]> Cypherpunks repositories - gostls13.git/log
gostls13.git
2 years agocmd/go/testdata/script: fix skip on list_replace_absolute_windows
Michael Matloob [Wed, 8 Jun 2022 23:14:11 +0000 (19:14 -0400)]
cmd/go/testdata/script: fix skip on list_replace_absolute_windows

The test should skip if it's not on windows *or* it's a short test, but
instead is now skipping if it's not on windows *and* it's a short test,
causing it to be run on non-windows longtest builders.

Change-Id: Ica011bab632b713b0564fefabd5b42878d401844
Reviewed-on: https://go-review.googlesource.com/c/go/+/411122
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Michael Matloob <matloob@golang.org>
Auto-Submit: Michael Matloob <matloob@golang.org>

2 years agoapi/next: minor reformat
Cherry Mui [Wed, 8 Jun 2022 22:08:30 +0000 (18:08 -0400)]
api/next: minor reformat

Add newline endings to files without them. Delete empty lines. So
it is consistent and easier to put them together.

Change-Id: I84e6b7a1fe59e9f4d7f00f61539f6449f19a5d40
Reviewed-on: https://go-review.googlesource.com/c/go/+/411121
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agoruntime: use pidleget for faketime jump
Michael Pratt [Wed, 8 Jun 2022 19:59:37 +0000 (15:59 -0400)]
runtime: use pidleget for faketime jump

In faketime mode, checkdead is responsible for jumping time forward to
the next timer expiration, and waking an M to handle the newly ready
timer.

Currently it pulls the exact P that owns the next timer off of the pidle
list. In theory this is efficient because that P is immediately eligible
to run the timer without stealing. Unfortunately it is also fraught with
peril because we are skipping all of the bookkeeping in pidleget:

* Skipped updates to timerpMask mean that our timers may not be eligible
  for stealing, as they should be.
* Skipped updates to idlepMask mean that our runq may not be eligible
  for stealing, as they should be.
* Skipped updates to sched.npidle may break tracking of spinning Ms,
  potentially resulting in lost work.
* Finally, as of CL 410122, skipped updates to p.limiterEvent may affect
  the GC limiter, or cause a fatal throw when another event occurs.

The last case has finally undercovered this issue since it quickly
results in a hard crash.

We could add all of these updates into checkdead, but it is much more
maintainable to keep this logic in one place and use pidleget here like
everywhere else in the runtime. This means we probably won't wake the
P owning the timer, meaning that the P will need to steal the timer,
which is less efficient, but faketime is not a performance-sensitive
build mode. Note that the M will automatically make itself a spinning M
to make it eligible to steal since it is the only one running.

Fixes #53294
For #52890

Change-Id: I4acc3d259b9b4d7dc02608581c8b4fd259f272e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/411119
Run-TryBot: Michael Pratt <mpratt@google.com>
Auto-Submit: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agocmd/go: clean paths before using them form index functions
Michael Matloob [Wed, 8 Jun 2022 18:26:44 +0000 (14:26 -0400)]
cmd/go: clean paths before using them form index functions

We use str.TrimFilePathPrefix to trim the module root prefix and get the
relative path of each package in the module when scanning the module
and in the RelPath function.  Make sure to clean the path before
indexing and in RelPath to ensure that each path starts with that
prefix, because walk will clean the root path before joining each
subdirectory path to it.

Change-Id: I1dc1eddbd42030eb6d5d8e76a8675f94216447c3
Reviewed-on: https://go-review.googlesource.com/c/go/+/411118
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
2 years agosyscall: remove unused setgroups on linux/loong64
Tobias Klauser [Wed, 8 Jun 2022 17:47:31 +0000 (19:47 +0200)]
syscall: remove unused setgroups on linux/loong64

Setgroups in syscall_linux.go already wraps the setgroups(2) syscall
with correct POSIX semantics (ref. CL 210639).

Change-Id: I961cd7c7fce1d70d23bf13cc82cad17854bbf40e
Reviewed-on: https://go-review.googlesource.com/c/go/+/411214
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

2 years agoruntime: skip TestGdbBacktrace on gdb bug
Austin Clements [Wed, 8 Jun 2022 16:24:55 +0000 (12:24 -0400)]
runtime: skip TestGdbBacktrace on gdb bug

Very rarely, GDB will successfully run the whole test and the inferior
will exit successfully, and then GDB itself hangs and never exits.
Detect this and skip the test as flaky. We could just continue the
test since all of the output we need is there, but by skipping it
we're less likely to notice serious regressions in this test.

Fixes #37405.

Change-Id: I016cbb06f48673f064733da3e3f1ddcbefd58159
Reviewed-on: https://go-review.googlesource.com/c/go/+/411117
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agonet: fix testHookDialTCP race
Damien Neil [Tue, 7 Jun 2022 23:53:53 +0000 (16:53 -0700)]
net: fix testHookDialTCP race

CL 410754 introduces a race accessing the global testHookDialTCP hook.
Avoiding this race is difficult, since Dial can return while
goroutines it starts are still running. Add a version of this
hook to sysDialer, so it can be set on a per-test basis.

(Perhaps other uses of this hook should be moved to use the
sysDialer-local hook, but this change fixes the immediate data race.)

For #52173.

Change-Id: I8fb9be13957e91f92919cae7be213c38ad2af75a
Reviewed-on: https://go-review.googlesource.com/c/go/+/410957
Run-TryBot: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/go: enable module index by default
Michael Matloob [Tue, 7 Jun 2022 17:53:40 +0000 (13:53 -0400)]
cmd/go: enable module index by default

This changes the module index to be enabled by default, rather than
disabled by default. The index can still be disabled by setting
GODEBUG=index=0.

Fixes #53290.

Change-Id: Ic3728fc69d96bb6ef56b56e8c9f2dce35f2923cc
Reviewed-on: https://go-review.googlesource.com/c/go/+/410821
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
2 years agocmd/go: properly call PackageModuleRoot to get modroot for index
Michael Matloob [Tue, 7 Jun 2022 17:54:53 +0000 (13:54 -0400)]
cmd/go: properly call PackageModuleRoot to get modroot for index

PackageModuleRoot needs to be called with the package's path, not
its directory on disk.

Change-Id: I080fe8ce2aeb72e1466624db81595a00915606bb
Reviewed-on: https://go-review.googlesource.com/c/go/+/410820
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>

2 years agocmd/go: set Root and target fields for packages in GOPATH
Michael Matloob [Tue, 7 Jun 2022 19:31:20 +0000 (15:31 -0400)]
cmd/go: set Root and target fields for packages in GOPATH

This change replicates the behavior filed in issue #37015 for packages
imported from the module index. That behavior is that packages that
happen to exist in a GOPATH src directory have p.Root and p.Target set
even when the packages are loaded from modules. This is likely
unintentional behavior because in module mode, packages shouldn't behave
differently depending on whether their directories exist in GOPATH. But
for uniformity, (and because two of our tests depend on this behavior),
this CL will implement this behavior. We can remove it from the module
index when we remove it from the go/build logic.

Change-Id: I3f501c92fbb76eaf86b6b9275539f2129b67f884
Reviewed-on: https://go-review.googlesource.com/c/go/+/410822
Reviewed-by: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
2 years agodoc/go1.19: delete remaining TODOs
Russ Cox [Wed, 8 Jun 2022 15:33:53 +0000 (11:33 -0400)]
doc/go1.19: delete remaining TODOs

The crypto ones were done in a separate CL and didn't merge well.
Same for runtime/debug.

The others are stale.

For #51400.

Change-Id: Iadb4de94d21cd6a20f52277a1c3d7800a729b81e
Reviewed-on: https://go-review.googlesource.com/c/go/+/411115
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
2 years agoruntime: gofmt
Michael Pratt [Wed, 8 Jun 2022 14:46:42 +0000 (10:46 -0400)]
runtime: gofmt

libfuzzerHookStrCmp is manually reformatted into a proper go doc list.

We don't always format testdata, but these test programs are standard Go
programs that can be formatted.

Change-Id: I4dde398bca225ae8c72e787e4d43fd0ccfd0a90b
Reviewed-on: https://go-review.googlesource.com/c/go/+/411114
Auto-Submit: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agodoc/go1.19: complete most remaining TODOs
Russ Cox [Tue, 7 Jun 2022 04:02:29 +0000 (00:02 -0400)]
doc/go1.19: complete most remaining TODOs

The ones I left behind are almost entirely ones that I see pending CLs for.
Also make various fixes to existing text.

For #51400.

Change-Id: I555e0074c9df82b5bdb345e21a08c8757ca147b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/410814
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
2 years agoruntime: remove unused pipe and setNonblock on linux/loong64
Tobias Klauser [Tue, 7 Jun 2022 05:25:10 +0000 (07:25 +0200)]
runtime: remove unused pipe and setNonblock on linux/loong64

CL 389354 removed the fallback to pipe on all platforms with pipe2. This
is the case for linux. Thus, pipe and setNonblock are no longer needed
on linux/loong64 too.

