Fix PMO to not scalarize empty tuple #59382
Merged
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
|
@swift-ci test |
meg-gupta
changed the title
|
@swift-ci smoke test |
meg-gupta
force-pushed
the
fixemptytuple
branch
2 times, most recently
from
f73dc78 to
4bcd799
Compare
Currently PMO tries to scalarize empty tuple, and ends up deleting store of an empty tuple. This causes a cascade of problems. Mem2Reg can end up seeing loads without stores of empty tuple type, and creates undef while optimizing the alloca away. This is bad, we don't want to create undef's unnecessarily in SIL. Some optimization can queiry the function or the block on the undef leading to nullptr and a compiler crash. Fixes rdar://94829482
|
Are we now saying that an undef address is illegal SIL? I'm ok with that as long as the SIL verifier enforces it. |
|
@atrick Maybe my PR description is misleading, I was trying to explain the sequence of problems that can happen. This PR is just trying to avoid deleting stores while there maybe address reads. This part of PMO is replacing a store of a tuple, into stores of tuple elements. It ends up inadvertently deleting the store of an empty tuple (because it doesn't have any elements). |
|
@swift-ci smoke test |
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Outdated suggestions cannot be applied.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.
Currently PMO tries to scalarize empty tuple, and ends up deleting
store of an empty tuple. This causes a cascade of problems.
Mem2Reg can end up seeing loads without stores of empty tuple type,
and creates undef while optimizing the alloca away.
This is bad, we don't want to create undef's unnecessarily in SIL.
Some optimization can query the function or the block on the undef leading to nullptr
and a compiler crash.
Fixes rdar://94829482