re-use self.get_all_attrs result for pass indirectly attribute

[folkertdev]

Could be a fix for a potential performance regression reported here #144529 (comment). Apparently the regression later disappeared. Nevertheless, this seems like a decent refactor.

r? @JonathanBrouwer (vaguely attribute-related, maybe there are other optimizations to that PR that we're missing)

[workingjubilee]

@bors try @rust-timer queue

[JonathanBrouwer]

get_all_attrs is pretty fast so I can't really imagine this being a performance improvement, but I think this improves the code anyways so lets merge it
@bors r+ rollup

[bors]

📌 Commit 1a9cc78 has been approved by JonathanBrouwer

It is now in the queue for this repository.

[folkertdev]

Does anything else in #144529 jump out to you?

[JonathanBrouwer]

Hmmm we do need to call pass_indirectly_in_non_rustic_abis on every argument, which does need to look through the flags, but I can't see that being significant and I can't see why it would specifically affect this benchmark.

I just took a look at the implementation of get_all_attrs, and it's just a btreemap lookup so it should not be that slow. I (and humans in general) are not good at predicting performance tho so lets let the numbers speak. We call get_all_attrs a ton of times throughout the compiler though so if that's slow we are leaving a lot of performance on the table.

Alternatively it could've just been noise, since the benchmark did just bounce back. I think it being noise is the most likely answer.

[rust-bors]

☀️ Try build successful (CI)
Build commit: 1869732 (18697328f3073a12811e91ab3af2d58b27d9689f, parent: c90bcb9571b7aab0d8beaa2ce8a998ffaf079d38)

[rust-timer]

Finished benchmarking commit (1869732): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible 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)

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

Cycles

Results (primary -2.4%, 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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.4%	[-2.4%, -2.4%]	1
Improvements ✅ (secondary)	-3.4%	[-3.4%, -3.4%]	1
All ❌✅ (primary)	-2.4%	[-2.4%, -2.4%]	1

Binary size

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

Bootstrap: 475.185s -> 475.633s (0.09%)
Artifact size: 390.81 MiB -> 390.80 MiB (-0.00%)

[JonathanBrouwer]

@bors rollup