go, branch go1.23.2

[release-branch.go1.23] go1.23.2

2024-10-01T17:24:29Z

Change-Id: I904d2e951796dd4142d6e9de4a55af07852bca51 Reviewed-on: https://go-review.googlesource.com/c/go/+/617019 LUCI-TryBot-Result: Go LUCI Reviewed-by: Michael Knyszek Auto-Submit: Gopher Robot Reviewed-by: David Chase

[release-branch.go1.23] runtime: if stop/reset races with running timer, return correct result

2024-09-27T18:23:47Z

The timer code is careful to ensure that if stop/reset is called while a timer is being run, we cancel the run. However, the code failed to ensure that in that case stop/reset returned true, meaning that the timer had been stopped. In the racing case stop/reset could see that t.when had been set to zero, and return false, even though the timer had not and never would fire. Fix this by tracking whether a timer run is in progress, and using that to reliably detect that the run was cancelled, meaning that stop/reset should return true. For #69312 Fixes #69333 Change-Id: I78e870063eb96650638f12c056e32c931417c84a Reviewed-on: https://go-review.googlesource.com/c/go/+/611496 Reviewed-by: David Chase Reviewed-by: Cuong Manh Le Reviewed-by: Michael Knyszek LUCI-TryBot-Result: Go LUCI Auto-Submit: Ian Lance Taylor (cherry picked from commit 2ebaff4890596ed6064e2dcbbe5e68bc93bed882) Reviewed-on: https://go-review.googlesource.com/c/go/+/616096 Reviewed-by: Ian Lance Taylor Commit-Queue: Ian Lance Taylor Auto-Submit: Ian Lance Taylor

[release-branch.go1.23] cmd/compile: fix wrong esacpe analysis for rangefunc

2024-09-25T16:22:40Z

CL 584596 "-range" suffix to the name of closure generated for a rangefunc loop body. However, this breaks the condition that escape analysis uses for checking whether a closure contains within function, which is "F.funcN" for outer function "F" and closure "funcN". Fixing this by adding new "-rangeN" to the condition. Updates #69434 Fixes #69511 Change-Id: I411de8f63b69a6514a9e9504d49d62e00ce4115d Reviewed-on: https://go-review.googlesource.com/c/go/+/614096 Reviewed-by: David Chase Reviewed-by: Cherry Mui LUCI-TryBot-Result: Go LUCI Auto-Submit: Cuong Manh Le Reviewed-on: https://go-review.googlesource.com/c/go/+/614195

[release-branch.go1.23] syscall: on exec failure, close pidfd

2024-09-18T19:41:05Z

For #69284 Fixes #69402 Change-Id: I6350209302778ba5e44fa03d0b9e680d2b4ec192 Reviewed-on: https://go-review.googlesource.com/c/go/+/611495 LUCI-TryBot-Result: Go LUCI Reviewed-by: roger peppe Reviewed-by: Tim King Auto-Submit: Ian Lance Taylor Reviewed-by: Dmitri Shuralyov (cherry picked from commit 8926ca9c5ec3ea0b51e413e87f737aeb1422ea48) Reviewed-on: https://go-review.googlesource.com/c/go/+/613616 Reviewed-by: Ian Lance Taylor Auto-Submit: Ian Lance Taylor Reviewed-by: Tobias Klauser

[release-branch.go1.23] unique: don't retain uncloned input as key

2024-09-11T18:11:20Z

Currently the unique package tries to clone strings that get stored in its internal map to avoid retaining large strings. However, this falls over entirely due to the fact that the original string is *still* stored in the map as a key. Whoops. Fix this by storing the cloned value in the map instead. This change also adds a test which fails without this change. For #69370. Fixes #69383. Change-Id: I1a6bb68ed79b869ea12ab6be061a5ae4b4377ddb Reviewed-on: https://go-review.googlesource.com/c/go/+/610738 Reviewed-by: Michael Pratt Auto-Submit: Michael Knyszek LUCI-TryBot-Result: Go LUCI (cherry picked from commit 21ac23a96f204dfb558a8d3071380c1d105a93ba) Reviewed-on: https://go-review.googlesource.com/c/go/+/612295 Auto-Submit: Tim King