Change-Id: I089adf918d0fd8de5d4d61a893707a2660f89183
Reviewed-on: https://go-review.googlesource.com/c/go/+/410736
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agodoc/go1.19: mention riscv64 supported regabi
Meng Zhuo [Sat, 28 May 2022 04:39:35 +0000 (12:39 +0800)]
doc/go1.19: mention riscv64 supported regabi

Change-Id: I715e53e4baf67f896fa9c240f7668ce11f7b33c3
Reviewed-on: https://go-review.googlesource.com/c/go/+/409195
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
2 years agoos: document that Chdir affects fs.FS returned by DirFS with a relative path
Dan Kortschak [Fri, 3 Jun 2022 11:21:26 +0000 (20:51 +0930)]
os: document that Chdir affects fs.FS returned by DirFS with a relative path

Fixes #47214.

Change-Id: I6fdc1c4340c0943b825ac22e311179ad1cf30915
Reviewed-on: https://go-review.googlesource.com/c/go/+/410334
Reviewed-by: Ian Lance Taylor <iant@google.com>
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: Cherry Mui <cherryyz@google.com>
2 years agosyscall: remove unused accept on linux/loong64
Tobias Klauser [Tue, 7 Jun 2022 05:25:36 +0000 (07:25 +0200)]
syscall: remove unused accept on linux/loong64

accept is no longer used on Linux since CL 346849 changed Accept to use
accept4 only.

This follows CL 386415 which already removed accept on all other Linux
platforms before the linux/loong64 port was submitted.

For #45964

Change-Id: I26945ff780e71174a0b0c2f5313c4bc1e1cbf786
Reviewed-on: https://go-review.googlesource.com/c/go/+/410737
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agonet: use synthetic network in TestDialParallel
Damien Neil [Mon, 6 Jun 2022 22:52:19 +0000 (15:52 -0700)]
net: use synthetic network in TestDialParallel

TestDialParallel is testing the Happy Eyeballs algorithm implementation,
which dials IPv4 and IPv6 addresses in parallel with the preferred
address family getting a head start. This test doesn't care about
the actual network operations, just the handling of the parallel
connections.

Use testHookDialTCP to replace socket creation with a function that
returns successfully, with an error, or after context cancellation
as required.

Limit tests of elapsed times to a check that the fallback deadline
has been exceeded in cases where this is expected.

This should fix persistent test flakiness.

Fixes #52173.

Change-Id: Ic93f270fccb63b24a91105a4d541479fc33a2de4
Reviewed-on: https://go-review.googlesource.com/c/go/+/410754
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agodoc/go1.19: document that the assembler requires -p
Cherry Mui [Tue, 7 Jun 2022 20:35:13 +0000 (16:35 -0400)]
doc/go1.19: document that the assembler requires -p

For #51400.

Change-Id: I50fb4313105ae6dbbbe2c98cbe4a8f8e2563eba9
Reviewed-on: https://go-review.googlesource.com/c/go/+/410824
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agodoc/go1.19: document linker CL that switches DWARF compressed section format
Cherry Mui [Tue, 7 Jun 2022 20:30:37 +0000 (16:30 -0400)]
doc/go1.19: document linker CL that switches DWARF compressed section format

For #51400.
Updates #50796.

Change-Id: Ica6c700a5b54e4712b09c43d1d7a9c3bba408b8b
Reviewed-on: https://go-review.googlesource.com/c/go/+/410823
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Fangrui Song <maskray@google.com>
2 years agogo/types, types2: better error message for invalid use of constraint type
Robert Griesemer [Tue, 7 Jun 2022 20:39:56 +0000 (13:39 -0700)]
go/types, types2: better error message for invalid use of constraint type

Fixes #42881.

Change-Id: If800c5f90c0034d192bf8b6649e5cfda96df48cb
Reviewed-on: https://go-review.googlesource.com/c/go/+/410954
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
2 years agogo/types, types2: better error message if type is not in type set
Robert Griesemer [Tue, 7 Jun 2022 17:43:51 +0000 (10:43 -0700)]
go/types, types2: better error message if type is not in type set

Fixes #40350.

Change-Id: Ia654d6b854971700ca618692a864265557122b23
Reviewed-on: https://go-review.googlesource.com/c/go/+/410876
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agogo/types, types2: use | rather than ∪ when printing term lists
Robert Griesemer [Tue, 7 Jun 2022 17:33:01 +0000 (10:33 -0700)]
go/types, types2: use | rather than ∪ when printing term lists

With this change, the termlist String() function prints termlists
in the usual Go notation and thus we can use it in error reporting.
Preparation for fixing #40350.

For #40350.

Change-Id: Ia28318841305de234a71af3146ce0c59f5e601a5
Reviewed-on: https://go-review.googlesource.com/c/go/+/410894
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
2 years agodoc/go1.19: add release notes for net/http and net/url
Damien Neil [Thu, 26 May 2022 22:09:34 +0000 (15:09 -0700)]
doc/go1.19: add release notes for net/http and net/url

For #51400

Change-Id: I6412132db79074eef7d2cb3d66456c48b0d745a3
Reviewed-on: https://go-review.googlesource.com/c/go/+/408877
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
2 years agodoc/go1.19: adjust runtime release notes
Michael Pratt [Mon, 6 Jun 2022 21:32:03 +0000 (17:32 -0400)]
doc/go1.19: adjust runtime release notes

This addresses comments from CL 410356.

For #48409.
For #51400.

Change-Id: I03560e820a06c0745700ac997b02d13bc03adfc6
Reviewed-on: https://go-review.googlesource.com/c/go/+/410735
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Chris Hines <chris.cs.guy@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
2 years agoruntime: document GOMEMLIMIT in environment variables section
Michael Pratt [Mon, 6 Jun 2022 21:20:44 +0000 (17:20 -0400)]
runtime: document GOMEMLIMIT in environment variables section

For #48409.

Change-Id: Ia6616a377bc4c871b7ffba6f5a59792a09b64809
Reviewed-on: https://go-review.googlesource.com/c/go/+/410734
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Chris Hines <chris.cs.guy@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
2 years agodoc/go1.19: document loong64 port
David Chase [Tue, 7 Jun 2022 16:28:54 +0000 (12:28 -0400)]
doc/go1.19: document loong64 port

Updates #46229
For #51400

Change-Id: Iedd5d3c4cd656b59ba2e1fe813851830849a8614
Reviewed-on: https://go-review.googlesource.com/c/go/+/410816
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2 years agosync/atomic: clarify that 8-byte alignment of variables is due to escape
Will Hawkins [Sat, 4 Jun 2022 07:44:18 +0000 (03:44 -0400)]
sync/atomic: clarify that 8-byte alignment of variables is due to escape

For #53223.

Change-Id: I79e9b920488581a4d850e4051ee0dd600b5bbcb1
Reviewed-on: https://go-review.googlesource.com/c/go/+/410102
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2 years agodoc/go1.19: some platforms are still on TSAN v2
Austin Clements [Tue, 7 Jun 2022 17:10:32 +0000 (13:10 -0400)]
doc/go1.19: some platforms are still on TSAN v2

For #51400

Change-Id: Ie6d6ac773aa81b105e15ef7399374f574197d775
Reviewed-on: https://go-review.googlesource.com/c/go/+/410817
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agodoc/go1.19: compiler section is complete, modulo TODOs
Austin Clements [Mon, 6 Jun 2022 21:05:49 +0000 (17:05 -0400)]
doc/go1.19: compiler section is complete, modulo TODOs

For #51400

Change-Id: I964e52e0a36e7bbe77175670e93ce8c99e7dab6d
Reviewed-on: https://go-review.googlesource.com/c/go/+/410367
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agodoc/go1.19: minor edits
Austin Clements [Mon, 6 Jun 2022 19:07:16 +0000 (15:07 -0400)]
doc/go1.19: minor edits

For #51400

Change-Id: I57565c1d79e0c5487d39d46f556b247d35f05d3c
Reviewed-on: https://go-review.googlesource.com/c/go/+/410674
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
2 years agodoc/go1.19: complete TODOs for go/types
Robert Findley [Tue, 7 Jun 2022 13:42:03 +0000 (09:42 -0400)]
doc/go1.19: complete TODOs for go/types

Fill in the details of outstanding TODO items for go/types changes.

For #51400

Change-Id: Ib40d75fa1018aa164022cb49b293795dd597d49d
Reviewed-on: https://go-review.googlesource.com/c/go/+/410815
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agodoc/go1.19: add various crypto release notes
Roland Shoemaker [Fri, 3 Jun 2022 18:48:43 +0000 (11:48 -0700)]
doc/go1.19: add various crypto release notes

For #51400

Change-Id: I908f53a54c6603e1bf2c9238cd51cf5c4a24407b
Reviewed-on: https://go-review.googlesource.com/c/go/+/410295
Reviewed-by: Russ Cox <rsc@golang.org>
Auto-Submit: Roland Shoemaker <roland@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Tatiana Bradley <tatiana@golang.org>
2 years agoruntime: fix inline assembly trampoline for arm64
Khaled Yakdan [Mon, 23 May 2022 23:20:00 +0000 (23:20 +0000)]
runtime: fix inline assembly trampoline for arm64

Use the program counter to compute the address of the first instruction
of the ret sled. The ret sled is located after 5 instructions from the
MOVD instruction saving the value of the program counter.

