musl: fix siginfo_t definition

[dybucc]

Description

#5345 messed up one thing about the musl bindings; Fields __first and
__second from __si_common in the untagged union were made records
instead of untagged unions themselves.

Checklist

• Relevant tests in libc-test/semver have been updated
• Commit messages permalink to headers for added or changed API
• Placeholder or unstable values like *LAST or *MAX have the
  standard doc comment
• Tested locally (cargo test -p libc-test --target mytarget);
  especially relevant for platforms that may not be checked in CI

@rustbot label +stable-nominated

[tgross35]

Huh, why didn't this show up in tests? I guess they're private and the structs here were still smaller than any other union variant?

In any case thanks for the fix, LGTM.

View changes since this review

[rustbot]

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[xtqqczze]

Huh, why didn't this show up in tests?

It showed up in nix-rust/nix#2821 (comment). Going forward, it might be worth testing libc against a few downstream crates as well, since I only happened to catch this because it surfaced there when compiling against the libc-0.2 branch of libc.

[tgross35]

I would love to have that kind of feedback, though I don't think we can commit to maintaining CI for downstream projects here. Maybe interested projects (nix, rustix, etc) could add a repository_dispatch trigger to their workflow that runs on pushes to libc-0.2 (use patch.crates-io to link to the git), then we link it here #5209.

[xtqqczze]

@tgross35 Regarding nix-rust/nix#2821 and bytecodealliance/rustix#1676, is there anyone you can reach out to who might be able to help get these merged soon? Otherwise, these crates will break with the next libc release, and users will need to pin libc.

[xtqqczze]

@tgross35 It would also help if you could backport these changes soon, so I can test against the libc-0.2 branch rather than my own cherry-picked libc-0.2-fix-siginfo_t branch.

[tgross35]

Pinged on the PR for one and on Zulip for the other. Thank you for doing the work to get that together! I'm loosely holding off the release waiting on the rustix fix, but that issue has been open for a long time so I'm not sure it's worth waiting much longer.

[dybucc]

Huh, why didn't this show up in tests? I guess they're private and the structs
here were still smaller than any other union variant?

In any case thanks for the fix, LGTM.

View changes since this
review

We don't test against C anonymous unions. The affected fields are anonymous
unions within anonymous unions (not directly the descendants of the unions
in the siginfo_t, though.)

Though this type seems to be popular enough to merit a test under
libc-test/tests.