/c/go/+/310591
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Heschi Kreinick <heschi@google.com>

3 years agoruntime: fix handling of SPWRITE functions in traceback
Russ Cox [Thu, 6 May 2021 15:38:46 +0000 (11:38 -0400)]
runtime: fix handling of SPWRITE functions in traceback

It is valid to see SPWRITE functions at the top of a GC stack traceback,
in the case where they self-preempted during the stack growth check
and haven't actually modified SP in a traceback-unfriendly manner yet.
The current check is therefore too aggressive.

isAsyncSafePoint is taking care of not async-preempting SPWRITE functions
because it doesn't async-preempt any assembly functions at all.
But perhaps it will in the future.

To keep a check that SPWRITE assembly functions are not async-preempted,
add one in preemptPark. Then relax the check in traceback to avoid
triggering on self-preempted SPWRITE functions.

The long and short of this is that the assembly we corrected in x/crypto
issue #44269 was incredibly dodgy but not technically incompatible with
the Go runtime. After this change, the original x/crypto assembly no longer
causes GC traceback crashes during "GOGC=1 go test -count=1000".
But we'll still leave the corrected assembly.

This also means that we don't need to worry about diagnosing SPWRITE
assembly functions that may exist in the wild. They will be skipped for
async preemption and no harm no foul.

Fixes #44269, which was open pending some kind of check for
bad SPWRITE functions in the wild. (No longer needed.)

Change-Id: I6000197b62812bbd2cd92da28eab422634cf75a8
Reviewed-on: https://go-review.googlesource.com/c/go/+/317669
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/link: check mmap error
Cherry Mui [Wed, 12 May 2021 02:40:41 +0000 (22:40 -0400)]
cmd/link: check mmap error

We already check mmap errors on some code paths, but we missed
one. Add error check there.

Change-Id: Ic0e9cb0eb03c805de40802cfc5d5500e3e065d99
Reviewed-on: https://go-review.googlesource.com/c/go/+/319290
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
3 years agocmd/link: don't cast end address to int32
Cherry Mui [Wed, 12 May 2021 02:40:02 +0000 (22:40 -0400)]
cmd/link: don't cast end address to int32

When linking a very large binary, the section address may not fit
in int32. Don't truncate it.

Change-Id: Ibcc8d74bf5662611949e547ce44ca8b973de383f
Reviewed-on: https://go-review.googlesource.com/c/go/+/319289
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go/testdata/script: fix test failing on nocgo builders
Manlio Perillo [Wed, 12 May 2021 13:16:20 +0000 (15:16 +0200)]
cmd/go/testdata/script: fix test failing on nocgo builders

The regression test introduced in https://golang.org/cl/318770 broke the
the nocgo builders.

Update the cgo package used in the test to ensure that it can be build
both with cgo enabled and disabled.

Change-Id: Iab0486f0b85ac5e5a22fdf8a1998edd50cbb4d96
Reviewed-on: https://go-review.googlesource.com/c/go/+/319210
Trust: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoruntime: fix typo in proc.go
Guilherme Souza [Tue, 11 May 2021 21:37:56 +0000 (21:37 +0000)]
runtime: fix typo in proc.go

Change-Id: I12c0befc5772a5c902a55aeb06a30ec7a34a3bd6
GitHub-Last-Rev: 7d41e1bcb9f6304e1b868701740279e845c99a66
GitHub-Pull-Request: golang/go#46112
Reviewed-on: https://go-review.googlesource.com/c/go/+/319053
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/go: ignore implicit imports when the -find flag is set
Manlio Perillo [Tue, 11 May 2021 15:02:20 +0000 (17:02 +0200)]
cmd/go: ignore implicit imports when the -find flag is set

The documentation of the go list -find flag says that the Deps list will
be empty. However the current implementation adds implicit imports when
supporting Cgo or SWIG and when linking a main package.

Update the documentation of PackageOpts.IgnoreImport to clarify that
both explicit and implicit imports are ignored.

Add a regression test.

