bsd: add BPF_WORDALIGN

[sbogomolov]

Description

Add the BPF_WORDALIGN macro on Apple and the freebsdlike platforms. DragonFly writes it as roundup2(x, BPF_ALIGNMENT), which expands to the same expression FreeBSD uses, so the two share one definition.

Sources

Apple BPF_WORDALIGN / BPF_ALIGNMENT (net/bpf.h):

• https://github.com/apple-oss-distributions/xnu/blob/f6217f891ac0bb64f3d375211650a4c1ff8ca1ea/bsd/net/bpf.h#L126-L127

FreeBSD BPF_WORDALIGN / BPF_ALIGNMENT (net/bpf.h):

• https://github.com/freebsd/freebsd-src/blob/2b1df6149e8a2d50a09d13c64d1574dad91e10b1/sys/net/bpf.h#L58-L59

DragonFly BPF_WORDALIGN / BPF_ALIGNMENT (net/bpf.h) and roundup2 (sys/param.h):

• https://github.com/DragonFlyBSD/DragonFlyBSD/blob/c5c76d9c10abfde5bdbf60b1535e3c5484bca042/sys/net/bpf.h#L61-L62
• https://github.com/DragonFlyBSD/DragonFlyBSD/blob/c5c76d9c10abfde5bdbf60b1535e3c5484bca042/sys/sys/param.h#L404

Checklist

• Relevant tests in libc-test/semver have been updated
• No placeholder or unstable values like *LAST or *MAX are included
• Tested locally (cd libc-test && cargo test --target mytarget)

@rustbot label +stable-nominated

[sbogomolov]

Two questions to the reviewer:

1. As I mentioned in the PR description, DragonFly uses a roundup2 helper macro. If you don't like skipping it here - I can move BPF_WORDALIGN under src/new/freebsd and leave DragonFly out of it.
2. Currently BPF_ALIGNMENT for Apple is defined with the wrong type (c_int instead of usize, sizeof(int32_t)). If you want, we can fix it in this PR (as a separate commit).

[tgross35]

Thanks!

Reusing for FreeBSD and Dragonfly seems fine to me. I think it would be good to fix BPF_ALIGNMENT at least in 1.0, please feel free to send a follow up!

View changes since this review

[sbogomolov]

@tgross35 merge queue errors do not seem to be related to my change.

[sbogomolov]

I think it would be good to fix BPF_ALIGNMENT at least in 1.0, please feel free to send a follow up!

#5348

[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.