runtime: factor out GC pause time CPU stats update

Currently this is done manually in two places. Replace these manual
updates with a method that also forces the caller to be mindful that the
number will be multiplied (and that it needs to be). This will make
follow-up changes simpler too.

Change-Id: I81ea844b47a40ff3470d23214b4b2fb5b71a4abe
Reviewed-on: https://go-review.googlesource.com/c/go/+/570255
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/link: close the file opened in the captureHostObj function

Change-Id: I940c63c3040035e507428798d50060cfd0d04e16
GitHub-Last-Rev: e27484cd43e21599ba6e798ab74a703ad290ae7c
GitHub-Pull-Request: golang/go#66706
Reviewed-on: https://go-review.googlesource.com/c/go/+/577055
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>

cmd/internal/obj/riscv: check immediate for rotation instructions

Ensure that the immediate for a RORI or RORIW instruction are within range,
adding test coverage. Also use a consistent "immediate out of range" error
for both rotations and shifts.

Change-Id: Id687d7c6e028786f607e9519bbb64dab62b6cf3d
Reviewed-on: https://go-review.googlesource.com/c/go/+/572735
Reviewed-by: M Zhuo <mengzhuo1203@gmail.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

all: update vendored golang.org/x/net

Pull in CL 576756:

b67a0f05 http2: send correct LastStreamID in stream-caused GOAWAY

It comes with newer x/crypto and x/sys requirements via CL 576516.

For #65051.
Fixes #66668.

Change-Id: Ib6df49b779ff5ea56966ae3de443e87ac7f4ec4f
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/576679
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

all: update vendored golang.org/x/net to v0.23.0

Pull in CL 576155:

ba872109 http2: close connections when receiving too many headers

Fixes CVE-2023-45288.
Fixes #65051.

Change-Id: I3f4e3d565189b4ed552935fe1d96f94ebac86d48
Reviewed-on: https://go-review.googlesource.com/c/go/+/576295
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@google.com>

go/types: fix bug in premature Unalias of alias cycle

Unalias memoizes the result of removing Alias constructors.
When Unalias is called too soon on a type in a cycle,
the initial value of the alias, Invalid, gets latched by
the memoization, causing it to appear Invalid forever.

This change disables memoization of Invalid, and adds
a regression test.

Fixes #66704
Updates #65294

Change-Id: I479fe14c88c802504a69f177869f091656489cd4
Reviewed-on: https://go-review.googlesource.com/c/go/+/576975
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Alan Donovan <adonovan@google.com>

runtime: push down systemstack requirement for tracer where possible

Currently lots of functions require systemstack because the trace buffer
might get flushed, but that will already switch to the systemstack for
the most critical bits (grabbing trace.lock). That means a lot of this
code is non-preemptible when it doesn't need to be. We've seen this
cause problems at scale, when dumping very large numbers of stacks at
once, for example.

Change-Id: I88340091a3c43f0513b5601ef5199c946aa56ed7
Reviewed-on: https://go-review.googlesource.com/c/go/+/572095
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>

runtime: take a stack trace during tracing only when we own the stack

Currently, the execution tracer may attempt to take a stack trace of a
goroutine whose stack it does not own. For example, if the goroutine is
in _Grunnable or _Gwaiting. This is easily fixed in all cases by simply
moving the emission of GoStop and GoBlock events to before the
casgstatus happens. The goroutine status is what is used to signal stack
ownership, and the GC may shrink a goroutine's stack if it can acquire
the scan bit.

Although this is easily fixed, the interaction here is very subtle,
because stack ownership is only implicit in the goroutine's scan status.
To make this invariant more maintainable and less error-prone in the
future, this change adds a GODEBUG setting that checks, at the point of
taking a stack trace, whether the caller owns the goroutine. This check
is not quite perfect because there's no way for the stack tracing code
to know that the _Gscan bit was acquired by the caller, so for
simplicity it assumes that it was the caller that acquired the scan bit.
In all other cases however, we can check for ownership precisely. At the
very least, this check is sufficient to catch the issue this change is
fixing.

To make sure this debug check doesn't bitrot, it's always enabled during
trace testing. This new mode has actually caught a few other issues
already, so this change fixes them.

One issue that this debug mode caught was that it's not safe to take a
stack trace of a _Gwaiting goroutine that's being unparked.

Another much bigger issue this debug mode caught was the fact that the
execution tracer could try to take a stack trace of a G that was in
_Gwaiting solely to avoid a deadlock in the GC. The execution tracer
already has a partial list of these cases since they're modeled as the
goroutine just executing as normal in the tracer, but this change takes
the list and makes it more formal. In this specific case, we now prevent
the GC from shrinking the stacks of goroutines in this state if tracing
is enabled. The stack traces from these scenarios are too useful to
discard, but there is indeed a race here between the tracer and any
attempt to shrink the stack by the GC.

Change-Id: I019850dabc8cede202fd6dcc0a4b1f16764209fb
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest,gotip-linux-amd64-longtest-race
Reviewed-on: https://go-review.googlesource.com/c/go/+/573155
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>

runtime: emit trace stacks for more goroutines in each generation

This change adds a new event, GoStatusStack, which is like GoStatus but
also carries a stack ID. The purpose of this event is to emit stacks in
more places, in particular for goroutines that may never emit a
stack-bearing event in a whole generation.

This CL targets one specific case: goroutines that were blocked or in a
syscall the entire generation. This particular case is handled at the
point that we scribble down the goroutine's status before the generation
transition. That way, when we're finishing up the generation and
emitting events for any goroutines we scribbled down, we have an
accurate stack for those goroutines ready to go, and we emit a
GoStatusStack instead of a GoStatus event. There's a small drawback with
the way we scribble down the stack though: we immediately register it in
the stack table instead of tracking the PCs. This means that if a
goroutine does run and emit a trace event in between when we scribbled
down its stack and the end of the generation, we will have recorded a
stack that never actually gets referenced in the trace. This case should
be rare.

There are two remaining cases where we could emit stacks for goroutines
but we don't.

One is goroutines that get unblocked but either never run, or run and
never block within a generation. We could take a stack trace at the
point of unblocking the goroutine, if we're emitting a GoStatus event
for it, but unfortunately we don't own the stack at that point. We could
obtain ownership by grabbing its _Gscan bit, but that seems a little
risky, since we could hold up the goroutine emitting the event for a
while. Something to consider for the future.

The other remaining case is a goroutine that was runnable when tracing
started and began running, but then ran until the end of the generation
without getting preempted or blocking. The main issue here is that
although the goroutine will have a GoStatus event, it'll only have a
GoStart event for it which doesn't emit a stack trace. This case is
rare, but still certainly possible. I believe the only way to resolve it
is to emit a GoStatusStack event instead of a GoStatus event for a
goroutine that we're emitting GoStart for. This case is a bit easier
than the last one because at the point of emitting GoStart, we have
ownership of the goroutine's stack.

We may consider dealing with these in the future, but for now, this CL
captures a fairly large class of goroutines, so is worth it on its own.

Fixes #65634.

Change-Id: Ief3b6df5848b426e7ee6794e98dc7ef5f37ab2d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/567076
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go: add tool names to go tool subcommand counters

When go tool <toolname> is run, increment the
go/subcommand:tool-<toolname> counter instead of just
go/subcommand:tool. go/subcommand:tool will be incremented
if go tool is run without a tool argument to list the
available tools.

Change-Id: I22b888fada1441389315a79f417c72b3f74070f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/571802
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>

testing/fstest: return structured errors in TestFS

TestFS now returns a structured error built with errors.Join to allow to
inspect errors using errors.Is and errors.As.

All errors are now wrapped using fmt.Errorf and %w.

Fixes #63675.