Fixes #46092

Change-Id: I37847528d84adb7a18eb6ff29e4af4b4318a66fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/318770
Trust: Jay Conrod <jayconrod@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agonet/http: check that Unicode-aware functions are not used
Filippo Valsorda [Mon, 10 May 2021 17:32:13 +0000 (19:32 +0200)]
net/http: check that Unicode-aware functions are not used

Change-Id: I398aff06bec95077bfff02bfb067aa949b70c184
Reviewed-on: https://go-review.googlesource.com/c/go/+/318429
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Roberto Clapis <roberto@golang.org>
Trust: Filippo Valsorda <filippo@golang.org>

3 years agoruntime: hold sched.lock across atomic pidleget/pidleput
Michael Pratt [Mon, 10 May 2021 20:50:32 +0000 (16:50 -0400)]
runtime: hold sched.lock across atomic pidleget/pidleput

As a cleanup, golang.org/cl/307914 unintentionally caused the idle GC
work recheck to drop sched.lock between acquiring a P and committing to
keep it (once a worker G was found).

This is unsafe, as releasing a P requires extra checks once sched.lock
is taken (such as for runSafePointFn). Since checkIdleGCNoP does not
perform these extra checks, we can now race with other users.

In the case of #45975, we may hang with this sequence:

1. M1: checkIdleGCNoP takes sched.lock, gets P1, releases sched.lock.
2. M2: forEachP takes sched.lock, iterates over sched.pidle without
   finding P1, releases sched.lock.
3. M1: checkIdleGCNoP puts P1 back in sched.pidle.
4. M2: forEachP waits forever for P1 to run the safePointFn.

Change back to the old behavior of releasing sched.lock only after we
are certain we will keep the P. Thus if we put it back its removal from
sched.pidle was never visible.

Fixes #45975
For #45916
For #45885
For #45884

Change-Id: I191a1800923b206ccaf96bdcdd0bfdad17b532e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/318569
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
3 years agoruntime,syscall: simplify openbsd related build tags
Joel Sing [Sun, 9 May 2021 17:16:28 +0000 (03:16 +1000)]
runtime,syscall: simplify openbsd related build tags

openbsd/mips64 is now the only openbsd port that uses non-libc syscall - revise
build tags to reflect this.

Update #36435

Change-Id: I357b2dd2926d058e25e618fcca42c388587598a8
Reviewed-on: https://go-review.googlesource.com/c/go/+/317919
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agonet/http: switch HTTP1 to ASCII equivalents of string functions
Roberto Clapis [Wed, 7 Apr 2021 12:36:40 +0000 (14:36 +0200)]
net/http: switch HTTP1 to ASCII equivalents of string functions

The current implementation uses UTF-aware functions
like strings.EqualFold and strings.ToLower.

This could, in some cases, cause http smuggling.

Change-Id: I0e76a993470a1e1b1b472f4b2859ea0a2b22ada0
Reviewed-on: https://go-review.googlesource.com/c/go/+/308009
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Roberto Clapis <roberto@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
3 years agocrypto/elliptic: upgrade from generic curve impl to specific if available
Roland Shoemaker [Thu, 14 May 2020 01:24:16 +0000 (18:24 -0700)]
crypto/elliptic: upgrade from generic curve impl to specific if available

This change alters the CurveParam methods to upgrade from the generic
curve implementation to the specific P224 or P256 implementations when
called on the embedded CurveParams. This removes the trap of using
elliptic.P224().Params() instead of elliptic.P224(), for example, which
results in using the generic implementation instead of the optimized
constant time one. For P224 this is done for all of the CurveParams
methods, except Params, as the optimized implementation covers all
these methods. For P256 this is only done for ScalarMult and
ScalarBaseMult, as despite having implementations of addition and
doubling they aren't exposed and instead the generic implementation is
used. For P256 an additional check that there actually is a specific
implementation is added, as unlike the P224 implementation the P256 one
is only available on certain platforms.

This change takes the simple, fast approach to checking this, it simply
compares pointers. This removes the most obvious class of mistakes
people make, but still allows edge cases where the embedded CurveParams
pointer has been dereferenced (as seen in the unit tests) or when someone
has manually constructed their own CurveParams that matches one of the
standard curves. A more complex approach could be taken to also address
these cases, but it would require directly comparing all of the
CurveParam fields which would, in the worst case, require comparing
against two standard CurveParam sets in the ScalarMult and
ScalarBaseMult paths, which are likely to be the hottest already.

