Remove redundant drop terminators ahead of mir_coroutine_witnesses by tmandry · Pull Request #155022 · rust-lang/rust

tmandry · 2026-04-09T02:12:07Z

This change is meant to fix #151942. Putting this up to see the perf effects of running it on every MIR body; then I'll work out a strategy to merge it.

Related issues to triage:

Async function's future not Send even when non-Send type is not alive across await expression #151942
Tracking issue for more precise coroutine captures #69663
Future non-Send although non-Send local is dropped before .await #104883
Dropped variables still included in generator type #57478
Moved value is considered as held across an await in async fn #87309
The rules for how non-Send local variables "infect" an async function, making its Future type non-Send also, are stricter than they need to be. #63768: This PR fixes at least part of the issue.
async/await: awaiting inside a match block captures borrow too eagerly #57017

Coroutine traits like Send and Sync are computed based on the mir_coroutine_witnesses query to get the types saved in the coroutine. This query runs the regular coroutine analysis pass on mir_promoted to compute which locals are saved across suspension points.

At this stage of MIR, drop terminators exist along every exit path of the function, even when a local has already been moved from. Those extra drop terminators causes locals (like guard in the case below) to appear live over suspension points, even when they have just been dropped.

async fn bar(mutex: &Mutex<()>) {
    let mut guard = mutex.lock().unwrap();
    loop {
        drop(guard);
        foo().await;
        guard = mutex.lock().unwrap();
    }
}

See this gist for MIR of a simplified version of this function.

Later MIR passes like drop_elaboration and remove_unneeded_drops remove the redundant drop terminators, but they depend on type info to check needs_drop on a place's type. Since mir_coroutine_witnesses query is run inside typeck, we cannot depend on this without introducing cycles.

Instead, we implement a simple pass that removes drop terminators for definitely-uninitialized places. It avoids the need for type info by skipping the check for whether types have destructors.

r? @ghost

tmandry · 2026-04-09T02:14:03Z

@bors try
@rust-timer queue

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rust-bors · 2026-04-09T04:22:22Z

☀️ Try build successful (CI)
Build commit: 5cfde26 (5cfde2679121005039f5e66aeb50f0105493903e, parent: 900485642855f4729d926ecf24680a791f9293cf)

rust-timer · 2026-04-09T05:06:49Z

Finished benchmarking commit (5cfde26): comparison URL.

Overall result: ❌ regressions - 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
@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)	1.1%	[0.2%, 5.1%]	55
Regressions ❌ (secondary)	1.5%	[0.1%, 5.6%]	96
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.1%	[0.2%, 5.1%]	55

Max RSS (memory usage)

Results (primary 6.7%, 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)	6.7%	[0.5%, 14.6%]	66
Regressions ❌ (secondary)	5.1%	[1.2%, 11.8%]	46
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	6.7%	[0.5%, 14.6%]	66

Cycles

Results (primary 2.9%, secondary 2.8%)

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

	mean	range	count
Regressions ❌ (primary)	2.9%	[2.1%, 5.5%]	13
Regressions ❌ (secondary)	3.1%	[1.3%, 4.9%]	16
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.3%	[-2.3%, -2.3%]	1
All ❌✅ (primary)	2.9%	[2.1%, 5.5%]	13

Binary size

Results (primary 0.1%, secondary -0.6%)

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

	mean	range	count
Regressions ❌ (primary)	0.1%	[0.1%, 0.1%]	5
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.6%	[-1.1%, -0.0%]	12
All ❌✅ (primary)	0.1%	[0.1%, 0.1%]	5

Bootstrap: 489.506s -> 497.499s (1.63%)
Artifact size: 395.50 MiB -> 395.57 MiB (0.02%)

tmandry · 2026-04-09T06:13:32Z

@bors try
@rust-timer queue

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rust-bors · 2026-04-09T08:25:52Z

☀️ Try build successful (CI)
Build commit: dc4f33d (dc4f33d8604405ae6c35c052336e3c3eb37de413, parent: d0442e2800d356ae282ddcdbe0eff8798fe648b6)

rust-timer · 2026-04-09T09:06:19Z

Finished benchmarking commit (dc4f33d): comparison URL.

Overall result: ❌ regressions - 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
@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%	[0.2%, 0.8%]	35
Regressions ❌ (secondary)	1.1%	[0.1%, 3.6%]	89
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.0%	[-0.0%, -0.0%]	1
All ❌✅ (primary)	0.5%	[0.2%, 0.8%]	35

Max RSS (memory usage)

Results (primary 6.7%, secondary 4.8%)

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

	mean	range	count
Regressions ❌ (primary)	6.7%	[0.5%, 14.2%]	65
Regressions ❌ (secondary)	5.3%	[0.9%, 11.8%]	42
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.3%	[-2.8%, -2.0%]	3
All ❌✅ (primary)	6.7%	[0.5%, 14.2%]	65

