[PromoteMem2Reg] Replace lifetime.start/end with store undef before promotion#932
Open
isoard-amd wants to merge 1 commit intoaie-publicfrom
Open
[PromoteMem2Reg] Replace lifetime.start/end with store undef before promotion#932isoard-amd wants to merge 1 commit intoaie-publicfrom
isoard-amd wants to merge 1 commit intoaie-publicfrom
Conversation
…romotion When PromoteMemToReg promotes an alloca, it erases all lifetime intrinsics via removeIntrinsicUsers(). This loses the semantic information that the alloca's content is undefined after lifetime.start/end, causing the SSA renaming pass to create spurious loop-carried phi nodes that carry a stale value from the previous iteration across a back-edge. Fix this by replacing each lifetime.start/end with an actual store of undef to the alloca, instead of simply deleting it. The standard SSA renaming then sees a real definition of undef at that point, and back-edge phi nodes are naturally resolved to undef — eliminating spurious loop-carried dependences without any special-casing in the renaming algorithm.
isoard-amd
requested review from
F-Stuckmann,
SagarMaheshwari99,
abhinay-anubola,
abnikant,
andcarminati,
katerynamuts,
khallouh,
konstantinschwarz,
martien-de-jong,
mludevid,
niwinanto and
stephenneuendorffer
as code owners
Collaborator
Author
|
I don't think it is fully ready, we should check that the byte-length in the lifetime intrinsic match the type store size. |
Collaborator
Author
|
We are attempting to push this change upstream llvm/llvm-project#191909 (mostly to get more eyeballs on it) |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When PromoteMemToReg promotes an alloca, it erases all lifetime intrinsics via removeIntrinsicUsers(). This loses the semantic information that the alloca's content is undefined after lifetime.start/end, causing the SSA renaming pass to create spurious loop-carried phi nodes that carry a stale value from the previous iteration across a back-edge.
Fix this by replacing each lifetime.start/end with an actual store of undef to the alloca, instead of simply deleting it. The standard SSA renaming then sees a real definition of undef at that point, and back-edge phi nodes are naturally resolved to undef — eliminating spurious loop-carried dependences without any special-casing in the renaming algorithm.