qiyuhuating opened PR #14981 from qiyuhuating:fix/aarch64-stack-limit-chain-offset-14946 to bytecodealliance:main:
Fixes #14946.
Problem
When loading a nested stack_limit global value on AArch64, x16 may hold the base of the next load. Materializing a large offset into x16 destroys that base and yields a bad address or a crash.
Change
Use the reserved x17 temporary for offset materialization when the base register is x16; otherwise retain x16.
Regression coverage
Add a Cranelift .clif run test for nested stack-limit loads with large negative offsets, covering AArch64 and x86_64.
Validation
Not compiled or run locally. Please validate with Cranelift filetests and AArch64 CI before marking ready.
qiyuhuating removed the review request for wasmtime-compiler-reviewers from PR #14981.
qiyuhuating requested fitzgen for a review on PR #14981.
qiyuhuating requested wasmtime-compiler-reviewers for a review on PR #14981.
github-actions[bot] added the label cranelift on PR #14981.
github-actions[bot] added the label cranelift:area:aarch64 on PR #14981.
fitzgen commented on PR #14981:
This PR description looks like it was written by an LLM, and then not even reviewed and completed by the human ("Not compiled or run locally. Please validate with Cranelift filetests and AArch64 CI before marking ready.").
I'm going to close this PR for not abiding by our AI policy but you may reopen after updating the PR description, thoroughly reviewing the code before asking maintainers to spend their time on it, and doing the bare minimum to check that the changes work, such as compiling and running tests locally.
:cross_mark: fitzgen closed without merge PR #14981.
Last updated: Oct 31 2026 at 23:03 UTC