Updates #34648

Change-Id: I82d752f979260394632905c15ffe4f65f4ffa376
Reviewed-on: https://go-review.googlesource.com/c/go/+/233939
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
3 years agocmd/internal/objfile: add objabi.SNOPTRDATA to "D"
Tao Qingyun [Sat, 12 Dec 2020 09:45:45 +0000 (09:45 +0000)]
cmd/internal/objfile: add objabi.SNOPTRDATA to "D"

Change-Id: I65913534a4a3e2cbc0d4b00454dd3092eb908cb5
GitHub-Last-Rev: 39dc0d21b81eb6aeec4c29d4ea72e6c1ef7fea0d
GitHub-Pull-Request: golang/go#43151
Reviewed-on: https://go-review.googlesource.com/c/go/+/277452
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: Heschi Kreinick <heschi@google.com>

3 years agoruntime/cgo,cmd/internal/obj/ppc64: fix signals with cgo
Lynn Boger [Wed, 31 Mar 2021 17:28:47 +0000 (12:28 -0500)]
runtime/cgo,cmd/internal/obj/ppc64: fix signals with cgo

Recently some tsan tests were enabled on ppc64le which had not
been enabled before. This resulted in failures on systems with
tsan available, and while debugging it was determined that
there were other issues related to the use of signals with cgo.

Signals were not being forwarded within programs linked against
libtsan because the nocgo sigaction was being called for ppc64le
with or without cgo. Adding callCgoSigaction and calling that
allows signals to be registered so that signal forwarding works.

For linux-ppc64 and aix-ppc64, this won't change. On linux-ppc64
there is no cgo. I can't test aix-ppc64 so those owners can enable
it if they want.

In reviewing comments about sigtramp in sys_linux_arm64 it was
noted that a previous issue in arm64 due to missing callee save
registers could also be a problem on ppc64x, so code was added
to save and restore those.

Also, the use of R31 as a temp register in some cases caused an
issue since it is a nonvolatile register in C and was being clobbered
in cases where the C code expected it to be valid. The code sequences to
load these addresses were changed to avoid the use of R31 when loading
such an address.

To get around a vet error, the stubs_ppc64x.go file in runtime
was split into stubs_ppc64.go and stubs_ppc64le.go.

Updates #45040

Change-Id: Ia4ecff950613cbe1b89471790b1d3819d5b5cfb9
Reviewed-on: https://go-review.googlesource.com/c/go/+/306369
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <carlos.seo@linaro.org>
3 years agogo/build: include files with parse errors in GoFiles and other lists
Jay Conrod [Wed, 8 Jul 2020 20:20:51 +0000 (16:20 -0400)]
go/build: include files with parse errors in GoFiles and other lists

go/build.ImportDir returns a *build.Package with various lists of
files. If a file is invalid for some reason, for example, because it
has a different package name than other files, it's added to
InvalidGoFiles in addition to GoFiles, TestGoFiles, or other lists.

Previously, files with parse errors or build constraint errors were
not included in these lists, which causes problems for tools that use
'go list' since InvalidGoFiles is not printed. With this change, files
with any kind of error are added to one of the GoFiles lists.

Fixes #39986

Change-Id: Iee007b5092293eb4420c8a39ce731805fe32135f
Reviewed-on: https://go-review.googlesource.com/c/go/+/241577
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agonet, runtime: drop macOS 10.12 skip conditions in tests
Tobias Klauser [Mon, 10 May 2021 12:52:22 +0000 (14:52 +0200)]
net, runtime: drop macOS 10.12 skip conditions in tests

Go 1.17 requires macOS 10.13 or later. Thus, drop the special cases for
the darwin-amd64-10_12 builder added in CL 202618.

Updates #22019
Updates #23011
Updates #32919

Change-Id: Idef11c213dfb25fd002b7cda6d425cf2e26a2e06
Reviewed-on: https://go-review.googlesource.com/c/go/+/318329
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/go: include packages with InvalidGoFiles when filtering main packages
Bryan C. Mills [Wed, 5 May 2021 21:08:39 +0000 (17:08 -0400)]
cmd/go: include packages with InvalidGoFiles when filtering main packages

If a package has files with conflicting package names, go/build
empirically populates the first name encountered and puts the
remaining files in InvalidGoFiles. That foiled our check for packages
whose name is either unpopulated or "main", since the "package main"
could be found in a source file after the first.