Change-Id: Ie7ae7a0807785d6fea035cf7a770dba7f37de0ec
GitHub-Last-Rev: 2719208c6a3b049e0f394e5311ce3282b58f8516
GitHub-Pull-Request: golang/go#53039
Reviewed-on: https://go-review.googlesource.com/c/go/+/407895
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
2 years agocmd/link: specify -Wl,-z params as documented
Motiejus Jakštys [Tue, 7 Jun 2022 05:55:40 +0000 (05:55 +0000)]
cmd/link: specify -Wl,-z params as documented

Both GNU and LLVM linkers de facto accept `-zPARAM`, and Go sometimes
does it. Inconsistently: there are more uses of `-z PARAM` than
`-zPARAM`:

    $ git grep -E -- '-Wl,-z[^,]' master | wc -l
    4
    $ git grep -E -- '-Wl,-z,' master | wc -l
    7

However, not adding a space between `-z` and the param is not
documented:

llvm-13:

    $ man ld.lld-13 | grep -E -A1 -w -- "^ +-z"
         -z option
                 Linker option extensions.

gnu ld:

    $ man ld | grep -E -A1 -w -- "^ +-z"
           -z keyword
               The recognized keywords are:
    --
           -z defs
               Report unresolved symbol references from regular object files.  This is done even if the linker is creating a non-symbolic
    --
           -z muldefs
               Normally when a symbol is defined multiple times, the linker will report a fatal error. These options allow multiple definitions
    --
           -z
           --imagic

... and thus should be avoided.

`zig cc`, when used as the C compiler (`CC="zig cc" go build ...`), will
bark, because `zig cc` accepts only `-z PARAM`, as documented.

Closes ziglang/zig#11669

Change-Id: I758054ecaa3ce01a72600bf65d7f7b5c3ec46d09
GitHub-Last-Rev: e068e007da9f2b0441ee0aa8b198a7ba3cd93ed3
GitHub-Pull-Request: golang/go#53030
Reviewed-on: https://go-review.googlesource.com/c/go/+/407834
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>

2 years agodoc/go1.19: delete boringcrypto TODO
Russ Cox [Sat, 4 Jun 2022 18:46:42 +0000 (14:46 -0400)]
doc/go1.19: delete boringcrypto TODO

Boringcrypto has never been officially supported and it remains unsupported.
It need not be mentioned in the release notes.

Change-Id: I24a08d424982615244d51c1d250035d85a602023
Reviewed-on: https://go-review.googlesource.com/c/go/+/410362
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
2 years agodoc/go1.19: add more TODOs from updated relnote
Russ Cox [Sat, 4 Jun 2022 18:33:41 +0000 (14:33 -0400)]
doc/go1.19: add more TODOs from updated relnote

CL 410244 changes relnote to look for api file changes as well
as references to proposal issues, finding various things that
were missing from the release notes.

This CL adds the TODOs that the updated relnote found.

For #51400.

Change-Id: I512a9b8f1349a6c68c8a6979f55a07964d630175
Reviewed-on: https://go-review.googlesource.com/c/go/+/410361
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
2 years agodoc/go1.19: add release notes for the soft memory limit and idle GC
Michael Anthony Knyszek [Sat, 4 Jun 2022 05:42:30 +0000 (05:42 +0000)]
doc/go1.19: add release notes for the soft memory limit and idle GC

This change resolves some TODOs in the release notes, and while we're
here, also clarifies how CPU profile samples are represented in runtime
traces.

Change-Id: Idaa36ccf65b03fd5463b2d5da682d3fa578d2f46
Reviewed-on: https://go-review.googlesource.com/c/go/+/410356
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2 years agoruntime, sync, sync/atomic: document happens-before guarantees
Russ Cox [Wed, 26 Jan 2022 21:53:50 +0000 (16:53 -0500)]
runtime, sync, sync/atomic: document happens-before guarantees

A few of these are copied from the memory model doc.
Many are entirely new, following discussion on #47141.
See https://research.swtch.com/gomm for background.

The rule we are establishing is that each type that is meant
to help synchronize a Go program should document its
happens-before guarantees.

For #50859.

Change-Id: I947c40639b263abe67499fa74f68711a97873a39
Reviewed-on: https://go-review.googlesource.com/c/go/+/381316
Auto-Submit: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
2 years agogo/doc/comment: add heuristics for common badly formatted comments
Russ Cox [Fri, 3 Jun 2022 16:44:58 +0000 (12:44 -0400)]
go/doc/comment: add heuristics for common badly formatted comments

In a set of 55M Go doc comments drawn from the latest version of
all public Go modules known to the module proxy in spring 2020,
the current Go 1.19 gofmt reformats about 1.57M of them.
Out of those 1.57M comments, inspection of random samples
shows that around 5% of the changed comments contain
unindented code snippets, multiline shell commands, or lists.
For example:

// Here is a greeting:
//
// func main() {
// fmt.Println("hello, world")
// }

// Run this command:
//
// path/to/your/program -flag1=longargument1 \
// -flag2=longargument2 \
// -flag3

// There are three possibilities:
//
// - Unindented code snippets (or JSON objects)
//    in which the first and last line are unindented
//    but end in { and start with }, respectively.
// - Unindented multiline shell commands
//    in which the lines end in \
// - Unindented lists, in which wrapped lines are indented.

All three of these cases involve unindented lines next to indented
lines that would according to the usual rules begin a pre block.
Before this CL, they'd be reformatted to:

// Here is a greeting:
//
// func main() {
//
// fmt.Println("hello, world")
//
// }

// Run this command:
//
// path/to/your/program -flag1=longargument1 \
//
// -flag2=longargument2 \
// -flag3

// There are three possibilities:
//
// - Unindented code snippets (or JSON objects)
//
// in which the first and last line are unindented
// but end in { and start with }, respectively.
//
// - Unindented multiline shell commands
//
// in which the lines end in \
//
// - Unindented lists, in which wrapped lines are indented.

The fact that they are not already in canonical format gives us
a signal that they might not mean what the usual rules would say.

This CL takes advantage of that opening to apply a few heuristics
to better handle these cases:

 1. If an indented code block immediately follows (without a blank line)
    an unindented line ending in { or \, include the unindented line
    in the code block.

 2. If an indented code block immediately precedes (without a blank line)
    an unindented line beginning with }, include the unindented line
    in the code block.

 3. If an indented line immediately follows (without a blank line)
    an unindented line that starts with a list marker, assume this is
    an unindented list with a wrapped indented line, and treat all
    adjacent unindented lines starting with list markers as part of
    the list, stopping at any surrounding blank lines.

This raises the fraction of “correctly” reformatted doc comments
in the corpus from approximately 87% to approximately 93%.

Change-Id: I7ac542eb085032d607a7caf3ba9020787b2978b5
Reviewed-on: https://go-review.googlesource.com/c/go/+/410360
Auto-Submit: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
2 years agodoc/go_mem: update revision date
Russ Cox [Mon, 6 Jun 2022 19:20:51 +0000 (15:20 -0400)]
doc/go_mem: update revision date

CL 381315 added major revisions but neglected to update the date.

For #50859.

Change-Id: I086a55f0c80579c479bca5268109c9f3ae680adf
Reviewed-on: https://go-review.googlesource.com/c/go/+/410675
Reviewed-by: Roland Shoemaker <roland@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Auto-Submit: Russ Cox <rsc@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
2 years agodoc/go1.19: gc requires -p=importpath
Austin Clements [Mon, 6 Jun 2022 19:45:00 +0000 (15:45 -0400)]
doc/go1.19: gc requires -p=importpath

For #51400

Change-Id: I07a805147a6aa0923331f3f940a9e6e5553cbea9
Reviewed-on: https://go-review.googlesource.com/c/go/+/410676
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2 years agodoc/go1.19: document Resolver.PreferGo
Brad Fitzpatrick [Sat, 4 Jun 2022 04:18:14 +0000 (21:18 -0700)]
doc/go1.19: document Resolver.PreferGo

Updates #51400

Change-Id: I61733574362d4cf3cb65122bd13361e5c0f6728c
Reviewed-on: https://go-review.googlesource.com/c/go/+/410375
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Auto-Submit: Michael Pratt <mpratt@google.com>

2 years agocmd/go: use index to match packages in dependency modules
Michael Matloob [Tue, 3 May 2022 17:17:08 +0000 (13:17 -0400)]
cmd/go: use index to match packages in dependency modules

If we're trying to search in a module in the module cache, instead
iterate over the packages in the index.
Change-Id: Ia94cbe6e9690110c28b93dbb33810680e3010381
Reviewed-on: https://go-review.googlesource.com/c/go/+/403756
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Peter Weinberger <pjw@google.com>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoall: boringcrypto post-merge cleanup
Russ Cox [Wed, 4 May 2022 12:58:02 +0000 (08:58 -0400)]
all: boringcrypto post-merge cleanup

This CL addresses the comments on CL 403154.

For #51940.

