encoding/pem: clarify Decode only works on lines

Fixes #53524

Change-Id: I929ee3c055c3ca564cd6cc374124f493aea2fbf6
Reviewed-on: https://go-review.googlesource.com/c/go/+/421636
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Junyang Shao <shaojunyang@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: improve issue62407_test

The original issue62407_test also passes with versions prior to 1.23.
The improvement makes it fail with versions prior to 1.23.

Change-Id: I94bfb9d1ac695c8e07997d7029fc2101535e14f8
GitHub-Last-Rev: 44be2a610a1a79d04dc3d228af2b313200f4d900
GitHub-Pull-Request: golang/go#70938
Reviewed-on: https://go-review.googlesource.com/c/go/+/638036
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: add tests for too-large shift amounts

Update #72018

Change-Id: I3188019658c37da3c31f06472023b39e13170ebf
Reviewed-on: https://go-review.googlesource.com/c/go/+/654316
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Keith Randall <khr@google.com>

testing: modify got,want equal comparison for unordered example output

This change eliminates sortLines function to avoid strings.Join calls.

It's not a performance problem, this change tries to make the comparison
more straightforward.

Change-Id: I3a7ae877c9fc927833ab9f143205f7e007197f60
GitHub-Last-Rev: a71aa58c58533fed24ba9c101664b977a094caf9
GitHub-Pull-Request: golang/go#72025
Reviewed-on: https://go-review.googlesource.com/c/go/+/653556
Reviewed-by: Junyang Shao <shaojunyang@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: Ian Lance Taylor <iant@google.com>

sync: document behavior of Map.Delete when key is not present

Change-Id: I6a6a465631e91141a5fcc9c04a7df0dc47780731
Reviewed-on: https://go-review.googlesource.com/c/go/+/643955
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

net/http: don't modify caller's tls.Config.NextProtos

Clone the input slice before adjusting NextProtos
to add or remove "http/1.1" and "h2" entries,
so as not to modify a slice that the caller might be using.
(We clone the tls.Config that contains the slice, but
that's a shallow clone.)

Fixes #72100

Change-Id: I9f228b8fb6f6f2ca5023179ec114929c002dbda9
Reviewed-on: https://go-review.googlesource.com/c/go/+/654875
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

context: skip allocs test with -asan

Allocates more with -asan after CL 653795.

Change-Id: Ib8cc1de1d649623713b6fc123c1c59a47528857b
Reviewed-on: https://go-review.googlesource.com/c/go/+/654876
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

all: update golang.org/x/net

For #71984

Change-Id: Ic15826f09ea818f8833bd3d979bffaede24d49df
Reviewed-on: https://go-review.googlesource.com/c/go/+/654717
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Junyang Shao <shaojunyang@google.com>

crypto/elliptic: drop hidden Inverse and CombinedMult methods

These methods were previously used by crypto/ecdsa, but now not even
ecdsa_legacy.go uses them. Neither were ever documented.

Inverse was available only on P256() and only on amd64 and arm64, so
hopefully no one used it. CombinedMult was always available on all
curves, so it's possible some application might have used it, but all
the samples on GitHub I can find copied the old crypto/ecdsa package,
which does a conditional interface upgrade with a fallback, so they
won't break.

Change-Id: I6a6a4656ee1ab98438ca0fb20bea53b229cd7e71
Reviewed-on: https://go-review.googlesource.com/c/go/+/640116
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

runtime: decorate anonymous memory mappings

Leverage the prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, ...) API to name
the anonymous memory areas.

This API has been introduced in Linux 5.17 to decorate the anonymous
memory areas shown in /proc/<pid>/maps.

This is already used by glibc. See:
* https://sourceware.org/git/?p=glibc.git;a=blob;f=malloc/malloc.c;h=27dfd1eb907f4615b70c70237c42c552bb4f26a8;hb=HEAD#l2434
* https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/setvmaname.c;h=ea93a5ffbebc9e5a7e32a297138f465724b4725f;hb=HEAD#l63

This can be useful when investigating the memory consumption of a
multi-language program.
On a 100% Go program, pprof profiler can be used to profile the memory
consumption of the program. But pprof is only aware of what happens
within the Go world.

On a multi-language program, there could be a doubt about whether the
suspicious extra-memory consumption comes from the Go part or the native
part.

With this change, the following Go program:

        package main

        import (
                "fmt"
                "log"
                "os"
        )

        /*
        #include <stdlib.h>

        void f(void)
        {
          (void)malloc(1024*1024*1024);
        }
        */
        import "C"

        func main() {
                C.f()

                data, err := os.ReadFile("/proc/self/maps")
                if err != nil {
                        log.Fatal(err)
                }
                fmt.Println(string(data))
        }

produces this output:

        $ GLIBC_TUNABLES=glibc.mem.decorate_maps=1 ~/doc/devel/open-source/go/bin/go run .
        00400000-00402000 r--p 00000000 00:21 28451768                           /home/lenaic/.cache/go-build/9f/9f25a17baed5a80d03eb080a2ce2a5ff49c17f9a56e28330f0474a2bb74a30a0-d/test_vma_name
        00402000-004a4000 r-xp 00002000 00:21 28451768                           /home/lenaic/.cache/go-build/9f/9f25a17baed5a80d03eb080a2ce2a5ff49c17f9a56e28330f0474a2bb74a30a0-d/test_vma_name
        004a4000-00574000 r--p 000a4000 00:21 28451768                           /home/lenaic/.cache/go-build/9f/9f25a17baed5a80d03eb080a2ce2a5ff49c17f9a56e28330f0474a2bb74a30a0-d/test_vma_name
        00574000-00575000 r--p 00173000 00:21 28451768                           /home/lenaic/.cache/go-build/9f/9f25a17baed5a80d03eb080a2ce2a5ff49c17f9a56e28330f0474a2bb74a30a0-d/test_vma_name
        00575000-00580000 rw-p 00174000 00:21 28451768                           /home/lenaic/.cache/go-build/9f/9f25a17baed5a80d03eb080a2ce2a5ff49c17f9a56e28330f0474a2bb74a30a0-d/test_vma_name
        00580000-005a4000 rw-p 00000000 00:00 0
        2e075000-2e096000 rw-p 00000000 00:00 0                                  [heap]
        c000000000-c000400000 rw-p 00000000 00:00 0                              [anon: Go: heap]
        c000400000-c004000000 ---p 00000000 00:00 0                              [anon: Go: heap reservation]
        777f40000000-777f40021000 rw-p 00000000 00:00 0                          [anon: glibc: malloc arena]
        777f40021000-777f44000000 ---p 00000000 00:00 0
        777f44000000-777f44021000 rw-p 00000000 00:00 0                          [anon: glibc: malloc arena]
        777f44021000-777f48000000 ---p 00000000 00:00 0
        777f48000000-777f48021000 rw-p 00000000 00:00 0                          [anon: glibc: malloc arena]
        777f48021000-777f4c000000 ---p 00000000 00:00 0
        777f4c000000-777f4c021000 rw-p 00000000 00:00 0                          [anon: glibc: malloc arena]
        777f4c021000-777f50000000 ---p 00000000 00:00 0
        777f50000000-777f50021000 rw-p 00000000 00:00 0                          [anon: glibc: malloc arena]
        777f50021000-777f54000000 ---p 00000000 00:00 0
        777f55afb000-777f55afc000 ---p 00000000 00:00 0
        777f55afc000-777f562fc000 rw-p 00000000 00:00 0                          [anon: glibc: pthread stack: 216378]
        777f562fc000-777f562fd000 ---p 00000000 00:00 0
        777f562fd000-777f56afd000 rw-p 00000000 00:00 0                          [anon: glibc: pthread stack: 216377]
        777f56afd000-777f56afe000 ---p 00000000 00:00 0
        777f56afe000-777f572fe000 rw-p 00000000 00:00 0                          [anon: glibc: pthread stack: 216376]
        777f572fe000-777f572ff000 ---p 00000000 00:00 0
        777f572ff000-777f57aff000 rw-p 00000000 00:00 0                          [anon: glibc: pthread stack: 216375]
        777f57aff000-777f57b00000 ---p 00000000 00:00 0
        777f57b00000-777f58300000 rw-p 00000000 00:00 0                          [anon: glibc: pthread stack: 216374]
        777f58300000-777f58400000 rw-p 00000000 00:00 0                          [anon: Go: page alloc index]
        777f58400000-777f5a400000 rw-p 00000000 00:00 0                          [anon: Go: heap index]
        777f5a400000-777f6a580000 ---p 00000000 00:00 0                          [anon: Go: scavenge index]
        777f6a580000-777f6a581000 rw-p 00000000 00:00 0                          [anon: Go: scavenge index]
        777f6a581000-777f7a400000 ---p 00000000 00:00 0                          [anon: Go: scavenge index]
        777f7a400000-777f8a580000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f8a580000-777f8a581000 rw-p 00000000 00:00 0                          [anon: Go: page alloc]
        777f8a581000-777f9c430000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f9c430000-777f9c431000 rw-p 00000000 00:00 0                          [anon: Go: page alloc]
        777f9c431000-777f9e806000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f9e806000-777f9e807000 rw-p 00000000 00:00 0                          [anon: Go: page alloc]
        777f9e807000-777f9ec00000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f9ec36000-777f9ecb6000 rw-p 00000000 00:00 0                          [anon: Go: immortal metadata]
        777f9ecb6000-777f9ecc6000 rw-p 00000000 00:00 0                          [anon: Go: gc bits]
        777f9ecc6000-777f9ecd6000 rw-p 00000000 00:00 0                          [anon: Go: allspans array]
        777f9ecd6000-777f9ece7000 rw-p 00000000 00:00 0                          [anon: Go: immortal metadata]
        777f9ece7000-777f9ed67000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f9ed67000-777f9ed68000 rw-p 00000000 00:00 0                          [anon: Go: page alloc]
        777f9ed68000-777f9ede7000 ---p 00000000 00:00 0                          [anon: Go: page summary]
        777f9ede7000-777f9ee07000 rw-p 00000000 00:00 0                          [anon: Go: page alloc]
        777f9ee07000-777f9ee0a000 rw-p 00000000 00:00 0                          [anon: glibc: loader malloc]
        777f9ee0a000-777f9ee2e000 r--p 00000000 00:21 48158213                   /usr/lib/libc.so.6
        777f9ee2e000-777f9ef9f000 r-xp 00024000 00:21 48158213                   /usr/lib/libc.so.6
        777f9ef9f000-777f9efee000 r--p 00195000 00:21 48158213                   /usr/lib/libc.so.6
        777f9efee000-777f9eff2000 r--p 001e3000 00:21 48158213                   /usr/lib/libc.so.6
        777f9eff2000-777f9eff4000 rw-p 001e7000 00:21 48158213                   /usr/lib/libc.so.6
        777f9eff4000-777f9effc000 rw-p 00000000 00:00 0
        777f9effc000-777f9effe000 rw-p 00000000 00:00 0                          [anon: glibc: loader malloc]
        777f9f00a000-777f9f04a000 rw-p 00000000 00:00 0                          [anon: Go: immortal metadata]
        777f9f04a000-777f9f04c000 r--p 00000000 00:00 0                          [vvar]
        777f9f04c000-777f9f04e000 r--p 00000000 00:00 0                          [vvar_vclock]
        777f9f04e000-777f9f050000 r-xp 00000000 00:00 0                          [vdso]
        777f9f050000-777f9f051000 r--p 00000000 00:21 48158204                   /usr/lib/ld-linux-x86-64.so.2
        777f9f051000-777f9f07a000 r-xp 00001000 00:21 48158204                   /usr/lib/ld-linux-x86-64.so.2
        777f9f07a000-777f9f085000 r--p 0002a000 00:21 48158204                   /usr/lib/ld-linux-x86-64.so.2
        777f9f085000-777f9f087000 r--p 00034000 00:21 48158204                   /usr/lib/ld-linux-x86-64.so.2
        777f9f087000-777f9f088000 rw-p 00036000 00:21 48158204                   /usr/lib/ld-linux-x86-64.so.2
        777f9f088000-777f9f089000 rw-p 00000000 00:00 0
        7ffc7bfa7000-7ffc7bfc8000 rw-p 00000000 00:00 0                          [stack]
        ffffffffff600000-ffffffffff601000 --xp 00000000 00:00 0                  [vsyscall]

The anonymous memory areas are now labelled so that we can see which
ones have been allocated by the Go runtime versus which ones have been
allocated by the glibc.

Fixes #71546

Change-Id: I304e8b4dd7f2477a6da794fd44e9a7a5354e4bf4
Reviewed-on: https://go-review.googlesource.com/c/go/+/646095
Auto-Submit: Alan Donovan <adonovan@google.com>
Commit-Queue: Alan Donovan <adonovan@google.com>
Reviewed-by: Felix Geisendörfer <felix.geisendoerfer@datadoghq.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

net/http/httputil: skip new test on Plan 9

TestReverseProxyWebSocketHalfTCP requires half closed connections,
which aren't supported on Plan 9.

For #35892
Fixes #72095

Change-Id: I64b458bc15ac3b8eda43dc871bf67ada32a59708
Reviewed-on: https://go-review.googlesource.com/c/go/+/654636
Reviewed-by: Ian Lance Taylor <iant@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: Damien Neil <dneil@google.com>
Reviewed-by: David du Colombier <0intro@gmail.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

os: guarantee min buffer size for ReadFile reads on /proc-like files

For instance, this fixes os.ReadFile on plan9's /net/iproute file.

But it's not necessarily plan9-specific; Linux /proc and /sys filesystems
can exhibit the same problems.

