MarinPostma requested abrown for a review on PR #10060.
MarinPostma opened PR #10060 from MarinPostma:fix-winch-atomic-and to bytecodealliance:main:
This PR fixes atomic
and/or/xor. The issue was that I was using the operand as a destination for theAtomicRmwSeq, but it is a macro-instruction, and the operand ended up being overwritten. This was caught when I enabled theatomic.wastspec tests in https://github.com/bytecodealliance/wasmtime/pull/10039#issuecomment-2603318507.Investigating into the issue I realized that
AtomicRmwSeqalso had requirement for one of it's registers, but that didn't show up initially, because I got lucky, and the value was popped in the right register every time I tested.The fix is to pop-push the operand to compute the address, and then pass the codegen context to masm to allocate the registers.
MarinPostma requested wasmtime-compiler-reviewers for a review on PR #10060.
MarinPostma requested wasmtime-core-reviewers for a review on PR #10060.
MarinPostma requested fitzgen for a review on PR #10060.
github-actions[bot] commented on PR #10060:
Subscribe to Label Action
cc @saulecabrera
<details>
This issue or pull request has been labeled: "winch"Thus the following users have been cc'd because of the following labels:
- saulecabrera: winch
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
MarinPostma updated PR #10060.
fitzgen submitted PR review:
LGTM, thanks!
fitzgen merged PR #10060.
Last updated: Dec 06 2025 at 06:05 UTC