Unschedule moved locals by cjgillot · Pull Request #159577 · rust-lang/rust

cjgillot · 2026-07-19T20:18:42Z

Alternative to #159537

This extracts part of the implementation from #125923.

Instead of maintaining a bit-set of locals that are moved from the innermost scope, this PR just removes the locals from the drop schedule.

r? @ghost for perf, and because of a borrowck corner case

And some more cases along the way. We have a specific optimization to avoid generating useless drops, use it. In particular, aggregate construction are very similar to function calls for which this is designed.

cjgillot · 2026-07-19T20:18:52Z

@bors try @rust-timer queue

Unschedule moved locals

rust-bors · 2026-07-19T22:13:05Z

☀️ Try build successful (CI)
Build commit: ab0430c (ab0430cfc09675383a04ad2a1b7877ec314aaac9)
Base parent: 9f36de7 (9f36de775bc636c8e88c31a173c2bcb6995956a0)

rust-timer · 2026-07-19T22:52:50Z

Finished benchmarking commit (ab0430c): comparison URL.

Overall result: ❌✅ regressions and improvements - 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%	[0.4%, 0.5%]	6
Regressions ❌ (secondary)	0.3%	[0.2%, 0.3%]	2
Improvements ✅ (primary)	-0.5%	[-1.0%, -0.2%]	53
Improvements ✅ (secondary)	-0.4%	[-1.0%, -0.2%]	44
All ❌✅ (primary)	-0.4%	[-1.0%, 0.5%]	59

Max RSS (memory usage)

Results (primary -0.7%, secondary -0.7%)

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)	-0.7%	[-0.9%, -0.5%]	3
Improvements ✅ (secondary)	-0.7%	[-0.7%, -0.7%]	1
All ❌✅ (primary)	-0.7%	[-0.9%, -0.5%]	3

Cycles

Results (primary -2.3%, secondary -1.4%)

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)	-2.3%	[-2.4%, -2.3%]	2
Improvements ✅ (secondary)	-1.4%	[-1.4%, -1.4%]	1
All ❌✅ (primary)	-2.3%	[-2.4%, -2.3%]	2

Binary size

Results (primary -0.4%, secondary -0.2%)

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

	mean	range	count
Regressions ❌ (primary)	0.6%	[0.6%, 0.6%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-1.9%, -0.0%]	42
Improvements ✅ (secondary)	-0.2%	[-1.9%, -0.0%]	30
All ❌✅ (primary)	-0.4%	[-1.9%, 0.6%]	43

Bootstrap: 484.26s -> 485.704s (0.30%)
Artifact size: 391.74 MiB -> 391.76 MiB (0.01%)

cjgillot · 2026-07-20T23:46:01Z

@bors try @rust-timer queue

Unschedule moved locals

rust-log-analyzer · 2026-07-21T00:19:38Z

The job aarch64-gnu-llvm-21-1 failed! Check out the build log: (web) (plain enhanced) (plain)

Click to see the possible cause of the failure (guessed by this bot)

test [mir-opt] tests/mir-opt/unusual_item_types.rs ... ok

failures:

---- [mir-opt] tests/mir-opt/nll/region_subtyping_basic.rs stdout ----
5 | '?1 | Local | ['?1]
6 |
7 | Inferred Region Values
- | '?0 | U0 | {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], '?0, '?1}
- | '?1 | U0 | {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], '?1}
+ | '?0 | U0 | {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], bb8[0], bb9[0], bb10[0], '?0, '?1}
+ | '?1 | U0 | {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], bb8[0], bb9[0], bb10[0], '?1}
10 | '?2 | U0 | {bb1[0..=8], bb2[0..=2]}
11 | '?3 | U0 | {bb1[1..=8], bb2[0..=2]}
12 | '?4 | U0 | {bb1[5..=8], bb2[0..=2]}

13 |
14 | Inference Constraints
- | '?0 live at {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0]}
- | '?1 live at {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0]}
+ | '?0 live at {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], bb8[0], bb9[0], bb10[0]}
+ | '?1 live at {bb0[0..=8], bb1[0..=8], bb2[0..=3], bb3[0..=3], bb4[0..=1], bb5[0..=2], bb6[0..=4], bb7[0], bb8[0], bb9[0], bb10[0]}
17 | '?2 live at {bb1[0]}
18 | '?3 live at {bb1[1..=4]}
19 | '?4 live at {bb1[5..=8], bb2[0..=2]}

