Stream: git-wasmtime

Topic: wasmtime / PR #14516 winch: restore aarch64 frame pointer...


view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 16:24):

macovedj opened PR #14516 from macovedj:fix-winch-aarch64-tail-call-sp to bytecodealliance:main:

When a function with stack arguments tail-calls a callee with no stack arguments, Winch’s AArch64 backend advances SP past the saved frame pointer before reloading it. A signal handler running between those instructions can overwrite the saved FP.

This PR computes the callee’s entry SP in a scratch register, restores FP while the old frame is still protected, and then updates SP.

Fixes #14503.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 16:24):

macovedj requested wasmtime-compiler-reviewers for a review on PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 16:24):

macovedj requested cfallin for a review on PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 16:24):

macovedj requested wasmtime-core-reviewers for a review on PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 19:35):

github-actions[bot] added the label winch on PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 03 2026 at 19:36):

github-actions[bot] commented on PR #14516:

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:

To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.

Learn more.
</details>

view this post on Zulip Wasmtime GitHub notifications bot (Oct 04 2026 at 21:17):

:memo: saulecabrera submitted PR review.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 04 2026 at 21:17):

:speech_balloon: saulecabrera created PR review comment:

I believe most of this code has a lot of commonalities with the one defined in with_tail_call_resize? I wonder if we can add a small helper to avoid duplicating this and more importantly to avoid it drifting?

view this post on Zulip Wasmtime GitHub notifications bot (Oct 04 2026 at 21:48):

macovedj updated PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 05 2026 at 16:09):

:thumbs_up: saulecabrera submitted PR review.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 05 2026 at 16:09):

saulecabrera added PR #14516 winch: restore aarch64 frame pointer before advancing SP to the merge queue.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 05 2026 at 16:32):

:check: saulecabrera merged PR #14516.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 05 2026 at 16:32):

saulecabrera removed PR #14516 winch: restore aarch64 frame pointer before advancing SP from the merge queue.


Last updated: Oct 11 2026 at 04:10 UTC