Parallel frontend: reproducible def ids for RPITITs

[susitsm]

Adds test and fixes #162202 by making sure that DefId generating queries in typecheck are called serially

r? @petrochenkov

[petrochenkov]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: bccd62b (bccd62bdc7801ec896a024c8948ac384caa22d0b)
Base parent: c4c4a57 (c4c4a576936e9e67717d0deb8e74e02dd5dd10de)

[rust-timer]

Finished benchmarking commit (bccd62b): comparison URL.

Overall result: ❌ regressions - 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)	0.5%	[0.1%, 1.1%]	66
Regressions ❌ (secondary)	0.7%	[0.0%, 2.8%]	63
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.5%	[0.1%, 1.1%]	66

Max RSS (memory usage)

Results (primary -0.9%, secondary -1.7%)

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

	mean	range	count
Regressions ❌ (primary)	0.7%	[0.5%, 1.0%]	2
Regressions ❌ (secondary)	3.4%	[3.1%, 3.6%]	2
Improvements ✅ (primary)	-1.6%	[-3.3%, -0.5%]	5
Improvements ✅ (secondary)	-4.3%	[-4.7%, -4.0%]	4
All ❌✅ (primary)	-0.9%	[-3.3%, 1.0%]	7

Cycles

Results (primary 0.1%, secondary 5.2%)

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

	mean	range	count
Regressions ❌ (primary)	0.7%	[0.4%, 1.6%]	10
Regressions ❌ (secondary)	5.2%	[5.2%, 5.2%]	1
Improvements ✅ (primary)	-0.7%	[-0.8%, -0.6%]	8
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[-0.8%, 1.6%]	18

Binary size

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

Bootstrap: 480.118s -> 480.397s (0.06%)
Artifact size: 403.78 MiB -> 403.73 MiB (-0.01%)

[petrochenkov]

These calls can be put under if tcx.sess.opts.jobs.frontend.is_some() to avoid regressing single-threaded performance.

View changes since the review

[petrochenkov]

Do assign_anon_assoc_item_def_ids/remap_opaque_captures create the def ids in the same order as single-threaded check_type_wf?

[susitsm]

Not sure, but likely not. The regression seems to come from calling resolve_bound_vars for everything.

[susitsm]

Modified remap_opaque_captures to only run resolve_bound_vars on the parents of opaques. It solved the problem locally on the biggest regression, typenum

[petrochenkov]

Not sure, but likely not.

Ideally, changing -j N to -j M should not change the produced binaries (including when N or M is 1).

[JonathanBrouwer]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 1fd98d3 (1fd98d3af1ab651fe641e1e2be84af00e279bda3)
Base parent: a59b87a (a59b87ab9929f0361031e0d0080a9bf2ea4cdbc0)

[rust-timer]

Finished benchmarking commit (1fd98d3): 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%	[0.2%, 0.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.0%	[-0.0%, -0.0%]	1
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary 1.3%, secondary 3.4%)

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

	mean	range	count
Regressions ❌ (primary)	1.6%	[0.7%, 2.8%]	7
Regressions ❌ (secondary)	3.4%	[2.2%, 4.7%]	3
Improvements ✅ (primary)	-0.4%	[-0.4%, -0.4%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.3%	[-0.4%, 2.8%]	8

Cycles

Results (primary 0.1%)

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

	mean	range	count
Regressions ❌ (primary)	1.0%	[0.5%, 1.6%]	7
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.7%	[-1.1%, -0.4%]	9
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[-1.1%, 1.6%]	16

Binary size

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

Bootstrap: 479.056s -> 485.246s (1.29%)
Artifact size: 403.79 MiB -> 403.80 MiB (0.00%)

[Zoxc]

Do note that I've not intended DefIds to be deterministic. Another strategy would be to make them deterministic when writing metadata. They should also never appear in code output.

[petrochenkov]

Another strategy would be to make them deterministic when writing metadata.

That would be #162809 (see the perf numbers in particular).