Change-Id: I99bb3530916d469077bfbd53095bfcd1d2aa82ef
Reviewed-on: https://go-review.googlesource.com/c/go/+/403976
Reviewed-by: Roland Shoemaker <roland@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agogo/doc/comment: do not turn ``` into “`
Russ Cox [Fri, 3 Jun 2022 13:18:57 +0000 (09:18 -0400)]
go/doc/comment: do not turn ``` into “`

``` is Markdown, not Go doc comment, but some small fraction of users get confused.

In a set of 55M Go doc comments drawn from the latest version of
all public Go modules known to the module proxy in spring 2020,
the current Go 1.19 gofmt reformats about 1.57M of them.
Out of those 1.57M comments, 8k of them (about 0.5%) contain ```.

Instead of rewriting ``` to “`, leave it alone.

For #51082.

Change-Id: I1c8c88aac7ef75ec03e1a396b84ffe711c46f941
Reviewed-on: https://go-review.googlesource.com/c/go/+/410359
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>

2 years agogo/doc/comment: add doc comment
Russ Cox [Thu, 2 Jun 2022 02:09:18 +0000 (22:09 -0400)]
go/doc/comment: add doc comment

A CL in the website repo will add go.dev/doc/comment.

One of the final steps for #51082.

Change-Id: I419b4f6dbb424a8a93a8d09db30f7321af9ae976
Reviewed-on: https://go-review.googlesource.com/c/go/+/410358
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agocmd/compile: cast riscv64 rewrite shifts to unsigned int
Joel Sing [Wed, 1 Jun 2022 19:09:09 +0000 (05:09 +1000)]
cmd/compile: cast riscv64 rewrite shifts to unsigned int

This appeases Go 1.4, making it possible to bootstrap GOARCH=riscv64 with
a Go 1.4 compiler.

Fixes #52583

Change-Id: Ib13c2afeb095b2bb1464dcd7f1502574209bc7ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/409974
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agostrconv: fix typo in atof.go
90364136+butterfly1924@users.noreply.github.com [Mon, 6 Jun 2022 05:10:38 +0000 (05:10 +0000)]
strconv: fix typo in atof.go

strings for 'NaN' -> string for 'NaN'

Change-Id: Ia415644a1b651e6ef9996ad24dd9708a60e57dfc
GitHub-Last-Rev: 877f1c3eb1dc885915ae75385c5d38ee6f5fd9b1
GitHub-Pull-Request: golang/go#53246
Reviewed-on: https://go-review.googlesource.com/c/go/+/410494
Auto-Submit: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agocmd/link/internal/loadpe: handle _main reference properly
Than McIntosh [Fri, 3 Jun 2022 11:50:58 +0000 (07:50 -0400)]
cmd/link/internal/loadpe: handle _main reference properly

When building CGO internal linking on windows 386, make sure to avoid
rewriting references to "_main" to "main" when reading symbols during
host object loading; the main routine defined by the Go runtime is
still named "_main" (not "main"). If we don't do this, we wind up with
an SXREF symbol named "main", which can then cause the loader to pull
an actual "main" symbol out of a host archive, which is undesirable.

Updates #35006.

Change-Id: I3768e3617b560552f4522e9e72af879c6adf7705
Reviewed-on: https://go-review.googlesource.com/c/go/+/410124
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Auto-Submit: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agogo/types, types2: simplify implementation of validType (fix TODO)
Robert Griesemer [Fri, 3 Jun 2022 03:40:17 +0000 (20:40 -0700)]
go/types, types2: simplify implementation of validType (fix TODO)

Now that validType is using the correct type nest (CL 409694),
the top entry of the type nest corresponds to the instantiated
type. Thus we can use that type instance to look up the value
of type parameters, there's no need anymore to create an environment
to look up type arguments.

Remove the need to pass around the environment and remove all
associated types and functions.

Updates #52698.

Change-Id: Ie37eace88896386e667ef93c77a4fc3cd0be6eb9
Reviewed-on: https://go-review.googlesource.com/c/go/+/410294
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>

2 years agogo/types, types2: use type nest to detect type cycles (fix validType)
Robert Griesemer [Wed, 1 Jun 2022 03:59:55 +0000 (20:59 -0700)]
go/types, types2: use type nest to detect type cycles (fix validType)

validType was using a global type info map to detect invalid recursive
types, which was incorrect. Instead, change the algorithm as follows:

- Rather than using a "seen" (or typeInfo) map which is cumbersome to
  update correctly, use the stack of embedding types (the type nest)
  to check whether a type is embedded within itself, directly or
  indirectly.

- Use Identical for type comparisons which correctly considers identity
  of instantiated generic types.

- As before, maintain the full path of types leading to a cycle. But
  unlike before, track the named types rather than their objects, for
  a smaller slice ([]*Named rather than []Object), and convert to an
  object list only when needed for error reporting.

- As an optimization, keep track of valid *Named types (Checker.valids).
  This prevents pathological cases from consuming excessive computation
  time.

- Add clarifying comments and document invariants.

Based on earlier insights by David Chase (see also CL 408818).

Fixes #52698.

Change-Id: I5e4598c58afcf4ab987a426c5c4b7b28bdfcf5ea
Reviewed-on: https://go-review.googlesource.com/c/go/+/409694
Reviewed-by: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Griesemer <gri@google.com>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agodoc/go1.19: add TODOs for changes to go/types
Robert Findley [Mon, 6 Jun 2022 15:48:29 +0000 (11:48 -0400)]
doc/go1.19: add TODOs for changes to go/types

Add TODO items for significant changes to go/types: the inclusion of
Origin methods for Var and Func, and a re-working of Named types to
ensure finiteness of reachable types via their API.

Updates #51400

Change-Id: I0f2a972023a5d5f995de3c33e9e2b0a4213e900a
Reviewed-on: https://go-review.googlesource.com/c/go/+/410614
Reviewed-by: Robert Griesemer <gri@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: fix breakpoint in ppc64x
Alejandro Sáez [Mon, 23 May 2022 12:07:47 +0000 (12:07 +0000)]
runtime: fix breakpoint in ppc64x

Currently runtime.Breakpoint generates a SIGSEGV in ppc64.
The solution is an unconditional trap similar to what clang and gcc do. It is documented in the section C.6 of the ABI Book 3.

Fixes #52101

Change-Id: I071d2f2679b695ef268445b04c9222bd74e1f9af
GitHub-Last-Rev: fff4e5e8ffe23bf0cef135b22abd2cc0a3838613
GitHub-Pull-Request: golang/go#52102
Reviewed-on: https://go-review.googlesource.com/c/go/+/397554
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Paul Murphy <murp@ibm.com>
Run-TryBot: Paul Murphy <murp@ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
2 years agotext/template/parse: fix data race on lexer initialization
Eli Bendersky [Sat, 4 Jun 2022 16:43:40 +0000 (09:43 -0700)]
text/template/parse: fix data race on lexer initialization

Before this change, `startParse` would write `lex.breakOK` and `lex.continueOK` when the lexer goroutine is already running, which is a potential race condition.

Makes `breakOK` and `continueOK` configuration flags passed when `lexer` is created, similarly to how `emitComment` works.

Fixes #53234

Change-Id: Ia65f6135509a758cd4c5a453b249a174f4fb3e21
Reviewed-on: https://go-review.googlesource.com/c/go/+/410414
Reviewed-by: Eli Bendersky <eliben@google.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>

2 years agogo/types, types2: ensure that named types never expand infinitely
Robert Findley [Sun, 8 May 2022 01:22:17 +0000 (21:22 -0400)]
go/types, types2: ensure that named types never expand infinitely

During type-checking, newly created instances share a type checking
Context which de-duplicates identical instances. However, when
unexpanded types escape the type-checking pass or are created via calls
to Instantiate, they lack this shared context. As reported in #52728,
this may lead to infinitely many identical but distinct types that are
reachable via the API.

This CL introduces a new invariant that ensures we don't create such
infinitely expanding chains: instances created during expansion share a
context with the type that led to their creation. During expansion, the
expanding type passes its Context to any newly created instances.

This ensures that cycles will eventually terminate with a previously
seen instance. For example, if we have an instantiation chain
T1[P]->T2[P]->T3[P]->T1[P], by virtue of this Context passing the
expansion of T3[P] will find the instantiation T1[P].

In general, storing a Context in a Named type could lead to pinning
types in memory unnecessarily, but in this case the Context pins only
those types that are reachable from the original instance. This seems
like a reasonable compromise between lazy and eager expansion.

Our treatment of Context was a little haphazard: Checker.bestContext