Fixes #72080

Change-Id: I60b035913f583a91c6d84df95a6ea7b7ec2b3c92
Reviewed-on: https://go-review.googlesource.com/c/go/+/654315
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net/http: speed up cookie and method validation

Fixes #67031

Change-Id: I1d764afdc7e50d61007f5f71a674eb6872ce507a
GitHub-Last-Rev: 869535e843d2133fa5279297b002dd96725384e0
GitHub-Pull-Request: golang/go#71798
Reviewed-on: https://go-review.googlesource.com/c/go/+/650195
Auto-Submit: Sean Liao <sean@liao.dev>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sean Liao <sean@liao.dev>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

mime/quotedprintable: accept LWSP-char after =

SP and HTAB are allowed after a = before the following CRLF.

RFC 2045 section 6.7 describes the ABNF for the quoted-printable encoding:

    qp-line := *(qp-segment transport-padding CRLF)
               qp-part transport-padding
    qp-segment := qp-section *(SPACE / TAB) "="
    transport-padding := *LWSP-char
                          ; Composers MUST NOT generate
                          ; non-zero length transport
                          ; padding, but receivers MUST
                          ; be able to handle padding
                          ; added by message transports.

RFC 822 defines LWSP-char as:

    LWSP-char   =  SPACE / HTAB

Dovecot's imaptest contains such a message in
src/tests/fetch-binary-mime-qp.mbox.

Fixes #70952

Change-Id: Ie05921088d7e4d6c92c4bf79b0f4a13586230753
GitHub-Last-Rev: e6e6eee8ebc2f629644a1d99129fb57cce58058f
GitHub-Pull-Request: golang/go#70951
Reviewed-on: https://go-review.googlesource.com/c/go/+/638276
Reviewed-by: Sean Liao <sean@liao.dev>
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Sean Liao <sean@liao.dev>

net/http: support TCP half-close when HTTP is upgraded in ReverseProxy

This CL propagates closing the write stream from either side of the
reverse proxy and ensures the proxy waits for both copy-to and the
copy-from the backend to complete.

The new unit test checks communication through the reverse proxy when
the backend or frontend closes either the read or write streams.
That closing the write stream is propagated through the proxy from
either the backend or the frontend. That closing the read stream is
not propagated through the proxy.

Fixes #35892

Change-Id: I83ce377df66a0f17b9ba2b53caf9e4991a95f6a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/637939
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Sean Liao <sean@liao.dev>
Auto-Submit: Sean Liao <sean@liao.dev>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Matej Kramny <matejkramny@gmail.com>

net/http: check server shutting down before processing the request

The root cause of issue #65802 is a small race condition that occurs between
two events:

1. During the HTTP server shutdown, a connection in an idle state is identified
and closed.
2. The connection, although idle, has just finished reading a complete request
before being closed and hasn't yet updated its state to active.

In this scenario, despite the connection being closed, the request continues to
be processed. This not only wastes server resources but also prevents the
client request from being retried.

Fixes #65802

Change-Id: Ic22abb4497be04f6c84dff059df00f2c319d8652
GitHub-Last-Rev: 426099a3e75f51b80f8ca866938f31417d75ff89
GitHub-Pull-Request: golang/go#65805
Reviewed-on: https://go-review.googlesource.com/c/go/+/565277
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Sean Liao <sean@liao.dev>
Reviewed-by: Damien Neil <dneil@google.com>
Auto-Submit: Sean Liao <sean@liao.dev>

cmd/compile,cmd/link: move to DWARF5-style location lists

This patch updates the compiler to generate DWARF5-style location
lists (e.g. entries that feed into .debug_loclists) as opposed to
DWARF4-style location lists (which wind up in .debug_loc). The DWARF5
format is much more compact, and can make indirect references to text
addresses via the .debug_addr section for further space savings.

Updates #26379.

Change-Id: If2e6fce1136d9cba5125ea51a71419596d1d1691
Reviewed-on: https://go-review.googlesource.com/c/go/+/635836
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

cmd/compile,cmd/link: move to DWARF5-style range lists

This patch updates the compiler to generate DWARF5-style range lists
(e.g. entries that feed into .debug_rnglists) as opposed to
DWARF4-style range lists (which wind up in .debug_ranges). The DWARF5
format is much more compact, and can make indirect references to text
address via the .debug_addr section for further space savings.

Updates #26379.

Change-Id: I273a6283484b7fe33d79d5412e31c5155b22a7c0
Reviewed-on: https://go-review.googlesource.com/c/go/+/635345
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>

cmd/compile: fix out of memory when inlining closure

CL 629195 strongly favor closure inlining, allowing closures to be
inlined more aggressively.

However, if the closure body contains a call to a function, which itself
is one of the call arguments, it causes the infinite inlining.

Fixing this by prevent this kind of functions from being inlinable.

Fixes #72063

Change-Id: I5fb5723a819b1e2c5aadb57c1023ec84ca9fa53c
Reviewed-on: https://go-review.googlesource.com/c/go/+/654195
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

crypto/internal/fips140/bigmod/_asm: update avo to v0.6.0

avo v0.4.0 x/tools dependency crashes while parsing with Go 1.25.

Change-Id: Ic951066b0b39b477887ad0e32be44f4d88d4c2f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/653175
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

net/textproto: document enforcement of RFC 9112 for headers

Fixes #68590

Change-Id: Ie7cf1fe8379182f86317d5ebb7f45a404ecd70e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/601555
Auto-Submit: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>

cmd/internal/obj/loong64: add F{MAXA/MINA}.{S/D} instructions

Go asm syntax:
F{MAXA/MINA}{F/D} FK, FJ, FD

Equivalent platform assembler syntax:
f{maxa/mina}.{s/d} fd, fj, fk

Ref: https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html

Change-Id: I6790657d2f36bdf5e6818b6c0aaa48117e782b8d
Reviewed-on: https://go-review.googlesource.com/c/go/+/653915
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>

cmd/internal/obj/loong64: add {V,XV}{SLL/SRL/SRA/ROTR}[I].{B/H/W/D} instructions support

Go asm syntax:
 V{SLL/SRL/SRA/ROTR}{B/H/W/V} $1, V2, V3
XV{SLL/SRL/SRA/ROTR}{B/H/W/V} $1, X2, X3
 V{SLL/SRL/SRA/ROTR}{B/H/W/V} VK, VJ, VD
XV{SLL/SRL/SRA/ROTR}{B/H/W/V} XK, XJ, XD

Equivalent platform assembler syntax:
 v{sll/srl/sra/rotr}i.{b/h/w/d} v3, v2, $1
xv{sll/srl/sra/rotr}i.{b/h/w/d} x3, x2, $1
 v{sll/srl/sra/rotr}.{b/h/w/d} vd, vj, vk
xv{sll/srl/sra/rotr}.{b/h/w/d} xd, xj, xk

Change-Id: Ie4f04de1c77491a71688d226f7d91cd1a699ab47
Reviewed-on: https://go-review.googlesource.com/c/go/+/637775
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

go/types, types2: remove remaining mentions of core type in error messages

The implementatiom still calls coreType in places and refers to
"core types" in comments, but user-visible error messages don't
know about core types anymore.

This brings the user-visible part of the implementation in sync with
the spec which doesn't have the notion of core types anymore.

For #70128.

Change-Id: I14bc6767a83e8f54b10ebe99a7df0b98cd9fca87
Reviewed-on: https://go-review.googlesource.com/c/go/+/654395
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>

context: use atomic operation in ctx.Err

oos: darwin
goarch: arm64
pkg: context
cpu: Apple M1 Pro
               │ /tmp/bench.0.mac │          /tmp/bench.1.mac           │
               │      sec/op      │   sec/op     vs base                │
ErrOK-10             13.750n ± 1%   2.080n ± 0%  -84.87% (p=0.000 n=10)
ErrCanceled-10       13.530n ± 1%   3.248n ± 1%  -76.00% (p=0.000 n=10)
geomean               13.64n        2.599n       -80.94%

goos: linux
goarch: amd64
pkg: context
cpu: Intel(R) Xeon(R) CPU @ 2.30GHz
               │ /tmp/bench.0.linux │         /tmp/bench.1.linux          │
               │       sec/op       │   sec/op     vs base                │
ErrOK-16               21.435n ± 0%   4.243n ± 0%  -80.21% (p=0.000 n=10)
ErrCanceled-16         21.445n ± 0%   5.070n ± 0%  -76.36% (p=0.000 n=10)
geomean                 21.44n        4.638n       -78.37%

Fixes #72040

Change-Id: I3b337ab1934689d2da4134492ee7c5aac8f92845
Reviewed-on: https://go-review.googlesource.com/c/go/+/653795
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Auto-Submit: Damien Neil <dneil@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>

os: only call GetConsoleMode for char devices

There is no need to call GetConsoleMode if we know that the file
type is not FILE_TYPE_CHAR. This is a tiny performance optimization,
as I sometimes see this call in profiles.

Change-Id: I9e9237908585d0ec8360930a0406b26f52699b92
Reviewed-on: https://go-review.googlesource.com/c/go/+/654155
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>

go/types, types2: remove remaining references to coreType in literals.go

For now, use commonUnder (formerly called sharedUnder) and update
error messages and comments. We can provide better error messages
in individual cases eventually.

For #70128.

Change-Id: I906ba9a0c768f6499c1683dc9be3ad27da8007a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/653156
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>

go/types, types2: remove most remaining references to coreType in builtin.go

For now, use commonUnder (formerly called sharedUnder) and update
error messages and comments. We can provide better error messages
in individual cases eventually.

Kepp using coreType for make built-in for now because it must accept
different channel types with non-conflicting directions and identical
element types. Added extra test cases.

While at it, rename sharedUnder, sharedUnderOrChan to commonUnder
and commonUnderOrChan, respectively (per suggestion from rfindley).

For #70128.

Change-Id: I11f3d5ce858746574f4302271d8cb763c2cdcf98
Reviewed-on: https://go-review.googlesource.com/c/go/+/653139
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>

crypto/rsa: use Div instead of GCD for trial division

Div is way faster. We could actually test a lot more primes and still
gain performance despite the diminishing returns, but necessarily it
would have marginal impact overall.

fips140: off
goos: linux
goarch: amd64
pkg: crypto/rsa
cpu: AMD Ryzen 7 PRO 8700GE w/ Radeon 780M Graphics
                    │  e325b41ad1  │             0f611af2e1              │
                    │    sec/op    │   sec/op     vs base                │
GenerateKey/2048-16   124.19m ± 0%   39.93m ± 0%  -67.85% (p=0.000 n=20)

Surprisingly, the performance gain is similar on ARM64, which doesn't
have intrinsified math.Div.

fips140: off
goos: darwin
goarch: arm64
pkg: crypto/rsa
cpu: Apple M2
                   │  e325b41ad1  │             6276161a7f              │
                   │    sec/op    │   sec/op     vs base                │
GenerateKey/2048-8   136.49m ± 0%   47.97m ± 1%  -64.86% (p=0.000 n=20)

Change-Id: I6a6a46560331198312bd09c1cbe4d2b3c370c552
Reviewed-on: https://go-review.googlesource.com/c/go/+/639955
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>

go/types, types2: remove references to core type in append

Writing explicit code for this case turned out to be simpler
and easier to reason about then relying on a helper functions
(except for typeset).

While at it, make append error messages more consistent.

For #70128.

Change-Id: I3dc79774249929de5061b4301ab2506d4b3da0d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/653095
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>

testing: fix testing.B.Loop doc on loop condition

As mentioned by
https://github.com/golang/go/issues/61515#issuecomment-2656656554,
the documentation should be relaxed.

Change-Id: I9f18301e1a4e4d9a72c9fa0b1132b1ba3cc57b03
Reviewed-on: https://go-review.googlesource.com/c/go/+/651435
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Commit-Queue: Junyang Shao <shaojunyang@google.com>

net: fix parsing of interfaces on plan9 without associated devices

Fixes #72060
Updates #39908

Change-Id: I7d5bda1654753acebc8aa9937d010b41c5722b36
Reviewed-on: https://go-review.googlesource.com/c/go/+/654055
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Russ Cox <rsc@golang.org>
Auto-Submit: Brad Fitzpatrick <bradfitz@golang.org>

net/http: allocate CloseNotifier channel lazily

The CloseNotifier interface is deprecated. We can defer allocating the
backing channel until the first use of CloseNotifier.

goos: linux
goarch: amd64
pkg: net/http
cpu: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz
                   │   before    │               after                │
                   │   sec/op    │   sec/op     vs base               │
Server-8             160.8µ ± 2%   160.1µ ± 1%       ~ (p=0.353 n=10)
CloseNotifier/h1-8   222.1µ ± 4%   226.4µ ± 7%       ~ (p=0.143 n=10)
geomean              189.0µ        190.4µ       +0.75%

                   │    before    │                after                │
                   │     B/op     │     B/op      vs base               │
Server-8             2.292Ki ± 0%   2.199Ki ± 0%  -4.07% (p=0.000 n=10)
CloseNotifier/h1-8   3.224Ki ± 0%   3.241Ki ± 0%  +0.51% (p=0.000 n=10)
geomean              2.718Ki        2.669Ki       -1.80%

                   │   before   │                after                │
                   │ allocs/op  │ allocs/op   vs base                 │
Server-8             21.00 ± 0%   20.00 ± 0%  -4.76% (p=0.000 n=10)
CloseNotifier/h1-8   50.00 ± 0%   50.00 ± 0%       ~ (p=1.000 n=10) ¹
geomean              32.40        31.62       -2.41%
¹ all samples are equal

Change-Id: I3f35d56b8356fb660589b7708a023e4480f32067
GitHub-Last-Rev: c75696b9b8498ae03a4ad9527b9b7c8337415456
GitHub-Pull-Request: golang/go#71163
Reviewed-on: https://go-review.googlesource.com/c/go/+/640598
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: add padding to m struct for 64 bit architectures

CL 652276 reduced the m struct by 8 bytes, which has changed the
allocation class on 64 bit OpenBSD platforms. This results in build
failures due to:

    M structure uses sizeclass 1792/0x700 bytes; incompatible with mutex flag mask 0x3ff

Add 128 bytes of padding when spinbitmutex is enabled on 64 bit
architectures, moving the size to the half point between the
1792 and 2048 allocation size.

