Add missing definitions on NetBSD

[0323pin]

This PR adds support for:
CLOCK_PROCESS_CPUTIME_ID
CLOCK_THREAD_CPUTIME_ID
sysctlnametomib

It replaces the following closed PRs:
#3926
#3923

Sorry for the back and forward actions.

[rustbot]

r? @JohnTitor

rustbot has assigned @JohnTitor.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[0323pin]

@rustbot label stable-nominated

[0323pin]

r? @tgross35

It looks like @JohnTitor is busy would you mind reviewing?

[tgross35]

I keep an eye on all the PRs even if they aren't assigned to me, just a bit of a slow turn time on this repo. It looks like other BSDs (Apple and NetBSD) have these already, could it instead just be moved to bsd/mod.rs?

Also, got a link to the relevant headers and/or docs?

@rustbot author (just comment @rustbot ready if you either make or decline the change)

[0323pin]

I keep an eye on all the PRs even if they aren't assigned to me

Thanks for reviewing the PR.

just a bit of a slow turn time on this repo.

Yeah, sorry I didn't meen to be pushy.

It looks like other BSDs (Apple and NetBSD) have these already, could it instead just be moved to bsd/mod.rs?

Hmm ... I think you are mixing these a bit.
The constants are defined for OpenBSD but, missing for NetBSD. Hence, this PR. As for sysctlnametomib function, this exists on NetBSD but, not on OpenBSD. So, adding these to bsd/mod.rs would break OpenBSD.

Also, got a link to the relevant headers and/or docs?

Sure,
CLOCK_...
sysctlnametomib

~> uname -rsv
NetBSD 10.99.12 NetBSD 10.99.12 (GENERIC) #0: Wed Sep 18 01:34:08 UTC 2024  mkrepro@mkrepro.NetBSD.org:/usr/src/sys/arch/amd64/compile/GENERIC
~> grep -r "sysctlnametomib" /usr/include/sys/
/usr/include/sys/sysctl.h:int	sysctlnametomib(const char *, int *, size_t *);
~> grep -r "CLOCK_PROCESS_CPUTIME" /usr/include/sys/
/usr/include/sys/time.h:#define CLOCK_PROCESS_CPUTIME_ID	0x40000000
~> grep -r "CLOCK_THREAD_CPUTIME" /usr/include/sys/
/usr/include/sys/time.h:#define CLOCK_THREAD_CPUTIME_ID		0x20000000

[0323pin]

@rustbot ready

[tgross35]

Thanks for confirming everything, and for the links to check against. Maybe sysctlnametomib should match the NetBSD names of (sname, name, namelenp) rather than (name, mibp, sizep) used by FreeBSD and Apple?

Either way lgtm.

[0323pin]

@tgross35

Maybe sysctlnametomib should match the NetBSD names of (sname, name, namelenp) rather than (name, mibp, sizep) used by FreeBSD and Apple?

Duh ... you're probably right. Let me check this tonight. If I push a change, you want the commits squashed, right?

[tgross35]

Duh ... you're probably right. Let me check this tonight. If I push a change, you want the commits squashed, right?

Yes please, GH's squash+merge doesn't work great with the merge queue.

@rustbot author

[0323pin]

@tgross35 Sorry, it seems like my git skills aren't that great 😞

Anything against if, I close the PR, redo it all and re-submit?

[tgross35]

Lol, git can be unfriendly :) usually you just do git rebase -i main which will open an editor where you should see your two commits. Then, for everything except the first line (oldest commit), replace pick with either f/fixup (= combine commits with the above but discard the message, which is what you want here) or s/squash (same thing but append the message). Save, and then git push --force-with-lease.

(of course a new PR is fine worst case, but squashing is usually easier if you can get it working)

[0323pin]

Not far from it :)
I think I've just missed git push --force-with-lease.

[0323pin]

@tgross35 That was it. Using --force-with-lease did the trick, thank you!

@rustbot ready

[tgross35]

Looks great, thanks!

[0323pin]

Thank you!