made it easy to get a context at any point, but made it harder to reason
about which context is being used. To fix this, replace bestContext with
Checker.context, which returns the type-checking context and panics on a
nil receiver. Update all call-sites to verify that the Checker is
non-nil when context is called.

Also make it a panic to call subst with a nil context. Instead, update
subst to explicitly accept a local (=instance) context along with a
global context, and require that one of them is non-nil. Thread this
through to the call to Checker.instance, and handle context updating
there.

Fixes #52728

Change-Id: Ib7f26eb8c406290325bc3212fda25421a37a1e8e
Reviewed-on: https://go-review.googlesource.com/c/go/+/404885
Reviewed-by: Robert Griesemer <gri@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>

2 years agogo/types, types2: store Named instance information separately
Robert Findley [Sat, 7 May 2022 23:26:35 +0000 (19:26 -0400)]
go/types, types2: store Named instance information separately

Separate instance information into an instance struct, to reduce memory
footprint for non-instance Named types. This may induce a sense of
deja-vu: we had a similar construct in the past that was removed as
unnecessary. With additional new fields being added that only apply to
instances, having a separate struct makes sense again.

Updates #52728

Change-Id: I0bb5982d71c27e6b574bfb4f7b886a6aeb9c5390
Reviewed-on: https://go-review.googlesource.com/c/go/+/404884
Reviewed-by: Robert Griesemer <gri@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agogo/types, types2: eliminate methodList in favor of just using Named.mu
Robert Findley [Sat, 7 May 2022 22:59:14 +0000 (18:59 -0400)]
go/types, types2: eliminate methodList in favor of just using Named.mu

In order to clean up context after fully expanding a type (in subsequent
CLs), we must use a common mutex. Eliminate the lazy methodList type,
which keeps a sync.Once per method, in favor of Named.mu.

Updates #52728

Change-Id: I2d13319276df1330ee53046ef1823b0167a258d8
Reviewed-on: https://go-review.googlesource.com/c/go/+/404883
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
2 years agogo/types, types2: remove Named.once in favor of monotonic state
Robert Findley [Sun, 15 May 2022 23:31:42 +0000 (19:31 -0400)]
go/types, types2: remove Named.once in favor of monotonic state

Introduce a monotonic state variable to track the lifecycle of a named
type, replacing the existing sync.Once. Having a single guard for the
state of underlying and methods will allow for cleaning-up when the type
is fully expanded. In the future, this state may also be used for
detecting access to information such as underlying or methods before the
type is fully set-up, though that will require rethinking our
type-checking of invalid cyclic types.

Also remove support for type-type inference. If we ever support this
feature in the future (inference of missing type arguments for named
type instances), it will likely involve additional machinery that does
not yet exist. Remove the current partial support to simplify our
internal APIs. In particular, this means that Named.resolver is only
used for lazy loading. As a result, we can revert the lazy loader
signature to its previous form.

A lot of exposition is added for how Named types work. Along the way,
the terminology we use to describe them is refined.

Some microbenchmarks are added that were useful in evaluating the
tradeoffs between synchronization mechanisms.

Updates #52728

Change-Id: I4e147360bc6e5d8cd4f37e32e86fece0530a6480
Reviewed-on: https://go-review.googlesource.com/c/go/+/404875
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/buildid: reject rewriting legacy buildids
Stephen Eckels [Sat, 4 Jun 2022 20:39:36 +0000 (20:39 +0000)]
cmd/buildid: reject rewriting legacy buildids

This resolves legacy go binaries crashing the buildid tool when the -w flag is specified.

Fixes #50809

Change-Id: I55a866f285a3c2cebcf2cdbb9cc30e5078e1d18f
GitHub-Last-Rev: 7169a58fd7ba17fdeb8037cef3f50080169dc137
GitHub-Pull-Request: golang/go#53163
Reviewed-on: https://go-review.googlesource.com/c/go/+/409535
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
2 years agostrconv: clarify ParseFloat accepts Go syntax for float literals
Ben Hoyt [Sat, 4 Jun 2022 06:31:40 +0000 (18:31 +1200)]
strconv: clarify ParseFloat accepts Go syntax for float literals

The documentation for strconv.ParseFloat mentions that it "accepts
decimal and hexadecimal floating-point number syntax", but it doesn't
specify what those formats entail. For example, "0x10" is not allowed;
you need an explicit exponent, as in "0x10p0".

This clarifies that ParseFloat accepts the Go syntax for floating-point
literals, and links to that spec section. I've also linked to the
relevant spec section for ParseInt's doc comment, which already said
"as defined by the Go syntax for integer literals".

Change-Id: Ib5d2b408bdd01ea0b9f69381a9dbe858f6d1d424
Reviewed-on: https://go-review.googlesource.com/c/go/+/410335
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

2 years agoruntime: fix typo in libfuzzer_arm64.s
Ikko Ashimine [Sat, 4 Jun 2022 03:41:06 +0000 (03:41 +0000)]
runtime: fix typo in libfuzzer_arm64.s

statment -> statement

Change-Id: Ia93a466fdc20157a7d6048903e359fe8717ecb8f
GitHub-Last-Rev: 0a9bc5cab0ec2ac8d76ede3722c8813372ac771e
GitHub-Pull-Request: golang/go#53231
Reviewed-on: https://go-review.googlesource.com/c/go/+/410374
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
2 years agodatabase/sql/driver: fix typo in driver.go
mstmdev [Sat, 4 Jun 2022 19:32:04 +0000 (19:32 +0000)]
database/sql/driver: fix typo in driver.go

ExecerContext -> ExecContext
QueryerContext -> QueryContext

Change-Id: Id3b1f44de5aa47372d59696523b4379e1fbfc11c
GitHub-Last-Rev: 571d01f8052cef27bd5f91a5e95c0735ddf2e327
GitHub-Pull-Request: golang/go#53235
Reviewed-on: https://go-review.googlesource.com/c/go/+/410415
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoregexp: avoid copying each instruction executed
Bryan Boreham [Wed, 13 Oct 2021 21:40:22 +0000 (22:40 +0100)]
regexp: avoid copying each instruction executed

Inst is a 40-byte struct, so avoiding the copy gives a decent speedup:

name                            old time/op    new time/op     delta
Find-8                             160ns ± 4%      109ns ± 4%  -32.22%  (p=0.008 n=5+5)
FindAllNoMatches-8                70.4ns ± 4%     53.8ns ± 0%  -23.58%  (p=0.016 n=5+4)
FindString-8                       154ns ± 6%      107ns ± 0%  -30.37%  (p=0.016 n=5+4)
FindSubmatch-8                     194ns ± 1%      135ns ± 1%  -30.56%  (p=0.008 n=5+5)
FindStringSubmatch-8               193ns ± 8%      131ns ± 0%  -31.82%  (p=0.008 n=5+5)
Literal-8                         42.8ns ± 2%     34.8ns ± 0%  -18.67%  (p=0.008 n=5+5)
NotLiteral-8                       917ns ± 2%      636ns ± 0%  -30.68%  (p=0.008 n=5+5)
MatchClass-8                      1.18µs ± 3%     0.91µs ± 1%  -22.27%  (p=0.016 n=5+4)
MatchClass_InRange-8              1.11µs ± 1%     0.87µs ± 2%  -21.38%  (p=0.008 n=5+5)
ReplaceAll-8                       659ns ± 6%      596ns ± 3%   -9.60%  (p=0.008 n=5+5)
AnchoredLiteralShortNonMatch-8    34.2ns ± 0%     30.4ns ± 1%  -11.20%  (p=0.016 n=4+5)
AnchoredLiteralLongNonMatch-8     38.7ns ± 0%     38.7ns ± 0%     ~     (p=0.579 n=5+5)
AnchoredShortMatch-8              67.0ns ± 1%     52.7ns ± 0%  -21.31%  (p=0.016 n=5+4)
AnchoredLongMatch-8                121ns ± 0%      124ns ±10%     ~     (p=0.730 n=5+5)
OnePassShortA-8                    392ns ± 0%      231ns ± 3%  -41.10%  (p=0.008 n=5+5)
NotOnePassShortA-8                 370ns ± 0%      282ns ± 1%  -23.81%  (p=0.008 n=5+5)
OnePassShortB-8                    280ns ± 0%      179ns ± 1%  -36.05%  (p=0.008 n=5+5)
NotOnePassShortB-8                 226ns ± 0%      185ns ± 3%  -18.26%  (p=0.008 n=5+5)
OnePassLongPrefix-8               51.7ns ± 0%     39.1ns ± 1%  -24.28%  (p=0.016 n=4+5)
OnePassLongNotPrefix-8             213ns ± 2%      132ns ± 1%  -37.86%  (p=0.008 n=5+5)
MatchParallelShared-8             25.3ns ± 3%     23.4ns ± 7%   -7.50%  (p=0.016 n=5+5)
MatchParallelCopied-8             26.5ns ± 7%     22.3ns ± 7%  -16.06%  (p=0.008 n=5+5)
QuoteMetaAll-8                    45.8ns ± 1%     45.8ns ± 1%     ~     (p=1.000 n=5+5)
QuoteMetaNone-8                   24.3ns ± 0%     24.3ns ± 0%     ~     (p=0.325 n=5+5)
Compile/Onepass-8                 1.98µs ± 0%     1.97µs ± 0%   -0.22%  (p=0.016 n=5+4)
Compile/Medium-8                  4.56µs ± 0%     4.55µs ± 1%     ~     (p=0.595 n=5+5)
Compile/Hard-8                    35.7µs ± 0%     35.3µs ± 3%     ~     (p=0.151 n=5+5)