Change-Id: I71623a1f75714543c302217e619d20cf0e717aeb
Reviewed-on: https://go-review.googlesource.com/c/go/+/653335
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/internal/obj/riscv: add riscv64 CSR map

The map is automatically generated by running the latest version of
parse.py from github.com/riscv/riscv-opcodes.

Change-Id: I05e00ab27ec583750752c25e1835c2578b339fbf
Reviewed-on: https://go-review.googlesource.com/c/go/+/630518
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Pengcheng Wang <wangpengcheng.pp@bytedance.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: Michael Pratt <mpratt@google.com>

cmd/go/internal/work: use par.Cache to cache tool IDs.

The tool IDs can be calculated once and reused across multiple
threads. This is a small optimization that helps optimize system
resources.

On a normal Windows machine with 12 virtual CPUs, the time to build
a hello world program is reduced from over 1 second, with spikes of 2
seconds, to a consistent 0.7 seconds.

Updates #71981.

Change-Id: I85f4a19f8ad4230afa32213780c761b7eb22fa29
Reviewed-on: https://go-review.googlesource.com/c/go/+/653715
Reviewed-by: Michael Matloob <matloob@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: use ABIInternal on syscall and other sys.stuff for loong64

Change-Id: I6b2942c413eab58c457980131022dace036cd76c
Reviewed-on: https://go-review.googlesource.com/c/go/+/623475
Reviewed-by: Meidan Li <limeidan@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: abner chenc <chenguoqi@loongson.cn>

cmd/internal/obj, cmd/asm: reclassify 32-bit immediate value of loong64

Change-Id: If9fd257ca0837a8c8597889c4f5ed3d4edc602c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/636995
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go/internal/modindex: clean modroot and pkgdir for openIndexPackage

GetPackage is sometimes called with a modroot or pkgdir that ends with a
path separator, and sometimes without. Clean them before passing them to
openIndexPackage to deduplicate calls to mcache.Do and action entry
files.

This shouldn't affect #71698 but was discovered while debugging that
issue.

Change-Id: I6a7fa4de93f45801504abea11bd97f6c6577f296
Reviewed-on: https://go-review.googlesource.com/c/go/+/652435
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/covdata: close output meta-data file

Change-Id: Idd2a324eb51ffa3f40cb3df03a82a1d6d882295a
GitHub-Last-Rev: 62e22b309d9f4b31b1fb426e4fdbabd04fcc8371
GitHub-Pull-Request: golang/go#71993
Reviewed-on: https://go-review.googlesource.com/c/go/+/653140
Reviewed-by: Than McIntosh <thanm@golang.org>
Commit-Queue: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>

cmd/link: use __got as the .got section name

The __nl_symbol_ptr is not a common section name anymore. LLVM prefers
__got for GOT symbols in the __DATA_CONST segment.

Note that the Go linker used to place the GOT section in the __DATA
segment, but since CL 644055 we place it in the __DATA_CONST segment.

Updates #71416.

Cq-Include-Trybots: luci.golang.try:gotip-darwin-amd64-longtest
Change-Id: Icb776e19855eaabb4777a9b1eb433497842413b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/652555
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

reflect: add more tests for Type.{CanSeq,CanSeq2}

For #71874.

Change-Id: I3850edfb3104305f3bf4847a73cdd826cc99837f
GitHub-Last-Rev: 574c1edb7a6152c71891fab011ac0aaeca955fc8
GitHub-Pull-Request: golang/go#71890
Reviewed-on: https://go-review.googlesource.com/c/go/+/651775
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime: increase timeout in TestSpuriousWakeupsNeverHangSemasleep

This change tries increasing the timeout in
TestSpuriousWakeupsNeverHangSemasleep. I'm not entirely sure of the
mechanism, but GODEBUG=gcstoptheworld=2 and GODEBUG=gccheckmark=1 can
cause this test to fail at it's regular timeout. It does not seem to
indicate a deadlock, because bumping the timeout 10x make the problem
go away. I suspect the problem is due to the long STW times these two
modes can induce, plus the fact this test runs in parallel with others.

Let's just bump the timeout. The test is fundamentally sound, and it's
unclear to me how else to test for a deadlock here.

Fixes #71691.
Fixes #71548.

Change-Id: I649531eeec8a8408ba90823ce5223f3a17863124
Reviewed-on: https://go-review.googlesource.com/c/go/+/652756
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

crypto/x509: avoid crypto/rand.Int to generate serial number

It's probabyl safe enough, but just reading bytes from rand and then
using SetBytes is simpler, and doesn't require allowing calls from
crypto into math/big's Lsh, Sub, and Cmp.

Change-Id: I6a6a4656761f7073f9e149f288c48e97048ab13c
Reviewed-on: https://go-review.googlesource.com/c/go/+/643278
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
Reviewed-by: Roland Shoemaker <roland@golang.org>

cmd/cgo: update generated headers for compatibility with latest MSVC C++ standards

Updates #71921

Change-Id: Idfbb72e259b169121c8ced6d89ee2f13d6254d0d
GitHub-Last-Rev: fcf12e5a221621f749841055df1d2c2ada3bf844
GitHub-Pull-Request: golang/go#72004
Reviewed-on: https://go-review.googlesource.com/c/go/+/653141
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
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>

flag: replace interface{} -> any for textValue.Get method

Make it literally match the Getter interface.

Change-Id: I73f03780ba1d3fd2230e0e5e2343d40530d9e6d8
GitHub-Last-Rev: 398b90b2fb04fdd401a1d719bf3ce19152a4cf6a
GitHub-Pull-Request: golang/go#71975
Reviewed-on: https://go-review.googlesource.com/c/go/+/652795
Reviewed-by: Ian Lance Taylor <iant@google.com>
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>

net/http: reject newlines in chunk-size lines

Unlike request headers, where we are allowed to leniently accept
a bare LF in place of a CRLF, chunked bodies must always use CRLF
line terminators. We were already enforcing this for chunk-data lines;
do so for chunk-size lines as well. Also reject bare CRs anywhere
other than as part of the CRLF terminator.

Fixes CVE-2025-22871
Fixes #71988

Change-Id: Ib0e21af5a8ba28c2a1ca52b72af8e2265ec79e4a
Reviewed-on: https://go-review.googlesource.com/c/go/+/652998
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/go: update c document

Fixes: #11875
Change-Id: I0ea2c3e94d7d1647c2aaa3d488ac3c1f5fb6cb18
GitHub-Last-Rev: 7512b33f055aa225d365d6c949a53778834e8dcd
GitHub-Pull-Request: golang/go#71966
Reviewed-on: https://go-review.googlesource.com/c/go/+/652675
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>

math/big: add tests for allocation during multiply

Test that big.Int.Mul reusing the same target is not allocating
temporary garbage during its computation. That code is going
to be modified in an upcoming CL.

Change-Id: I3ed55c06da030282233c29cd7af2a04f395dc7a2
Reviewed-on: https://go-review.googlesource.com/c/go/+/652056
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>

math/big: move multiplication to natmul.go

No code changes.

This CL moves the multiplication (and squaring) code into natmul.go,
in preparation for cleaning up Karatsuba and then adding Toom-Cook
and FFT-based multiplication.

Change-Id: I7f84328284cc4e1ca4da0ebb9f666a5535e8d7f2
Reviewed-on: https://go-review.googlesource.com/c/go/+/652055
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Russ Cox <rsc@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>

math/big: optimize atoi of base 2, 4, 16

Avoid multiplies when converting base 2, 4, 16 inputs,
reducing conversion time from O(N²) to O(N).

The Base8 and Base10 code paths should be unmodified,
but the base-2,4,16 changes tickle the compiler to generate
better (amd64) or worse (arm64) when really it should not.
This is described in detail in #71868 and should be ignored
for the purposes of this CL.

goos: linux
goarch: amd64
pkg: math/big
cpu: Intel(R) Xeon(R) CPU @ 3.10GHz
                      │     old      │                 new                 │
                      │    sec/op    │   sec/op     vs base                │
Scan/10/Base2-16         324.4n ± 0%   258.7n ± 0%  -20.25% (p=0.000 n=15)
Scan/100/Base2-16        2.376µ ± 0%   1.968µ ± 0%  -17.17% (p=0.000 n=15)
Scan/1000/Base2-16       23.89µ ± 0%   19.16µ ± 0%  -19.80% (p=0.000 n=15)
Scan/10000/Base2-16      311.5µ ± 0%   190.4µ ± 0%  -38.86% (p=0.000 n=15)
Scan/100000/Base2-16    10.508m ± 0%   1.904m ± 0%  -81.88% (p=0.000 n=15)
Scan/10/Base8-16         138.3n ± 0%   127.9n ± 0%   -7.52% (p=0.000 n=15)
Scan/100/Base8-16        886.1n ± 0%   790.2n ± 0%  -10.82% (p=0.000 n=15)
Scan/1000/Base8-16       9.227µ ± 0%   8.234µ ± 0%  -10.76% (p=0.000 n=15)
Scan/10000/Base8-16      165.8µ ± 0%   155.6µ ± 0%   -6.19% (p=0.000 n=15)
Scan/100000/Base8-16     9.044m ± 0%   8.935m ± 0%   -1.20% (p=0.000 n=15)
Scan/10/Base10-16        129.9n ± 0%   120.0n ± 0%   -7.62% (p=0.000 n=15)
Scan/100/Base10-16       816.3n ± 0%   730.0n ± 0%  -10.57% (p=0.000 n=15)
Scan/1000/Base10-16      8.518µ ± 0%   7.628µ ± 0%  -10.45% (p=0.000 n=15)
Scan/10000/Base10-16     158.6µ ± 0%   149.4µ ± 0%   -5.80% (p=0.000 n=15)
Scan/100000/Base10-16    8.962m ± 0%   8.855m ± 0%   -1.20% (p=0.000 n=15)
Scan/10/Base16-16        114.5n ± 0%   108.6n ± 0%   -5.15% (p=0.000 n=15)
Scan/100/Base16-16       648.3n ± 0%   525.0n ± 0%  -19.02% (p=0.000 n=15)
Scan/1000/Base16-16      7.375µ ± 0%   5.636µ ± 0%  -23.58% (p=0.000 n=15)
Scan/10000/Base16-16    171.18µ ± 0%   66.99µ ± 0%  -60.87% (p=0.000 n=15)
Scan/100000/Base16-16   9490.9µ ± 0%   682.8µ ± 0%  -92.81% (p=0.000 n=15)
geomean                  20.11µ        13.69µ       -31.94%

goos: linux
goarch: amd64
pkg: math/big
cpu: Intel(R) Xeon(R) Platinum 8481C CPU @ 2.70GHz
                      │      old      │                 new                 │
                      │    sec/op     │   sec/op     vs base                │
Scan/10/Base2-88          275.4n ± 0%   215.0n ± 0%  -21.93% (p=0.000 n=15)
Scan/100/Base2-88         1.869µ ± 0%   1.629µ ± 0%  -12.84% (p=0.000 n=15)
Scan/1000/Base2-88        18.56µ ± 0%   15.81µ ± 0%  -14.82% (p=0.000 n=15)
Scan/10000/Base2-88       270.0µ ± 0%   157.2µ ± 0%  -41.77% (p=0.000 n=15)
Scan/100000/Base2-88     11.518m ± 0%   1.571m ± 0%  -86.36% (p=0.000 n=15)
Scan/10/Base8-88          108.9n ± 0%   106.0n ± 0%   -2.66% (p=0.000 n=15)
Scan/100/Base8-88         655.2n ± 0%   594.9n ± 0%   -9.20% (p=0.000 n=15)
Scan/1000/Base8-88        6.467µ ± 0%   5.966µ ± 0%   -7.75% (p=0.000 n=15)
Scan/10000/Base8-88       151.2µ ± 0%   147.4µ ± 0%   -2.53% (p=0.000 n=15)
Scan/100000/Base8-88      10.33m ± 0%   10.30m ± 0%   -0.25% (p=0.000 n=15)
Scan/10/Base10-88        100.20n ± 0%   98.53n ± 0%   -1.67% (p=0.000 n=15)
Scan/100/Base10-88        596.9n ± 0%   543.3n ± 0%   -8.98% (p=0.000 n=15)
Scan/1000/Base10-88       5.904µ ± 0%   5.485µ ± 0%   -7.10% (p=0.000 n=15)
Scan/10000/Base10-88      145.7µ ± 0%   142.0µ ± 0%   -2.55% (p=0.000 n=15)
Scan/100000/Base10-88     10.26m ± 0%   10.24m ± 0%   -0.18% (p=0.000 n=15)
Scan/10/Base16-88         90.33n ± 0%   87.60n ± 0%   -3.02% (p=0.000 n=15)
Scan/100/Base16-88        506.4n ± 0%   437.7n ± 0%  -13.57% (p=0.000 n=15)
Scan/1000/Base16-88       5.056µ ± 0%   4.007µ ± 0%  -20.75% (p=0.000 n=15)
Scan/10000/Base16-88     163.35µ ± 0%   65.37µ ± 0%  -59.98% (p=0.000 n=15)
Scan/100000/Base16-88   11027.2µ ± 0%   735.1µ ± 0%  -93.33% (p=0.000 n=15)
geomean                   17.13µ        11.74µ       -31.46%