Change-Id: I8fc3363f8ae70085af4afdb84c16be9ca70d7731
Reviewed-on: https://go-review.googlesource.com/c/go/+/537015
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd/compile: export/import materialized aliases

This CL changes the compiler's type import/export logic
to create and preserve materialized Alias types
when GODEBUG=gotypesaliases=1.

In conjunction with CL 574717, it allows the x/tools
tests to pass with GODEBUG=gotypesaliases=1.

Updates #65294
Updates #64581
Fixes #66550

Change-Id: I70b9279f4e0ae7a1f95ad153c4e6909a878915a4
Reviewed-on: https://go-review.googlesource.com/c/go/+/574737
Auto-Submit: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

encoding/csv: port the go-fuzz function to native fuzzing

Beyond the required file move and refactor to use the testing package,
a number of changes were made to get the fuzzing working properly.

First, add more logs to see what is going on.

Second, some option combinations set Comma to the null character,
which simply never worked at all. I suspect the author meant to leave
the comma character as the default instead.
This was spotted thanks to the added logging.

Third, the round-trip DeepEqual check did not work at all
when any comments were involved, as the writer does not support them.

Fourth and last, massage the first and second parsed records before
comparing them with DeepEqual, as the nature of Reader and Writer
causes empty quoted records and CRLF sequences to change.

With all the changes above, the fuzzing function appears to work
normally on my laptop now. I fuzzed for a solid five minutes and
could no longer encounter any errors or panics.

Change-Id: Ie27f65f66099bdaa076343cee18b480803d2e4d3
Reviewed-on: https://go-review.googlesource.com/c/go/+/576375
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Ian Lance Taylor <iant@google.com>

strings: intrinsify and optimize Compare

slices.SortFunc requires a three-way comparison and we need an
efficient strings.Compare to perform three-way string comparisons.
This new implementation adds bytealg.CompareString as a wrapper of
runtime_cmpstring and changes Compare to use bytealg.CompareString.

The new implementation of Compare with runtime_cmpstring is about
28% faster than the previous one.

Fixes #61725

                           │ /tmp/gobench-sort-cmp.txt │    /tmp/gobench-sort-strings.txt    │
                           │          sec/op           │   sec/op     vs base                │
SortFuncStruct/Size16-48                   918.8n ± 1%   726.6n ± 0%  -20.92% (p=0.000 n=10)
SortFuncStruct/Size32-48                   2.666µ ± 1%   2.003µ ± 1%  -24.85% (p=0.000 n=10)
SortFuncStruct/Size64-48                   1.934µ ± 1%   1.331µ ± 1%  -31.22% (p=0.000 n=10)
SortFuncStruct/Size128-48                  3.560µ ± 1%   2.423µ ± 0%  -31.94% (p=0.000 n=10)
SortFuncStruct/Size512-48                 13.019µ ± 0%   9.071µ ± 0%  -30.33% (p=0.000 n=10)
SortFuncStruct/Size1024-48                 25.61µ ± 0%   17.75µ ± 0%  -30.70% (p=0.000 n=10)
geomean                                    4.217µ        3.018µ       -28.44%

Change-Id: I2513b6f8c1b9b273ef2d23f0a86f691e2d097eb6
Reviewed-on: https://go-review.googlesource.com/c/go/+/532195
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
Reviewed-by: Keith Randall <khr@google.com>

net: update the doc for TCPConn.SetKeepAlivePeriod on Windows

The method comment of TCPConn.SetKeepAlivePeriod had become
obsolete and inaccurate since CL 565495 and CL 570077 were merged.

For #65817

Change-Id: Ide99b2949676d452a505ba6fd634088f05c9df44
Reviewed-on: https://go-review.googlesource.com/c/go/+/576435
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>

internal/poll: eliminate the redundant conditional branch for isKernelVersionGE53

Follow up CL 573755

Change-Id: I27c7571d3ef1274cf2c6892e678f946f9b65de33
Reviewed-on: https://go-review.googlesource.com/c/go/+/576416
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>

cmd/internal/obj/ppc64: on Power10, use xxspltidp for float constants

Any normal float32 constant can be generated by this instruction;
use xxspltidp when possible. This prefixed instruction is much
faster than the two instruction load sequence from the
float32/float64 constant pool.

Change-Id: Id751d9ffdae71463adbde66427b986f0b2ef74c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/575555
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Paul Murphy <murp@ibm.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>

expvar: add the missing deletion step for keys

In CL575777 I forgot to remove the key from the "sync.Map".
This did not cause the test to fail due to the lack of an associated
testcase. Now delete the key correctly and add the testcase.

Change-Id: I26f770966a828caa02f1766675756b67894dc195
GitHub-Last-Rev: a351ce095b5256fd1351d71f09ba38fdcdfce3d9
GitHub-Pull-Request: golang/go#66675
Reviewed-on: https://go-review.googlesource.com/c/go/+/576395
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>

all: fix some comments

Change-Id: I0ee85161846c13d938213ef04d3a34f690a93e48
Reviewed-on: https://go-review.googlesource.com/c/go/+/553435
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>

archive/zip: fix a broken URL in a comment in writeDataDescriptor

The comment contains a dead link to a OpenJDK bug ID 7073588
this change fixes the link.

Change-Id: Ib9b10362c707507e59bb6f340d52a0025f65e292
GitHub-Last-Rev: 37af15b947a59346e23191dad264efa0aaa461af
GitHub-Pull-Request: golang/go#66669
Reviewed-on: https://go-review.googlesource.com/c/go/+/576335
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

time: add available godoc link

Change-Id: Idfe9cf2f2e4750d6673455f98deef2efc2d292a6
Reviewed-on: https://go-review.googlesource.com/c/go/+/539837
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>

strconv: add available godoc link

Change-Id: Iad58155f29a101fb72768b170c0c2c9d76b6041a
Reviewed-on: https://go-review.googlesource.com/c/go/+/539357
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
Run-TryBot: shuang cui <imcusg@gmail.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

go/types: don't fail fast on Go version errors

Many tools (especially in the IDE) rely on type information
being computed even for packages that have some type errors.
Previously, there were two early (error) exits in checkFiles
that violated this invariant, one related to FakeImportC
and one related to a too-new Config.GoVersion.
(The FakeImportC one is rarely encountered in practice,
but the GoVersion one, which was recently downgraded from
a panic by CL 507975, was a source of crashes
due to incomplete type information.)

This change moves both of those errors out of checkFiles
so that they report localized errors and don't obstruct
type checking. A test exercises the errors, and that
type annotations are produced.

Also, we restructure and document checkFiles to make clear
that it is never supposed to stop early.

Updates #66525

Change-Id: I9c6210e30bbf619f32a21157f17864b09cfb5cf2
Reviewed-on: https://go-review.googlesource.com/c/go/+/574495
Reviewed-by: Tim King <taking@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

errors: return early for Is(nil, ...)

If err is nil it wouldn't match any given target error except for nil,
so we can return early to speed up cases where Is is used without a
preceding err != nil check.

Change-Id: Ib33cff50453fe070f06871ce8074694c81ab787b
Reviewed-on: https://go-review.googlesource.com/c/go/+/576015
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

net/http: speed up go test

Optimize the execution speed of go test ./net/http from ~38s to ~28s.
This is achieved by shortening the sleep interval utilized for
identifying goroutine leaks.

This optimization is motivated by noticing significant periods of
inactivity in the -trace output. Even after applying this CL, many
Off-CPU wait periods seem to remain:

$ go test ./net/http
ok      net/http        27.744s

real    0m28.204s
user    0m4.991s
sys     0m1.797s

Change-Id: I6108ebbb715c33900f1506d810c0a8f8ed674d35
Reviewed-on: https://go-review.googlesource.com/c/go/+/575975
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Damien Neil <dneil@google.com>

slices: update doc for Repeat

To use (len(x) * count) consistently.