Match/Easy0/16-8                  2.18ns ± 2%     2.19ns ± 5%     ~     (p=0.690 n=5+5)
Match/Easy0/32-8                  27.4ns ± 2%     27.6ns ± 4%     ~     (p=1.000 n=5+5)
Match/Easy0/1K-8                   246ns ± 0%      252ns ± 7%     ~     (p=0.238 n=5+5)
Match/Easy0/32K-8                 4.58µs ± 7%     4.64µs ± 5%     ~     (p=1.000 n=5+5)
Match/Easy0/1M-8                   235µs ± 0%      235µs ± 0%     ~     (p=0.886 n=4+4)
Match/Easy0/32M-8                 7.86ms ± 0%     7.86ms ± 1%     ~     (p=0.730 n=4+5)
Match/Easy0i/16-8                 2.15ns ± 0%     2.15ns ± 0%     ~     (p=0.246 n=5+5)
Match/Easy0i/32-8                  507ns ± 2%      466ns ± 4%   -8.03%  (p=0.008 n=5+5)
Match/Easy0i/1K-8                 14.7µs ± 0%     13.6µs ± 2%   -7.63%  (p=0.008 n=5+5)
Match/Easy0i/32K-8                 571µs ± 1%      570µs ± 1%     ~     (p=0.556 n=4+5)
Match/Easy0i/1M-8                 18.2ms ± 0%     18.8ms ±11%     ~     (p=0.548 n=5+5)
Match/Easy0i/32M-8                 581ms ± 0%      590ms ± 1%   +1.52%  (p=0.016 n=4+5)
Match/Easy1/16-8                  2.17ns ± 0%     2.15ns ± 0%   -0.90%  (p=0.000 n=5+4)
Match/Easy1/32-8                  25.1ns ± 0%     25.4ns ± 4%     ~     (p=0.651 n=5+5)
Match/Easy1/1K-8                   462ns ± 1%      431ns ± 4%   -6.56%  (p=0.008 n=5+5)
Match/Easy1/32K-8                 18.8µs ± 0%     18.8µs ± 1%     ~     (p=1.000 n=5+5)
Match/Easy1/1M-8                   658µs ± 0%      658µs ± 1%     ~     (p=0.841 n=5+5)
Match/Easy1/32M-8                 21.0ms ± 1%     21.0ms ± 2%     ~     (p=0.841 n=5+5)
Match/Medium/16-8                 2.15ns ± 0%     2.16ns ± 0%     ~     (p=0.714 n=4+5)
Match/Medium/32-8                  561ns ± 1%      512ns ± 5%   -8.69%  (p=0.008 n=5+5)
Match/Medium/1K-8                 16.9µs ± 0%     15.2µs ± 1%  -10.40%  (p=0.008 n=5+5)
Match/Medium/32K-8                 632µs ± 0%      631µs ± 1%     ~     (p=0.421 n=5+5)
Match/Medium/1M-8                 20.3ms ± 1%     20.1ms ± 0%     ~     (p=0.190 n=5+4)
Match/Medium/32M-8                 650ms ± 1%      646ms ± 0%   -0.58%  (p=0.032 n=5+4)
Match/Hard/16-8                   2.15ns ± 0%     2.15ns ± 1%     ~     (p=0.111 n=5+5)
Match/Hard/32-8                    870ns ± 2%      667ns ± 1%  -23.28%  (p=0.008 n=5+5)
Match/Hard/1K-8                   26.9µs ± 0%     21.0µs ± 2%  -21.83%  (p=0.008 n=5+5)
Match/Hard/32K-8                   833µs ± 0%      833µs ± 1%     ~     (p=0.548 n=5+5)
Match/Hard/1M-8                   26.6ms ± 0%     26.8ms ± 1%     ~     (p=0.905 n=4+5)
Match/Hard/32M-8                   856ms ± 0%      851ms ± 0%   -0.65%  (p=0.016 n=5+4)
Match/Hard1/16-8                  2.96µs ±12%     1.81µs ± 3%  -38.68%  (p=0.008 n=5+5)
Match/Hard1/32-8                  5.62µs ± 3%     3.48µs ± 0%  -38.07%  (p=0.016 n=5+4)
Match/Hard1/1K-8                   175µs ± 5%      108µs ± 0%  -37.85%  (p=0.016 n=5+4)
Match/Hard1/32K-8                 4.09ms ± 2%     4.05ms ± 0%   -0.85%  (p=0.016 n=5+4)
Match/Hard1/1M-8                   131ms ± 0%      131ms ± 3%     ~     (p=0.151 n=5+5)
Match/Hard1/32M-8                  4.19s ± 0%      4.20s ± 1%     ~     (p=1.000 n=5+5)
Match_onepass_regex/16-8           262ns ± 2%      170ns ± 2%  -35.13%  (p=0.008 n=5+5)
Match_onepass_regex/32-8           463ns ± 0%      306ns ± 0%  -33.90%  (p=0.008 n=5+5)
Match_onepass_regex/1K-8          13.3µs ± 2%      8.8µs ± 0%  -33.84%  (p=0.008 n=5+5)
Match_onepass_regex/32K-8          424µs ± 3%      280µs ± 1%  -33.93%  (p=0.008 n=5+5)
Match_onepass_regex/1M-8          13.4ms ± 0%      9.0ms ± 1%  -32.80%  (p=0.016 n=4+5)
Match_onepass_regex/32M-8          427ms ± 0%      288ms ± 1%  -32.60%  (p=0.008 n=5+5)

Change-Id: I02c54176ed5c9f5b5fc99524a2d5eb1c490f0ebf
Reviewed-on: https://go-review.googlesource.com/c/go/+/355789
Reviewed-by: Peter Weinberger <pjw@google.com>
Reviewed-by: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Auto-Submit: Russ Cox <rsc@golang.org>

2 years agodoc: update Go memory model
Russ Cox [Wed, 26 Jan 2022 21:42:05 +0000 (16:42 -0500)]
doc: update Go memory model

Following discussion on #47141, make the following changes:

- Document Go's overall approach.
- Document that multiword races can cause crashes.
- Document happens-before for runtime.SetFinalizer.
- Document (or link to) happens-before for more sync types.
- Document happens-before for sync/atomic.
- Document disallowed compiler optimizations.

See also https://research.swtch.com/gomm for background.

Fixes #50859.

Change-Id: I17d837756a77f4d8569f263489c2c45de20a8778
Reviewed-on: https://go-review.googlesource.com/c/go/+/381315
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agodoc/go1.19: remove TODO about LimitedReader
Russ Cox [Sat, 4 Jun 2022 13:44:11 +0000 (09:44 -0400)]
doc/go1.19: remove TODO about LimitedReader

Rolled back in CL 410133.

For #51115.

Change-Id: I009c557acf98a98a9e5648fa82d998d41974ae60
Reviewed-on: https://go-review.googlesource.com/c/go/+/410357
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoio: revert: add an Err field to LimitedReader
Russ Cox [Fri, 3 Jun 2022 20:00:16 +0000 (16:00 -0400)]
io: revert: add an Err field to LimitedReader

We are having a hard time deciding the exact semantics
of the Err field, and we need to ship the beta.
So revert the Err field change; it can wait for Go 1.20.

For #51115.

This reverts CL 396215.

Change-Id: I7719386567d3da10a614058a11f19dbccf304b4d
Reviewed-on: https://go-review.googlesource.com/c/go/+/410133
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>

2 years agocmd/go: index standard library packages
Michael Matloob [Fri, 29 Apr 2022 18:45:46 +0000 (14:45 -0400)]
cmd/go: index standard library packages

Change-Id: I07594303a1e9833723522d5ff94577a5510ca6f0
Reviewed-on: https://go-review.googlesource.com/c/go/+/404714
Run-TryBot: Michael Matloob <matloob@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
2 years agodoc/go1.19: add TODOs for undocumented runtime features
Michael Pratt [Thu, 2 Jun 2022 20:58:28 +0000 (16:58 -0400)]
doc/go1.19: add TODOs for undocumented runtime features

As of this CL, release notes for all packages owned by @golang/runtime
on https://dev.golang.org/owners are either complete or have explicit
TODOs.

For #51400

Change-Id: I5b6affd43883991a3b8a065b4aa211efce7427f0
Reviewed-on: https://go-review.googlesource.com/c/go/+/410118
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>

2 years agodoc/go1.19: add release notes for runtime packages
Michael Pratt [Thu, 2 Jun 2022 20:56:41 +0000 (16:56 -0400)]
doc/go1.19: add release notes for runtime packages

This documents most of the changes in runtime packages, which the major
exception of GC changes, which will be documented in a future CL.

For #51400

Change-Id: Ibcf501e1b4f7caa3397db6b9136daec07aac5a65
Reviewed-on: https://go-review.googlesource.com/c/go/+/410117
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agoruntime: track total idle time for GC CPU limiter
Michael Anthony Knyszek [Thu, 2 Jun 2022 21:26:49 +0000 (21:26 +0000)]
runtime: track total idle time for GC CPU limiter

Currently the GC CPU limiter only tracks idle GC work time. However, in
very undersubscribed situations, it's possible that all this extra idle
time prevents the enabling of the limiter, since it all gets account for
as mutator time. Fix this by tracking all idle time via pidleget and
pidleput. To support this, pidleget and pidleput also accept and return
"now" parameters like the timer code.

While we're here, let's clean up some incorrect assumptions that some of
the scheduling code makes about "now."

Fixes #52890.

Change-Id: I4a97893d2e5ad1e8c821f8773c2a1d449267c951
Reviewed-on: https://go-review.googlesource.com/c/go/+/410122
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: avoid string allocation in printDebuglog
Michael Anthony Knyszek [Thu, 2 Jun 2022 21:10:26 +0000 (21:10 +0000)]
runtime: avoid string allocation in printDebuglog

Either due to a new nowritebarrierrec annotation or a change in escape
analysis, printDebuglog can't be called from sighandler anymore.

Fix this by avoiding a string allocation that's the primary culprit.

Change-Id: Ic84873a453f45852b0443a46597ed3ab8c9443fd
Reviewed-on: https://go-review.googlesource.com/c/go/+/410121
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: only use CPU time from the current window in the GC CPU limiter
Michael Anthony Knyszek [Thu, 2 Jun 2022 19:06:27 +0000 (19:06 +0000)]
runtime: only use CPU time from the current window in the GC CPU limiter

Currently the GC CPU limiter consumes CPU time from a few pools, but
because the events that flush to those pools may overlap, rather than be
strictly contained within, the update window for the GC CPU limiter, the
