Verify that query keys result in unique dep nodes

[Zoxc]

This implements checking that query keys result into unique dep nodes as mentioned in #112469.

We could do a perf check to see how expensive this is.

r? @michaelwoerister

[oli-obk]

@bors try @rust-timer queue

[bors]

⌛ Trying commit eae7593 with merge d8a5ccf...

[bors]

☀️ Try build successful - checks-actions
Build commit: d8a5ccf (d8a5ccf04db2dbce42c32925c5651bd56b6852b1)

[rust-timer]

Finished benchmarking commit (d8a5ccf): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.0%	[0.2%, 7.6%]	200
Regressions ❌ (secondary)	2.4%	[0.2%, 7.9%]	146
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.0%	[0.2%, 7.6%]	200

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.6%	[0.9%, 7.5%]	32
Regressions ❌ (secondary)	4.7%	[2.2%, 9.2%]	31
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.2%	[-2.2%, -2.2%]	1
All ❌✅ (primary)	3.6%	[0.9%, 7.5%]	32

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.4%	[0.9%, 7.4%]	86
Regressions ❌ (secondary)	3.9%	[1.1%, 7.6%]	55
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.1%	[-2.1%, -2.1%]	1
All ❌✅ (primary)	2.4%	[0.9%, 7.4%]	86

Binary size

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

Bootstrap: 649.534s -> 651.184s (0.25%)
Artifact size: 172.59 MiB -> 173.04 MiB (0.26%)

[Zoxc]

I've made -Z incremental_verify_ich or cfg!(debug_assertions) enable these checks.

[oli-obk]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 8099206 with merge 8455f36...

[bors]

☀️ Try build successful - checks-actions
Build commit: 8455f36 (8455f36495442921566585d0db4fd9d0b344e255)

[rust-timer]

Finished benchmarking commit (8455f36): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.0%	[0.2%, 7.8%]	193
Regressions ❌ (secondary)	2.7%	[0.4%, 7.8%]	129
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.0%	[0.2%, 7.8%]	193

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.6%	[0.8%, 9.2%]	42
Regressions ❌ (secondary)	4.3%	[0.9%, 15.1%]	115
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	3.6%	[0.8%, 9.2%]	42

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.3%	[0.9%, 8.4%]	98
Regressions ❌ (secondary)	4.2%	[1.2%, 8.5%]	61
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.3%	[0.9%, 8.4%]	98

Binary size

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

Bootstrap: 648.949s -> 648.36s (-0.09%)
Artifact size: 310.01 MiB -> 310.47 MiB (0.15%)

[lqd]

(random passerby note about these last results, they show the worst case: for stability reasons the perf runs are done with incremental hash verification turned on all the time, instead of the default lower verification frequency)

[michaelwoerister]

Thanks for the PR, @Zoxc! Looks good to me.

One thing we could try to reduce the perf impact is skipping any DefId query keys because for these we already do exhaustive collision checking when constructing the DefPathHash map. I'm not sure if we are as thorough with other query keys that map to a DefPathHash. But we could do a perf run that skips queries with FingerprintStyle::reconstructible() == true to see what impact that has.

[michaelwoerister]

(random passerby note about these last results, they show the worst case: for stability reasons the perf runs are done with incremental hash verification turned on all the time, instead of the default lower verification frequency)

Thanks for the hint, @lqd! That's really good to know.

[Zoxc]

One thing we could try to reduce the perf impact

Not sure if it's worth optimizing given that it's opt-in.

[michaelwoerister]

Yes, I agree. The overall perf impact is low enough not to make a real difference for compilers with debug assertions enabled and regular compilers won't be affected at all.

@bors r+

As per @lqd's comment, the performance cost will be visible to perf.rlo but that is an artificial side effect of perf.rlo enabling fingerprint verification. Regular users of the compiler won't be affected.

@rustbot label: +perf-regression-triaged

[bors]

📌 Commit 55ba7a7 has been approved by michaelwoerister

It is now in the queue for this repository.

[bors]

⌛ Testing commit 55ba7a7 with merge d3555f3...

[bors]

☀️ Test successful - checks-actions
Approved by: michaelwoerister
Pushing d3555f3 to master...

[rust-timer]

Finished benchmarking commit (d3555f3): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.0%	[0.2%, 7.8%]	193
Regressions ❌ (secondary)	2.4%	[0.3%, 7.8%]	153
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.3%	[-1.3%, -1.3%]	1
All ❌✅ (primary)	2.0%	[0.2%, 7.8%]	193

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.3%	[0.9%, 7.2%]	30
Regressions ❌ (secondary)	7.1%	[2.4%, 11.7%]	10
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	3.3%	[0.9%, 7.2%]	30

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.3%	[0.7%, 7.9%]	117
Regressions ❌ (secondary)	4.3%	[1.4%, 8.1%]	60
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.3%	[0.7%, 7.9%]	117

Binary size

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

Bootstrap: 673.298s -> 674.552s (0.19%)
Artifact size: 310.29 MiB -> 310.77 MiB (0.15%)