Change-Id: I45fa4383de7e05846152a8090ea5c04bc72a1727
Reviewed-on: https://go-review.googlesource.com/c/go/+/575955
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

reflect: add missing String case in Value.UnsafePointer doc

CL 516596 changes Value.UnsafePointer to handle String case. However,
the method's doc comment is not updated to reflect this change.

Updates #61308

Change-Id: I84e02fd969ae0244184e1a2f05cac4651cdf7bff
Reviewed-on: https://go-review.googlesource.com/c/go/+/575956
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

os: convert poll.ErrFileClosed to ErrClosed for Stat

Fixes #66665

Change-Id: I3e3b7433d245daa997d7d502c2ef8978af6664fb
Reviewed-on: https://go-review.googlesource.com/c/go/+/576119
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>

expvar: use slices to simplify the code

No effect on benchmarks.

Change-Id: I7454c21b25d5e44b9c4a39922ed470522f81872d
GitHub-Last-Rev: 5801b30dac33712c2e3d282b2d3a9fbe04779b2e
GitHub-Pull-Request: golang/go#66660
Reviewed-on: https://go-review.googlesource.com/c/go/+/575777
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

go/types: ObjectString: remove only 1 Alias for "type A = RHS"

As we migrate towards materialized Alias types, the ObjectString
for a type A such as
   type A = B
   type B = int
should be "type A = B", removing exactly one Alias constructor
from the type of A. (The previous behavior was "type A = int".)

I suspect the existing Alias.{Unalias,Underlying} API is
inadequate and that we will need an Alias.RHS accessor that
removes exactly one Alias. Other clients such as the import/
export packages will need it, because aliases are not
isomorphic to defined types, in which, given
   type A B
   type B int
the Underlying of A is indeed int. See #66559.

Change-Id: I11a4aacbe6dbeeafc3aee31b3c096296b5970cd8
Reviewed-on: https://go-review.googlesource.com/c/go/+/574716
Auto-Submit: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>

go/types: include link to source in generated files

This makes navigation a little easier.
Also, make the "inconsistent generated files" test's
error message more explicit about which file is which.

Change-Id: Iae1fb9e496fb68325448d807469a5b0c55ceda99
Reviewed-on: https://go-review.googlesource.com/c/go/+/574715
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Alan Donovan <adonovan@google.com>

slices: simplify slice expression for Replace

A slice expression of the form:
s[a:len(s)]
will be simplified to:
s[a:]
This is one of the simplifications that "gofmt -s" applies.

See https://pkg.go.dev/golang.org/x/tools/gopls/internal/analysis/simplifyslice

Change-Id: Ib3c01ecf24b84333fd1993a343450fc57fb8ac84
Reviewed-on: https://go-review.googlesource.com/c/go/+/575335
Reviewed-by: Ian Lance Taylor <iant@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Valentin Deleplace <deleplace@google.com>

fmt: add available godoc link

Change-Id: Ia7eaa654b44625983d09284d906a7b67ef589696
Reviewed-on: https://go-review.googlesource.com/c/go/+/535082
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

text/template: more godoc links

Change-Id: Ib8ea569b3cd92bbc9fa8a54f7da61207ba978bbe
Reviewed-on: https://go-review.googlesource.com/c/go/+/574796
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

all: add reflect.SliceAt function

Fixes #61308

Change-Id: Ic17d737fda055a60779985d5da497745c80d5cfa
Reviewed-on: https://go-review.googlesource.com/c/go/+/516597
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>

reflect: handle String kind in Value.{Pointer,UnsafePointer}

Updates #61308

Change-Id: I92d459383c520d137787ce5c8f135d205af74e5d
Reviewed-on: https://go-review.googlesource.com/c/go/+/516596
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

internal/poll, net, os: remove poll.Splice syscall name return value

The sc return value of internal/poll.Splice is always set to the same
value "splice" in the error case and then passed to wrapSyscallError.
Move that value to the wrapSyscallError calls to simplify the code a
bit.

Change-Id: I98104d755da68ff9f301fabc43c2618fda21a175
Reviewed-on: https://go-review.googlesource.com/c/go/+/575655
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>

cmd/go/internal/modload: close modules.txt

Change-Id: Ieb4cf76f4d7050d2f3052c2f90e47901de74d641
GitHub-Last-Rev: 1e8dcfd6517201c3a467109ae133846cc66e053a
GitHub-Pull-Request: golang/go#66629
Reviewed-on: https://go-review.googlesource.com/c/go/+/575475
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>

cmd/compile: add missing OASOP case in mayModifyPkgVar

CL 395541 made staticopy safe, stop applying the optimization once
seeing an expression that may modify global variables. However, it
misses the case for OASOP expression, causing the static init
mis-recognizes the modification and think it's safe.

Fixing this by adding missing OASOP case.

Fixes #66585

Change-Id: I603cec018d3b5a09825c14e1f066a0e16f8bde23
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/575216
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile: check ODEREF for safe lhs in assignment during static init

For #66585

Change-Id: Iddc407e3ef4c3b6ecf5173963b66b3e65e43c92d
Reviewed-on: https://go-review.googlesource.com/c/go/+/575336
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

internal/abi: clean up type of Kind and Type.Kind_

Currently, Type.Kind_ is a uint8, Kind is a uint, and some of the
abi.Kind consts are not of type Kind. Clean this all up by making Kind
a uint8, then making Type.Kind a Kind, and finally making all Kind
consts actually have type Kind. This has some ripple effect, but I
think all of the changes are improvements.

Change-Id: If39be74699c2cdb52bf0ad7092d392bc8fb68d15
Reviewed-on: https://go-review.googlesource.com/c/go/+/575579
Auto-Submit: Austin Clements <austin@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile/internal/ssa: fix typos in comment and log

Change-Id: Ic872bac2989ea1c83f31456eb334e6d756ebd7d1
GitHub-Last-Rev: d4098849791d510b9a1f21c3fbce395cd75a04c3
GitHub-Pull-Request: golang/go#66612
Reviewed-on: https://go-review.googlesource.com/c/go/+/575296
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>

reflect: add available godoc link

Change-Id: Ib199ce1a781e8e3a66d3dc8bda617e6bc30b290e
Reviewed-on: https://go-review.googlesource.com/c/go/+/539578
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: shuang cui <imcusg@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>

all: use kind* of abi

For #59670

Change-Id: Id66e102f13e529dd041b68ce869026a56f0a1b9b
GitHub-Last-Rev: 43aa9376f72bc02a9d86518cdc99494a6b2f8573
GitHub-Pull-Request: golang/go#65564
Reviewed-on: https://go-review.googlesource.com/c/go/+/562298
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Austin Clements <austin@google.com>

os: define wrapSyscallError only on linux

wrapSyscallError by now is only used on linux in the methods defined in
os/zero_copy_linux.go. Move the definition there.

Change-Id: I0ca5749adaac44e8d095b8452458b647f595d3c8
Reviewed-on: https://go-review.googlesource.com/c/go/+/575595
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>

encoding/json: prevent duplicate slicebytetostring

When storing literal to JSON number v, if s is valid number, the slicebytetostring operation will be performed twice. In fact, the operation is unavoidable on any code path, so just perform it at the very beginning.

