x64: Fix lowering rules for shld
#12321
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
These failed to account for the edge case of 0/width(type) shifts where the `shld` instruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero. This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year. Closes bytecodealliance#12318
cfallin
approved these changes
Jan 12, 2026
Member
Author
|
@cfallin mind double-checking the Winch-specific commit I added? Looks like Winch's encoding of shift-by-32 was using the immediate form of the instruction but that's not compatible with AArch64's encoding so it ended up encoding an illegal instruction. |
cfallin
approved these changes
Jan 12, 2026
Member
cfallin
left a comment
There was a problem hiding this comment.
Latest commit to Winch looks fine too!
alexcrichton
added a commit
to alexcrichton/wasmtime
that referenced
this pull request
Jan 12, 2026
* x64: Fix lowering rules for `shld` These failed to account for the edge case of 0/width(type) shifts where the `shld` instruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero. This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year. Closes bytecodealliance#12318 * Fix Winch's encoding of shift-by-32
alexcrichton
added a commit
that referenced
this pull request
Jan 12, 2026
* x64: Fix lowering rules for `shld` These failed to account for the edge case of 0/width(type) shifts where the `shld` instruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero. This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year. Closes #12318 * Fix Winch's encoding of shift-by-32
alexcrichton
added a commit
to alexcrichton/wasmtime
that referenced
this pull request
Jan 14, 2026
* x64: Fix lowering rules for `shld` These failed to account for the edge case of 0/width(type) shifts where the `shld` instruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero. This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year. Closes bytecodealliance#12318 * Fix Winch's encoding of shift-by-32
cfallin
pushed a commit
that referenced
this pull request
Jan 14, 2026
* Fix ISLE optimization for vector inputs (#12179) Add some more `fits_in_64` constraints to types to handle the fact that `iconst_u` can't construct a 128-bit output. Closes #12170 * x64: Fix lowering rules for `shld` (#12321) * x64: Fix lowering rules for `shld` These failed to account for the edge case of 0/width(type) shifts where the `shld` instruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero. This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year. Closes #12318 * Fix Winch's encoding of shift-by-32 * Fix ISLE icmp optimization rules for vector inputs (#12335) Closes #12328 * Add release notes --------- Co-authored-by: Zijie Zhao <zzjas98@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
These failed to account for the edge case of 0/width(type) shifts where the
shldinstruction is no longer applicable. Guards are added to ensure that the shift amounts are both greater than zero.This in theory shouldn't have much practical impact since shift-by-zero and shift-by-type-width are both optimized away in the mid-end. That means that this is only possible to expose with opt-level=0 which may help explain why this went undiscovered for ~1 year.
Closes #12318