Enable DestinationPropagation by default

[cjgillot]

This PR proposes to perform destination propagation on MIR. Most of the pass was fully rewritten by @JakobDegen in #96451.

This pass is quite heavy, as it needs to perform and save the results of a full liveness dataflow analysis. This accounts for ~50% of the pass' runtime.

Perf sees a few decent savings in later llvm passes, but also sizeable régressions when there are no savings to balance this pass' runtime.

[cjgillot]

@bors try @rust-timer queue

[bors]

⌛ Trying commit b835228 with merge 2989fd8...

[bors]

☀️ Try build successful - checks-actions
Build commit: 2989fd8 (2989fd868755417871c65b8864528f20125af5bc)

[rust-timer]

Finished benchmarking commit (2989fd8): comparison URL.

Overall result: ❌✅ regressions and improvements - please read the text below

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.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@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)	0.7%	[0.2%, 3.1%]	27
Regressions ❌ (secondary)	0.6%	[0.2%, 2.5%]	36
Improvements ✅ (primary)	-0.4%	[-1.7%, -0.1%]	60
Improvements ✅ (secondary)	-0.7%	[-2.9%, -0.2%]	71
All ❌✅ (primary)	-0.0%	[-1.7%, 3.1%]	87

Max RSS (memory usage)

Results (primary 0.7%, secondary 1.6%)

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

	mean	range	count
Regressions ❌ (primary)	5.6%	[1.7%, 10.5%]	5
Regressions ❌ (secondary)	1.6%	[1.4%, 2.2%]	9
Improvements ✅ (primary)	-4.3%	[-5.9%, -2.5%]	5
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.7%	[-5.9%, 10.5%]	10

Cycles

Results (primary 0.3%, secondary -1.0%)

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%, 1.4%]	5
Regressions ❌ (secondary)	3.1%	[1.8%, 4.4%]	3
Improvements ✅ (primary)	-1.5%	[-1.7%, -1.3%]	2
Improvements ✅ (secondary)	-2.6%	[-3.5%, -1.5%]	8
All ❌✅ (primary)	0.3%	[-1.7%, 1.4%]	7

Binary size

Results (primary -0.0%, secondary 0.1%)

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

	mean	range	count
Regressions ❌ (primary)	0.2%	[0.0%, 0.4%]	22
Regressions ❌ (secondary)	0.4%	[0.0%, 2.0%]	94
Improvements ✅ (primary)	-0.1%	[-1.0%, -0.0%]	54
Improvements ✅ (secondary)	-1.0%	[-13.0%, -0.0%]	28
All ❌✅ (primary)	-0.0%	[-1.0%, 0.4%]	76

Bootstrap: 688.878s -> 687.306s (-0.23%)
Artifact size: 371.92 MiB -> 371.62 MiB (-0.08%)

[rustbot]

r? @petrochenkov

rustbot has assigned @petrochenkov.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[rustbot]

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

[petrochenkov]

r? mir

[oli-obk]

r=me

[cjgillot]

There is a pathological behaviour with cranelift-codegen I'd like to fix first. There are many locals that can be merged with _0. The current version needs to recompute all candidates and all conflicts each time it merges _0 with something.

[cjgillot]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 9298dfa with merge 00ffd2b...

[bors]

☀️ Try build successful - checks-actions
Build commit: 00ffd2b (00ffd2bf2b816d1f0df05611fccdd7dad0a8db22)

[rust-timer]

Finished benchmarking commit (58abde7): comparison URL.

Overall result: ❌✅ regressions and improvements - please read the text below

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.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@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)	0.3%	[0.1%, 1.2%]	62
Regressions ❌ (secondary)	0.4%	[0.0%, 1.6%]	98
Improvements ✅ (primary)	-0.5%	[-2.0%, -0.1%]	29
Improvements ✅ (secondary)	-0.9%	[-2.5%, -0.1%]	46
All ❌✅ (primary)	0.0%	[-2.0%, 1.2%]	91

Max RSS (memory usage)

Results (primary 1.8%)

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

	mean	range	count
Regressions ❌ (primary)	3.3%	[2.2%, 4.5%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.3%	[-1.3%, -1.3%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.8%	[-1.3%, 4.5%]	3

Cycles

Results (primary -2.3%, secondary -4.3%)

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)	2.9%	[2.9%, 2.9%]	1
Improvements ✅ (primary)	-2.3%	[-2.3%, -2.3%]	1
Improvements ✅ (secondary)	-4.8%	[-7.8%, -1.9%]	13
All ❌✅ (primary)	-2.3%	[-2.3%, -2.3%]	1

Binary size

Results (primary -0.1%, 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)	0.1%	[0.0%, 0.4%]	21
Regressions ❌ (secondary)	0.3%	[0.0%, 1.1%]	57
Improvements ✅ (primary)	-0.1%	[-2.4%, -0.0%]	73
Improvements ✅ (secondary)	-0.6%	[-2.4%, -0.0%]	24
All ❌✅ (primary)	-0.1%	[-2.4%, 0.4%]	94

Bootstrap: 467.433s -> 465.765s (-0.36%)
Artifact size: 387.54 MiB -> 387.18 MiB (-0.09%)

[lqd]

It's possible destprop caused #146383, so we may want to hold on this PR until we have more information, or a fix.

[saethlin]

It's possible

I am certain that DestinationPropagation is the cause. I've added some more text to make that clear.

[rustbot]

This PR was rebased onto a different master 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.

[cjgillot]

#146383 is fixed.

[saethlin]

I don't think this is particularly risky; this pass has had a fair bit of fuzzing by Rustlantis because I run with -Zmir-opt-level=4.