limiter's accounting is ultimately sloppy.

This sloppiness complicates accounting for idle time more completely,
and makes reasoning about the transient behavior of the GC CPU limiter
much more difficult.

To remedy this, this CL adds a field to the P struct that tracks the
start time of any in-flight event the limiter might care about, along
with information about the nature of that event. This timestamp is
managed atomically so that the GC CPU limiter can come in and perform a
read of the partial CPU time consumed by a given event. The limiter also
updates the timestamp so that only what's left over is flushed by the
event itself when it completes.

The end result of this change is that, since the GC CPU limiter is aware
of all past completed events, and all in-flight events, it can much more
accurately collect the CPU time of events since the last update. There's
still the possibility for skew, but any leftover time will be captured
in the following update, and the magnitude of this leftover time is
effectively bounded by the update period of the GC CPU limiter, which is
much easier to consider.

One caveat of managing this timestamp-type combo atomically is that they
need to be packed in 64 bits. So, this CL gives up the top 3 bits of the
timestamp and places the type information there. What this means is we
effectively have only a 61-bit resolution timestamp. This is fine when
the top 3 bits are the same between calls to nanotime, but becomes a
problem on boundaries when those 3 bits change. These cases may cause
hiccups in the GC CPU limiter by not accounting for some source of CPU
time correctly, but with 61 bits of resolution this should be extremely
rare. The rate of update is on the order of milliseconds, so at worst
the runtime will be off of any given measurement by only a few
CPU-milliseconds (and this is directly bounded by the rate of update).
We're probably more inaccurate from the fact that we don't measure real
CPU time but only approximate it.

For #52890.

Change-Id: I347f30ac9e2ba6061806c21dfe0193ef2ab3bbe9
Reviewed-on: https://go-review.googlesource.com/c/go/+/410120
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/compile/internal/escape: escape values with >PtrSize alignment
Michael Anthony Knyszek [Fri, 3 Jun 2022 19:15:24 +0000 (19:15 +0000)]
cmd/compile/internal/escape: escape values with >PtrSize alignment

After CL 381317 there exist values that may have an alignment greater
than the pointer size for that platform. Specifically, atomic.{Ui|I}nt64
may be aligned to 8 bytes on a 32-bit platform. If such a value, or
a container for the value, gets stack-allocated, it's possible that it
won't be aligned correctly, because the maximum alignment we enforce on
stacks is governed by the pointer size. Changing that would be a
significant undertaking, so just escape these values to the heap
instead, where we're sure they'll actually be aligned correctly.

Change is by rsc@, I'm just shepherding it through code review.

For #50860.

Change-Id: I51669561c0a13ecb84f821020e144c58cb528418
Reviewed-on: https://go-review.googlesource.com/c/go/+/410131
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2 years agocmd/go: changes to use modindex
Michael Matloob [Tue, 5 Apr 2022 22:47:23 +0000 (18:47 -0400)]
cmd/go: changes to use modindex

This CL makes the changes to actually use the module index when loading
packages and instead of scanning their directories to see if they
contain go files or to extract imports.

Change-Id: I70106181cf64d6fd5a416644ba518b6b90030e0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/403778
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>

2 years agocmd/go: add functions to read index file
Michael Matloob [Tue, 5 Apr 2022 22:47:23 +0000 (18:47 -0400)]
cmd/go: add functions to read index file

The data read is used for three primary functions: ImportPackage,
IsDirWithGoFiles and ScanDir. Functions are also provided to get this
information from the intermediate package representation to cache
the information from reads for non-indexed packages.

Change-Id: I5eed629bb0d6ee5b88ab706d06b074475004c081
Reviewed-on: https://go-review.googlesource.com/c/go/+/403975
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agosync/atomic: note that alignment responsibility doesn't apply to types
Michael Pratt [Fri, 3 Jun 2022 17:47:10 +0000 (13:47 -0400)]
sync/atomic: note that alignment responsibility doesn't apply to types

For #50860.

Change-Id: I8e117f00c5da230d0dc398aaed417fe5e64a5b22
Reviewed-on: https://go-review.googlesource.com/c/go/+/410127
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agodoc/go1.19: add release notes for syscall
Michael Pratt [Thu, 2 Jun 2022 20:53:36 +0000 (16:53 -0400)]
doc/go1.19: add release notes for syscall

For #51400

Change-Id: If5fc131df254d47a989ff61c8e584cb8149cbd09
Reviewed-on: https://go-review.googlesource.com/c/go/+/410116
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>

2 years agodoc/go1.19: add release notes for sync/atomic
Michael Pratt [Thu, 2 Jun 2022 20:43:12 +0000 (16:43 -0400)]
doc/go1.19: add release notes for sync/atomic

For #51400

Change-Id: I32a3aedf1e8a52148a9d78c4f7dae1ea59c810b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/410115
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
2 years agocmd/link: fix TestLargeText
eric fang [Thu, 2 Jun 2022 06:54:50 +0000 (06:54 +0000)]
cmd/link: fix TestLargeText

Do not need to add single quotes '' when passing the parameter value of
 the -ldflags option, otherwise the following error will be reported:
invalid value "'-linkmode=external'" for flag -ldflags: parameter may
not start with quote character.

Change-Id: I322fa7079ac24c8a68d9cb0872b0a20dbc4893d1
Reviewed-on: https://go-review.googlesource.com/c/go/+/410074
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Benny Siegert <bsiegert@gmail.com>
Run-TryBot: Eric Fang <eric.fang@arm.com>

2 years agodoc/go1.19: add a release note for CL 404134
Bryan C. Mills [Fri, 27 May 2022 20:51:06 +0000 (16:51 -0400)]
doc/go1.19: add a release note for CL 404134

For #51473.

Change-Id: I01a35e5ebc83b8b72e414ed1730e9147ea590959
Reviewed-on: https://go-review.googlesource.com/c/go/+/409176
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
2 years agodoc/go1.19: add release note for CL 398058
Bryan C. Mills [Fri, 27 May 2022 20:39:54 +0000 (16:39 -0400)]
doc/go1.19: add release note for CL 398058

Change-Id: I483de9a6a2016432df13e030c675d42470db9ed1
Reviewed-on: https://go-review.googlesource.com/c/go/+/409175
Run-TryBot: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
2 years agoarchive/zip: use bufio.Reset rather than NewReader
Ian Lance Taylor [Thu, 2 Jun 2022 16:58:42 +0000 (09:58 -0700)]
archive/zip: use bufio.Reset rather than NewReader

A clean up for CL 408734, suggested by Joe Tsai.

Change-Id: Ida9db0b8d31785d5640938c286c9c6c82c27f457
Reviewed-on: https://go-review.googlesource.com/c/go/+/410154
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
2 years agocmd/dist: use gohostarch for ssa rewrite check
Christian Stewart [Wed, 1 Jun 2022 20:52:12 +0000 (20:52 +0000)]
cmd/dist: use gohostarch for ssa rewrite check

Fix a build failure when bootstrapping the Go compiler with go-bootstrap 1.4
while the environment contains GOARCH=riscv64.

Building Go toolchain1 using go-1.4-bootstrap-20171003.
src/cmd/compile/internal/ssa/rewriteRISCV64.go:4814
invalid operation: y << x (shift count type int64, must be unsigned integer)

This is because:

 - buildtool.go:198: calls bootstrapRewriteFile(src)
 - bootstrapRewriteFile: buildtool.go:283 calls:
 - isUnneededSSARewriteFile: checks os.Getenv("GOARCH")
 - isUnneededSSARewriteFile: returns "", false
 - bootstrapRewriteFile: calls bootstrapFixImports
 - boostrapFixImports: generates code go1.4 cannot compile

Instead of checking "GOARCH" in the environment, use the gohostarch variable.

Change-Id: Ie9c190498555c4068461fead6278a62e924062c6
GitHub-Last-Rev: 300d7a7fea0a67c696970fd271e2ce709674a658
GitHub-Pull-Request: golang/go#52362
Reviewed-on: https://go-review.googlesource.com/c/go/+/400376
Reviewed-by: Bryan Mills <bcmills@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Bryan Mills <bcmills@google.com>
Run-TryBot: Bryan Mills <bcmills@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>

2 years agoarchive/zip: if non-zero base offset fails, fall back to zero
Ian Lance Taylor [Wed, 25 May 2022 21:12:29 +0000 (14:12 -0700)]
archive/zip: if non-zero base offset fails, fall back to zero

This permits us to read files that earlier Go releases could read.
It is also compatible with other zip programs.

Change-Id: I7e2999f1073c4db5ba3f51f92681e0b149d55b3e
Reviewed-on: https://go-review.googlesource.com/c/go/+/408734
