Refactor proc_macro_decls_static

[aerooneqq]

During experiments in #158024 I've noticed that we iterate over all hir_free_items and find RustcProcMacroDecls on each of them for rewrite of a single Option which is the result of the query.

In this PR:

• The seach for RustcProcMacroDecls attribute is moved into hir_crate_items and result is cached in ModuleItems, now we do not find RustcProcMacroDecls for each free item, we stop when it is found,
• self.crate_collector is replaced with generic const CRATE_COLLECTOR, maybe will give perf. improvements.

r? @petrochenkov

[petrochenkov]

@bors try @rust-timer queue

[petrochenkov]

Does it need to be a query now when it just reads a field in ModuleItems?

View changes since the review

[aerooneqq]

It can be seen as a projection query over hir_crate_items, so if anything else changes in hir_crate_items and we will use tcx.hir_crate_items(()).rustc_proc_macro_decls() instead of projection query then we would invalidate calling query as hir_crate_items will be red. proc_macro_decls_static it is not used in many places, but I can't predict how it will affect perf if we remove it.

[aerooneqq]

But all accessors (free_items, trait_items, ...) of hir_crate_items are regular methods, not queries, maybe they were queries in the past and were made regular methods for perf reasons, or maybe we can try to make them all queries and see what happens.

[rust-bors]

☀️ Try build successful (CI)
Build commit: 4efc489 (4efc48944e8dc10589dee0844efac3a8737040fe)
Base parent: 8e15021 (8e150217bafcaaaa0c45bf685c55fd56cec48598)

[rust-timer]

Finished benchmarking commit (4efc489): 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)	1.4%	[1.4%, 1.4%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.2%	[-0.3%, -0.2%]	4
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary 1.2%, secondary 5.1%)

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

	mean	range	count
Regressions ❌ (primary)	1.2%	[1.2%, 1.2%]	1
Regressions ❌ (secondary)	5.1%	[2.2%, 7.2%]	7
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.2%	[1.2%, 1.2%]	1

Cycles

Results (primary -0.0%, secondary 3.3%)

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

	mean	range	count
Regressions ❌ (primary)	1.5%	[1.5%, 1.5%]	1
Regressions ❌ (secondary)	3.3%	[2.4%, 4.3%]	6
Improvements ✅ (primary)	-1.5%	[-1.5%, -1.5%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.0%	[-1.5%, 1.5%]	2

Binary size

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

Bootstrap: 481.812s -> 481.916s (0.02%)
Artifact size: 390.71 MiB -> 390.73 MiB (0.01%)

[rustbot]

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[aerooneqq]

Given insignificant perf changes I returned self.crate_collector.
@rustbot ready

[petrochenkov]

@bors r+ rollup

[rust-bors]

📌 Commit 3d19869 has been approved by petrochenkov

It is now in the queue for this repository.