Instead, we now treat any package with a nonzero set of InvalidGoFiles
as potentially a main package. This biases toward over-reporting
errors, but we would rather over-report than under-report.

If we fix #45999, we will be able to make these error checks more
precise.

Updates #42088
Fixes #45827
Fixes #39986

Change-Id: I588314341b17961b38660192c2130678dc03023e
Reviewed-on: https://go-review.googlesource.com/c/go/+/317300
Trust: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/go: add a test that reproduces #45827
Bryan C. Mills [Mon, 3 May 2021 21:03:01 +0000 (17:03 -0400)]
cmd/go: add a test that reproduces #45827

For #45827

Change-Id: I4d3268d66fb0927161f44b353faef11aa4551e40
Reviewed-on: https://go-review.googlesource.com/c/go/+/317298
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agogo/build: avoid duplicates in InvalidGoFiles
Bryan C. Mills [Wed, 5 May 2021 21:05:51 +0000 (17:05 -0400)]
go/build: avoid duplicates in InvalidGoFiles

For #45827
For #39986
Updates #45999

Change-Id: I0c919b6a2e56e7003b90425487eafe0f0eadc609
Reviewed-on: https://go-review.googlesource.com/c/go/+/317299
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agocmd/compile,reflect: allow longer type names
Keith Randall [Sun, 9 May 2021 18:06:17 +0000 (11:06 -0700)]
cmd/compile,reflect: allow longer type names

Encode the length of type names and tags in a varint encoding
instead of a fixed 2-byte encoding. This allows lengths longer
than 65535 (which can happen for large unnamed structs).

Removed the alignment check for #14962, it isn't relevant any more
since we're no longer reading pointers directly out of this data
(it is encoded as an offset which is copied out bytewise).

Fixes #44155
Update #14962

Change-Id: I6084f6027e5955dc16777c87b0dd5ea2baa49629
Reviewed-on: https://go-review.googlesource.com/c/go/+/318249
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoruntime: skip TestCrashDumpsAllThreads on openbsd/arm
Joel Sing [Wed, 17 Feb 2021 06:12:03 +0000 (17:12 +1100)]
runtime: skip TestCrashDumpsAllThreads on openbsd/arm

This test is also now flakey on this platform.

Updates #36435
Updates #42464

Change-Id: Idedb81478178ffffe7a9c125a6e8bbd83458f9ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/315794
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime,syscall: convert syscall on openbsd/arm to libc
Joel Sing [Thu, 4 Feb 2021 17:22:18 +0000 (04:22 +1100)]
runtime,syscall: convert syscall on openbsd/arm to libc

Convert the syscall package on openbsd/arm to use libc rather than performing
direct system calls.

Updates #36435

Change-Id: Iff3a91c959292cbf4e0a09c7fd34efc8e88ff83f
Reviewed-on: https://go-review.googlesource.com/c/go/+/315793
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime: switch runtime to libc for openbsd/arm
Joel Sing [Wed, 3 Feb 2021 07:17:35 +0000 (18:17 +1100)]
runtime: switch runtime to libc for openbsd/arm

Use libc rather than performing direct system calls for the runtime on
openbsd/arm.

Updates #36435

Change-Id: If64a96a61c80b9748792f8a85a8f16ed6ebca91f
Reviewed-on: https://go-review.googlesource.com/c/go/+/315792
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime: switch openbsd/arm locking to libc
Joel Sing [Sun, 31 Jan 2021 16:51:02 +0000 (03:51 +1100)]
runtime: switch openbsd/arm locking to libc

Switch openbsd/arm to locking via libc, rather than performing direct
system calls.

Update #36435

Change-Id: I190abb1aa544d2cb406fe412960ec106c9716f87
Reviewed-on: https://go-review.googlesource.com/c/go/+/315791
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime: switch openbsd/arm to pthreads
Joel Sing [Sat, 30 Jan 2021 17:21:47 +0000 (04:21 +1100)]
runtime: switch openbsd/arm to pthreads

This switches openbsd/arm to thread creation via pthreads, rather than doing
direct system calls.

Update #36435

Change-Id: Ia8749e3723a9967905c33b6d93dfd9be797a486c
Reviewed-on: https://go-review.googlesource.com/c/go/+/315790
Trust: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>