@bors r+ rollup=never

[bors]

📌 Commit 44c1a00 has been approved by saethlin

It is now in the queue for this repository.

[bors]

⌛ Testing commit 44c1a00 with merge ce6daf3...

[bors]

☀️ Test successful - checks-actions
Approved by: saethlin
Pushing ce6daf3 to master...

[github-actions]

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing 2ebb126 (parent) -> ce6daf3 (this PR)

Test differences

Show 16 test diffs

Stage 1

• [mir-opt] tests/mir-opt/dest-prop/nrvo_borrowed.rs: [missing] -> pass (J1)
• [mir-opt] tests/mir-opt/dest-prop/nrvo_miscompile_111005.rs: [missing] -> pass (J1)
• [mir-opt] tests/mir-opt/nrvo_miscompile_111005.rs: pass -> [missing] (J1)
• [mir-opt] tests/mir-opt/nrvo_simple.rs: pass -> [missing] (J1)

Stage 2

• [mir-opt] tests/mir-opt/dest-prop/nrvo_borrowed.rs: [missing] -> pass (J0)
• [mir-opt] tests/mir-opt/dest-prop/nrvo_miscompile_111005.rs: [missing] -> pass (J0)
• [mir-opt] tests/mir-opt/nrvo_miscompile_111005.rs: pass -> [missing] (J0)
• [mir-opt] tests/mir-opt/nrvo_simple.rs: pass -> [missing] (J0)

Additionally, 8 doctest diffs were found. These are ignored, as they are noisy.

Job group index

• J0: aarch64-apple, aarch64-gnu, aarch64-gnu-llvm-20-1, aarch64-msvc-1, arm-android, armhf-gnu, dist-i586-gnu-i586-i686-musl, i686-gnu-1, i686-gnu-nopt-1, i686-msvc-1, test-various, x86_64-gnu, x86_64-gnu-llvm-20, x86_64-gnu-llvm-20-2, x86_64-gnu-nopt, x86_64-gnu-stable, x86_64-mingw-1, x86_64-msvc-1
• J1: x86_64-gnu-llvm-20-3

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard ce6daf3d5a5bffb2a00264197f92dc31608df0da --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

1. dist-apple-various: 4430.5s -> 3369.0s (-24.0%)
2. aarch64-apple: 5511.1s -> 6624.5s (20.2%)
3. x86_64-rust-for-linux: 3107.5s -> 2645.6s (-14.9%)
4. x86_64-gnu-llvm-20-1: 3720.3s -> 3228.5s (-13.2%)
5. aarch64-gnu-llvm-20-1: 3787.9s -> 3328.0s (-12.1%)
6. i686-gnu-nopt-1: 8261.1s -> 7258.6s (-12.1%)
7. pr-check-1: 1537.1s -> 1358.2s (-11.6%)
8. i686-gnu-2: 6028.3s -> 5400.6s (-10.4%)
9. x86_64-gnu-debug: 7132.4s -> 6532.8s (-8.4%)
10. arm-android: 6173.2s -> 5670.9s (-8.1%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

[rust-timer]

Finished benchmarking commit (ce6daf3): comparison URL.

Overall result: ❌✅ regressions and improvements - please read the text below

Our benchmarks found a performance regression caused by this PR.
This might be an actual regression, but it can also be just noise.

Next Steps:

• If the regression was expected or you think it can be justified,
  please write a comment with sufficient written justification, and add
  @rustbot label: +perf-regression-triaged to it, to mark the regression as triaged.
• If you think that you know of a way to resolve the regression, try to create
  a new PR with a fix for the regression.
• If you do not understand the regression or you think that it is just noise,
  you can ask the @rust-lang/wg-compiler-performance working group for help (members of this group
  were already notified of this PR).

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

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)	0.3%	[0.1%, 1.1%]	62
Regressions ❌ (secondary)	0.4%	[0.1%, 1.6%]	98
Improvements ✅ (primary)	-0.4%	[-1.7%, -0.1%]	31
Improvements ✅ (secondary)	-0.7%	[-2.5%, -0.1%]	60
All ❌✅ (primary)	0.0%	[-1.7%, 1.1%]	93

Max RSS (memory usage)

Results (primary 1.5%, secondary -2.0%)

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

	mean	range	count
Regressions ❌ (primary)	2.8%	[0.9%, 4.6%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.1%	[-1.1%, -1.1%]	1
Improvements ✅ (secondary)	-2.0%	[-2.4%, -1.5%]	3
All ❌✅ (primary)	1.5%	[-1.1%, 4.6%]	3

Cycles

Results (primary -1.4%, secondary -0.3%)

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)	3.5%	[2.1%, 5.3%]	6
Improvements ✅ (primary)	-1.4%	[-1.4%, -1.4%]	1
Improvements ✅ (secondary)	-3.7%	[-6.1%, -1.3%]	7
All ❌✅ (primary)	-1.4%	[-1.4%, -1.4%]	1

Binary size

Results (primary -0.1%, 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)	0.1%	[0.0%, 0.4%]	25
Regressions ❌ (secondary)	0.3%	[0.0%, 1.0%]	62
Improvements ✅ (primary)	-0.1%	[-2.0%, -0.0%]	56
Improvements ✅ (secondary)	-1.0%	[-3.5%, -0.0%]	23
All ❌✅ (primary)	-0.1%	[-2.0%, 0.4%]	81

Bootstrap: 472.767s -> 472.939s (0.04%)
Artifact size: 390.62 MiB -> 387.93 MiB (-0.69%)

[Kobzol]

The compile-time effects seem to be a wash, while this nicely reduced the binary size of the compiler itself, and is expected to improve runtime performance.

@rustbot label: +perf-regression-triaged