Add wrapper for mremap

[sporksmith]

Fixes #1295

[sporksmith]

It looks like the libc crate doesn't have mremap in the FreeBSD CI environment:

error[E0425]: cannot find function `mremap` in module `libc`
   --> src/sys/mman.rs:330:21
    |
330 |     let ret = libc::mremap(addr, old_size, new_size, flags.bits(), new_address);
    |                     ^^^^^^
help: a function with a similar name exists
    |
330 |     let ret = libc::mmap(addr, old_size, new_size, flags.bits(), new_address);
    |                     ^^^^
help: possible candidate is found in another module, you can import it into scope
    |
1   | use crate::sys::mman::mremap;
    |

I assume I need to config-guard this wrapper, but I'm not sure how to tell which configs ought to include it. I don't see any such config-guard in the libc crate itself: https://docs.rs/libc/0.2.68/src/libc/unix/linux_like/linux/mod.rs.html#2910

Any ideas?

[sporksmith]

I assume I need to config-guard this wrapper, but I'm not sure how to tell which configs ought to include it. I don't see any such config-guard in the libc crate itself: https://docs.rs/libc/0.2.68/src/libc/unix/linux_like/linux/mod.rs.html#2910

Added a guard to make it Linux-only, since the man page notes that it's Linux-specific.

[sporksmith]

Hmm, I'm fairly certain the CI failure is spurious. It doesn't look related to this PR, and it previously passed on 7a46b52, which should be exactly the same.

I'm happy to rerun if desired; is there a nicer way to force a rerun than pushing an empty commit?

[asomers]

Shouldn't new_address be optional?

[sporksmith]

Yes; I wasn't sure the best way to handle this since Rust doesn't support variadic functions. The argument is ignored if MAP_FIXED isn't in flags; i.e. if MAP_FIXED isn't set the caller can put anything there, and it'll be ignored.

Maybe it'd be a bit clearer to make it an Option<size_t>?

[sporksmith]

I just realized the flags enum is wrong; will fix that and change new_address to be an Option<size_t> (but lmk if there's another solution you recommend)

[sporksmith]

Done. Also added a few tests, though in hindsight maybe these should just be doc tests (and drop the second mremap test)?

[asomers]

Nice tests.

[asomers]

How about a URL to a man page?

[sporksmith]

Done

[asomers]

Looks good. Now it just needs a squash.

[sporksmith]

Looks good. Now it just needs a squash.

Done

[asomers-ax]

bors r+

[bors]

🔒 Permission denied

Existing reviewers: click here to make asomers-ax a reviewer

[asomers]

bors r+

[bors]

Build succeeded:

• continuous-integration/travis-ci/push
• FreeBSD 11.4