This is not a big optimization, but better than nothing:

    $ ../bin/go test ./encoding/json/ -bench UnmarshalNumber -run NOTEST -benchtime 10000000x -count 16  > old.txt
    $ ../bin/go test ./encoding/json/ -bench UnmarshalNumber -run NOTEST -benchtime 10000000x -count 16  > new.txt
    $ benchstat old.txt new.txt
                      │   old.txt   │              new.txt               │
                      │   sec/op    │   sec/op     vs base               │
    UnmarshalNumber-8   234.5n ± 3%   228.2n ± 4%  -2.67% (p=0.033 n=16)

                      │  old.txt   │            new.txt             │
                      │    B/op    │    B/op     vs base            │
    UnmarshalNumber-8   168.0 ± 0%   168.0 ± 0%  ~ (p=1.000 n=16) ¹
    ¹ all samples are equal

                      │  old.txt   │            new.txt             │
                      │ allocs/op  │ allocs/op   vs base            │
    UnmarshalNumber-8   2.000 ± 0%   2.000 ± 0%  ~ (p=1.000 n=16) ¹
    ¹ all samples are equal

Change-Id: I1dfdb1ed0883e385f753b2046b7f047c792aa4e3
GitHub-Last-Rev: d236dd7265f110dbb6e0b9b0a824aab9ba7c36be
GitHub-Pull-Request: golang/go#61242
Reviewed-on: https://go-review.googlesource.com/c/go/+/508556
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Joseph Tsai <joetsai@digital-static.net>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

regexp/syntax: cleanup code generation in perl_groups.go

Cleanup code generation of perl_groups.go:
* Fix the generated code header to follow the standard https://go.dev/s/generatedcode
* Apply gofmt as last step of code generation
* Add //go:generate lines in parse.go to trigger code generation
* Adapt make_perl_groups.pl to handle writing directly to the output
  file (as we can't use shell redirection in go:generate lines)
* use strict; use warnings;

Change-Id: I675241da03dd3f6facc9eb9de00999bd9203ad70
Reviewed-on: https://go-review.googlesource.com/c/go/+/555995
Reviewed-by: Bryan Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

internal/zstd: optimize window reset using make for buffer allocation

Optimize the zstd window's reset function to use `make` for buffer
allocation when needed. This ensures efficient memory usage by
conditionally reallocating the buffer based on its capacity, improving
performance in scenarios where the existing buffer is insufficient.

```
goos: darwin
goarch: amd64
pkg: internal/zstd
cpu: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
         │   old.out   │            new.out            │
         │   sec/op    │   sec/op     vs base          │
Large-12   7.299m ± 5%   7.235m ± 1%  ~ (p=0.353 n=10)

         │   old.out    │            new.out             │
         │     B/s      │     B/s       vs base          │
Large-12   35.56Mi ± 4%   35.87Mi ± 1%  ~ (p=0.353 n=10)

         │   old.out    │               new.out                │
         │     B/op     │     B/op      vs base                │
Large-12   68.86Ki ± 4%   15.87Ki ± 1%  -76.95% (p=0.000 n=10)

         │  old.out   │            new.out             │
         │ allocs/op  │ allocs/op   vs base            │
Large-12   0.000 ± 0%   0.000 ± 0%  ~ (p=1.000 n=10) ¹
¹ all samples are equal
```

Change-Id: Icd9c278b42ea149107a9861d006f40129c59831b
GitHub-Last-Rev: c00344cca415342eb563c62e325f1c2f7f28327d
GitHub-Pull-Request: golang/go#64221
Reviewed-on: https://go-review.googlesource.com/c/go/+/543375
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd: update vendored x/tools to 904c6ba

This fixes one of the many obstacles to enabling
types.Alias by updating vet.

The 'loopclosure' checker (formerly 'rangeloop') no longer
reports any findings with go1.22, so the test needed work
to ensure that it still runs on files with go1.21 semantics.

Updates #65294

Change-Id: I987ff529d4e165b56b7241563c6003e63bf92bb1
Reviewed-on: https://go-review.googlesource.com/c/go/+/575315
Auto-Submit: Alan Donovan <adonovan@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

database/sql: refactor unused variables to use blank identifiers

Hi
I've replaced unused variables in database/sql with blank identifiers to improve code readability.

This change has no impact on the functionality of the code, but makes it easier to read and understand.

Change-Id: I701e93fd9bf86725f411085bf2e8f8e6b235af14
GitHub-Last-Rev: eb40bd5e2ef9e28871fae2e97b4433e8d6e2c652
GitHub-Pull-Request: golang/go#58986
Reviewed-on: https://go-review.googlesource.com/c/go/+/475675
Commit-Queue: Ian Lance Taylor <iant@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

cmd: update and vendor golang.org/x/telemetry@3640ba5

Commands run:
go get golang.org/x/telemetry@3640ba5
go mod vendor
go mod tidy

This pulls in golang.org/cl/574815, fixing #66344

Fixes #66344

Change-Id: Ib5a0c4fc1e0bd44dbc13453ec1dade89a8d96f70
Reviewed-on: https://go-review.googlesource.com/c/go/+/575575
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>

cmd/compile: support float min/max instructions on PPC64

This enables efficient use of the builtin min/max function
for float64 and float32 types on GOPPC64 >= power9.

Extend the assembler to support xsminjdp/xsmaxjdp and use
them to implement float min/max.

Simplify the VSX xx3 opcode rules to allow FPR arguments,
if all arguments are an FPR.

Change-Id: I15882a4ce5dc46eba71d683cf1d184dc4236a328
Reviewed-on: https://go-review.googlesource.com/c/go/+/574535
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Paul Murphy <murp@ibm.com>
Reviewed-by: Than McIntosh <thanm@google.com>

cmd/compile: close files in the startProfile function

Change-Id: I803d625249890c88fbf35394a85e0b2952123620
GitHub-Last-Rev: 239b28e85741db29d852975b0d6350f8a5d68193
GitHub-Pull-Request: golang/go#66532
Reviewed-on: https://go-review.googlesource.com/c/go/+/574336
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

os: support relative paths in fixLongPath

(This CL takes the tests and some ideas from the abandoned CL 263538).

fixLongPath is used on Windows to process all path names
before syscalls to switch them to extended-length format
(with prefix \\?\) to workaround a historical limit
of 260-ish characters.

This CL updates fixLongPath to convert relative paths to absolute
paths if the working directory plus the relative path exceeds
MAX_PATH. This is necessary because the Windows API does not
support extended-length paths for relative paths.

This CL also adds support for fixing device paths (\\.\-prefixed),
which were not previously normalized.

Fixes #41734
Fixes #21782
Fixes #36375

Cq-Include-Trybots: luci.golang.try:gotip-windows-amd64-longtest,gotip-windows-amd64-race,gotip-windows-arm64
Co-authored-by: Giovanni Bajo <rasky@develer.com>
Change-Id: I63cfb79f3ae6b9d42e07deac435b730d97a6f492
Reviewed-on: https://go-review.googlesource.com/c/go/+/574695
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>

database/sql: use slices to simplify the code

Change-Id: Ia198272330626271ee7d4e1ae46afca819ab2933
GitHub-Last-Rev: e713ac31638671f60cc3cf62fa514f784e834e66
GitHub-Pull-Request: golang/go#66572
Reviewed-on: https://go-review.googlesource.com/c/go/+/574995
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>

cmd/cgo: update outdated docs about cgocheck2

Setting GODEBUG=cgocheck=2 now panics with a message
such as "fatal error: cgocheck > 1 mode is no longer supported at runtime.
Use GOEXPERIMENT=cgocheck2 at build time instead."