goos: linux
goarch: arm64
pkg: math/big
                      │     old      │                 new                  │
                      │    sec/op    │    sec/op     vs base                │
Scan/10/Base2-16         324.7n ± 0%    348.4n ± 0%   +7.30% (p=0.000 n=15)
Scan/100/Base2-16        2.604µ ± 0%    3.031µ ± 0%  +16.40% (p=0.000 n=15)
Scan/1000/Base2-16       26.15µ ± 0%    29.94µ ± 0%  +14.52% (p=0.000 n=15)
Scan/10000/Base2-16      334.3µ ± 0%    298.8µ ± 0%  -10.64% (p=0.000 n=15)
Scan/100000/Base2-16    10.664m ± 0%    2.991m ± 0%  -71.95% (p=0.000 n=15)
Scan/10/Base8-16         144.4n ± 1%    162.2n ± 1%  +12.33% (p=0.000 n=15)
Scan/100/Base8-16        917.2n ± 0%   1084.0n ± 0%  +18.19% (p=0.000 n=15)
Scan/1000/Base8-16       9.367µ ± 0%   10.901µ ± 0%  +16.38% (p=0.000 n=15)
Scan/10000/Base8-16      164.2µ ± 0%    181.2µ ± 0%  +10.34% (p=0.000 n=15)
Scan/100000/Base8-16     8.871m ± 1%    9.140m ± 0%   +3.04% (p=0.000 n=15)
Scan/10/Base10-16        134.6n ± 1%    148.3n ± 1%  +10.18% (p=0.000 n=15)
Scan/100/Base10-16       837.1n ± 0%    986.6n ± 0%  +17.86% (p=0.000 n=15)
Scan/1000/Base10-16      8.563µ ± 0%    9.936µ ± 0%  +16.03% (p=0.000 n=15)
Scan/10000/Base10-16     156.5µ ± 1%    171.3µ ± 0%   +9.41% (p=0.000 n=15)
Scan/100000/Base10-16    8.863m ± 1%    9.011m ± 0%   +1.66% (p=0.000 n=15)
Scan/10/Base16-16        115.7n ± 2%    129.1n ± 1%  +11.58% (p=0.000 n=15)
Scan/100/Base16-16       708.6n ± 0%    796.8n ± 0%  +12.45% (p=0.000 n=15)
Scan/1000/Base16-16      7.314µ ± 0%    7.554µ ± 0%   +3.28% (p=0.000 n=15)
Scan/10000/Base16-16    149.05µ ± 0%    74.60µ ± 0%  -49.95% (p=0.000 n=15)
Scan/100000/Base16-16   9091.6µ ± 0%    741.5µ ± 0%  -91.84% (p=0.000 n=15)
geomean                  20.39µ         17.65µ       -13.44%

goos: darwin
goarch: arm64
pkg: math/big
cpu: Apple M3 Pro
                      │     old      │                 new                 │
                      │    sec/op    │   sec/op     vs base                │
Scan/10/Base2-12         193.8n ± 2%   157.3n ± 1%  -18.83% (p=0.000 n=15)
Scan/100/Base2-12        1.445µ ± 2%   1.362µ ± 1%   -5.74% (p=0.000 n=15)
Scan/1000/Base2-12       14.28µ ± 0%   13.51µ ± 0%   -5.42% (p=0.000 n=15)
Scan/10000/Base2-12      177.1µ ± 0%   134.6µ ± 0%  -24.04% (p=0.000 n=15)
Scan/100000/Base2-12     5.429m ± 1%   1.333m ± 0%  -75.45% (p=0.000 n=15)
Scan/10/Base8-12         75.52n ± 2%   76.09n ± 1%        ~ (p=0.010 n=15)
Scan/100/Base8-12        528.4n ± 1%   532.1n ± 1%        ~ (p=0.003 n=15)
Scan/1000/Base8-12       5.423µ ± 1%   5.427µ ± 0%        ~ (p=0.183 n=15)
Scan/10000/Base8-12      89.26µ ± 1%   89.37µ ± 0%        ~ (p=0.237 n=15)
Scan/100000/Base8-12     4.543m ± 2%   4.560m ± 1%        ~ (p=0.595 n=15)
Scan/10/Base10-12        69.87n ± 1%   70.51n ± 0%        ~ (p=0.002 n=15)
Scan/100/Base10-12       488.4n ± 1%   491.2n ± 0%        ~ (p=0.060 n=15)
Scan/1000/Base10-12      5.014µ ± 1%   5.008µ ± 0%        ~ (p=0.783 n=15)
Scan/10000/Base10-12     84.90µ ± 0%   85.10µ ± 0%        ~ (p=0.109 n=15)
Scan/100000/Base10-12    4.516m ± 1%   4.521m ± 1%        ~ (p=0.713 n=15)
Scan/10/Base16-12        59.21n ± 1%   57.70n ± 1%   -2.55% (p=0.000 n=15)
Scan/100/Base16-12       380.0n ± 1%   360.7n ± 1%   -5.08% (p=0.000 n=15)
Scan/1000/Base16-12      3.775µ ± 0%   3.421µ ± 0%   -9.38% (p=0.000 n=15)
Scan/10000/Base16-12     80.62µ ± 0%   34.44µ ± 1%  -57.28% (p=0.000 n=15)
Scan/100000/Base16-12   4826.4µ ± 2%   450.9µ ± 2%  -90.66% (p=0.000 n=15)
geomean                  11.05µ        8.448µ       -23.52%

Change-Id: Ifdb2049545f34072aa75cdbb72bed4cf465f0ad7
Reviewed-on: https://go-review.googlesource.com/c/go/+/650640
Auto-Submit: Russ Cox <rsc@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>

math/big: improve scan test and benchmark

Add a few more test cases for scanning (integer conversion),
which were helpful in debugging some upcoming changes.

BenchmarkScan currently times converting the value 10**N
represented in base B back into []Word form.
When B = 10, the text is 1 followed by many zeros, which
could hit a "multiply by zero" special case when processing
many digit chunks, misrepresenting the actual time required
depending on whether that case is optimized.

Change the benchmark to use 9**N, which is about as big and
will not cause runs of zeros in any of the tested bases.

The benchmark comparison below is not showing faster code,
since of course the code is not changing at all here. Instead,
it is showing that the new benchmark work is roughly the same
size as the old benchmark work.

goos: darwin
goarch: arm64
pkg: math/big
cpu: Apple M3 Pro
                      │     old     │                new                 │
                      │   sec/op    │   sec/op     vs base               │
ScanPi-12               43.35µ ± 1%   43.59µ ± 1%       ~ (p=0.069 n=15)
Scan/10/Base2-12        202.3n ± 2%   193.7n ± 1%  -4.25% (p=0.000 n=15)
Scan/100/Base2-12       1.512µ ± 3%   1.447µ ± 1%  -4.30% (p=0.000 n=15)
Scan/1000/Base2-12      15.06µ ± 2%   14.33µ ± 0%  -4.83% (p=0.000 n=15)
Scan/10000/Base2-12     188.0µ ± 5%   177.3µ ± 1%  -5.65% (p=0.000 n=15)
Scan/100000/Base2-12    5.814m ± 3%   5.382m ± 1%  -7.43% (p=0.000 n=15)
Scan/10/Base8-12        78.57n ± 2%   75.02n ± 1%  -4.52% (p=0.000 n=15)
Scan/100/Base8-12       548.2n ± 2%   526.8n ± 1%  -3.90% (p=0.000 n=15)
Scan/1000/Base8-12      5.674µ ± 2%   5.421µ ± 0%  -4.46% (p=0.000 n=15)
Scan/10000/Base8-12     94.42µ ± 1%   88.61µ ± 1%  -6.15% (p=0.000 n=15)
Scan/100000/Base8-12    4.906m ± 2%   4.498m ± 3%  -8.31% (p=0.000 n=15)
Scan/10/Base10-12       73.42n ± 1%   69.56n ± 0%  -5.26% (p=0.000 n=15)
Scan/100/Base10-12      511.9n ± 1%   488.2n ± 0%  -4.63% (p=0.000 n=15)
Scan/1000/Base10-12     5.254µ ± 2%   5.009µ ± 0%  -4.66% (p=0.000 n=15)
Scan/10000/Base10-12    90.22µ ± 2%   84.52µ ± 0%  -6.32% (p=0.000 n=15)
Scan/100000/Base10-12   4.842m ± 3%   4.471m ± 3%  -7.65% (p=0.000 n=15)
Scan/10/Base16-12       62.28n ± 1%   58.70n ± 1%  -5.75% (p=0.000 n=15)
Scan/100/Base16-12      398.6n ± 0%   377.9n ± 1%  -5.19% (p=0.000 n=15)
Scan/1000/Base16-12     4.108µ ± 1%   3.782µ ± 0%  -7.94% (p=0.000 n=15)
Scan/10000/Base16-12    83.78µ ± 2%   80.51µ ± 1%  -3.90% (p=0.000 n=15)
Scan/100000/Base16-12   5.080m ± 3%   4.698m ± 3%  -7.53% (p=0.000 n=15)
geomean                 12.41µ        11.74µ       -5.36%

Change-Id: If3ce290ecc7f38672f11b42fd811afb53dee665d
Reviewed-on: https://go-review.googlesource.com/c/go/+/650639
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Russ Cox <rsc@golang.org>

math/big: replace nat pool with Word stack

In the early days of math/big, algorithms that needed more space
grew the result larger than it needed to be and then used the
high words as extra space. This made results their own temporary
space caches, at the cost that saving a result in a data structure
might hold significantly more memory than necessary.
Specifically, new(big.Int).Mul(x, y) returned a big.Int with a
backing slice 3X as big as it strictly needed to be.
If you are storing many multiplication results, or even a single
large result, the 3X overhead can add up.

This approach to storage for temporaries also requires being able
to analyze the algorithms to predict the exact amount they need,
which can be difficult.

For both these reasons, the implementation of recursive long division,
which came later, introduced a “nat pool” where temporaries could be
stored and reused, or reclaimed by the GC when no longer used.
This avoids the storage and bookkeeping overheads but introduces a
per-temporary sync.Pool overhead. divRecursiveStep takes an array
of cached temporaries to remove some of that overhead.
The nat pool was better but is still not quite right.

This CL introduces something even better than the nat pool
(still probably not quite right, but the best I can see for now):
a sync.Pool holding stacks for allocating temporaries.
Now an operation can get one stack out of the pool and then
allocate as many temporaries as it needs during the operation,
eventually returning the stack back to the pool. The sync.Pool
operations are now per-exported-operation (like big.Int.Mul),
not per-temporary.

This CL converts both the pre-allocation in nat.mul and the
uses of the nat pool to use stack pools instead. This simplifies
some code and sets us up better for more complex algorithms
(such as Toom-Cook or FFT-based multiplication) that need
more temporaries. It is also a little bit faster.

goos: linux
goarch: amd64
pkg: math/big
cpu: Intel(R) Xeon(R) CPU @ 3.10GHz
                         │     old     │                 new                 │
                         │   sec/op    │   sec/op     vs base                │
Div/20/10-16               23.68n ± 0%   22.21n ± 0%   -6.21% (p=0.000 n=15)
Div/40/20-16               23.68n ± 0%   22.21n ± 0%   -6.21% (p=0.000 n=15)
Div/100/50-16              56.65n ± 0%   55.53n ± 0%   -1.98% (p=0.000 n=15)
Div/200/100-16             194.6n ± 1%   172.8n ± 0%  -11.20% (p=0.000 n=15)
Div/400/200-16             232.1n ± 0%   206.7n ± 0%  -10.94% (p=0.000 n=15)
Div/1000/500-16            405.3n ± 1%   383.8n ± 0%   -5.30% (p=0.000 n=15)
Div/2000/1000-16           810.4n ± 1%   795.2n ± 0%   -1.88% (p=0.000 n=15)
Div/20000/10000-16         25.88µ ± 0%   25.39µ ± 0%   -1.89% (p=0.000 n=15)
Div/200000/100000-16       931.5µ ± 0%   924.3µ ± 0%   -0.77% (p=0.000 n=15)
Div/2000000/1000000-16     37.77m ± 0%   37.75m ± 0%        ~ (p=0.098 n=15)
Div/20000000/10000000-16    1.367 ± 0%    1.377 ± 0%   +0.72% (p=0.003 n=15)
NatMul/10-16               168.5n ± 3%   164.0n ± 4%        ~ (p=0.751 n=15)
NatMul/100-16              6.086µ ± 3%   5.380µ ± 3%  -11.60% (p=0.000 n=15)
NatMul/1000-16             238.1µ ± 3%   228.3µ ± 1%   -4.12% (p=0.000 n=15)
NatMul/10000-16            8.721m ± 2%   8.518m ± 1%   -2.33% (p=0.000 n=15)
NatMul/100000-16           369.6m ± 0%   371.1m ± 0%   +0.42% (p=0.000 n=15)
geomean                    19.57µ        18.74µ        -4.21%

                 │     old      │                  new                   │
                 │     B/op     │     B/op      vs base                  │
NatMul/10-16         192.0 ± 0%     192.0 ± 0%        ~ (p=1.000 n=15) ¹
NatMul/100-16      4.750Ki ± 0%   1.751Ki ± 0%  -63.14% (p=0.000 n=15)
NatMul/1000-16     48.16Ki ± 0%   16.02Ki ± 0%  -66.73% (p=0.000 n=15)
NatMul/10000-16    482.9Ki ± 1%   165.4Ki ± 3%  -65.75% (p=0.000 n=15)
NatMul/100000-16   5.747Mi ± 7%   4.197Mi ± 0%  -26.97% (p=0.000 n=15)
geomean            41.42Ki        20.63Ki       -50.18%
¹ all samples are equal

