[pull] master from golang:master - #196
Merged
Merged
Conversation
randHeapBasePrefixMask clears the top byte at heapAddrBits-8 (bit 40 on
amd64), but hint generation places the randomized prefix byte at
randHeapAddrBits-8 (bit 38): randHeapAddrBits is heapAddrBits-1 on most
platforms and heapAddrBits-2 on amd64. The randHeapBase bits between
those two positions survive the mask and are OR'd into the prefix's low
bits in
p = prefix | (randHeapBase & randHeapBasePrefixMask)
so wherever one of those stray bits is 1 (~75% of process launches),
the corresponding bit of every generated prefix is forced to 1.
Distinct prefix values randHeapBasePrefix+i then collapse to the same
address in pairs or quads, and the arena hint chain carries duplicate
and non-monotonically-spaced hint addresses at doubled or quadrupled
spacing. The heap still works because hints are only mmap placement
fallbacks, but the layout the randomization intends (64 distinct,
evenly spaced heap hints) is silently violated, duplicate hints waste
mmap retries, and two bits of intended address-space structure are
lost.
Define the mask from a hoisted randHeapAddrBits constant instead, and
drop mallocinit's now-duplicate local.
The new TestArenaHintChainGeneration asserts the hint chain holds
pairwise distinct addresses with at most one non-ascending step (the
prefix byte's single mod-256 wraparound); it fails on 10/10 launches
with the old mask and passes with the fix. The test is skipped when
randomizeHeapBase is false: in race mode the hint chain deliberately
repeats addresses within TSAN's narrow heap range, and on 32-bit the
chain is built dynamically by sysAlloc from adjacent up/down hints.
Change-Id: I9aec248b1b606a95f0ff88e7083734a982aa36d6
Reviewed-on: https://go-review.googlesource.com/c/go/+/823944
Reviewed-by: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Add Sub to the scalable SVE vector types, lowering to ZSUB and ZFSUB. It is the first non-commutative operation with a merging-predicated form, which folds more restrictively: only a select whose "else" operand is the minuend matches, with no mirrored and no MOVPRFX-prefixed rule, and ZSUBMerging is pinned resultInArg0; see sveMergingPeephole in gen_simdrules.go. Also assert in the shared ssa-to-prog helper that a resultInArg0 op never reaches the source-swapping path, which would silently compute y-x for a non-commutative op; see simdZ2kvPred in arm64/ssa.go. For #79781. Change-Id: Ifa2ee4dee0bbd11a928284f48a9254a25cb91211 Reviewed-on: https://go-review.googlesource.com/c/go/+/820760 Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
…dicate Add Abs to the scalable SVE vector types, lowering to ZABS. It is the first operation with no unpredicated encoding: the Go API runs it under a synthesized all-true predicate, and a select over it folds by putting the select's mask in that predicate's place, (ZSELB (ZABSB x <all-true>) z mask) => (ZABSMergingB z x mask) so IfElse costs no extra instruction and Masked costs only its zero vector. ABS is constructive, so its merging form carries the merge value as a resultInArg0 operand and needs no MOVPRFX. Only the merging encoding is used: zeroing predication here is an Armv9.6-A extension that baseline SVE hardware traps, and nothing the XML parser exposes distinguishes it; see groupPredicatedOnly in sve/load.go. For #79781. Change-Id: I1547a3c513edca52b11039bcce8023d900027dce Reviewed-on: https://go-review.googlesource.com/c/go/+/820761 Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com>
Lower to ZSQSUB and ZUQSUB. Both have unpredicated and merging-predicated encodings, so the existing machinery derives the non-commutative merging fold; the peepholes are exercised by the Add/Sub/Abs tests already. For #79781. Change-Id: I1561e53ce9970207ca068d9931193b9dd163e13f Reviewed-on: https://go-review.googlesource.com/c/go/+/822925 Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Lower to ZNEG and ZFNEG, both predicated-only like ZABS: the Go API runs them under a synthesized all-true predicate, and selects fold into the merging form via the existing machinery. For #79781. Change-Id: I115580252857cae0a522278371dac122f0624fc9 Reviewed-on: https://go-review.googlesource.com/c/go/+/822926 Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com>