Change-Id: If9a5a96933973e14a1a60b8e9fb6a1f4a818cf59
GitHub-Last-Rev: 787d62454c31565d0828777bf5ba24c99345af4d
GitHub-Pull-Request: golang/go#65224
Reviewed-on: https://go-review.googlesource.com/c/go/+/557597
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go: close elf file in the readpkglist function

Change-Id: Ief08e311598152f047878fc0fe6a6e37df372ee9
GitHub-Last-Rev: daec402b390e8d1b9882e3fc7c1814ff8d3782c8
GitHub-Pull-Request: golang/go#66588
Reviewed-on: https://go-review.googlesource.com/c/go/+/575156
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

Revert "cmd/compile/internal: merge stack slots for selected local auto vars"

This reverts CL 553055.

Reason for revert: causes crypto/ecdsa failures on linux ppc64/s390x builders

Change-Id: I9266b030693a5b6b1e667a009de89d613755b048
Reviewed-on: https://go-review.googlesource.com/c/go/+/575236
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile/internal: merge stack slots for selected local auto vars

Preliminary compiler support for merging/overlapping stack
slots of local variables whose access patterns are disjoint.

This patch includes changes in AllocFrame to do the actual
merging/overlapping based on information returned from a new
liveness.MergeLocals helper. The MergeLocals helper identifies
candidates by looking for sets of AUTO variables that either A) have
the same size and GC shape (if types contain pointers), or B) have the
same size (but potentially different types as long as those types have
no pointers). Variables must be greater than (3*types.PtrSize) in size
to be considered for merging.

After forming candidates, MergeLocals collects variables into "can be
overlapped" equivalence classes or partitions; this process is driven
by an additional liveness analysis pass. Ideally it would be nice to
move the existing stackmap liveness pass up before AllocFrame
and "widen" it to include merge candidates so that we can do just a
single liveness as opposed to two passes, however this may be difficult
given that the merge-locals liveness has to take into account
writes corresponding to dead stores.

This patch also required a change to the way ssa.OpVarDef pseudo-ops
are generated; prior to this point they would only be created for
variables whose type included pointers; if stack slot merging is
enabled then the ssagen code creates OpVarDef ops for all auto vars
that are merge candidates.

Note that some temporaries created late in the compilation process
(e.g. during ssa backend) are difficult to reason about, especially in
cases where we take the address of a temp and pass it to the runtime.
For the time being we mark most of the vars created post-ssagen as
"not a merge candidate".

Stack slot merging for locals/autos is enabled by default if "-N" is
not in effect, and can be disabled via "-gcflags=-d=mergelocals=0".

Fixmes/todos/restrictions:
- try lowering size restrictions
- re-evaluate the various skips that happen in SSA-created autotmps

Fixes #62737.
Updates #65532.
Updates #65495.

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Change-Id: Ibc22e8a76c87e47bc9fafe4959804d9ea923623d
Reviewed-on: https://go-review.googlesource.com/c/go/+/553055
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

crypto/tls,regexp: remove always-nil error results

These were harmless, but added unnecessary verbosity to the code.
This can happen as a result of refactors: for example,
the method sessionState used to return errors in some cases.

Change-Id: I4e6dacc01ae6a49b528c672979f95cbb86795a85
Reviewed-on: https://go-review.googlesource.com/c/go/+/528995
Reviewed-by: Leo Isla <islaleo93@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Olivier Mengué <olivier.mengue@gmail.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
Reviewed-by: Quim Muntal <quimmuntal@gmail.com>

cmd/compile/internal: refactor coverage ir.Name flags

Minor refactoring to eliminate one of the ir.Name flag values used
when building in coverage mode (no changes to functionality). This is
intended to free up a bit in the uint16 flags field to be used in a
subsequent patch.

Change-Id: I4aedb9a55fde24c808ff3f7b077ee0552aa979af
Reviewed-on: https://go-review.googlesource.com/c/go/+/572055
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile/internal/liveness: introduce "live intervals" utility

Introduce a helper type "Intervals" that contains sets of sorted
disjoint ranges corresponding to live ranges within a function.
Example: the Intervals set "{ [0,1), [4,10) }" would indicate that
something is live starting at instruction 0, then up to but not
including instruction 1, then dead from 1-3, then live again at
instruction 4 up to (but not including) instruction 10.

This patch provides APIs for constructing interval sets, testing to
see whether two sets overlap, and unioning/merging together two
intervals sets.

Updates #62737.
Updates #65532.
Updates #65495.

Change-Id: I7140a5989eba93bf3b8762d9224261f5eba0646d
Reviewed-on: https://go-review.googlesource.com/c/go/+/566177
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

cmd/compiler/internal/ssagen: refactor code to sort stack vars

Minor refactoring of the code that sorts stack variables to move
from sort.Stable to sort.SliceStable. No change in semantics; this
is intended to lay the groundwork for a future change.

Change-Id: I9eb920e3b3029a734fbe0e0e88c0d57ea3452599
Reviewed-on: https://go-review.googlesource.com/c/go/+/566176
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

Revert "cmd/compile: add missing OINLCAll case in mayModifyPkgVar"

This reverts CL 575175.

Reason for revert: causes crypto/ecdh failures on longtest builders.

Change-Id: Ieed326fedf91760ac73095a42ba0237cf969843b
Reviewed-on: https://go-review.googlesource.com/c/go/+/575316
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>

cmd/compile: add missing OINLCAll case in mayModifyPkgVar

CL 395541 made staticopy safe, stop applying the optimization once
seeing an expression that may modify global variables.

However, if a call expression was inlined, the analyzer mis-recognizes
and think that the expression is safe. For example:

var x = 0
var a = f()
var b = x

are re-written to:

var x = 0
var a = ~r0
var b = 0

even though it's not safe because "f()" may modify "x".

Fixing this by recognizing OINLCALL and mark the initialization as
not safe for staticopy.

Fixes #66585

Change-Id: Id930c0b7e74274195f54a498cc4c5a91c4e6d84d
Reviewed-on: https://go-review.googlesource.com/c/go/+/575175
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

regexp/syntax: use the Regexp.Equal static method directly

A follow-up for the recent https://go.dev/cl/573978.

Change-Id: I0e75ca0b37d9ef063bbdfb88d4d2e34647e0ee50
Reviewed-on: https://go-review.googlesource.com/c/go/+/574677
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

runtime: make use of builtin clear in tests

This is a follow-up to CL 574675.

Change-Id: I98c3ea968e9c7dc61472849c385a1e697568aa30
Reviewed-on: https://go-review.googlesource.com/c/go/+/574975
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Jes Cok <xigua67damn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd/cgo: close the file opened in the dynimport function

Change-Id: Ic457e994b50a0e8c1769311937eede1e710a1293
GitHub-Last-Rev: a8af7b5e6710687273203235c19d5402d7f28832
GitHub-Pull-Request: golang/go#66549
Reviewed-on: https://go-review.googlesource.com/c/go/+/574537
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: simplify timers.siftDown

No effect on benchmarks, but the code is clearer.