                 │     old     │                 new                  │
                 │  allocs/op  │  allocs/op   vs base                 │
NatMul/10-16       1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/100-16      1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/1000-16     1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/10000-16    1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/100000-16   7.000 ± 14%   7.000 ± 14%       ~ (p=0.668 n=15)
geomean            1.476         1.476        +0.00%
¹ all samples are equal

goos: linux
goarch: amd64
pkg: math/big
cpu: Intel(R) Xeon(R) Platinum 8481C CPU @ 2.70GHz
                         │     old     │                 new                 │
                         │   sec/op    │   sec/op     vs base                │
Div/20/10-88               15.84n ± 1%   13.12n ± 0%  -17.17% (p=0.000 n=15)
Div/40/20-88               15.88n ± 1%   13.12n ± 0%  -17.38% (p=0.000 n=15)
Div/100/50-88              26.42n ± 0%   25.47n ± 0%   -3.60% (p=0.000 n=15)
Div/200/100-88             132.4n ± 0%   114.9n ± 0%  -13.22% (p=0.000 n=15)
Div/400/200-88             150.1n ± 0%   135.6n ± 0%   -9.66% (p=0.000 n=15)
Div/1000/500-88            275.5n ± 0%   264.1n ± 0%   -4.14% (p=0.000 n=15)
Div/2000/1000-88           586.5n ± 0%   581.1n ± 0%   -0.92% (p=0.000 n=15)
Div/20000/10000-88         25.87µ ± 0%   25.72µ ± 0%   -0.59% (p=0.000 n=15)
Div/200000/100000-88       772.2µ ± 0%   779.0µ ± 0%   +0.88% (p=0.000 n=15)
Div/2000000/1000000-88     33.36m ± 0%   33.63m ± 0%   +0.80% (p=0.000 n=15)
Div/20000000/10000000-88    1.307 ± 0%    1.320 ± 0%   +1.03% (p=0.000 n=15)
NatMul/10-88               140.4n ± 0%   148.8n ± 4%   +5.98% (p=0.000 n=15)
NatMul/100-88              4.663µ ± 1%   4.388µ ± 1%   -5.90% (p=0.000 n=15)
NatMul/1000-88             207.7µ ± 0%   205.8µ ± 0%   -0.89% (p=0.000 n=15)
NatMul/10000-88            8.456m ± 0%   8.468m ± 0%   +0.14% (p=0.021 n=15)
NatMul/100000-88           295.1m ± 0%   297.9m ± 0%   +0.94% (p=0.000 n=15)
geomean                    14.96µ        14.33µ        -4.23%

                 │     old      │                   new                   │
                 │     B/op     │     B/op       vs base                  │
NatMul/10-88         192.0 ± 0%     192.0 ±  0%        ~ (p=1.000 n=15) ¹
NatMul/100-88      4.750Ki ± 0%   1.758Ki ±  0%  -62.99% (p=0.000 n=15)
NatMul/1000-88     48.44Ki ± 0%   16.08Ki ±  0%  -66.80% (p=0.000 n=15)
NatMul/10000-88    489.7Ki ± 1%   166.1Ki ±  3%  -66.08% (p=0.000 n=15)
NatMul/100000-88   5.546Mi ± 0%   3.819Mi ± 60%  -31.15% (p=0.000 n=15)
geomean            41.29Ki        20.30Ki        -50.85%
¹ all samples are equal

                 │     old     │                 new                  │
                 │  allocs/op  │  allocs/op   vs base                 │
NatMul/10-88       1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/100-88      1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/1000-88     1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/10000-88    1.000 ±  0%   1.000 ±  0%       ~ (p=1.000 n=15) ¹
NatMul/100000-88   5.000 ± 20%   6.000 ± 67%       ~ (p=0.672 n=15)
geomean            1.380         1.431        +3.71%
¹ all samples are equal

goos: linux
goarch: arm64
pkg: math/big
                         │     old     │                 new                 │
                         │   sec/op    │   sec/op     vs base                │
Div/20/10-16               15.85n ± 0%   15.23n ± 0%   -3.91% (p=0.000 n=15)
Div/40/20-16               15.88n ± 0%   15.22n ± 0%   -4.16% (p=0.000 n=15)
Div/100/50-16              29.69n ± 0%   26.39n ± 0%  -11.11% (p=0.000 n=15)
Div/200/100-16             149.2n ± 0%   123.3n ± 0%  -17.36% (p=0.000 n=15)
Div/400/200-16             160.3n ± 0%   139.2n ± 0%  -13.16% (p=0.000 n=15)
Div/1000/500-16            271.0n ± 0%   256.1n ± 0%   -5.50% (p=0.000 n=15)
Div/2000/1000-16           545.3n ± 0%   527.0n ± 0%   -3.36% (p=0.000 n=15)
Div/20000/10000-16         22.60µ ± 0%   22.20µ ± 0%   -1.77% (p=0.000 n=15)
Div/200000/100000-16       889.0µ ± 0%   892.2µ ± 0%   +0.35% (p=0.000 n=15)
Div/2000000/1000000-16     38.01m ± 0%   38.12m ± 0%   +0.30% (p=0.000 n=15)
Div/20000000/10000000-16    1.437 ± 0%    1.444 ± 0%   +0.50% (p=0.000 n=15)
NatMul/10-16               166.4n ± 2%   169.5n ± 1%   +1.86% (p=0.000 n=15)
NatMul/100-16              5.733µ ± 1%   5.570µ ± 1%   -2.84% (p=0.000 n=15)
NatMul/1000-16             232.6µ ± 1%   229.8µ ± 0%   -1.22% (p=0.000 n=15)
NatMul/10000-16            9.039m ± 1%   8.969m ± 0%   -0.77% (p=0.000 n=15)
NatMul/100000-16           367.0m ± 0%   368.8m ± 0%   +0.48% (p=0.000 n=15)
geomean                    16.15µ        15.50µ        -4.01%

                 │     old      │                  new                   │
                 │     B/op     │     B/op      vs base                  │
NatMul/10-16         192.0 ± 0%     192.0 ± 0%        ~ (p=1.000 n=15) ¹
NatMul/100-16      4.750Ki ± 0%   1.751Ki ± 0%  -63.14% (p=0.000 n=15)
NatMul/1000-16     48.33Ki ± 0%   16.02Ki ± 0%  -66.85% (p=0.000 n=15)
NatMul/10000-16    536.5Ki ± 1%   165.7Ki ± 3%  -69.12% (p=0.000 n=15)
NatMul/100000-16   6.078Mi ± 6%   4.197Mi ± 0%  -30.94% (p=0.000 n=15)
geomean            42.81Ki        20.64Ki       -51.78%
¹ all samples are equal

                 │     old     │                  new                  │
                 │  allocs/op  │  allocs/op   vs base                  │
NatMul/10-16       1.000 ±  0%   1.000 ±  0%        ~ (p=1.000 n=15) ¹
NatMul/100-16      1.000 ±  0%   1.000 ±  0%        ~ (p=1.000 n=15) ¹
NatMul/1000-16     1.000 ±  0%   1.000 ±  0%        ~ (p=1.000 n=15) ¹
NatMul/10000-16    2.000 ± 50%   1.000 ±  0%  -50.00% (p=0.001 n=15)
NatMul/100000-16   9.000 ± 11%   8.000 ± 12%  -11.11% (p=0.001 n=15)
geomean            1.783         1.516        -14.97%
¹ all samples are equal

goos: darwin
goarch: arm64
pkg: math/big
cpu: Apple M3 Pro
                         │     old      │                new                 │
                         │    sec/op    │   sec/op     vs base               │
Div/20/10-12                9.850n ± 1%   9.405n ± 1%  -4.52% (p=0.000 n=15)
Div/40/20-12                9.858n ± 0%   9.403n ± 1%  -4.62% (p=0.000 n=15)
Div/100/50-12               16.40n ± 1%   14.81n ± 0%  -9.70% (p=0.000 n=15)
Div/200/100-12              88.48n ± 2%   80.88n ± 0%  -8.59% (p=0.000 n=15)
Div/400/200-12             107.90n ± 1%   99.28n ± 1%  -7.99% (p=0.000 n=15)
Div/1000/500-12             188.8n ± 1%   178.6n ± 1%  -5.40% (p=0.000 n=15)
Div/2000/1000-12            399.9n ± 0%   389.1n ± 0%  -2.70% (p=0.000 n=15)
Div/20000/10000-12          13.94µ ± 2%   13.81µ ± 1%       ~ (p=0.574 n=15)
Div/200000/100000-12        523.8µ ± 0%   521.7µ ± 0%  -0.40% (p=0.000 n=15)
Div/2000000/1000000-12      21.46m ± 0%   21.48m ± 0%       ~ (p=0.067 n=15)
Div/20000000/10000000-12    812.5m ± 0%   812.9m ± 0%       ~ (p=0.061 n=15)
NatMul/10-12                77.14n ± 0%   78.35n ± 1%  +1.57% (p=0.000 n=15)
NatMul/100-12               2.999µ ± 0%   2.871µ ± 1%  -4.27% (p=0.000 n=15)
NatMul/1000-12              126.2µ ± 0%   126.8µ ± 0%  +0.51% (p=0.011 n=15)
NatMul/10000-12             5.099m ± 0%   5.125m ± 0%  +0.51% (p=0.000 n=15)
NatMul/100000-12            206.7m ± 0%   208.4m ± 0%  +0.80% (p=0.000 n=15)
geomean                     9.512µ        9.236µ       -2.91%

                 │     old      │                   new                    │
                 │     B/op     │      B/op       vs base                  │
NatMul/10-12         192.0 ± 0%     192.0 ±   0%        ~ (p=1.000 n=15) ¹
NatMul/100-12      4.750Ki ± 0%   1.750Ki ±   0%  -63.16% (p=0.000 n=15)
NatMul/1000-12     48.13Ki ± 0%   16.01Ki ±   0%  -66.73% (p=0.000 n=15)
NatMul/10000-12    483.5Ki ± 1%   163.2Ki ±   2%  -66.24% (p=0.000 n=15)
NatMul/100000-12   5.480Mi ± 4%   1.532Mi ± 104%  -72.05% (p=0.000 n=15)
geomean            41.03Ki        16.82Ki         -59.01%
¹ all samples are equal

                 │    old     │                  new                   │
                 │ allocs/op  │  allocs/op    vs base                  │
NatMul/10-12       1.000 ± 0%   1.000 ±   0%        ~ (p=1.000 n=15) ¹
NatMul/100-12      1.000 ± 0%   1.000 ±   0%        ~ (p=1.000 n=15) ¹
NatMul/1000-12     1.000 ± 0%   1.000 ±   0%        ~ (p=1.000 n=15) ¹
NatMul/10000-12    1.000 ± 0%   1.000 ±   0%        ~ (p=1.000 n=15) ¹
NatMul/100000-12   5.000 ± 0%   1.000 ± 400%  -80.00% (p=0.007 n=15)
geomean            1.380        1.000         -27.52%
¹ all samples are equal

Change-Id: I7efa6fe37971ed26ae120a32250fcb47ece0a011
Reviewed-on: https://go-review.googlesource.com/c/go/+/650638
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>

math/big: report allocs in BenchmarkNatMul, BenchmarkNatSqr

Change-Id: I112f55c0e3ee3b75e615a06b27552de164565c04
Reviewed-on: https://go-review.googlesource.com/c/go/+/650637
Reviewed-by: Robert Griesemer <gri@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: Russ Cox <rsc@golang.org>

math/big: clean up GCD a little

The GCD code was setting one *Int to the value of another
by smashing one struct on top of the other, instead of using Set.
That was safe in this one case, but it's not idiomatic in math/big
nor safe in general, so rewrite the code not to do that.
(In one case, by swapping variables around; in another, by calling Set.)

The added Set call does slow down GCDs by a small amount,
since the answer has to be copied out. To compensate for that,
optimize a bit: remove the s, t temporaries entirely and handle
vector x word multiplication directly. The net result is that almost
all GCDs are faster, except for small ones, which are a few
nanoseconds slower.

