Appropriately zero extend registers when a guard fails. #1595
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Outdated suggestions cannot be applied.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 handed registers back to AOT with possibly undefined upper bits. As far as we can tell so far, LLVM expects registers to be zero extended [that said, I must admit that I expect that in some cases they should be sign extended, but that's not something I'm sure about].
This commit has the register allocator return a
GuardSnapshot
at each guard in a trace, where we work out what registers would need to be zero extended if that guard fails. When we generate the "get ready for deopt" code, we pass thatGuardSnapshot
back to the register allocator, which generates the necessary zero extension code.