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │  s7base.txt  │               s7.txt                │
                                   │    sec/op    │    sec/op     vs base               │
AdjustTimers10000-32                 195.9µ ± 13%   198.1µ ±  2%       ~ (p=0.436 n=10)
AdjustTimers10000SingleThread-32     1.573m ± 13%   1.566m ± 10%       ~ (p=0.739 n=10)
AdjustTimers10000NoReset-32          170.6µ ±  1%   170.4µ ±  1%       ~ (p=0.912 n=10)
AdjustTimers10000NoSleep-32          183.9µ ±  2%   181.4µ ±  2%  -1.39% (p=0.045 n=10)
AdjustTimers10000NoResetNoSleep-32   151.3µ ±  1%   150.0µ ±  1%  -0.90% (p=0.007 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt               │
                                   │   sec/op    │   sec/op     vs base               │
AdjustTimers10000-12                 234.2µ ± 1%   234.5µ ± 1%       ~ (p=0.796 n=10)
AdjustTimers10000SingleThread-12     1.191m ± 1%   1.272m ± 1%  +6.81% (p=0.000 n=10)
AdjustTimers10000NoReset-12          239.6µ ± 2%   236.3µ ± 9%       ~ (p=0.971 n=10)
AdjustTimers10000NoSleep-12          223.3µ ± 2%   221.4µ ± 3%       ~ (p=0.579 n=10)
AdjustTimers10000NoResetNoSleep-12   209.2µ ± 2%   209.0µ ± 4%       ~ (p=0.796 n=10)

Change-Id: Id48aa893235d652814b7fa4605037f09b0b4d73b
Reviewed-on: https://go-review.googlesource.com/c/go/+/574897
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

runtime: move whenHeap into heap slice

This CL changes the timer heap from a slice of timers to a slice
of {timer, when} pairs, moving timer.whenHeap into the heap
backing store itself. This makes the ownership clearer (t.whenHeap
was always protected by the heap lock, not the timer's lock)
and also avoids an indirection when doing heap operations,
which look at the when field quite a lot.

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │  s7base.txt  │                s7.txt                │
                                   │    sec/op    │    sec/op     vs base                │
AdjustTimers10000-32                 244.1µ ±  5%   195.9µ ± 13%  -19.76% (p=0.001 n=10)
AdjustTimers10000SingleThread-32     1.674m ±  6%   1.573m ± 13%   -6.03% (p=0.001 n=10)
AdjustTimers10000NoReset-32          194.0µ ±  2%   170.6µ ±  1%  -12.06% (p=0.000 n=10)
AdjustTimers10000NoSleep-32          223.0µ ±  7%   183.9µ ±  2%  -17.54% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-32   181.1µ ±  1%   151.3µ ±  1%  -16.45% (p=0.000 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt                │
                                   │   sec/op    │   sec/op     vs base                │
AdjustTimers10000-12                 266.6µ ± 2%   234.2µ ± 1%  -12.16% (p=0.000 n=10)
AdjustTimers10000SingleThread-12     1.194m ± 0%   1.191m ± 1%   -0.33% (p=0.029 n=10)
AdjustTimers10000NoReset-12          260.9µ ± 2%   239.6µ ± 2%   -8.15% (p=0.000 n=10)
AdjustTimers10000NoSleep-12          247.8µ ± 2%   223.3µ ± 2%   -9.90% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-12   231.7µ ± 2%   209.2µ ± 2%   -9.68% (p=0.000 n=10)

Change-Id: I546f077068476d7a7b855889419b292525fb2bc5
Reviewed-on: https://go-review.googlesource.com/c/go/+/574896
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

net/http: correct doc for ServeFileFS

The documentation of ServeFileFS was partly copied from ServeFile
in CL 513956, however it's not exact. This CL fixes some typos, also
removes obsolete comment for name param.

For consistency, also adds godoc link for ServeFile and ServeContent.

Fixes #66578

Change-Id: I87147d72c533d46284f06ef20b37fdafa8706710
Reviewed-on: https://go-review.googlesource.com/c/go/+/575016
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Mauri de Souza Meneguzzo <mauri870@gmail.com>

cmd/compile: put constants before variables in initialization order

Fixes #66575

Change-Id: I03f4d4577b88ad0a92b260b2efd0cb9fe5082b2f
Reviewed-on: https://go-review.googlesource.com/c/go/+/575075
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

go/ast: more godoc links

Change-Id: I1290ad66d3c758c3b89caf0a217cb3d5358c5dd2
Reviewed-on: https://go-review.googlesource.com/c/go/+/574696
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

runtime: fix timer race introduced in CL 573455

There is a short window when timers.adjust could miss a
timer update. Close that window. Does not change benchmark.

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │ s7base.txt  │               s7.txt                │
                                   │   sec/op    │    sec/op     vs base               │
AdjustTimers10000-32                 239.9µ ± 5%   237.6µ ±  8%       ~ (p=0.631 n=10)
AdjustTimers10000SingleThread-32     1.686m ± 8%   1.710m ±  5%       ~ (p=0.481 n=10)
AdjustTimers10000NoReset-32          194.1µ ± 1%   190.8µ ±  2%  -1.69% (p=0.023 n=10)
AdjustTimers10000NoSleep-32          226.2µ ± 3%   222.9µ ±  3%       ~ (p=0.143 n=10)
AdjustTimers10000NoResetNoSleep-32   182.9µ ± 1%   180.9µ ±  2%       ~ (p=0.165 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt               │
                                   │   sec/op    │   sec/op     vs base               │
AdjustTimers10000-12                 269.3µ ± 2%   267.0µ ± 2%       ~ (p=0.529 n=10)
AdjustTimers10000SingleThread-12     1.176m ± 1%   1.213m ± 1%  +3.15% (p=0.000 n=10)
AdjustTimers10000NoReset-12          262.6µ ± 2%   261.4µ ± 3%       ~ (p=0.123 n=10)
AdjustTimers10000NoSleep-12          247.8µ ± 1%   246.5µ ± 1%       ~ (p=0.393 n=10)
AdjustTimers10000NoResetNoSleep-12   231.0µ ± 1%   232.3µ ± 1%       ~ (p=0.684 n=10)

Change-Id: Ifdfcdd5a25046027912a8b306644bde7ec2d3214
Reviewed-on: https://go-review.googlesource.com/c/go/+/574741
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

runtime: adjust when timers.adjust is called

This new logic more closely mimics what we did before my CL stack.
I had reasoned that certainly

ts.adjust(now, force=true)
ts.run(now)

would be faster than

ts.adjust(now, force=false)
ts.run(now)
ts.adjust(now, force=true)

But certainty is just an emotion, and that turns out not to be the case.

I don't really understand why the second sequence is faster,
but it definitely is, so put it back.

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │ s7base.txt  │               s7.txt               │
                                   │   sec/op    │   sec/op     vs base               │
AdjustTimers10000-32                 263.3µ ± 4%   239.9µ ± 5%  -8.87% (p=0.000 n=10)
AdjustTimers10000SingleThread-32     1.742m ± 3%   1.686m ± 8%       ~ (p=0.105 n=10)
AdjustTimers10000NoReset-32          192.2µ ± 2%   194.1µ ± 1%  +1.00% (p=0.009 n=10)
AdjustTimers10000NoSleep-32          237.0µ ± 2%   226.2µ ± 3%  -4.55% (p=0.001 n=10)
AdjustTimers10000NoResetNoSleep-32   185.2µ ± 1%   182.9µ ± 1%  -1.23% (p=0.003 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt               │
                                   │   sec/op    │   sec/op     vs base               │
AdjustTimers10000-12                 272.6µ ± 3%   269.3µ ± 2%       ~ (p=0.063 n=10)
AdjustTimers10000SingleThread-12     1.126m ± 1%   1.176m ± 1%  +4.42% (p=0.000 n=10)
AdjustTimers10000NoReset-12          255.1µ ± 2%   262.6µ ± 2%  +2.96% (p=0.000 n=10)
AdjustTimers10000NoSleep-12          250.2µ ± 2%   247.8µ ± 1%       ~ (p=0.063 n=10)
AdjustTimers10000NoResetNoSleep-12   230.3µ ± 1%   231.0µ ± 1%       ~ (p=0.280 n=10)

Change-Id: I67b5765f97dfca0142ee38e15a9904b520f51e83
Reviewed-on: https://go-review.googlesource.com/c/go/+/574740
Auto-Submit: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: optimize timers.cleanHead