goos: darwin
goarch: arm64
pkg: math/big
cpu: Apple M3 Pro
                              │ bench.before │             bench.after             │
                              │    sec/op    │   sec/op     vs base                │
GCD10x10/WithoutXY-12            23.80n ± 1%   31.71n ± 1%  +33.24% (p=0.000 n=10)
GCD10x10/WithXY-12              100.40n ± 0%   92.14n ± 1%   -8.22% (p=0.000 n=10)
GCD10x100/WithoutXY-12           63.70n ± 0%   70.73n ± 0%  +11.05% (p=0.000 n=10)
GCD10x100/WithXY-12              278.6n ± 0%   233.1n ± 1%  -16.35% (p=0.000 n=10)
GCD10x1000/WithoutXY-12          153.4n ± 0%   162.2n ± 1%   +5.74% (p=0.000 n=10)
GCD10x1000/WithXY-12             456.0n ± 0%   411.8n ± 1%   -9.69% (p=0.000 n=10)
GCD10x10000/WithoutXY-12         1.002µ ± 1%   1.036µ ± 0%   +3.39% (p=0.000 n=10)
GCD10x10000/WithXY-12            2.330µ ± 1%   2.210µ ± 0%   -5.13% (p=0.000 n=10)
GCD10x100000/WithoutXY-12        8.894µ ± 0%   8.889µ ± 1%        ~ (p=0.754 n=10)
GCD10x100000/WithXY-12           20.84µ ± 0%   20.24µ ± 0%   -2.84% (p=0.000 n=10)
GCD100x100/WithoutXY-12          373.3n ± 3%   314.4n ± 0%  -15.76% (p=0.000 n=10)
GCD100x100/WithXY-12             662.5n ± 0%   572.4n ± 1%  -13.59% (p=0.000 n=10)
GCD100x1000/WithoutXY-12         641.8n ± 0%   598.1n ± 1%   -6.81% (p=0.000 n=10)
GCD100x1000/WithXY-12            1.123µ ± 0%   1.019µ ± 1%   -9.26% (p=0.000 n=10)
GCD100x10000/WithoutXY-12        2.870µ ± 0%   2.831µ ± 0%   -1.38% (p=0.000 n=10)
GCD100x10000/WithXY-12           4.930µ ± 1%   4.675µ ± 0%   -5.16% (p=0.000 n=10)
GCD100x100000/WithoutXY-12       24.08µ ± 0%   23.97µ ± 0%   -0.48% (p=0.007 n=10)
GCD100x100000/WithXY-12          43.66µ ± 0%   42.52µ ± 0%   -2.61% (p=0.001 n=10)
GCD1000x1000/WithoutXY-12        3.999µ ± 0%   3.569µ ± 1%  -10.75% (p=0.000 n=10)
GCD1000x1000/WithXY-12           6.397µ ± 0%   5.534µ ± 0%  -13.49% (p=0.000 n=10)
GCD1000x10000/WithoutXY-12       6.875µ ± 0%   6.450µ ± 0%   -6.18% (p=0.000 n=10)
GCD1000x10000/WithXY-12          20.75µ ± 1%   19.17µ ± 1%   -7.64% (p=0.000 n=10)
GCD1000x100000/WithoutXY-12      36.38µ ± 0%   35.60µ ± 1%   -2.13% (p=0.000 n=10)
GCD1000x100000/WithXY-12         172.1µ ± 0%   174.4µ ± 3%        ~ (p=0.052 n=10)
GCD10000x10000/WithoutXY-12      79.89µ ± 1%   75.16µ ± 2%   -5.92% (p=0.000 n=10)
GCD10000x10000/WithXY-12         160.1µ ± 0%   150.0µ ± 0%   -6.33% (p=0.000 n=10)
GCD10000x100000/WithoutXY-12     213.2µ ± 1%   209.0µ ± 1%   -1.98% (p=0.000 n=10)
GCD10000x100000/WithXY-12        1.399m ± 0%   1.342m ± 3%   -4.08% (p=0.002 n=10)
GCD100000x100000/WithoutXY-12    5.463m ± 1%   5.504m ± 2%        ~ (p=0.190 n=10)
GCD100000x100000/WithXY-12       11.36m ± 0%   11.46m ± 1%   +0.86% (p=0.000 n=10)
geomean                          6.953µ        6.695µ        -3.71%

goos: linux
goarch: amd64
pkg: math/big
cpu: AMD Ryzen 9 7950X 16-Core Processor
                              │ bench.before │             bench.after             │
                              │    sec/op    │   sec/op     vs base                │
GCD10x10/WithoutXY-32           39.66n ±  4%   44.34n ± 4%  +11.77% (p=0.000 n=10)
GCD10x10/WithXY-32              156.7n ± 12%   130.8n ± 2%  -16.53% (p=0.000 n=10)
GCD10x100/WithoutXY-32          115.8n ±  5%   120.2n ± 2%   +3.89% (p=0.000 n=10)
GCD10x100/WithXY-32             465.3n ±  3%   368.1n ± 2%  -20.91% (p=0.000 n=10)
GCD10x1000/WithoutXY-32         201.1n ±  1%   210.8n ± 2%   +4.82% (p=0.000 n=10)
GCD10x1000/WithXY-32            652.9n ±  4%   605.0n ± 1%   -7.32% (p=0.002 n=10)
GCD10x10000/WithoutXY-32        1.046µ ±  2%   1.143µ ± 1%   +9.33% (p=0.000 n=10)
GCD10x10000/WithXY-32           3.360µ ±  1%   3.258µ ± 1%   -3.04% (p=0.000 n=10)
GCD10x100000/WithoutXY-32       9.391µ ±  3%   9.997µ ± 1%   +6.46% (p=0.000 n=10)
GCD10x100000/WithXY-32          27.92µ ±  1%   28.21µ ± 0%   +1.04% (p=0.043 n=10)
GCD100x100/WithoutXY-32         443.7n ±  5%   320.0n ± 2%  -27.88% (p=0.000 n=10)
GCD100x100/WithXY-32            789.9n ±  2%   690.4n ± 1%  -12.60% (p=0.000 n=10)
GCD100x1000/WithoutXY-32        718.4n ±  3%   600.0n ± 1%  -16.48% (p=0.000 n=10)
GCD100x1000/WithXY-32           1.388µ ±  4%   1.175µ ± 1%  -15.28% (p=0.000 n=10)
GCD100x10000/WithoutXY-32       2.750µ ±  1%   2.668µ ± 1%   -2.96% (p=0.000 n=10)
GCD100x10000/WithXY-32          6.016µ ±  1%   5.590µ ± 1%   -7.09% (p=0.000 n=10)
GCD100x100000/WithoutXY-32      21.40µ ±  1%   22.30µ ± 1%   +4.21% (p=0.000 n=10)
GCD100x100000/WithXY-32         47.02µ ±  4%   48.80µ ± 0%   +3.78% (p=0.015 n=10)
GCD1000x1000/WithoutXY-32       3.417µ ±  4%   3.020µ ± 1%  -11.65% (p=0.000 n=10)
GCD1000x1000/WithXY-32          5.752µ ±  0%   5.418µ ± 2%   -5.81% (p=0.000 n=10)
GCD1000x10000/WithoutXY-32      6.150µ ±  0%   6.246µ ± 1%   +1.55% (p=0.000 n=10)
GCD1000x10000/WithXY-32         24.68µ ±  3%   25.07µ ± 1%        ~ (p=0.051 n=10)
GCD1000x100000/WithoutXY-32     34.60µ ±  2%   36.85µ ± 1%   +6.51% (p=0.000 n=10)
GCD1000x100000/WithXY-32        209.5µ ±  4%   227.4µ ± 0%   +8.56% (p=0.000 n=10)
GCD10000x10000/WithoutXY-32     90.69µ ±  0%   88.48µ ± 0%   -2.44% (p=0.000 n=10)
GCD10000x10000/WithXY-32        197.1µ ±  0%   200.5µ ± 0%   +1.73% (p=0.000 n=10)
GCD10000x100000/WithoutXY-32    239.1µ ±  0%   242.5µ ± 0%   +1.42% (p=0.000 n=10)
GCD10000x100000/WithXY-32       1.963m ±  3%   2.028m ± 0%   +3.28% (p=0.000 n=10)
GCD100000x100000/WithoutXY-32   7.466m ±  0%   7.412m ± 0%   -0.71% (p=0.000 n=10)
GCD100000x100000/WithXY-32      16.10m ±  2%   16.47m ± 0%   +2.25% (p=0.000 n=10)
geomean                         8.388µ         8.127µ        -3.12%

Change-Id: I161dc409bad11bcc553bc8116449905ae5b06742
Reviewed-on: https://go-review.googlesource.com/c/go/+/650636
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Russ Cox <rsc@golang.org>

cmd/internal/obj/riscv: implement vector load/store instructions

Implement vector unit stride, vector strided, vector indexed and
vector whole register load and store instructions.

The vector unit stride instructions take an optional vector mask
register, which if specified must be register V0. If only two
operands are given, the instruction is encoded as unmasked.

The vector strided and vector indexed instructions also take an
optional vector mask register, which if specified must be register
V0. If only three operands are given, the instruction is encoded as
unmasked.

Cq-Include-Trybots: luci.golang.try:gotip-linux-riscv64
Change-Id: I35e43bb8f1cf6ae8826fbeec384b95ac945da50f
Reviewed-on: https://go-review.googlesource.com/c/go/+/631937
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Pengcheng Wang <wangpengcheng.pp@bytedance.com>

cmd/compile: simplify intrinsification of TrailingZeros16 and TrailingZeros8

Decompose Ctz16 and Ctz8 within the SSA rules for LOONG64, MIPS, PPC64
and S390X, rather than having a custom intrinsic. Note that for PPC64 this
actually allows the existing Ctz16 and Ctz8 rules to be used.

Change-Id: I27a5e978f852b9d75396d2a80f5d7dfcb5ef7dd4
Reviewed-on: https://go-review.googlesource.com/c/go/+/651816
Reviewed-by: Paul Murphy <murp@ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>

runtime/cgo: use standard ABI call setg_gcc in crosscall1 on loong64

Change-Id: Ie38583d667d579751d643b2da2aa56390b69904c
Reviewed-on: https://go-review.googlesource.com/c/go/+/652255
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Meidan Li <limeidan@loongson.cn>

runtime: use correct memory barrier in exitThread function on loong64

In the runtime.exitThread function, a storeRelease barrier
is required instead of a full barrier.

Change-Id: I2815ddb03e4984c891d71811ccf650a82325e10d
Reviewed-on: https://go-review.googlesource.com/c/go/+/631915
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

runtime: remove ret field from gobuf

It's not used for anything.

Change-Id: I031b3cdfe52b6b1cff4b3cb6713ffe588084542f
Reviewed-on: https://go-review.googlesource.com/c/go/+/652276
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime/cgo: avoid errors from -Wdeclaration-after-statement

CL 652181 accidentally missed this iPhone only code.

For #71961

Change-Id: I567f8bb38958907442e69494da330d5199d11f54
Reviewed-on: https://go-review.googlesource.com/c/go/+/653135
Auto-Submit: Ian Lance Taylor <iant@google.com>
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>
Reviewed-by: Michael Pratt <mpratt@google.com>

net: return proper error from Context

Sadly err was a named parameter so this did not cause
compile error.

Fixes #71974

Change-Id: I10cf29ae14c52d48a793c9a6cb01b01d79b1b356
GitHub-Last-Rev: 4dc0e6670a9265612b8ec26dbc378219b25156b4
GitHub-Pull-Request: golang/go#71976
Reviewed-on: https://go-review.googlesource.com/c/go/+/652815
Auto-Submit: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Commit-Queue: Ian Lance Taylor <iant@google.com>

cmd/link: require cgo for -linkmode=external test

For #71416
Fixes #71957

Change-Id: I2180dada34d9dd2d3f5b0aaf8525951fd2e86a27
Reviewed-on: https://go-review.googlesource.com/c/go/+/652277
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Quim Muntal <quimmuntal@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>

crypto/internal/fips140/edwards25519/field: optimize AMD64

Replace constant multiplication with shift and adds,
this reduces pressure on multiplications, making things overall
faster.

goos: windows
goarch: amd64
pkg: crypto/internal/fips140/edwards25519/field
cpu: AMD Ryzen Threadripper 2950X 16-Core Processor
            │   v0.log~   │              v1.log~               │
            │   sec/op    │   sec/op     vs base               │
Add-32        4.768n ± 1%   4.763n ± 0%       ~ (p=0.683 n=20)
Multiply-32   20.93n ± 0%   19.48n ± 0%  -6.88% (p=0.000 n=20)
Square-32     15.88n ± 0%   15.00n ± 0%  -5.51% (p=0.000 n=20)
Invert-32     4.291µ ± 0%   4.072µ ± 0%  -5.10% (p=0.000 n=20)
Mult32-32     5.184n ± 0%   5.169n ± 0%  -0.30% (p=0.032 n=20)
Bytes-32      11.36n ± 0%   11.34n ± 0%       ~ (p=0.106 n=20)
geomean       27.15n        26.32n       -3.06%

Change-Id: I9c2f588fad29d89c3e6c712c092b32b66479f596
Reviewed-on: https://go-review.googlesource.com/c/go/+/652716
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Commit-Queue: Junyang Shao <shaojunyang@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>

crypto/internal/fips140/edwards25519/field/_asm: update avo dependency

v0.4.0 does not work with Go 1.25.

Change-Id: Ib9081b0ee78cad7974b038d89fa92d9ccbfa305a
Reviewed-on: https://go-review.googlesource.com/c/go/+/652715
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
Commit-Queue: Junyang Shao <shaojunyang@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

sync: don't keep func alive after OnceFunc panics

This moves the f = nil assignment to the defer statement,
so that in case the functions panics, the f func is not
referenced anymore.

Change-Id: I3e53b90a10f21741e26602270822c8a75679f163
GitHub-Last-Rev: bda01100c6d48d1b0ca3e1baefef4d592cca1fee
GitHub-Pull-Request: golang/go#68636
Reviewed-on: https://go-review.googlesource.com/c/go/+/601240
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

runtime: document that cleanups can run concurrently with each other

Fixes #71825.

Change-Id: I25af19eb72d75f13cf661fc47ee5717782785326
Reviewed-on: https://go-review.googlesource.com/c/go/+/650696
Reviewed-by: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

runtime/cgo: avoid errors from -Wdeclaration-after-statement

It's used by the SWIG CI build, at least, and it's an easy fix.

Fixes #71961

Change-Id: Id21071a5aef216b35ecf0e9cd3e05d08972d92fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/652181
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Pratt <mpratt@google.com>

reflect: let Value.Seq return the iteration value correct type

Fixes #71905

Change-Id: I50a418f8552e071c6e5011af5b9accc7d41548d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/651855
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@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

cmd/compile: don't pull constant offsets out of pointer arithmetic

This could lead to manufacturing a pointer that points outside
its original allocation.

Bug was introduced in CL 629858.

Fixes #71932

Change-Id: Ia86ab0b65ce5f80a8e0f4f4c81babd07c5904f8d
Reviewed-on: https://go-review.googlesource.com/c/go/+/652078
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>

runtime: Added usage example for the runtime.AddCleanup() function.

The existing description of the function lacks usage examples, which makes it difficult to understand, so I added one.

There is no open issue about this, since the implementation seems trivial.

Change-Id: I96b29f0b21d1c7fda04128239633c8a2fc36fef2
Reviewed-on: https://go-review.googlesource.com/c/go/+/649995
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Auto-Submit: Carlos Amedee <carlos@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile/internal/test: fix noopt builder

The function argument passed to hash function escaped to heap when
optimization is disabled, causing the builder failed.