Lower to ZFABS, predicated-only like ZABS, for parity with the NEON and amd64 Abs API surface. For #79781. Change-Id: I259c705758f5b6ae0b40139e0dee0e2f689b3ec0 Reviewed-on: https://go-review.googlesource.com/c/go/+/822927 Reviewed-by: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com>
Lower to ZFSQRT, predicated-only like ZFABS. For #79781. Change-Id: I71a88f0e3135dae3905dab04c248e8336be0060e Reviewed-on: https://go-review.googlesource.com/c/go/+/822928 Reviewed-by: David Chase <drchase@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Lower to ZFRINTP, predicated-only like ZFSQRT. For #79781. Change-Id: Ic405ba37bde1b4890a85b9edeb8b772a7af98814 Reviewed-on: https://go-review.googlesource.com/c/go/+/822929 Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Lower to ZFRINTM, predicated-only like ZFSQRT. For #79781. Change-Id: Ibc60df151b44ad3327d5ab10cf66d431ef5d730d Reviewed-on: https://go-review.googlesource.com/c/go/+/822930 Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Lower to ZFRINTZ, predicated-only like ZFSQRT. For #79781. Change-Id: If1d5329bbcd156e596611cfe72f7e41f40148462 Reviewed-on: https://go-review.googlesource.com/c/go/+/822931 Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Lower to ZFRINTN (round to nearest, ties to even), predicated-only like ZFSQRT, matching the NEON Round mapping. For #79781. Change-Id: Ib73cefd4313e1382a73a2e0f6d20dc96a89fab56 Reviewed-on: https://go-review.googlesource.com/c/go/+/822932 Reviewed-by: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com>
Lower to ZCMPHI and ZFCMGT, predicated-only like ZCMPGT, for parity with the NEON Greater API surface. For #79781. Change-Id: Ibe5fb3f9061ee890c783f326b2c57be45b3e6051 Reviewed-on: https://go-review.googlesource.com/c/go/+/822933 Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com> TryBot-Bypass: David Chase <drchase@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Lower to ZCMPEQ for both integer signednesses and ZFCMEQ for floats, predicated-only like ZCMPGT. For #79781. Change-Id: I5a98a1adfd2b4afe8f7b6c4266de0ff006ca59bb Reviewed-on: https://go-review.googlesource.com/c/go/+/822934 Reviewed-by: David Chase <drchase@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Lower to ZCMPNE for both integer signednesses and ZFCMNE for floats, predicated-only like ZCMPGT. NEON has no native not-equal compare; SVE does, so the API is provided directly. For #79781. Change-Id: Id58641731b0ea682b699015c3a2eda0d8563b568 Reviewed-on: https://go-review.googlesource.com/c/go/+/822935 Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com>
Lower to ZCMPGE (signed), ZCMPHS (unsigned), and ZFCMGE (float), predicated-only like ZCMPGT. For #79781. Change-Id: Ica18c8987d099afaf2dad5f95196faea3258d4f0 Reviewed-on: https://go-review.googlesource.com/c/go/+/822936 Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com> TryBot-Bypass: David Chase <drchase@google.com>
Lower to ZAND. A bitwise vector operation's unpredicated encoding is a single width-agnostic .D instruction, while its merging-predicated sibling is per-<T>; the loader now emits a def per element width for such operations (recognized by the spec's "Bitwise " brief, as the NEON loader does) and simdgen collapses their unpredicated machine op back to the one .D instruction, keeping the per-width merging forms: (AndInt8s x y) => (ZANDD x y) (ZSELB (ZANDD x y) x mask) => (ZANDMergingB x y mask) For #79781. Change-Id: I5e6caa56ab72c868e58775127d4441345ed5bede Reviewed-on: https://go-review.googlesource.com/c/go/+/822937 Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com> Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> TryBot-Bypass: David Chase <drchase@google.com>
Shared promoted wrapper symbols used the outer receiver's package and a
short name for the wrapped receiver. Same-named receiver types from
separate packages could therefore collide, causing the linker to use a
wrapper with an incompatible method signature.
For example, both a.W.M and b.W.M produced:
main..embed.W.M.1.0
Put wrappers in the wrapped receiver's package, and end their names with
the ordinary wrapped method symbol, producing:
foo.bar/a..embed.1.0.foo.bar/a.W.M
foo.bar/b..embed.1.0.foo.bar/b.W.M
The trailing method symbol also lets consumers such as go-cmp's NameOf
recover bytes.Buffer.Write instead of returning an empty name.
