linux/gnu: add utmp(5) default paths

[lucab]

This adds utmp(5) default paths on Linux with GNU libc.

Ref: https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/paths.h;h=1342ab3a96ab12065311b718e57e7235e07587f8;hb=HEAD

[rust-highfive]

r? @gnzlbg

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

[gnzlbg]

These should be extern statics without a default value.

[lucab]

@gnzlbg thanks for the reviews!

I have a couple of doubts when trying to apply your suggestion above (and I didn't spot any similar example in libc):

• what's the actual type signature? I mean, is it a pointer or an actual array of bytes? If the latter, is the size harcoded based on the current observed value?
• for the final linkage, where would the symbols actually be defined?

I tried defining a bare extern static, but then consumers fail to link with undefined reference to _PATH_UTMP.
I ran an objdump -Ttx on my local glibc, and indeed I cannot see the symbol in there.

That seems consistent to me with the fact that those are C pre-processor definitions and not actual constants, so they may not end up anywhere in the (public) ABI.

I also run latest bindgen on the header and its suggestion seems to agree on re-defining them:

pub const _PATH_UTMP: &'static [u8; 14usize] = b"/var/run/utmp\0";

[gnzlbg]

What bindgen suggests sounds correct.

[lucab]

Ack. I've updated the PR with that.

[gnzlbg]

There is an assertion triggering in ctest, which I believe is maybe due to ctest not supporting byte strings, e.g., you probably want to add a test for byte strings here: https://github.com/gnzlbg/ctest/blob/master/testcrate/src/t1.rs#L6 and then work up your way to adding support to ctest for these. Shouldn't be too hard since &'static str is already supported.

[joshtriplett]

Because these are preprocessor definitions, they don't really have a precise type. There isn't a symbol being bound to here, and nothing will break if we give them a better type for Rust. As a result, I think it'd make more sense to define these as &'static str, and adapt ctest to match.

(Also, as a side note, whatever's depending on utmp/wtmp/btmp paths, please make it cope with those files not existing, as they aren't required to exist.)

I don't think we should add all of these, unless someone specifically needs them. Adding the ones people specifically ask for, that aren't historical relics, seems reasonable.

[bors]

☔ The latest upstream changes (presumably #1999) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[JohnTitor]

Closing as this is inactive for years and in favor of Josh's comment. Feel free to re-open if you really want them, we can re-consider it anytime.