do not add noalias in return position

[RalfJung]

noalias as a return attribute in LLVM indicates that the returned pointer does not alias anything else that is reachable from the caller, including things reachable before this function call. This is clearly not the case with a function like fn id(Box<T>) -> Box<T>, so we cannot use this attribute.

Fixes rust-lang/unsafe-code-guidelines#385 (including an actual miscompilation that @comex managed to produce).

[rustbot]

r? @jackh726

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

[nikic]

I think this is the right thing to do, and I intuitively wouldn't expect this to have a huge impact, because we still have noalias on the __rustc_alloc return, which is where it really matters. Still, let's do a sanity check...

@bors try @rust-timer queue

[bors]

⌛ Trying commit e7cad62 with merge 4921783924d104b60b4219521643e1d57e96d94e...

[bors]

☀️ Try build successful - checks-actions
Build commit: 4921783924d104b60b4219521643e1d57e96d94e (4921783924d104b60b4219521643e1d57e96d94e)

[rust-timer]

Finished benchmarking commit (4921783924d104b60b4219521643e1d57e96d94e): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

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

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-3.0%	[-3.6%, -2.6%]	5
Improvements ✅ (secondary)	-2.6%	[-3.6%, -0.8%]	31
All ❌✅ (primary)	-3.0%	[-3.6%, -2.6%]	5

Cycles

This benchmark run did not return any relevant results for this metric.

[nikic]

Okay, that was a pretty clear answer...

@bors r+

[bors]

📌 Commit e7cad62 has been approved by nikic

It is now in the queue for this repository.

[bors]

⌛ Testing commit e7cad62 with merge 442f997...

[bors]

☀️ Test successful - checks-actions
Approved by: nikic
Pushing 442f997 to master...

[rust-timer]

Finished benchmarking commit (442f997): comparison URL.

Overall result: ✅ improvements - no action needed

@rustbot label: -perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.2%	[-1.3%, -1.2%]	3
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.6%	[2.6%, 2.6%]	1
Regressions ❌ (secondary)	2.3%	[1.2%, 3.5%]	21
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.0%	[-3.0%, -3.0%]	1
All ❌✅ (primary)	2.6%	[2.6%, 2.6%]	1

Cycles

This benchmark run did not return any relevant results for this metric.

[pnkfelix]

I am now curious: We briefly touched on this PR (#106371) while discussing rust-lang/unsafe-code-guidelines#196, and I cannot immediately tell from a cursory look: Did this change end up removing noalias from every return position, include the return position of allocation functions?

@RalfJung said in the meeting that they believed we are still emitting noalias for return position of allocation functions, but skimming over this PR it is not obvious to me that is still the case. (Its possible there's a completely separate path that is being exercised for those functions, which would explain my confusion here.)

((I now also see there was significant discussion of the original motivations for return-position noalias being pretty narrow cases involving malloc and thin wrappers around malloc... as @RalfJung notes, "Any time that there is a "life after deallocation" inside the abstract machine, we cannot use this attribute." ... I haven't worked through the implications of that, in terms of what it means for whether we can continue to apply the attribute to Rust allocation methods themselves...))

[RalfJung]

__rust_alloc has the #[rustc_allocator] attribute which still lets us add noalias (even though for custom allocators this rests in somewhat shaky ground):

rust/compiler/rustc_codegen_llvm/src/attributes.rs

Line 380 in 2a643b2

attributes::apply_to_llfn(llfn, AttributePlace::ReturnValue, &[no_alias]);

[bjorn3]

I believe I broke applying this attribute for #[global_allocator] in #86844.