53         _3 = const ConstValue(Scalar(0x0000000000000000): usize);
54         FakeRead(ForIndex, _1);
55         _4 = Lt(copy _3, const ValTree(Leaf(0x0000000000000003): usize));
-         assert(move _4, "index out of bounds: the length is {} but the index is {}", const ValTree(Leaf(0x0000000000000003): usize), copy _3) -> [success: bb1, unwind: bb7];
+         assert(move _4, "index out of bounds: the length is {} but the index is {}", const ValTree(Leaf(0x0000000000000003): usize), copy _3) -> [success: bb1, unwind: bb8];
57     }
58 
59     bb1: {

72         StorageLive(_7);
73         StorageLive(_8);
74         _8 = copy (*_5);
-         _7 = ConstValue(ZeroSized: fn(usize) -> bool {use_x})(move _8) -> [return: bb3, unwind: bb7];
+         _7 = ConstValue(ZeroSized: fn(usize) -> bool {use_x})(move _8) -> [return: bb3, unwind: bb9];
76     }
77 
78     bb3: {

84 
85     bb4: {
86         StorageLive(_9);
-         _9 = ConstValue(ZeroSized: fn(usize) -> bool {use_x})(const ConstValue(Scalar(0x0000000000000016): usize)) -> [return: bb5, unwind: bb7];
+         _9 = ConstValue(ZeroSized: fn(usize) -> bool {use_x})(const ConstValue(Scalar(0x0000000000000016): usize)) -> [return: bb5, unwind: bb10];
88     }
89 
90     bb5: {

103 
---
107 }
108 


thread '[mir-opt] tests/mir-opt/nll/region_subtyping_basic.rs' panicked at src/tools/compiletest/src/runtest/mir_opt.rs:73:21:
Actual MIR output differs from expected MIR output /checkout/tests/mir-opt/nll/region_subtyping_basic.main.nll.0.64bit.mir
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
---- [mir-opt] tests/mir-opt/nll/region_subtyping_basic.rs stdout end ----

failures:
    [mir-opt] tests/mir-opt/nll/region_subtyping_basic.rs

rust-bors · 2026-07-21T01:42:14Z

☀️ Try build successful (CI)
Build commit: 563ed1f (563ed1fa91907743b81f0cbaac99dc8c8ad29077)
Base parent: 87e5904 (87e5904f5eb6398af6b22eac2802c78934260c48)

rust-timer · 2026-07-21T02:23:07Z

Finished benchmarking commit (563ed1f): comparison URL.

Overall result: ❌✅ regressions and improvements - 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)	1.3%	[0.2%, 11.7%]	65
Regressions ❌ (secondary)	0.7%	[0.1%, 2.0%]	66
Improvements ✅ (primary)	-0.3%	[-0.6%, -0.2%]	10
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	2
All ❌✅ (primary)	1.0%	[-0.6%, 11.7%]	75

Max RSS (memory usage)

Results (primary 1.7%, secondary 2.4%)

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

	mean	range	count
Regressions ❌ (primary)	6.9%	[6.6%, 7.2%]	2
Regressions ❌ (secondary)	3.2%	[2.1%, 6.0%]	4
Improvements ✅ (primary)	-1.7%	[-3.5%, -0.8%]	3
Improvements ✅ (secondary)	-0.7%	[-0.7%, -0.7%]	1
All ❌✅ (primary)	1.7%	[-3.5%, 7.2%]	5

Cycles

Results (primary 10.9%, secondary -1.4%)

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

	mean	range	count
Regressions ❌ (primary)	10.9%	[1.9%, 23.2%]	6
Regressions ❌ (secondary)	5.9%	[5.9%, 5.9%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.8%	[-4.2%, -3.5%]	3
All ❌✅ (primary)	10.9%	[1.9%, 23.2%]	6

Binary size

Results (primary -0.4%, secondary -0.2%)

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

	mean	range	count
Regressions ❌ (primary)	0.6%	[0.6%, 0.6%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-1.9%, -0.0%]	42
Improvements ✅ (secondary)	-0.2%	[-1.9%, -0.0%]	29
All ❌✅ (primary)	-0.4%	[-1.9%, 0.6%]	43

Bootstrap: 486.983s -> 486.82s (-0.03%)
Artifact size: 391.82 MiB -> 391.79 MiB (-0.01%)

cjgillot added 3 commits July 18, 2026 20:50

Mark locals used to construct Aggregates as moved.

60dabb4

And some more cases along the way. We have a specific optimization to avoid generating useless drops, use it. In particular, aggregate construction are very similar to function calls for which this is designed.

Add test.

d11ceca

Use a GrowableBitSet for moved_locals.

e8296a0

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Jul 19, 2026