bsd: Add missing minherit system call

[oherrala]

Description

Add a missing minherit() system call and constants related to it.

minherit(2) from macOS describes this call as:

The minherit() system call changes the specified pages to have the inheritance characteristic inherit, which can be set to VM_INHERIT_NONE, VM_INHERIT_COPY, or VM_INHERIT_SHARE. Not all implementations will guarantee that the inheritance characteristic can be set on a page basis; the granularity of changes may be as large as an entire region.

Sources

• OpenBSD:
  • https://man.openbsd.org/minherit
  • https://github.com/openbsd/src/blob/4ae131efb048309284c5ee06d14cc10f03403763/sys/sys/mman.h#L111-L117
• FreeBSD:
  • https://man.freebsd.org/cgi/man.cgi?query=minherit
  • https://github.com/freebsd/freebsd-src/blob/de1aa3dab23c06fec962a14da3e7b4755c5880cf/sys/sys/mman.h#L42-L48
• NetBSD:
  • https://man.netbsd.org/minherit.2
  • https://github.com/NetBSD/src/blob/33371e687adff4ba1ef3d243f008a674001b7b51/sys/sys/mman.h#L205-L214
• DragonFly:
  • https://man.bsd.lv/DragonFly-5.6.1/minherit.2
  • https://github.com/DragonFlyBSD/DragonFlyBSD/blob/fcabea4d653da65056585e7572aa19f7a81583ce/sys/sys/mman.h#L55-L60
• MacOS:
  • https://github.com/apple-oss-distributions/xnu/blob/f6217f891ac0bb64f3d375211650a4c1ff8ca1ea/bsd/sys/mman.h#L290
  • https://github.com/apple-oss-distributions/xnu/blob/f6217f891ac0bb64f3d375211650a4c1ff8ca1ea/osfmk/mach/vm_inherit.h#L85-L88

Checklist

• Relevant tests in libc-test/semver have been updated
• No placeholder or unstable values like *LAST or *MAX are
  included (see #3131)
• Tested locally (cd libc-test && cargo test --target mytarget);
  especially relevant for platforms that may not be checked in CI

@rustbot label +stable-nominated

[oherrala]

The arm-linux-androidabi test failed maybe because "The job has exceeded the maximum execution time of 25m0s".

[tgross35]

Would you be able to use mach2 instead? https://docs.rs/mach2/latest/mach2/vm_inherit/index.html. We're tryig to transition away from binding platform API like mach/ and linux/.

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[oherrala]

@tgross35 FWIW minherit is not macOS specific. It originates from OpenBSD and is also in FreeBSD, NetBSD and DragonFly:

• https://man.openbsd.org/minherit
• https://man.netbsd.org/minherit.2
• https://man.freebsd.org/cgi/man.cgi?query=minherit
• https://man.bsd.lv/DragonFly-5.6.1/minherit.2

[tgross35]

Thanks for the links. In that case I think it’s fine here, but would you mind enabling it for those other platforms as well? At least FreeBSD and NetBSD get CI here.

[tgross35]

The non-Apple BSDs seem to use MAP_ as the prefix rather than VM_. I think you'll need to add those constants and update these files.

[tgross35]

@oherrala this seems reasonably fine to me, any reason it's marked as a draft still?

[rustbot]

Some changes occurred in an OpenBSD module

cc @semarie

Some changes occurred in a NetBSD-like module

cc @semarie

[oherrala]

@tgross35 this is needed for one of my free time projects that has been on hold and I've forgotten about this. Sorry about that. Now marked as ready for review.

[semarie]

MAP_INHERIT_ZERO exists in OpenBSD with value 3.
here you omit it, and still declaring it in the semver openbsd.txt file.

View changes since the review

[oherrala]

@semarie Good catch! Fixed:

libc/src/unix/bsd/netbsdlike/mod.rs

Lines 220 to 221 in c538525

	#[cfg(target_os = "openbsd")]
	pub const MAP_INHERIT_ZERO: c_int = 3;

[tgross35]

Could you rebase this as well? There have been some changes to freebsd CI.

[oherrala]

@tgross35 Rebased and also fixed the one issue semarie pointed out.

[semarie]

LGTM.
I would personally prefer to not use #[cfg(target_os = "xxx")] and use proper netbsd/mod.rs and openbsd/mod.rs files.

View changes since this review

[tgross35]

@rustbot author, CI failures are because the semver files aren't accurate

I would personally prefer to not use #[cfg(target_os = "xxx")] and use proper netbsd/mod.rs and openbsd/mod.rs files.

+1 here, could you move these to the separate files? It's fine that SHARE/COPY/NONE will be duplicated.

[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.

[oherrala]

+1 here, could you move these to the separate files? It's fine that SHARE/COPY/NONE will be duplicated.

@tgross35 this is now done.