Remove deprecation from SA_ONSTACK; fix hexagon-unknown-linux-musl

[androm3da]

In ad3f860 the RLIM_* values were removed, but leaving behind a deprecation about RLIM_NLIMITS caused it to incorrectly apply instead to SA_ONSTACK.

[rustbot]

r? @JohnTitor

(rustbot has picked a reviewer for you, use r? to override)

[androm3da]

cc @lancethepants
cc @asomers

[asomers]

It looks like ad3f860 also accidentally removed the deprecation notice from RLIM_NLIMITS. Can you please restore it?

[androm3da]

It looks like ad3f860 also accidentally removed the deprecation notice from RLIM_NLIMITS. Can you please restore it?

I didn't look closely enough there, thanks. Will do.

[androm3da]

It looks like ad3f860 also accidentally removed the deprecation notice from RLIM_NLIMITS. Can you please restore it?

Do I apply it to the RLIM_NLIMITS in src/unix/linux_like/linux/arch/generic/mod.rs or other instances as well?

[asomers]

Yes, you should apply the deprecation to all instances of RLIM_NLIMITS.

[androm3da]

Yes, you should apply the deprecation to all instances of RLIM_NLIMITS.

But I suppose applying it causes the build to fail. I could locally bypass this warning but is that really what we want?

error: use of deprecated constant `unix::linux_like::linux::arch::generic::RLIM_NLIMITS`: Not stable across OS versions
   --> src/unix/linux_like/linux/arch/generic/mod.rs:260:59
    |
260 |         pub const RLIMIT_NLIMITS: ::__rlimit_resource_t = RLIM_NLIMITS;
    |                                                           ^^^^^^^^^^^^
    |
note: the lint level is defined here
   --> src/lib.rs:20:38
    |
20  | #![cfg_attr(libc_deny_warnings, deny(warnings))]
    |                                      ^^^^^^^^
    = note: `#[deny(deprecated)]` implied by `#[deny(warnings)]`

[asomers]

Yes, you'll have to put a #[allow(deprecated)] at each one of those sites.

[lancethepants]

There are also references to RLIM_NLIMITS in sparc, mips and powerpc if those need to be labeled deprecated as well.

[asomers]

Yeah, all of them do. There's no way to reliably use that constant, given that the OS can change it at any time.

[androm3da]

Yeah, all of them do. There's no way to reliably use that constant, given that the OS can change it at any time.

Okay - so I'll apply the deprecation to RLIM{IT,}_NLIMITS for to all of the targets, then. And then #[allow(deprecated)] for any uses of either within the libc crate.

[androm3da]

Yeah, all of them do. There's no way to reliably use that constant, given that the OS can change it at any time.

Okay - so I'll apply the deprecation to RLIM{IT,}_NLIMITS for to all of the targets, then. And then #[allow(deprecated)] for any uses of either within the libc crate.

Also: RLIMIT_NLIMITS / RLIM_NLIMITS appears a few times within libc-test/semver/ - but no changes are required here?

[asomers]

Also: RLIMIT_NLIMITS / RLIM_NLIMITS appears a few times within libc-test/semver/ - but no changes are required here?

I'm not sure. I don't know how that part works. But if CI passes, then I wouldn't worry about it.

[androm3da]

Not changed: references to RLIM*_NLIMITS in fuchsia, aix, emscripten - but I suppose these are bound by the same limitation and I should mark them deprecated there too?

[asomers]

Yeah, it ought to be deprecated everywhere.

[androm3da]

cc @cramertj
cc @ecnelises

[asomers]

What about src/unix/bsd ?

[ecnelises]

Looks good to me about AIX. AIX has backward compatible RLIM_NLIMITS_* for elder versions, but not actually changed for very long time.

[androm3da]

What about src/unix/bsd ?

Looks like that's already as-expected.

$ rg -N -B3 '(RLIM|RLIMIT)_NLIM' src/unix/bsd/
src/unix/bsd/netbsdlike/openbsd/mod.rs
pub const AT_REMOVEDIR: ::c_int = 0x08;

#[deprecated(since = "0.2.64", note = "Not stable across OS versions")]
pub const RLIM_NLIMITS: ::c_int = 9;

src/unix/bsd/netbsdlike/netbsd/mod.rs
pub const RLIMIT_NTHR: ::c_int = 11;

#[deprecated(since = "0.2.64", note = "Not stable across OS versions")]
pub const RLIM_NLIMITS: ::c_int = 12;

src/unix/bsd/apple/mod.rs
pub const RLIMIT_NPROC: ::c_int = 7;
pub const RLIMIT_NOFILE: ::c_int = 8;
#[deprecated(since = "0.2.64", note = "Not stable across OS versions")]
pub const RLIM_NLIMITS: ::c_int = 9;

src/unix/bsd/freebsdlike/freebsd/mod.rs
pub const RLIMIT_KQUEUES: ::c_int = 13;
pub const RLIMIT_UMTXP: ::c_int = 14;
#[deprecated(since = "0.2.64", note = "Not stable across OS versions")]
pub const RLIM_NLIMITS: ::rlim_t = 15;

src/unix/bsd/freebsdlike/dragonfly/mod.rs
pub const ELAST: ::c_int = 99;
pub const RLIMIT_POSIXLOCKS: ::c_int = 11;
#[deprecated(since = "0.2.64", note = "Not stable across OS versions")]
pub const RLIM_NLIMITS: ::rlim_t = 12;

[asomers]

What about src/unix/bsd ?

Looks like that's already as-expected.

Ahh, good. But what about Haiku and nto?

[androm3da]

Thanks, @asomers

Are you able to remove the S-waiting-on-review label and replace it with whatever's appropriate? I tried but I don't think I have permission to change this.

[androm3da]

Thanks, @asomers

Are you able to remove the S-waiting-on-review label and replace it with whatever's appropriate? I tried but I don't think I have permission to change this.

gentle ping @asomers

[asomers]

No, @androm3da I don't have any permissions in this repo.

[ecnelises]

@rustbot ready

[ecnelises]

Maybe @JohnTitor ?

[androm3da]

@rustbot review

[androm3da]

I have updated the commit message for ba6e03b to clarfiy that this is a fix for the libstd build for the hexagon-unknown-linux-musl target.

[androm3da]

r? @JohnTitor

[rustbot]

Could not assign reviewer from: JohnTitor.
User(s) JohnTitor are either the PR author, already assigned, or on vacation, and there are no other candidates.
Use r? to specify someone else to assign.

[JohnTitor]

Could you squash commits into one?

[androm3da]

Could you squash commits into one?

Done. Sorry, I'd composed the commits independently in order to make it so that if the RLIM_LIMITS deprecation caused problems, that part could be reverted without reverting the un-deprecation of SA_ONSTACK. But I probably should've looked closer at the contribution guidelines.

[androm3da]

Thanks @JohnTitor, let me know if there's anything else I need to fix. 😁

[JohnTitor]

Thanks! @bors r+

[bors]

📌 Commit b3cf87f has been approved by JohnTitor

It is now in the queue for this repository.

[bors]

⌛ Testing commit b3cf87f with merge 14ef2e5...

[bors]

☀️ Test successful - checks-actions, checks-cirrus-freebsd-12, checks-cirrus-freebsd-13, checks-cirrus-freebsd-14
Approved by: JohnTitor
Pushing 14ef2e5 to main...