Only honor -Zincremental-verify-ich for incremental compiles

[xmakro]

For non-incremental compiles the flag has no purpose. It triggers a query key verification sweep that costs ~1.3% of instructions on non-incr full measurements. This is overhead that exists only in the benchmark, not in real builds.

[Kobzol]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: f2d657e (f2d657ea99c440e3e341998de02b54cba54d84bc)
Base parent: 1498f99 (1498f9968f75e8e5ded6892b9c97bfe66b5a966f)

[rust-timer]

Finished benchmarking commit (f2d657e): comparison URL.

Overall result: ✅ improvements - no action needed

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.

@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)	0.2%	[0.2%, 0.3%]	16
Improvements ✅ (primary)	-3.2%	[-6.7%, -0.5%]	71
Improvements ✅ (secondary)	-2.9%	[-10.3%, -0.3%]	105
All ❌✅ (primary)	-3.2%	[-6.7%, -0.5%]	71

Max RSS (memory usage)

Results (primary -2.4%, secondary -3.0%)

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)	2.1%	[2.1%, 2.1%]	2
Improvements ✅ (primary)	-2.4%	[-3.8%, -1.1%]	14
Improvements ✅ (secondary)	-3.3%	[-6.3%, -1.6%]	46
All ❌✅ (primary)	-2.4%	[-3.8%, -1.1%]	14

Cycles

Results (primary -3.9%, secondary -4.5%)

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)	-3.9%	[-7.1%, -2.0%]	31
Improvements ✅ (secondary)	-4.5%	[-10.5%, -1.8%]	46
All ❌✅ (primary)	-3.9%	[-7.1%, -2.0%]	31

Binary size

Results (primary 0.0%, secondary 0.0%)

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

	mean	range	count
Regressions ❌ (primary)	0.0%	[0.0%, 0.0%]	3
Regressions ❌ (secondary)	0.0%	[0.0%, 0.0%]	3
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.0%	[0.0%, 0.0%]	3

Bootstrap: 486.272s -> 487.245s (0.20%)
Artifact size: 387.67 MiB -> 388.11 MiB (0.11%)

[bjorn3]

with_stable_hashing_context is also used in the legacy symbol mangling, for type id hashes and most importantly the crate hash which should depend on all compiler inputs that could potentially affect the output, which includes spans.

View changes since the review

[xmakro]

Thanks! I think get_symbol_hash and type_id_hash already hash with spans disabled, but I missed the crate hash which needs the spans. I'll follow #94878 (crate hash from the encoded metadata) first and might revisit this afterwards.

I reduced this PR now to just the -Zincremental-verify-ich change, which turned out to be a good part of the perf win, but is misleading because it is a perf artifact, see rust-lang/rustc-perf#2510

[xmakro]

r? @Zalathar

[rustbot]

Zalathar is not on the review rotation at the moment.
They may take a while to respond.