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │ s7base.txt  │               s7.txt                │
                                   │   sec/op    │   sec/op     vs base                │
AdjustTimers10000-32                 291.4µ ± 5%   263.3µ ± 4%   -9.64% (p=0.000 n=10)
AdjustTimers10000SingleThread-32     1.728m ± 5%   1.742m ± 3%        ~ (p=0.796 n=10)
AdjustTimers10000NoReset-32          253.1µ ± 6%   192.2µ ± 2%  -24.07% (p=0.000 n=10)
AdjustTimers10000NoSleep-32          277.3µ ± 3%   237.0µ ± 2%  -14.54% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-32   241.4µ ± 2%   185.2µ ± 1%  -23.30% (p=0.000 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt                │
                                   │   sec/op    │   sec/op     vs base                │
AdjustTimers10000-12                 288.1µ ± 1%   272.6µ ± 3%   -5.38% (p=0.000 n=10)
AdjustTimers10000SingleThread-12     1.195m ± 1%   1.126m ± 1%   -5.74% (p=0.000 n=10)
AdjustTimers10000NoReset-12          280.8µ ± 1%   255.1µ ± 2%   -9.14% (p=0.000 n=10)
AdjustTimers10000NoSleep-12          292.5µ ± 1%   250.2µ ± 2%  -14.47% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-12   279.8µ ± 2%   230.3µ ± 1%  -17.69% (p=0.000 n=10)

Change-Id: I36edb40ee2cd11ab44d20bff045fa77609dca648
Reviewed-on: https://go-review.googlesource.com/c/go/+/574739
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: avoid unnecessary netpoll wakeups during timer creation

goos: linux
goarch: amd64
pkg: time
cpu: AMD Ryzen 9 7950X 16-Core Processor
                                   │ s7base.txt  │               s7.txt               │
                                   │   sec/op    │   sec/op     vs base               │
AdjustTimers10000-32                 315.1µ ± 4%   291.4µ ± 5%  -7.52% (p=0.001 n=10)
AdjustTimers10000SingleThread-32     1.728m ± 6%   1.728m ± 5%       ~ (p=0.971 n=10)
AdjustTimers10000NoReset-32          263.0µ ± 3%   253.1µ ± 6%  -3.75% (p=0.023 n=10)
AdjustTimers10000NoSleep-32          306.8µ ± 5%   277.3µ ± 3%  -9.62% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-32   245.3µ ± 3%   241.4µ ± 2%       ~ (p=0.529 n=10)

goos: darwin
goarch: arm64
pkg: time
cpu: Apple M3 Pro
                                   │ m3base.txt  │               m3.txt                │
                                   │   sec/op    │   sec/op     vs base                │
AdjustTimers10000-12                 385.5µ ± 3%   288.1µ ± 1%  -25.26% (p=0.000 n=10)
AdjustTimers10000SingleThread-12     1.213m ± 1%   1.195m ± 1%   -1.45% (p=0.001 n=10)
AdjustTimers10000NoReset-12          346.4µ ± 5%   280.8µ ± 1%  -18.94% (p=0.000 n=10)
AdjustTimers10000NoSleep-12          362.6µ ± 1%   292.5µ ± 1%  -19.33% (p=0.000 n=10)
AdjustTimers10000NoResetNoSleep-12   317.3µ ± 8%   279.8µ ± 2%  -11.79% (p=0.000 n=10)

Change-Id: Id3e10abfb1024f8e7883c8c366d941e4f1c2a894
Reviewed-on: https://go-review.googlesource.com/c/go/+/574738
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Russ Cox <rsc@golang.org>

encoding/asn1: simplify appendFourDigits

The new code does not need a for-loop and is easier to read.

Change-Id: Ic182d63c4779c2179b721fcfaec362681284cc16
GitHub-Last-Rev: b3ee265df7eb4d51c945bf5453eb1b09f91912d9
GitHub-Pull-Request: golang/go#63879
Reviewed-on: https://go-review.googlesource.com/c/go/+/538721
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>

internal/sysinfo: implement CPUName on bsd systems

sysctl machdep.cpu.brand_string seems to be standard
across the BSDs. There does not seem to be a standard
way to get the CPU frequency.

Change-Id: Ic986d6c81dd54e1b84544317f2a53ce16801319b
Reviewed-on: https://go-review.googlesource.com/c/go/+/520636
Auto-Submit: Russ Cox <rsc@golang.org>
TryBot-Bypass: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

cmd/go: replace reflect.DeepEqual with slices.Equal and maps.Equal

All of these maps and slices are made up of comparable types,
so we can avoid the overhead of reflection entirely.

Change-Id: If77dbe648a336ba729c171e84c9ff3f7e160297d
Reviewed-on: https://go-review.googlesource.com/c/go/+/574597
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

cmd/compile: rename cmd/compile/internal/pgo to cmd/compile/internal/pgoir

This helps reduce confusion with cmd/internal/pgo, which performs
compilation-independent analysis. pgoir associates that data with the
IR from the current package compilation.

For #58102.

Change-Id: I9ef1c8bc41db466d3340f41f6d071b95c09566de
Reviewed-on: https://go-review.googlesource.com/c/go/+/569338
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile,cmd/preprofile: move logic to shared common package

The processing performed in cmd/preprofile is a simple version of the
same initial processing performed by cmd/compile/internal/pgo. Refactor
this processing into the new IR-independent cmd/internal/pgo package.

Now cmd/preprofile and cmd/compile run the same code for initial
processing of a pprof profile, guaranteeing that they always stay in
sync.

Since it is now trivial, this CL makes one change to the serialization
format: the entries are ordered by weight. This allows us to avoid
sorting ByWeight on deserialization.

Impact on PGO parsing when compiling cmd/compile with PGO:

* Without preprocessing: PGO parsing ~13.7% of CPU time
* With preprocessing (unsorted): ~2.9% of CPU time (sorting ~1.7%)
* With preprocessing (sorted): ~1.3% of CPU time

The remaining 1.3% of CPU time approximately breaks down as:

* ~0.5% parsing the preprocessed profile
* ~0.7% building weighted IR call graph
  * ~0.5% walking function IR to find direct calls
  * ~0.2% performing lookups for indirect calls targets

For #58102.

Change-Id: Iaba425ea30b063ca195fb2f7b29342961c8a64c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/569337
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

os: make readdir more robust on Windows

On Windows, File.readdir currently fails if the volume information
can't be retrieved via GetVolumeInformationByHandle and if the
directory path is relative and can't be converted to an absolute
path.

This change makes readdir more robust by not failing in these cases,
as these steps are just necessary to support a potential call to
os.SameFile, but not for the actual readdir operation. os.SameFile
will still fail in these cases, but that's a separate issue tracked
in #62042.

Change-Id: I8d98d8379bdac4b2832fa433432a5f027756abaa
Reviewed-on: https://go-review.googlesource.com/c/go/+/574155
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>

test: issue16016: use fewer goroutines for gccgo

For https://gcc.gnu.org/PR114453

Change-Id: If41d9fca6288b18ed47b0f21ff224c74ddb34958
Reviewed-on: https://go-review.googlesource.com/c/go/+/574536
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

all: make use of builtin clear

Change-Id: I1df0685c75fc1044ba46003a69ecc7dfc53bbc2b
Reviewed-on: https://go-review.googlesource.com/c/go/+/574675
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd/preprofile: drop output directory check

This check serves only to provide a more descriptive error if the output
directory doesn't exist. That isn't useless, but I don't see why this tool
specifically should do this when no other part of the toolchain does.

For #58102.

Change-Id: I01cf9db2cc1dad85c3afd8a6b008c53f26cb877a
Reviewed-on: https://go-review.googlesource.com/c/go/+/569336
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/preprofile: clean up error handling

This CL adjusts error handling to be a bit more idiomatic. The
processing function returns errors, leaving main to log and exit on
error.

This CL contains no functional changes.

For #58102.