Fixes #81286
Change-Id: Ibd7ab2767c0c1a07f9b9445d326258ff5507a644
Reviewed-on: https://go-review.googlesource.com/c/go/+/826084
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: David Chase <drchase@google.com>
Lower to ZORR, width-agnostic bitwise like ZAND. For #79781. Change-Id: I7936182383cd2b71eef542ecd965e57feaa29fcf Reviewed-on: https://go-review.googlesource.com/c/go/+/822938 Reviewed-by: David Chase <drchase@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Junyang Shao <shaojunyang@google.com> Auto-Submit: Junyang Shao <shaojunyang@google.com>
Lower to ZEOR, width-agnostic bitwise like ZAND. For #79781. Change-Id: I01eaa7c8aa8c3b77fabe7d42320043d20c7961b5 Reviewed-on: https://go-review.googlesource.com/c/go/+/822939 Auto-Submit: Junyang Shao <shaojunyang@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Junyang Shao <shaojunyang@google.com>
Change-Id: Ie68c00da89df118ebda2a49f773686c6a7b64167 Reviewed-on: https://go-review.googlesource.com/c/go/+/806825 Reviewed-by: Keith Randall <khr@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Keith Randall <khr@google.com> Auto-Submit: Keith Randall <khr@golang.org> Reviewed-by: David Chase <drchase@google.com>
Fix the misspelling of "instruction" in the identical comments on simdVgpvImm and simdVgpImm. Comment-only change; no functional changes. Change-Id: Ib4ccd1950876050c588e9be64c8e030f6180bb77 Reviewed-on: https://go-review.googlesource.com/c/go/+/818600 Reviewed-by: Keith Randall <khr@golang.org> Reviewed-by: David Chase <drchase@google.com> Auto-Submit: Keith Randall <khr@golang.org> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: Keith Randall <khr@google.com>
…rompt TestAfterQueuing dates to CL 3416043, which reimplemented After to keep a queue of pending deadlines instead of spending a goroutine and a host process per call. The property it means to check is that timers created in one order but with deadlines in another fire in deadline order. It never checked that. It started 13 goroutines, each blocking on one After channel and then sending the result on a shared channel, and required the sends to arrive in slot order. Which goroutine reaches the shared channel first is up to the scheduler, so on a machine where a woken goroutine can be delayed the test fails even though the timers were queued and fired correctly. That is what has been happening on linux/s390x: 94 failures since 2025-05-13, every one an ordering complaint, and in every one all five attempts failed, so the retry loop never helped. The arrival time check that follows was never reached in any of them. There were also previous attempts at deflaking in CL 4335042 (2011) and CL 9795 (2015). Sort the results by the time each timer reported instead. That is the order the timers fired in, it does not depend on the scheduler, and it lets the arrival time check run. The retry loop stays, since that check does depend on the machine keeping up, but it is no longer papering over the ordering check. Also pass the after parameter through to the goroutines. CL 3366171 added it so the test could cover both the channel and the AfterFunc implementations, but the call kept using the package level After, so impl=func has been a second copy of impl=chan and newTimerFunc was never reached. Verified by making newTimerFunc panic: with the old call impl=func passes and never reaches it, with this change it does. For #9903. For #10680. For #58901. For #80294. Change-Id: I7b3a5db3268612c3d515a090714fca1b6a6a6964 Reviewed-on: https://go-review.googlesource.com/c/go/+/822724 Reviewed-by: Cherry Mui <cherryyz@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com>
The test constructed its synthetic tool directory using GOOS and GOARCH, while testgo located installed tools using TESTGO_GOHOSTOS and TESTGO_GOHOSTARCH. When these architectures differed, `go install` failed with `go: no such tool "compile"`. Preserve the installed host tool directory as the symlink source, then make testgo use its runtime tuple within the synthetic GOROOT. Change-Id: Ia132fffa8acb842ec6de98c2c166294886b8a231 Reviewed-on: https://go-review.googlesource.com/c/go/+/826364 Reviewed-by: Michael Matloob <matloob@google.com> LUCI-TryBot-Result: golang-scoped@luci-project-accounts.iam.gserviceaccount.com <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Reviewed-by: David Chase <drchase@google.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )