Mark const ptr methods and free functions as inline(always) to match *mut

[zakrad]

Closes #160809. The *const read/read_volatile/read_unaligned methods were #[inline] while the *mut equivalents are #[inline(always)] and verified the *const forms now inline in debug instead of emitting an out of line call.

[rustbot]

r? @nia-e

rustbot has assigned @nia-e.
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

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: libs
• libs expanded to 12 candidates
• Random selection from JohnTitor, Mark-Simulacrum, clarfonthey, nia-e

[nia-e]

this should probably also cover std::ptr::read & co, as per the linked issue, but otherwise looks good ^^

@rustbot author

[rustbot]

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

[zakrad]

this should probably also cover std::ptr::read & co, as per the linked issue, but otherwise looks good ^^

@rustbot author

Thanks! Added #[inline(always)] to the free functions too

[nia-e]

I did a thorough check and there's a fair few other places (though these should also be quite trivial):

• copy_to, copy_to_nonoverlapping, offset_from, and (on slices) len are #[inline(always)] only on *mut T but not on *const T;
• write, write_unaligned, write_volatile, replace, and swap are #[inline(always)] on *mut T but not on the functions in ptr/mod.rs;
• same with as_ptr vs as_mut_ptr and get_unchecked vs get_unchecked_mut on *const T & *mut T;
• the impls of PartialEq and PartialOrd also have different inline attrs on *const T and *mut T.

once these are all inline(always), that should actually be eveything :D ty!

[nia-e]

@bors r+ rollup

[rust-bors]

📌 Commit 1b7548a has been approved by nia-e

It is now in the queue for this repository.

[zakrad]

Thanks for the help, I actually spotted a few of these and was planning for a follow up, but since you listed them all here it is.

[saethlin]

This PR is likely perf sensitive because of the attribute changes and is not a good candidate for adding to a roll-up. I would have marked it iffy and perfed it before merge. But given the recent queue slowness I won't r- it, someone can just perf it after merge.

The always variant of the inline attribute is much less of a hint, and it tends to produce the desired inlining even in unoptimized builds and it changes unoptimized monomorphization to do this. I think it should be used very delicately in libraries.

[nia-e]

I'm happy to do a perf run, but the benchmarks in question were mostly run already in #85218 so I didn't see the case for marking this as perf-sensitive independent of that - I'd be surprised if it's not just more of the same.

if it's useful:
@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 0f1a713 (0f1a713f5caf1917f14cfe76880d8de8a3551187)
Base parent: 969b803 (969b803cbe1d4499f841ae0a49c637d8c70a0458)

[rust-timer]

Finished benchmarking commit (0f1a713): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

@bors rollup=never rustc-perf
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.4%	[0.2%, 0.5%]	7
Improvements ✅ (primary)	-0.7%	[-0.7%, -0.7%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.7%	[-0.7%, -0.7%]	1

Max RSS (memory usage)

Results (primary -0.6%, secondary 0.9%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	2.8%	[2.1%, 3.5%]	2
Regressions ❌ (secondary)	3.6%	[0.4%, 12.5%]	5
Improvements ✅ (primary)	-4.0%	[-5.4%, -2.6%]	2
Improvements ✅ (secondary)	-1.2%	[-4.2%, -0.4%]	6
All ❌✅ (primary)	-0.6%	[-5.4%, 3.5%]	4

Cycles

Results (primary -2.4%, secondary 0.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	1.2%	[0.4%, 4.3%]	10
Improvements ✅ (primary)	-2.4%	[-2.4%, -2.4%]	1
Improvements ✅ (secondary)	-1.2%	[-3.2%, -0.4%]	9
All ❌✅ (primary)	-2.4%	[-2.4%, -2.4%]	1

Binary size

Results (primary 0.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.2%	[0.0%, 0.2%]	5
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-0.4%, -0.4%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.0%	[-0.4%, 0.2%]	7

Bootstrap: 458.441s -> 464.25s (1.27%)
Artifact size: 398.59 MiB -> 398.66 MiB (0.02%)

[nia-e]

yeah, this reads as noise to me. I'm not sure if there's much to read into here ^^