Change-Id: I9074127cc675e177d046474b7f01fbc37d0bd4c0
Reviewed-on: https://go-review.googlesource.com/c/go/+/569335
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

crypto/internal/boring: don't shadow named returns

In setupRSA we use named returns so that we can defer freeing of the
boring private key and context, but were using returns of the form
`return nil, nil, ...` which nil'd the named returns, preventing them
from actually being freed.

Update all of the returns to not shadow the named variables.

Thanks to Quim Muntal of Microsoft for reporting this issue.

Change-Id: Iaf0f0b17e123a7df730cb1e91a324fe622611f66
Reviewed-on: https://go-review.googlesource.com/c/go/+/574195
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>

runtime,hash/maphash: reuse hashSets to save memory pressure

Might help with OOMs on 32-bit platforms

Change-Id: Idd5129c61ecdfeedd5a9a18fce85dbba27cab946
Reviewed-on: https://go-review.googlesource.com/c/go/+/574475
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Than McIntosh <thanm@google.com>

crypto/aes: fix regression for ppc64x

CL 561080 introduced some failures for CBC and GCM crypto tests that are run during a build resulting in dashboard failures.

Tests that are failing:
- TestCBCEncrypterAES
- TestCBCDecrypterAES
- TestAESGCM
- TestGCMCounterWrap
- TestGCMAsm

This fixes those failures.

Change-Id: I90d6f8d279a3051cf951a1a9628afb8d0b5c2f17
Reviewed-on: https://go-review.googlesource.com/c/go/+/574076
Reviewed-by: Marten Seemann <martenseemann@gmail.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Paul Murphy <murp@ibm.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Eli Bendersky <eliben@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

all: fix a large number of comments

Partial typo corrections, following https://go.dev/wiki/Spelling

Change-Id: I2357906ff2ea04305c6357418e4e9556e20375d1
Reviewed-on: https://go-review.googlesource.com/c/go/+/573776
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: shuang cui <imcusg@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

cmd/internal/ssa: fix typo in comment

Change-Id: Ib66830792ebb15538596f4bf4a51d2b3f05c7fe7
GitHub-Last-Rev: 947445916abc91331f51e29c0541c510bd9c623c
GitHub-Pull-Request: golang/go#66460
Reviewed-on: https://go-review.googlesource.com/c/go/+/573557
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Keith Randall <khr@google.com>

cmd/link: close memory profile

Change-Id: If17e982dd0272e90972d2d4a94b1a744343ae2ec
GitHub-Last-Rev: c8d03566b3e66a6de58e778102fa60f5d46ab2d9
GitHub-Pull-Request: golang/go#66531
Reviewed-on: https://go-review.googlesource.com/c/go/+/574335
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Than McIntosh <thanm@google.com>

regexp/syntax: simplify the code

Use the slices package and the built-in max to simplify the code.
There's no noticeable performance change in this modification.

Change-Id: I96e46ba8ab1323f1ba0b8c9b827836e217772cf2
GitHub-Last-Rev: f0111ac7e220f7dac03290125a3a83831012f235
GitHub-Pull-Request: golang/go#66511
Reviewed-on: https://go-review.googlesource.com/c/go/+/573978
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>

runtime,hash/maphash: make tests use less memory

Use a list + sorting instead of a map to keep track of collisions.
Should save a fair amount of memory.

Change-Id: I7b48cdb5f7060d12ebbb950d56b5c479f131c027
Reviewed-on: https://go-review.googlesource.com/c/go/+/574275
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: qiulaidongfeng <2645477756@qq.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Keith Randall <khr@google.com>

net/http: fix typo in log message

Change-Id: I399c65b44706cf5407099e867de27b3acdaf4825
Reviewed-on: https://go-review.googlesource.com/c/go/+/574235
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Auto-Submit: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

slices: add examples

For Clone, Grow, Clip, Concat, Contains, Repeat.

Fixes #66435

Change-Id: Ife8f61427e9cd18b7106c100de8f82f9d7840c9b
Reviewed-on: https://go-review.googlesource.com/c/go/+/573255
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

net/http: panic early in do when Client is nil

Callers who invoke `*http.Client.Do` with a nil *Client will now panic
at the top of c.Do, instead of panicking when `deadline` attempts to
read `c.Timeout`.

Errors inside of net/http can be difficult to track down because the
caller is often invoking the standard library code via an SDK. This
can mean that there are many places to check when code panics, and
raises the importance of being clear about error messages.

If nil receiver calls panic during the `deadline()` call, callers
may confuse the error with a more common timeout or deadline
misconfiguration, which may lead a caller who passed a nil receiver
(the author, for example) down the wrong rabbit hole, or cause them to
suspect their timeout/deadline logic. It is less common to configure
client.Jar, so the probability of detecting the actual problem, given
the underlying error cause, is higher.

Fixes #53521.

Change-Id: If102d17bed56fdd950da6e87762166fd29724654
Reviewed-on: https://go-review.googlesource.com/c/go/+/413975
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: don't call lockOSThread for every syscall call on Windows

Windows syscall.SyscallN currently calls lockOSThread for every syscall.
This can be expensive and produce unnecessary context switches,
especially when the syscall is called frequently under high contention.

The lockOSThread was necessary to ensure that cgocall wouldn't
reschedule the goroutine to a different M, as the syscall return values
are reported back in the M struct.

This CL instructs cgocall to copy the syscall return values into the
the M that will see the caller on return, so the caller no longer needs
to call lockOSThread.

Updates #58336.

Cq-Include-Trybots: luci.golang.try:gotip-windows-arm64,gotip-windows-amd64-longtest
Change-Id: If6644fd111dbacab74e7dcee2afa18ca146735da
Reviewed-on: https://go-review.googlesource.com/c/go/+/562915
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emmanuel@orijtech.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/internal/obj/riscv: use native rotation instructions for rva22u64

When rva22u64 is available, we can now use the native rotation instructions
from the Zbb extension. Use these instead of synthesising rotation
instructions.

This provides a significant performance gain for SHA-512, the following
benchmarked on a StarFive VisionFive 2:

                    │ sha512.rva20u64 │            sha512.rva22u64            │
                    │       B/s       │      B/s       vs base                │
Hash8Bytes/New-4         859.4Ki ± 0%   1337.9Ki ± 0%  +55.68% (p=0.000 n=10)
Hash8Bytes/Sum384-4      888.7Ki ± 1%   1308.6Ki ± 1%  +47.25% (p=0.000 n=10)
Hash8Bytes/Sum512-4      869.1Ki ± 0%   1269.5Ki ± 1%  +46.07% (p=0.000 n=10)
Hash1K/New-4             19.83Mi ± 0%    29.03Mi ± 0%  +46.38% (p=0.000 n=10)
Hash1K/Sum384-4          20.00Mi ± 0%    28.86Mi ± 0%  +44.30% (p=0.000 n=10)
Hash1K/Sum512-4          19.93Mi ± 0%    28.72Mi ± 0%  +44.11% (p=0.000 n=10)
Hash8K/New-4             23.85Mi ± 0%    34.12Mi ± 0%  +43.09% (p=0.000 n=10)
Hash8K/Sum384-4          23.88Mi ± 0%    34.09Mi ± 0%  +42.77% (p=0.000 n=10)
Hash8K/Sum512-4          23.87Mi ± 0%    34.07Mi ± 0%  +42.71% (p=0.000 n=10)
geomean                  7.399Mi         10.78Mi       +45.77%

Change-Id: I9dca8e3f311eea101684c806cb998872dc697288
Reviewed-on: https://go-review.googlesource.com/c/go/+/572716
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: M Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Auto-Submit: Emmanuel Odeke <emmanuel@orijtech.com>