bootstrap: use nix crate to get rusage on unix

[GrigorenkoPV]

Part of #109859

[rustbot]

r? @Kobzol

rustbot has assigned @Kobzol.
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

[Kobzol]

Hi, thanks for the PR! While removing unsafe from bootstrap in general is a good idea, this PR adds 4 more dependencies that need to be compiled on every bootstrap rebuild on Linux. I'm not sure if that's a good trade-off, especially when RUSTC_PRINT_STEP_RUSAGE seems kind of niche. I don't know about others, but I haven't personally ever used it, and it also doesn't seem to be used on CI by default.

The dependency cost could be amortized if we used it for more things, of course. I checked #109859 and it seems like nix could also help with getuid, but not with setpriority (based on what is currently offered by the nix crate). I'm still unsure whether it's a good enough value proposition for rusage and getuid.

@onur-ozkan What do you think?

[Kobzol]

For context, prior discussion: #116476

[onur-ozkan]

I am not afraid of having unsafe blocks in bootstrap, but it's definitely an improvement if we can replace them with safe blocks. However, adding dependencies to achieve this brings more problems than solutions. Also, many of these dependencies follow the logic we already do in bootstrap, which is why we decided not to merge #116476.

It seems we should update some items of #109859 as it is leading to incorrect directions that we don't want to follow on bootstrap.

Thanks for the ping! @Kobzol

[GrigorenkoPV]

The dependency cost could be amortized if we used it for more things, of course. I checked #109859 and it seems like nix could also help with getuid, but not with setpriority (based on what is currently offered by the nix crate). I'm still unsure whether it's a good enough value proposition for rusage and getuid.

@onur-ozkan, do I get it right that it's a no from you even in this case?

It seems we should update some items of #109859 as it is leading to incorrect directions that we don't want to follow on bootstrap.

That's what I've been trying to do:
#109859 (comment)

In all honesty, I did not really expect this PR get accepted, I mostly wanted a confirmation that unsafe for rusage on unix is a non-issue and will not be addressed.

[JohnCSimon]

Triage:

I did not really expect this PR get accepted

@GrigorenkoPV Maybe close this PR?

[GrigorenkoPV]

@GrigorenkoPV Maybe close this PR?

I would still love to get a reply to my question above first.

The dependency cost could be amortized if we used it for more things, of course. I checked #109859 and it seems like nix could also help with getuid, but not with setpriority (based on what is currently offered by the nix crate). I'm still unsure whether it's a good enough value proposition for rusage and getuid.

@onur-ozkan, do I get it right that it's a no from you even in this case?

[onur-ozkan]

@onur-ozkan, do I get it right that it's a no from you even in this case?

Yeah:

rust/src/bootstrap/src/lib.rs

Lines 290 to 294 in f04f6ca

	// SAFETY: getuid() system call is always successful and no return value is reserved
	// to indicate an error.
	//
	// For more context, see https://man7.org/linux/man-pages/man2/geteuid.2.html
	let uid = unsafe { libc::getuid() };