AMDGPU: Fix treating unknown mem operands as uniform#170309
Merged
Conversation
This was referenced Dec 2, 2025
Contributor
Author
This was referenced Dec 2, 2025
Contributor
Author
|
Repost of #170309, which I accidentally merged into the non-main branch |
Base automatically changed from
users/arsenm/amdgpu/use-constant-pool-kernarg-loads
to
main
December 2, 2025 15:48
arsenm
changed the title
The test changes are mostly GlobalISel specific regressions. GlobalISel is still relying on isUniformMMO, but it doesn't really have an excuse for doing so. These should be avoidable with new regbankselect. There is an additional regression for addrspacecast for cov4. We probably ought to be using a separate PseudoSourceValue for the access of the queue pointer.
arsenm
force-pushed
the
users/arsenm/arsenm/fix-treating-unknown-mmo-uniform-2
branch
from
3b09937 to
8ab7ed2
Compare
arsenm
deleted the
users/arsenm/arsenm/fix-treating-unknown-mmo-uniform-2
branch
kcloudy0717
pushed a commit
to kcloudy0717/llvm-project
that referenced
this pull request
Dec 4, 2025
The test changes are mostly GlobalISel specific regressions. GlobalISel is still relying on isUniformMMO, but it doesn't really have an excuse for doing so. These should be avoidable with new regbankselect. There is an additional regression for addrspacecast for cov4. We probably ought to be using a separate PseudoSourceValue for the access of the queue pointer.
honeygoyal
pushed a commit
to honeygoyal/llvm-project
that referenced
this pull request
Dec 9, 2025
The test changes are mostly GlobalISel specific regressions. GlobalISel is still relying on isUniformMMO, but it doesn't really have an excuse for doing so. These should be avoidable with new regbankselect. There is an additional regression for addrspacecast for cov4. We probably ought to be using a separate PseudoSourceValue for the access of the queue pointer.
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.
The test changes are mostly GlobalISel specific regressions.
GlobalISel is still relying on isUniformMMO, but it doesn't really
have an excuse for doing so. These should be avoidable with new
regbankselect.
There is an additional regression for addrspacecast for cov4. We
probably ought to be using a separate PseudoSourceValue for the
access of the queue pointer.