Fix TypeOutlives fast-path

[ShoyuVanilla]

Oh, the fast path is scuffed here. So, I feel fairly confident that we can encounter cases where stalled_on is wrong due to region vars without there being a bug, as in, your change is correct and desirable, but this specific test is just a bug in the fast path:

<MaybeVerify as Service<?infer>>::Future: 'static should simply not be considered stalled. What's the cost of either entirely removing this trivially_stalled_on fast path or fixing it to bail when encountering non-rigid aliases

This change is correct. Please separately do a PR to fix the type-outlives fastpath 😊

Originally posted by @lcnr in #162782 (comment)

The first perf-run result is for always returning Outcome::NoFastPath on any infer var and the second one is for the current HEAD.

We shouldn't stall the outlives goal if the goal contains a non-rigid alias even though the goal contains a non-region infer. Non-fast path will normalize that non-rigid alias and that make the evaluation progress, and I think in theory the fast path shouldn't make observable difference outside the solver.

I'm not entirely sure on disabling fast path only in the presence of non-rigid opaques instead of disabling it entirely for non-region infer, but..

• It's no-less-correct than the status quo
• It roughly matches the actual non-fast path:
  

  rust/compiler/rustc_next_trait_solver/src/solve/mod.rs

  Lines 128 to 129 in dba8825

  	if ty.has_non_region_infer() || ty.has_non_rigid_aliases() {
  	self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS)

• The later has some perf impact hard to ignore for typenum

I couldn't conjure up any case fixed by this PR other than the one in #162782 😅

r? lcnr

[ShoyuVanilla]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: ff34d4b (ff34d4bb1acbe2d82c4a8089354f983739266bc7)
Base parent: 21b707e (21b707e3f97e0b522ebd2f277a862339625ad83f)

[rust-timer]

Finished benchmarking commit (ff34d4b): 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)	2.3%	[0.2%, 3.1%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	1
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

This perf run didn't have relevant results for this metric.

Cycles

Results (primary -2.7%, secondary -20.1%)

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)	-	-	0
Improvements ✅ (primary)	-2.7%	[-3.1%, -2.2%]	2
Improvements ✅ (secondary)	-20.1%	[-20.9%, -19.4%]	2
All ❌✅ (primary)	-2.7%	[-3.1%, -2.2%]	2

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 487.961s -> 491.627s (0.75%)
Artifact size: 406.51 MiB -> 406.65 MiB (0.03%)

[ShoyuVanilla]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 5b8f953 (5b8f953674ed9146b762b22f2c98f0ec7bbba552)
Base parent: a5c10c3 (a5c10c39125788787370a1ab11c954b9dff37859)

[rust-timer]

Finished benchmarking commit (5b8f953): comparison URL.

Overall result: ❌✅ regressions and improvements - no action needed

Benchmarking means the PR may be perf-sensitive. Consider adding rollup=never if this change is not fit for rolling up.

@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)	0.2%	[0.1%, 0.3%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.1%]	1
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (secondary -4.0%)

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)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-4.0%	[-4.1%, -4.0%]	2
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary 4.6%)

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)	10.0%	[7.8%, 12.3%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-6.1%	[-7.1%, -5.4%]	3
All ❌✅ (primary)	-	-	0

Binary size

Results (primary 0.1%, 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)	0.1%	[0.0%, 0.2%]	107
Regressions ❌ (secondary)	0.1%	[0.0%, 0.2%]	91
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[0.0%, 0.2%]	107

Bootstrap: 490.07s -> 491.114s (0.21%)
Artifact size: 406.35 MiB -> 407.26 MiB (0.22%)

[rustbot]

Some changes occurred to the core trait solver

cc @rust-lang/initiative-trait-system-refactor

[lcnr]

this seems right, cool

the remaining perf impact seems like noise

@bors r+ rollup

[rust-bors]

📌 Commit 9270acb has been tentatively approved by lcnr

It will be put into the queue for this repository once PR CI succeeds.