[AArch64][SVE] Avoid extra pop of "FixedObject" with FPAfterSVECalleeSaves #156452
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
…Saves Previously, we would pop `FixedObject`-bytes after deallocating the SVE area, then again as part of the "AfterCSRPopSize". This could be seen in the tests `@f6` and `@f9`. This patch removes the erroneous pop, and refactors `FPAfterSVECalleeSaves` to reuse more of the existing GPR deallocation logic, which allows for post-decrements.
MacDue
commented
Sep 2, 2025
Comment on lines
+2601
to
+2607
| // If not, and FPAfterSVECalleeSaves is enabled, deallocate callee-save | ||
| // non-SVE registers to move the stack pointer to the start of the SVE | ||
| // area. | ||
| emitFrameOffset(MBB, std::next(Pop), DL, AArch64::SP, AArch64::SP, | ||
| StackOffset::getFixed(ProloguePopSize), TII, | ||
| MachineInstr::FrameDestroy, false, NeedsWinCFI, | ||
| &HasWinCFI); |
There was a problem hiding this comment.
I've been unable to come up with a reasonable test for this case. With !FPAfterSVECalleeSaves we only get here if:
- We have hazard padding (not applicable)
- Or use some somewhat niche
swiftccortailccfunctionality, which I'm not sure works with SVE.
I wonder if it'd be okay to just do if (FPAfterSVECalleeSaves) reportFatalInternalError(...) instead?
There was a problem hiding this comment.
I wouldn't be surprised if tailcc handling doesn't work with SVE argument/return types, but it shouldn't interact with SVE spill slots outside of this code.
Comment on lines
+2601
to
+2607
| // If not, and FPAfterSVECalleeSaves is enabled, deallocate callee-save | ||
| // non-SVE registers to move the stack pointer to the start of the SVE | ||
| // area. | ||
| emitFrameOffset(MBB, std::next(Pop), DL, AArch64::SP, AArch64::SP, | ||
| StackOffset::getFixed(ProloguePopSize), TII, | ||
| MachineInstr::FrameDestroy, false, NeedsWinCFI, | ||
| &HasWinCFI); |
There was a problem hiding this comment.
I wouldn't be surprised if tailcc handling doesn't work with SVE argument/return types, but it shouldn't interact with SVE spill slots outside of this code.
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.
Previously, we would pop
FixedObject-bytes after deallocating the SVE area, then again as part of the "AfterCSRPopSize". This could be seen in the tests@f6and@f9.This patch removes the erroneous pop, and refactors
FPAfterSVECalleeSavesto reuse more of the existing GPR deallocation logic, which allows for post-decrements.