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
2 years agoencoding/json: mention SyntaxError in Unmarshal doc comment
Lukas Joisten [Tue, 8 Mar 2022 11:21:00 +0000 (12:21 +0100)]
encoding/json: mention SyntaxError in Unmarshal doc comment

Change-Id: I71c9d9ef9d21a7ae9466d8c7b283fdfbba01f5a8
Reviewed-on: https://go-review.googlesource.com/c/go/+/390734
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/go: add index creation methods
Michael Matloob [Tue, 5 Apr 2022 22:47:23 +0000 (18:47 -0400)]
cmd/go: add index creation methods

This change functions to scan modules and packages into an intermediate
RawPackage struct and also functions to write them out to and index.

Change-Id: Ia1a3b58b992e9be52c5d1397e85c642f902011cb
Reviewed-on: https://go-review.googlesource.com/c/go/+/398415
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Bryan Mills <bcmills@google.com>
2 years agocmd/go: set GIT_TRACE_CURL for tests on builders
Bryan C. Mills [Tue, 31 May 2022 21:03:29 +0000 (17:03 -0400)]
cmd/go: set GIT_TRACE_CURL for tests on builders

We have noticed a pattern of connection timeouts connecting to
github.com on the builders. Adding tracing may shed some light on the
underlying cause.

For #52545.

Change-Id: Ic73b494be9a3ea8ce3c7631ee8f62bd3d0d8e7bd
Reviewed-on: https://go-review.googlesource.com/c/go/+/409575
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agocmd/compile: fix wrong unsafe.Offsetof evaluation inside generic function
Cuong Manh Le [Mon, 30 May 2022 14:34:01 +0000 (21:34 +0700)]
cmd/compile: fix wrong unsafe.Offsetof evaluation inside generic function

For instantiated generic functions, all implicit dot operations are
resolved. Thus unsafe.Offsetof may calculating the offset against the
wrong base selector.

To fix it, we must remove any implicit dot operations to find the first
non-implicit one, which is the right base selector for calculating the
offset.

Fixes #53137

Change-Id: I38504067ce0f274615b306edc8f7d7933bdb631a
Reviewed-on: https://go-review.googlesource.com/c/go/+/409355
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
2 years agocmd/link: call syscall.FlushFileBuffers on outbuf Unmap
Than McIntosh [Tue, 17 May 2022 12:17:38 +0000 (08:17 -0400)]
cmd/link: call syscall.FlushFileBuffers on outbuf Unmap

In the windows version of OutBuf.munmap, call syscall.FlushFileBuffers
after the call to syscall.FlushViewOfFile, on the theory that this
will help flush all associated meta-data for the file the linker is
writing.

Updates #44817.

Change-Id: Ibff7d05008a91eeed7634d2760153851e15e1c18
Reviewed-on: https://go-review.googlesource.com/c/go/+/406814
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
2 years agolib/time, time/tzdata: update to 2022a
Dmitri Shuralyov [Mon, 30 May 2022 17:20:51 +0000 (13:20 -0400)]
lib/time, time/tzdata: update to 2022a

Version 2022a was released on 2022-03-15 and we can start using it for
Go 1.19. Its release announcement was:

https://mm.icann.org/pipermail/tz-announce/2022-March/000070.html

For #22487.

Change-Id: Ie89b90927a251413e4f67c9e0ed3a67d4161529b
Reviewed-on: https://go-review.googlesource.com/c/go/+/409374
Auto-Submit: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Benny Siegert <bsiegert@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Run-TryBot: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
2 years agonet: permit use of Resolver.PreferGo, netgo on Windows and Plan 9
Brad Fitzpatrick [Sat, 28 May 2022 21:06:43 +0000 (14:06 -0700)]
net: permit use of Resolver.PreferGo, netgo on Windows and Plan 9

This reverts commit CL 401754 (440c9312c8) which reverted CL 400654,
thus reapplying CL 400654, re-adding the func init() { netGo = true }
to cgo_stub.go CL 400654 had originally removed (mistakenly during
development?) that had broken the darwin nocgo builder.

Fixes #33097

Change-Id: I90f59746d2ceb6b5d2bd832c9fc90068f8ff7417
Reviewed-on: https://go-review.googlesource.com/c/go/+/409234
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Keith Randall <khr@google.com>
2 years agodoc: add release note for reflect.Value.{Bytes,Len,Cap}
Keith Randall [Thu, 26 May 2022 16:52:44 +0000 (09:52 -0700)]
doc: add release note for reflect.Value.{Bytes,Len,Cap}

Update #47066
Update #52411

Change-Id: I85139d774c16c9e6d1a2592a5abba58a49338674
Reviewed-on: https://go-review.googlesource.com/c/go/+/408874
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
2 years agomime: ignore .js => text/plain in Windows registry
Ian Lance Taylor [Tue, 17 May 2022 19:28:28 +0000 (12:28 -0700)]
mime: ignore .js => text/plain in Windows registry

This seems to be a common registry misconfiguration on Windows.

Fixes #32350

Change-Id: I68c617c42a6e72948e2acdf335ff8e7df569432d
Reviewed-on: https://go-review.googlesource.com/c/go/+/406894
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
2 years agodoc/go1.19: mention runtime/trace's CPU samples
Rhys Hiltner [Fri, 27 May 2022 00:41:07 +0000 (17:41 -0700)]
doc/go1.19: mention runtime/trace's CPU samples

Change-Id: Ia8becc3e28523e9b1da4f7a274c800309add331a
Reviewed-on: https://go-review.googlesource.com/c/go/+/408994
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Rhys Hiltner <rhys@justin.tv>

2 years agotest/heapsampling.go: slow down allocation rate and reduce iterations
Michael Anthony Knyszek [Fri, 27 May 2022 18:14:58 +0000 (18:14 +0000)]
test/heapsampling.go: slow down allocation rate and reduce iterations

As far as I can tell, this test suffers from #52433. For some reason,
this seems to become more of a problem on the windows/386 than anywhere
else. This CL is an attempt at a mitigation by slowing down the
allocation rate by inserting runtime.Gosched call in the inner loop. It
also cuts the iteration count which should help too (as less memory is
allocated in total), but the main motivation is to make sure the test
doesn't take too long to run.

Fixes #49564.

Change-Id: I8cc622b06a69cdfa66f680a30e1ccf334eea2164
Reviewed-on: https://go-review.googlesource.com/c/go/+/408825
Reviewed-by: Bryan Mills <bcmills@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agoruntime: cancel mark and scavenge assists if the limiter is enabled
Michael Anthony Knyszek [Wed, 25 May 2022 22:51:21 +0000 (22:51 +0000)]
runtime: cancel mark and scavenge assists if the limiter is enabled

This change forces mark and scavenge assists to be cancelled early if
the limiter is enabled. This avoids goroutines getting stuck in really
long assists if the limiter happens to be disabled when they first come
into the assist. This can get especially bad for mark assists, which, in
dire situations, can end up "owing" the GC a really significant debt.

For #52890.

Change-Id: I4bfaa76b8de3e167d49d2ffd8bc2127b87ea566a
Reviewed-on: https://go-review.googlesource.com/c/go/+/408816
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

2 years agomisc/cgo/testsanitizers: buffer the signal channel in TestTSAN/tsan11
Bryan C. Mills [Fri, 27 May 2022 15:42:59 +0000 (11:42 -0400)]
misc/cgo/testsanitizers: buffer the signal channel in TestTSAN/tsan11

This fix is analogous to the one in CL 407888.

'go vet' catches the error, but it is not run on this file because the
file is (only) compiled when running testsanitizers/TestTSAN.

Fixes #53113.

Change-Id: I74f7b7390a9775ff00a06214c1019ba28846dd11
Reviewed-on: https://go-review.googlesource.com/c/go/+/409094
Auto-Submit: Bryan Mills <bcmills@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Bryan Mills <bcmills@google.com>

2 years agodoc: fix up race detector release note
Keith Randall [Thu, 26 May 2022 16:58:18 +0000 (09:58 -0700)]
doc: fix up race detector release note

It's already mentioned under runtime/race, we don't need it
also mentioned under runtime.

Change-Id: I538322b32d75b9642f3ead03539fccb95988ef8c
Reviewed-on: https://go-review.googlesource.com/c/go/+/408875
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Rhys Hiltner <rhys@justin.tv>
2 years agodoc: lint Markdown for trailing spaces and code blocks
John Bampton [Tue, 17 May 2022 22:33:02 +0000 (22:33 +0000)]
doc: lint Markdown for trailing spaces and code blocks

MD009 Trailing spaces.
MD031 Fenced code blocks should be surrounded by blank lines.

https://github.com/markdownlint/markdownlint/blob/master/docs/RULES.md

Change-Id: I6a2fff4afa8224442d90b30f84444b4a888fcef2
GitHub-Last-Rev: 67cf146b8501096949faf0155239770d4f6a73b4
GitHub-Pull-Request: golang/go#44285
Reviewed-on: https://go-review.googlesource.com/c/go/+/292409
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Alex Rakoczy <alex@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Alex Rakoczy <alex@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>