Repository navigation
internal/loopir: reject a constant offset that puts an index below zero - #1
Merged
Merged
Conversation
ubuntu-latest is now ubuntu-26.04, which has no qemu-user-static package, so every run fails at the install step and skips the arm64 tsvc run, the golden check, fmt and lint. qemu-user ships the static emulators there; link qemu-aarch64-static to it when the name is missing, as the Makefile runs that name.
thesyncim
force-pushed
the
fix-negative-constant-offset
branch
2 times, most recently
from
October 31, 2026 19:56
4d49970 to
922738c
Compare
thesyncim
changed the title
This was referenced Oct 31, 2026
Go tip writes export data in iexport version 4, which encodes float constants exactly. x/tools v0.51.0, and the v0.52.0 release, read only up to version 3, so on tip go/packages fails for every package: cannot import "math" (unstable iexport format version 4, just rebuild compiler and std library), export data is newer version - update tool and every script that runs loopvec fails in the tip job. x/tools master reads version 4 (golang/tools@966056e2637c). Update to it until a release includes it. The pinned toolchain is unaffected.
thesyncim
force-pushed
the
fix-negative-constant-offset
branch
from
October 31, 2026 20:50
922738c to
1aa6273
Compare
go1.27.3 is the latest 1.27 release. Its simd package is the same as go1.27.1's, so the capability table only changes in its header, and the -methods crash (golang/go#80657) is still there, as flag_methods_ice.txt checks. go.mod keeps go 1.27.1: that is the first release whose simd package loopvec's output compiles against (go1.27.0's ShiftAllRight takes a uint8).
A read at a negative constant offset in a loop with a constant limit, such as dst[i] = a[i-20] + 1 in a loop that stops before 16, was rewritten into code that does not compile: the length check of the vector loop is _ = a[16-21], a negative constant index. -split wrote that file, and the toolexec wrapper fell back to the original package. Such a loop panics as soon as it runs, as does one whose first index is a negative constant (a[i-1] in a loop from 0). Both are now left alone with their own reason, the way a constant limit of zero or less already is. A read whose indexes can all be zero or more (a[i-1] from 1) is still rewritten. The checks read the start and limit as text, so a named constant or a constant expression (const s0 = 0; for i := s0; ...) slipped past them. A constant start or limit is now spelled as its value, which also lets the short-loop rule see a named constant trip count: goav1's for x := range restorationExtraHorz, three iterations, is no longer rewritten. The vectorized set of std and gonum does not change.
thesyncim
force-pushed
the
fix-negative-constant-offset
branch
from
October 31, 2026 21:50
1aa6273 to
087660e
Compare
This was referenced Oct 31, 2026
mauri870
pushed a commit
that referenced
this pull request
Nov 4, 2026
The runners have moved to ubuntu-26.04, so this is the qemu part of #1 you said you'd take when that happened. `qemu-user-static` has no installation candidate there, so every PR now fails at that step and the arm64 runs, the golden check, fmt and lint get skipped. You can see it on #7, #9 and #19, where Build, Test and the amd64 tsvc run all pass first. master's last run predates the move, so the next push will hit it too. `qemu-user` has the static emulators on 26.04, just without the `-static` names, so this installs that and links `qemu-aarch64-static` when it's missing, since the Makefile runs that name. It's been passing on my fork's ubuntu-26.04 runners, and this PR's own CI runs it too.
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 join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Stacked on #4.
loopvec writes a
_simd.gothat doesn't compile for a loop like this:The length check comes out as
_ = a[16-21]:toolexec falls back, but
-splitand-wleave the broken file behind.A loop like that panics as soon as it runs, and so does one whose first index is a negative constant (
a[i-1]starting at 0). Both are now rejected with their own reason, the same way a constant limit of zero or less already is.a[i-1]starting at 1 is still rewritten.The checks parsed the start and limit as text, so named constants and constant expressions (
const s0 = 0; for i := s0; ...,for i := 1 - 1; ...) slipped through. Constant bounds are now written out as their value. A side effect is that the short-loop rule now sees named constant trip counts too: in goav1, a 3-iterationfor x := range restorationExtraHorzis no longer vectorized. The vectorized set of std and gonum doesn't change.