To fix this, skip the test on noopt builder.

Updates #71943
Fixes #71965

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-noopt
Change-Id: I3a9ece09bfa10bf5eb102a7da3ade65634565cb4
Reviewed-on: https://go-review.googlesource.com/c/go/+/652735
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: Cuong Manh Le <cuong.manhle.vn@gmail.com>

cmd/compile: simplify intrinsification of BitLen16 and BitLen8

Decompose BitLen16 and BitLen8 within the SSA rules for architectures that
support BitLen32 or BitLen64, rather than having a custom intrinsic.

Change-Id: Ie4188ce69d1021e63cec27a8e7418efb0714812b
Reviewed-on: https://go-review.googlesource.com/c/go/+/651817
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Junyang Shao <shaojunyang@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

runtime: route calls to msan_memmove through cgo

This avoids problems when the C linker doesn't want to see the Go relocation.

Fixes #71954

Change-Id: I7cf884c4059d596cad6074ade02020d5a724f20e
Reviewed-on: https://go-review.googlesource.com/c/go/+/652180
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>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>

go/scanner: report specific error for UCS-2 encoded files

Windows text files may be encoded as UCS-2 (i.e. 2-byte UTF-16).
This CL causes the scanner to emit a better error when it reads
a file in this encoding.

+ test

Fixes #71950

Change-Id: Ia65bbf9a60e36984b0f3e4865591aa6978d2bde2
Reviewed-on: https://go-review.googlesource.com/c/go/+/652515
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Robert Griesemer <gri@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Alan Donovan <adonovan@google.com>
Commit-Queue: Alan Donovan <adonovan@google.com>

cmd/compile: document -embedcfg flag

Fixes #71942

Change-Id: Ie7e795506a9c8781f0e0963012233a7ed1093855
Reviewed-on: https://go-review.googlesource.com/c/go/+/652475
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>

cmd/compile, runtime: optimize concatbytes

CL 527935 optimized []byte(string1 + string2) to use runtime.concatbytes
to prevent concatenating of strings before converting to slices.
However, the optimization is implemented without allowing temporary
buffer for slice on stack, causing un-necessary allocations.

To fix this, optimize concatbytes to use temporary buffer if the result
string length fit to the buffer size.

Fixes #71943

Change-Id: I1d3c374cd46aad8f83a271b8a5ca79094f9fd8db
Reviewed-on: https://go-review.googlesource.com/c/go/+/652395
Reviewed-by: Keith Randall <khr@google.com>
Auto-Submit: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>

os, syscall: use unix build tag where appropriate

These newly added files may use the unix build tag instead of explitly
listing all unix-like GOOS values.

For #51572

Change-Id: I31c71d2b5533b39bbccd89bf616a99b8e33565d3
Reviewed-on: https://go-review.googlesource.com/c/go/+/651996
Auto-Submit: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

crypto/internal/fips140/edwards25519/field: inline carryPropagate

Manually inlining carryPropagate seems to help quite a bit.

goos: linux
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
                              │     OLD     │                NEW                 │
                              │   sec/op    │   sec/op     vs base               │
EncodingDecoding-4              141.4µ ± 0%   133.1µ ± 0%  -5.90% (p=0.000 n=10)
ScalarBaseMult-4                260.5µ ± 0%   254.0µ ± 0%  -2.49% (p=0.000 n=10)
ScalarMult-4                    916.6µ ± 0%   892.5µ ± 0%  -2.63% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-4   909.6µ ± 0%   886.6µ ± 0%  -2.52% (p=0.000 n=10)
geomean                         418.6µ        404.4µ       -3.40%

pkg: crypto/internal/fips140/edwards25519/field
           │     OLD     │                NEW                 │
           │   sec/op    │   sec/op     vs base               │
Add-4        33.50n ± 0%   33.52n ± 0%  +0.04% (p=0.013 n=10)
Multiply-4   406.8n ± 0%   400.0n ± 0%  -1.66% (p=0.000 n=10)
Square-4     246.4n ± 0%   234.4n ± 0%  -4.85% (p=0.000 n=10)
Invert-4     67.37µ ± 0%   63.53µ ± 0%  -5.69% (p=0.000 n=10)
Mult32-4     78.68n ± 0%   78.67n ± 0%       ~ (p=0.367 n=10)
Bytes-4      110.6n ± 0%   110.8n ± 0%       ~ (p=0.568 n=10)
geomean      354.0n        346.8n       -2.03%

goos: darwin
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
cpu: Apple M1 Pro
                               │     OLD     │                NEW                 │
                               │   sec/op    │   sec/op     vs base               │
EncodingDecoding-10              7.610µ ± 2%   7.459µ ± 0%  -1.98% (p=0.000 n=10)
ScalarBaseMult-10                11.54µ ± 0%   11.36µ ± 0%  -1.53% (p=0.000 n=10)
ScalarMult-10                    38.59µ ± 2%   37.09µ ± 0%  -3.88% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-10   37.10µ ± 0%   35.79µ ± 0%  -3.54% (p=0.000 n=10)
geomean                          18.83µ        18.31µ       -2.74%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                NEW                 │
            │   sec/op    │   sec/op     vs base               │
Add-10        5.116n ± 5%   5.168n ± 5%       ~ (p=0.669 n=10)
Multiply-10   18.00n ± 2%   16.90n ± 1%  -6.09% (p=0.000 n=10)
Square-10     13.66n ± 0%   13.48n ± 0%  -1.28% (p=0.000 n=10)
Invert-10     3.629µ ± 0%   3.508µ ± 2%  -3.33% (p=0.000 n=10)
Mult32-10     6.165n ± 2%   6.324n ± 1%  +2.57% (p=0.000 n=10)
Bytes-10      10.33n ± 8%   10.28n ± 4%       ~ (p=0.516 n=10)
geomean       25.74n        25.40n       -1.31%

tags: purego
goos: windows
goarch: amd64
pkg: crypto/internal/fips140/edwards25519
cpu: AMD Ryzen Threadripper 2950X 16-Core Processor
                               │     OLD     │                NEW                 │
                               │   sec/op    │   sec/op     vs base               │
EncodingDecoding-32              9.557µ ± 1%   8.966µ ± 0%  -6.18% (p=0.000 n=10)
ScalarBaseMult-32                19.14µ ± 2%   19.00µ ± 1%       ~ (p=0.190 n=10)
ScalarMult-32                    64.61µ ± 1%   65.83µ ± 2%  +1.88% (p=0.003 n=10)
VarTimeDoubleScalarBaseMult-32   62.36µ ± 1%   62.14µ ± 1%       ~ (p=0.529 n=10)
geomean                          29.30µ        28.89µ       -1.39%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                NEW                 │
            │   sec/op    │   sec/op     vs base               │
Add-32        4.879n ± 1%   4.880n ± 1%       ~ (p=0.739 n=10)
Multiply-32   22.75n ± 2%   22.03n ± 3%  -3.14% (p=0.000 n=10)
Square-32     16.46n ± 2%   15.38n ± 1%  -6.59% (p=0.000 n=10)
Invert-32     4.466µ ± 1%   4.168µ ± 1%  -6.67% (p=0.000 n=10)
Mult32-32     5.311n ± 1%   5.492n ± 1%  +3.40% (p=0.000 n=10)
Bytes-32      11.51n ± 1%   11.53n ± 1%       ~ (p=0.303 n=10)
geomean       28.16n        27.54n       -2.20%

Change-Id: I6e60d2f1220661df4b4f2bf2d810864c19c03012
Reviewed-on: https://go-review.googlesource.com/c/go/+/650279
Auto-Submit: Filippo Valsorda <filippo@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: Michael Pratt <mpratt@google.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>

crypto/internal/fips140/edwards25519/field: optimize carryPropagate

Using pure Go solution for ARM64 seems to perform better when the
operation order is slightly tweaked.

goos: linux
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
                              │     OLD      │                 NEW                 │
                              │    sec/op    │   sec/op     vs base                │
EncodingDecoding-4               158.7µ ± 0%   141.4µ ± 0%  -10.88% (p=0.000 n=10)
ScalarBaseMult-4                 281.2µ ± 0%   260.5µ ± 0%   -7.35% (p=0.000 n=10)
ScalarMult-4                    1008.9µ ± 0%   916.6µ ± 0%   -9.15% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-4   1003.4µ ± 0%   909.6µ ± 0%   -9.36% (p=0.000 n=10)
geomean                          461.0µ        418.6µ        -9.19%

pkg: crypto/internal/fips140/edwards25519/field
           │     OLD     │                 NEW                 │
           │   sec/op    │   sec/op     vs base                │
Add-4        45.22n ± 0%   33.50n ± 0%  -25.91% (p=0.000 n=10)
Multiply-4   454.0n ± 0%   406.8n ± 0%  -10.41% (p=0.000 n=10)
Square-4     278.2n ± 0%   246.4n ± 0%  -11.43% (p=0.000 n=10)
Invert-4     75.83µ ± 0%   67.37µ ± 0%  -11.16% (p=0.000 n=10)
Mult32-4     78.66n ± 0%   78.68n ± 0%   +0.02% (p=0.022 n=10)
Bytes-4      120.6n ± 0%   110.6n ± 0%   -8.25% (p=0.000 n=10)
geomean      400.2n        354.0n       -11.54%

goos: darwin
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
cpu: Apple M1 Pro
                               │     OLD      │                 NEW                 │
                               │    sec/op    │   sec/op     vs base                │
EncodingDecoding-10              10.095µ ± 0%   7.610µ ± 2%  -24.62% (p=0.000 n=10)
ScalarBaseMult-10                 12.65µ ± 0%   11.54µ ± 0%   -8.80% (p=0.000 n=10)
ScalarMult-10                     51.49µ ± 0%   38.59µ ± 2%  -25.06% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-10    49.41µ ± 0%   37.10µ ± 0%  -24.92% (p=0.000 n=10)
geomean                           23.88µ        18.83µ       -21.14%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                 NEW                  │
            │   sec/op    │    sec/op     vs base                │
Add-10        6.009n ± 1%    5.116n ± 5%  -14.85% (p=0.000 n=10)
Multiply-10   19.59n ± 0%    18.00n ± 2%   -8.14% (p=0.000 n=10)
Square-10     18.14n ± 0%    13.66n ± 0%  -24.70% (p=0.000 n=10)
Invert-10     4.854µ ± 0%    3.629µ ± 0%  -25.24% (p=0.000 n=10)
Mult32-10     6.151n ± 0%    6.165n ± 2%        ~ (p=0.224 n=10)
Bytes-10      7.463n ± 1%   10.330n ± 8%  +38.43% (p=0.000 n=10)
geomean       27.94n         25.74n        -7.89%

tags: purego
goos: windows
goarch: amd64
pkg: crypto/internal/fips140/edwards25519
cpu: AMD Ryzen Threadripper 2950X 16-Core Processor
                               │     OLD      │                 NEW                 │
                               │    sec/op    │   sec/op     vs base                │
EncodingDecoding-32              12.856µ ± 0%   9.557µ ± 1%  -25.66% (p=0.000 n=10)
ScalarBaseMult-32                 21.28µ ± 1%   19.14µ ± 2%  -10.04% (p=0.000 n=10)
ScalarMult-32                     74.83µ ± 1%   64.61µ ± 1%  -13.65% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-32    73.85µ ± 0%   62.36µ ± 1%  -15.56% (p=0.000 n=10)
geomean                           35.06µ        29.30µ       -16.44%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                 NEW                 │
            │   sec/op    │   sec/op     vs base                │
Add-32        5.700n ± 1%   4.879n ± 1%  -14.40% (p=0.000 n=10)
Multiply-32   29.24n ± 2%   22.75n ± 2%  -22.21% (p=0.000 n=10)
Square-32     23.06n ± 1%   16.46n ± 2%  -28.60% (p=0.000 n=10)
Invert-32     5.952µ ± 2%   4.466µ ± 1%  -24.97% (p=0.000 n=10)
Mult32-32     5.240n ± 1%   5.311n ± 1%   +1.35% (p=0.006 n=10)
Bytes-32      12.39n ± 1%   11.51n ± 1%   -7.10% (p=0.000 n=10)
geomean       33.78n        28.16n       -16.63%

Change-Id: I71fa40307e803caec56227607ee666198e4c0b03
Reviewed-on: https://go-review.googlesource.com/c/go/+/650278
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>

crypto/internal/fips140/edwards25519/field: optimize *19

Using a `x*19 == x + (x + x<<3)<<1` gives a significant performance
improvement for arm devices that have a slow multiply.
Surprisingly it also seems to help Mac M1 and AMD64+purgo a bit.