[release-branch.go1.23] os: dup pidfd if caller sets PidFD manually

2024-09-06T16:06:09Z

For #68984. Fixes #69119. Change-Id: I16d25777cb38a337cd4204a8147eaf866c3df9e1 Reviewed-on: https://go-review.googlesource.com/c/go/+/607695 Reviewed-by: Kirill Kolyshkin Reviewed-by: Michael Pratt LUCI-TryBot-Result: Go LUCI Commit-Queue: Ian Lance Taylor Reviewed-by: David Chase Auto-Submit: Ian Lance Taylor (cherry picked from commit 239666cd7343d46c40a5b929c8bec8b532dbe83f) Reviewed-on: https://go-review.googlesource.com/c/go/+/611415 Auto-Submit: Dmitri Shuralyov Reviewed-by: Dmitri Shuralyov TryBot-Bypass: Dmitri Shuralyov

[release-branch.go1.23] internal/weak: shade pointer in weak-to-strong conversion

2024-09-06T14:34:16Z

There's a bug in the weak-to-strong conversion in that creating the *only* strong pointer to some weakly-held object during the mark phase may result in that object not being properly marked. The exact mechanism for this is that the new strong pointer will always point to a white object (because it was only weakly referenced up until this point) and it can then be stored in a blackened stack, hiding it from the garbage collector. This "hide a white pointer in the stack" problem is pretty much exactly what the Yuasa part of the hybrid write barrier is trying to catch, so we need to do the same thing the write barrier would do: shade the pointer. Added a test and confirmed that it fails with high probability if the pointer shading is missing. For #69210. Fixes #69240. Change-Id: Iaae64ae95ea7e975c2f2c3d4d1960e74e1bd1c3f Reviewed-on: https://go-review.googlesource.com/c/go/+/610396 LUCI-TryBot-Result: Go LUCI Reviewed-by: Cherry Mui Reviewed-by: Michael Pratt Auto-Submit: Michael Knyszek (cherry picked from commit 79fd633632cdbaf9ca38f7559e5abb5c07fbbd9d) Reviewed-on: https://go-review.googlesource.com/c/go/+/610696 Auto-Submit: Dmitri Shuralyov

[release-branch.go1.23] runtime: size maps.Clone destination bucket array safely

2024-09-06T14:32:57Z

In rare situations, like during same-sized grows, the source map for maps.Clone may be overloaded (has more than 6.5 entries per bucket). This causes the runtime to allocate a larger bucket array for the destination map than for the source map. The maps.Clone code walks off the end of the source array if it is smaller than the destination array. This is a pretty simple fix, ensuring that the destination bucket array is never longer than the source bucket array. Maybe a better fix is to make the Clone code handle shorter source arrays correctly, but this fix is deliberately simple to reduce the risk of backporting this fix. Fixes #69156 Change-Id: I824c93d1db690999f25a3c43b2816fc28ace7509 Reviewed-on: https://go-review.googlesource.com/c/go/+/610377 Reviewed-by: Dmitri Shuralyov Reviewed-by: Cuong Manh Le LUCI-TryBot-Result: Go LUCI Auto-Submit: Dmitri Shuralyov Reviewed-by: Keith Randall

[release-branch.go1.23] cmd/cgo: correct padding required by alignment

2024-09-05T21:57:44Z

If the aligned offset isn't sufficient for the field offset, we were padding based on the aligned offset. We need to pad based on the original offset instead. Also set the Go alignment correctly for int128. We were defaulting to the maximum alignment, but since we translate int128 into an array of uint8 the correct Go alignment is 1. For #69086 Fixes #69219 Change-Id: I23ce583335c81beac2ac51f7f9336ac97ccebf09 Reviewed-on: https://go-review.googlesource.com/c/go/+/608815 Reviewed-by: Damien Neil LUCI-TryBot-Result: Go LUCI Reviewed-by: Cherry Mui Auto-Submit: Ian Lance Taylor (cherry picked from commit c2098929056481d0dc09f5f42b8959f4db8878f2) Reviewed-on: https://go-review.googlesource.com/c/go/+/611296 Reviewed-by: Ian Lance Taylor Auto-Submit: Ian Lance Taylor Commit-Queue: Ian Lance Taylor

[release-branch.go1.23] go1.23.1

2024-09-05T15:20:23Z

Change-Id: I1f2dab5560d3214c8934074a53f7750d8d431936 Reviewed-on: https://go-review.googlesource.com/c/go/+/611196 Reviewed-by: Michael Pratt Reviewed-by: Dmitri Shuralyov Commit-Queue: Gopher Robot TryBot-Bypass: Dmitri Shuralyov Auto-Submit: Gopher Robot