gh-139757: Treat call specially in JIT assembly backend optimizer on x86-64 and AArch64#142907
Uh oh!
There was an error while loading. Please reload this page.
Conversation
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
Fidget-Spinner requested review from brandtbucher, diegorusso and savannahostrowski as code owners
Uh oh!
There was an error while loading. Please reload this page.
Fidget-Spinner changed the title Fidget-Spinner changed the title Uh oh!
There was an error while loading. Please reload this page.
Co-authored-by: Savannah Ostrowski <savannah@python.org>
…n into treat_call_as_jump
Fidget-Spinner requested review from AA-Turner, ezio-melotti and hugovk as code owners
savannahostrowski approved these changes Dec 17, 2025
Member
savannahostrowski left a comment
savannahostrowski left a comment There was a problem hiding this comment.
LGTM - thanks for the back and forth on this!
MemberAuthor
Fidget-Spinner commented Dec 17, 2025
I'm merging this as it's fixing an actual bug that causes malformed code in the JIT. We can refine it later if we want, but I have gone through a lot of discussion with Savannah about the correctness of this PR, and I feel confident on it. |
Fidget-Spinner merged commit 8b64dd8 into python:main 77 checks passed
Uh oh!
There was an error while loading. Please reload this page.
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
On x86-64 and AArch64, a call is effectively a jump to a label with some caveats. As such, the assembly flow optimizer must treat it as such, or it risks emitting invalid optimizations.
The bug triggers in the following code:
The assembly optimizer currently doesn't see
PyStackRef_CLOSEas being used, and just optimizes the whole thing away.The fix informs the assembly optimizer we might jump to
PyStackRef_CLOSE, thus the label is used and to preserve it.I verified it fixes the crash on FT builds on my system.
The reason why we didn't see this earlier is that our stencils were at -O3 and that squashed them all into one function. After we switched to -Os, bigger stencils get outlined. It just so happens the biggest stencils are on FT.