goos: linux
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
                              │     OLD     │                NEW                 │
                              │   sec/op    │   sec/op     vs base               │
EncodingDecoding-4              166.3µ ± 0%   158.7µ ± 0%  -4.57% (p=0.000 n=10)
ScalarBaseMult-4                286.0µ ± 0%   281.2µ ± 0%  -1.70% (p=0.000 n=10)
ScalarMult-4                    1.042m ± 0%   1.009m ± 0%  -3.22% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-4   1.042m ± 0%   1.003m ± 0%  -3.66% (p=0.000 n=10)
geomean                         476.7µ        461.0µ       -3.29%

pkg: crypto/internal/fips140/edwards25519/field
           │     OLD     │                NEW                 │
           │   sec/op    │   sec/op     vs base               │
Add-4        45.24n ± 0%   45.22n ± 0%       ~ (p=0.166 n=10)
Multiply-4   447.5n ± 0%   454.0n ± 0%  +1.46% (p=0.000 n=10)
Square-4     289.7n ± 0%   278.2n ± 0%  -3.99% (p=0.000 n=10)
Invert-4     79.45µ ± 0%   75.83µ ± 0%  -4.55% (p=0.000 n=10)
Mult32-4     78.67n ± 0%   78.66n ± 0%       ~ (p=0.272 n=10)
Bytes-4      120.5n ± 0%   120.6n ± 0%       ~ (p=0.390 n=10)
geomean      405.0n        400.2n       -1.20%

goos: darwin
goarch: arm64
pkg: crypto/internal/fips140/edwards25519
cpu: Apple M1 Pro
                               │     OLD     │                NEW                 │
                               │   sec/op    │   sec/op     vs base               │
EncodingDecoding-10              10.04µ ± 0%   10.10µ ± 0%  +0.54% (p=0.000 n=10)
ScalarBaseMult-10                12.72µ ± 0%   12.65µ ± 0%  -0.50% (p=0.000 n=10)
ScalarMult-10                    51.82µ ± 0%   51.49µ ± 0%  -0.63% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-10   50.63µ ± 2%   49.41µ ± 0%  -2.41% (p=0.001 n=10)
geomean                          24.06µ        23.88µ       -0.75%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                NEW                 │
            │   sec/op    │   sec/op     vs base               │
Add-10        6.327n ± 2%   6.009n ± 1%  -5.03% (p=0.000 n=10)
Multiply-10   19.12n ± 0%   19.59n ± 0%  +2.48% (p=0.000 n=10)
Square-10     17.88n ± 0%   18.14n ± 0%  +1.40% (p=0.000 n=10)
Invert-10     4.816µ ± 0%   4.854µ ± 0%  +0.78% (p=0.000 n=10)
Mult32-10     6.188n ± 0%   6.151n ± 0%  -0.61% (p=0.001 n=10)
Bytes-10      7.460n ± 0%   7.463n ± 1%       ~ (p=0.795 n=10)
geomean       27.99n        27.94n       -0.19%

tags: purego
goos: windows
goarch: amd64
pkg: crypto/internal/fips140/edwards25519
cpu: AMD Ryzen Threadripper 2950X 16-Core Processor
                               │     OLD     │                NEW                 │
                               │   sec/op    │   sec/op     vs base               │
EncodingDecoding-32              13.61µ ± 1%   12.86µ ± 0%  -5.54% (p=0.000 n=10)
ScalarBaseMult-32                22.88µ ± 2%   21.28µ ± 1%  -6.98% (p=0.000 n=10)
ScalarMult-32                    79.29µ ± 3%   74.83µ ± 1%  -5.63% (p=0.000 n=10)
VarTimeDoubleScalarBaseMult-32   77.91µ ± 2%   73.85µ ± 0%  -5.22% (p=0.000 n=10)
geomean                          37.24µ        35.06µ       -5.85%

pkg: crypto/internal/fips140/edwards25519/field
            │     OLD     │                NEW                 │
            │   sec/op    │   sec/op     vs base               │
Add-32        5.723n ± 2%   5.700n ± 1%       ~ (p=0.218 n=10)
Multiply-32   30.63n ± 1%   29.24n ± 2%  -4.52% (p=0.000 n=10)
Square-32     24.30n ± 1%   23.06n ± 1%  -5.10% (p=0.000 n=10)
Invert-32     6.368µ ± 1%   5.952µ ± 2%  -6.53% (p=0.000 n=10)
Mult32-32     5.303n ± 2%   5.240n ± 1%  -1.17% (p=0.041 n=10)
Bytes-32      12.47n ± 1%   12.39n ± 1%       ~ (p=0.137 n=10)
geomean       34.86n        33.78n       -3.10%

Change-Id: I889b322bf49293516574d3e9514734a49cca1f86
Reviewed-on: https://go-review.googlesource.com/c/go/+/650277
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>

all: use testenv.GoToolPath instead of hardcoding go

Change-Id: I84ec73d3ddef913a87cb9b48147c44ac3e7c8a8d
Reviewed-on: https://go-review.googlesource.com/c/go/+/651957
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

runtime/cgo: use CreateThread instead of _beginthread

_beginthread is intended to be used together with the C runtime.
The cgo runtime doesn't use it, so better use CreateThread directly,
which is the Windows API for creating threads.

Cq-Include-Trybots: luci.golang.try:gotip-windows-amd64-longtest,gotip-windows-arm64
Change-Id: Ic6cf75f69f62a3babf5e74155da1aac70961886c
Reviewed-on: https://go-review.googlesource.com/c/go/+/651995
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

sync: reduce OnceFunc (and variants) heap allocations

The lifetime of the variables are identical; capture
them in a single struct to avoid individual allocations.
The inner closure can also avoid allocation by using the
capture of the outer closure.

Escape analysis for OnceValues:

  /go/src/sync/oncefunc.go:74:29: moved to heap: sync.f
  /go/src/sync/oncefunc.go:76:3: moved to heap: sync.once
  /go/src/sync/oncefunc.go:77:3: moved to heap: sync.valid
  /go/src/sync/oncefunc.go:78:3: moved to heap: sync.p
  /go/src/sync/oncefunc.go:79:3: moved to heap: sync.r1
  /go/src/sync/oncefunc.go:80:3: moved to heap: sync.r2
  /go/src/sync/oncefunc.go:82:7: func literal escapes to heap
  /go/src/sync/oncefunc.go:83:9: func literal does not escape
  /go/src/sync/oncefunc.go:93:9: func literal escapes to heap

After provided changes:

  /go/src/sync/oncefunc.go:86:2: moved to heap: sync.d
  /go/src/sync/oncefunc.go:96:9: func literal escapes to heap
  /go/src/sync/oncefunc.go:99:13: func literal does not escape
  /go/src/sync/oncefunc.go:100:10: func literal does not escape

Change-Id: Ib06e650fd427b57e0bdbdf1fe759fe436104ff79
Reviewed-on: https://go-review.googlesource.com/c/go/+/601596
Auto-Submit: Austin Clements <austin@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>

cmd/link: put .got section in __DATA_CONST segment

On Darwin, the .got section can be placed in a read-only segment. Only the dynamic linker should modify it at start-up time.

Other read-only sections, like .typelink and .itablink, are already placed in the __DATA_CONST segment. Do the same for the .got section.

Fixes #71416.

Cq-Include-Trybots: luci.golang.try:gotip-darwin-amd64-longtest
Change-Id: I9cd9c20da63b655fabb61d742feb086c3ef3bea7
Reviewed-on: https://go-review.googlesource.com/c/go/+/644055
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

crypto/tls: require EMS in FIPS 140-3 mode

See Implementation Guidance D.Q.

Change-Id: I6a6a465607da94f2bb249934f0561ae04a55e7b7
Reviewed-on: https://go-review.googlesource.com/c/go/+/650575
Reviewed-by: Daniel McCarney <daniel@binaryparadox.net>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>

all: gofmt -w

Change-Id: Ie30a780cbd98bab1e80035b3dfddf92eb281759e
GitHub-Last-Rev: 369ada24ffc297efb47768e430b1bd0216706998
GitHub-Pull-Request: golang/go#71898
Reviewed-on: https://go-review.googlesource.com/c/go/+/651795
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Rob Pike <r@golang.org>
Auto-Submit: Ian Lance Taylor <iant@google.com>

encoding/json: use builtin min function in appendString

To make code a bit simpler.

Change-Id: I59fca1d5760e304abd53873ecf9ca8b2903e02e8
GitHub-Last-Rev: 1369df6da16121c342a4e678efe3e5b082485b74
GitHub-Pull-Request: golang/go#71873
Reviewed-on: https://go-review.googlesource.com/c/go/+/651355
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: Michael Pratt <mpratt@google.com>

all: surround -test.run arguments with ^$

If the -test.run value is not surrounded by ^$ then any test that
matches the -test.run value will be run. This is normally not the
desired behavior, as it can lead to unexpected tests being run.

Change-Id: I3447aaebad5156bbef7f263cdb9f6b8c32331324
Reviewed-on: https://go-review.googlesource.com/c/go/+/651956
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

cmd/compile: ensure we don't reuse temporary register

Before this CL, we could use the same register for both a temporary
register and for moving a value in the output register out of the way.

Fixes #71857

Change-Id: Iefbfd9d4139136174570d8aadf8a0fb391791ea9
Reviewed-on: https://go-review.googlesource.com/c/go/+/651221
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>

unicode/utf8: use builtin max function to simplify code

Change-Id: I6a73b645d074baaa4d09480bdf4192816a8c2450
GitHub-Last-Rev: 202d498eb019c18b9ba30bccc2cb169c9eb79366
GitHub-Pull-Request: golang/go#71945
Reviewed-on: https://go-review.googlesource.com/c/go/+/652177
Auto-Submit: Keith Randall <khr@golang.org>
Auto-Submit: Ian Lance Taylor <iant@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>
Reviewed-by: Ian Lance Taylor <iant@google.com>

crypto/internal/fips140test: add a test for a non-approved function

Change-Id: I6a6a46560bfca588e5874f3769f33b561c33096a
Reviewed-on: https://go-review.googlesource.com/c/go/+/652415
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Damien Neil <dneil@google.com>

net: properly wrap context cancellation errors and return DNSErrors consistently

Fixes #71939

Change-Id: Id7cd720fcca2812ffca2b1b20fe923914422d994
GitHub-Last-Rev: 4671f338c91b5826c669fbd113c176e22f5020e5
GitHub-Pull-Request: golang/go#71941
Reviewed-on: https://go-review.googlesource.com/c/go/+/652275
Commit-Queue: 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>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>

spec: remove notion of core types

This CL removes the notion of core types from the spec.

Instead of referring to core types, each section that did
so before is reverted to approx. the pre-generics (1.17)
prose, and additional paragraphs cover the type parameter
cases as needed.

The hope is that this makes it easier to read the spec.
When type parameters are involved, the extra prose is
local to the language feature in question and thus more
readily available. When no type parameters are present,
readers do not have to concern themselves with core types.

In contrast to CL 621919, this change is not intended to
loosen the spec in any way and therefore does not change
the language (if the new prose implies otherwise, we will
correct it).

Except for adjustments to compiler error messages
(no mention of core types anymore), no other changes
to the compiler or tools are required.

Future CLs may selectively relax requirements on a language
construct by language construct basis; each such change can
be discussed and proposed independently.

For #70128.

Change-Id: I6ed879a472c615d7c8dbdc7b6bd7eef3d12eff7e
Reviewed-on: https://go-review.googlesource.com/c/go/+/645716
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
TryBot-Bypass: Robert Griesemer <gri@google.com>

go/parser: require label after goto

Fixes #70957

Change-Id: Ied7cf29ea3e02bb71ddce8a19ddd381ce5991ed1
GitHub-Last-Rev: 310bd1537b7a36758f3fbf8db476fa68e1a11599
GitHub-Pull-Request: golang/go#70958
Reviewed-on: https://go-review.googlesource.com/c/go/+/638395
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Alan Donovan <adonovan@google.com>

weak: test the use of runtime.AddCleanup

This change adds a test case for runtime.AddCleanup.

Updates #70907

Change-Id: I29cba9dc5b40cec8e610215974e61ee47e10d00f
Reviewed-on: https://go-review.googlesource.com/c/go/+/649459
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>

cmd/compile, runtime: use PC of deferreturn for panic transfer

this removes the old conditional-on-register-value
handshake from the deferproc/deferprocstack logic.

The "line" for the recovery-exit frame itself (not the defers
that it runs) is the closing brace of the function.

Reduces code size slightly (e.g. go command is 0.2% smaller)

Sample output showing effect of this change, also what sort of
code it requires to observe the effect:
```
package main

import "os"

func main() {
g(len(os.Args) - 1)           // stack[0]
}

var gi int
var pi *int = &gi

//go:noinline
func g(i int) {
switch i {
case 0:
defer func() {
println("g0", i)
q()                  // stack[2] if i == 0
}()
for j := *pi; j < 1; j++ {
defer func() {
println("recover0", recover().(string))
}()
}
default:
for j := *pi; j < 1; j++ {
defer func() {
println("g1", i)
q()              // stack[2] if i == 1
}()
}
defer func() {
println("recover1", recover().(string))
}()
}
p()
}                                // stack[1] (deferreturn)

//go:noinline
func p() {
panic("p()")
}

//go:noinline
func q() {
panic("q()")                 // stack[3]
}

/* Sample output for "./foo foo":
recover1 p()
g1 1
panic: q()

goroutine 1 [running]:
main.q()
.../main.go:46 +0x2c
main.g.func3()
.../main.go:29 +0x48
main.g(0x1?)
.../main.go:37 +0x68
main.main()
.../main.go:6 +0x28
*/
```

Change-Id: Ie39ea62ecc244213500380ea06d44024cadc2317
Reviewed-on: https://go-review.googlesource.com/c/go/+/650795
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

go/types, types2: remove coreType call in lookup

For #70128.

Change-Id: I7d16ad7fdc6b07a2632b4eaefaedfa2bcceffe1d
Reviewed-on: https://go-review.googlesource.com/c/go/+/652215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>

internal/abi: use arch family instead of arch string

No point in using string comparison when we can use integer comparison instead.

Unify the constants in cmd/internal/sys and internal/goarch while
we are at it.

Change-Id: I5681a601030307b7b286f958a8965559cb43506d
Reviewed-on: https://go-review.googlesource.com/c/go/+/652175
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Stapelberg <stapelberg@google.com>
Reviewed-by: Keith Randall <khr@google.com>

go/types, types2: remove another coreType call in type checking range clause

For #70128.

Change-Id: I5949bccbfaaebc435ae8ac7c70580d9740de6f00
Reviewed-on: https://go-review.googlesource.com/c/go/+/652136
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Auto-Submit: Robert Griesemer <gri@google.com>