Cycles

Results (primary 1.8%, 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)	1.8%	[1.3%, 2.4%]	2
Regressions ❌ (secondary)	3.4%	[2.6%, 5.4%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.8%	[1.3%, 2.4%]	2

Binary size

Results (secondary -0.6%)

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
Improvements ✅ (secondary)	-0.6%	[-1.1%, -0.0%]	12
All ❌✅ (primary)	-	-	0

Bootstrap: 489.468s -> 489.718s (0.05%)
Artifact size: 395.51 MiB -> 395.60 MiB (0.02%)

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rust-bors · 2026-04-10T18:54:28Z

☀️ Try build successful (CI)
Build commit: 0af2e44 (0af2e4420b164861a451f4f1efe42f951dc8709c, parent: 8317fef20409adedaa7c385fa6e954867bf626fc)

rust-timer · 2026-04-10T19:41:18Z

Finished benchmarking commit (0af2e44): comparison URL.

Overall result: ❌ regressions - 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
@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)	1.1%	[0.1%, 11.9%]	47
Regressions ❌ (secondary)	1.2%	[0.0%, 3.8%]	84
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.1%	[0.1%, 11.9%]	47

Max RSS (memory usage)

Results (primary 5.4%, 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)	5.6%	[0.7%, 15.2%]	82
Regressions ❌ (secondary)	5.3%	[0.7%, 11.0%]	47
Improvements ✅ (primary)	-2.3%	[-2.4%, -2.2%]	2
Improvements ✅ (secondary)	-3.1%	[-3.1%, -3.1%]	1
All ❌✅ (primary)	5.4%	[-2.4%, 15.2%]	84

Cycles

Results (primary 7.5%, secondary -0.4%)

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

	mean	range	count
Regressions ❌ (primary)	7.5%	[2.8%, 12.0%]	4
Regressions ❌ (secondary)	3.8%	[3.6%, 4.3%]	3
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-4.6%	[-5.3%, -4.0%]	3
All ❌✅ (primary)	7.5%	[2.8%, 12.0%]	4

Binary size

Results (secondary -0.6%)

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
Improvements ✅ (secondary)	-0.6%	[-1.1%, -0.0%]	12
All ❌✅ (primary)	-	-	0

Bootstrap: 489.232s -> 489.714s (0.10%)
Artifact size: 395.57 MiB -> 395.73 MiB (0.04%)

tmandry · 2026-04-10T22:53:08Z

@bors try
@rust-timer queue

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rust-bors · 2026-04-11T01:04:55Z

☀️ Try build successful (CI)
Build commit: a0e885e (a0e885e3aa4ea171798eea1b7cfb6285f4a7e1d1, parent: 02c7f9bec0fd583160f8bcccb830216023b07bee)

rust-timer · 2026-04-11T02:16:48Z

Finished benchmarking commit (a0e885e): comparison URL.

Overall result: ❌ regressions - 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
@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)	1.1%	[0.2%, 11.9%]	40
Regressions ❌ (secondary)	1.2%	[0.2%, 3.8%]	75
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.1%	[0.2%, 11.9%]	40

Max RSS (memory usage)

Results (primary 6.5%, secondary 5.4%)

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

	mean	range	count
Regressions ❌ (primary)	6.6%	[0.6%, 14.6%]	71
Regressions ❌ (secondary)	5.5%	[0.6%, 10.8%]	47
Improvements ✅ (primary)	-0.8%	[-0.8%, -0.8%]	1
Improvements ✅ (secondary)	-3.7%	[-3.7%, -3.7%]	1
All ❌✅ (primary)	6.5%	[-0.8%, 14.6%]	72

Cycles

Results (primary 6.6%, secondary 2.8%)

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

	mean	range	count
Regressions ❌ (primary)	6.6%	[2.1%, 13.2%]	5
Regressions ❌ (secondary)	6.1%	[4.0%, 8.1%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.8%	[-3.8%, -3.8%]	1
All ❌✅ (primary)	6.6%	[2.1%, 13.2%]	5

Binary size

Results (primary 0.1%, secondary -0.6%)

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

	mean	range	count
Regressions ❌ (primary)	0.1%	[0.1%, 0.1%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.6%	[-1.1%, -0.0%]	12
All ❌✅ (primary)	0.1%	[0.1%, 0.1%]	2

Bootstrap: 491.759s -> 492.365s (0.12%)
Artifact size: 394.19 MiB -> 394.40 MiB (0.05%)

This reverts commit 619036b.

This should fix performance issues, but includes at least one test change that is likely unacceptable (pass => fail).

tmandry · 2026-04-13T23:05:33Z

@bors try
@rust-timer queue

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rust-bors · 2026-04-14T01:12:35Z

☀️ Try build successful (CI)
Build commit: 764a690 (764a690366084d99756efd2887abda48831d4299, parent: 17584a181979f04f2aaad867332c22db1caa511a)

rust-timer · 2026-04-14T02:28:33Z

Finished benchmarking commit (764a690): 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
@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)	2.7%	[0.2%, 25.9%]	31
Regressions ❌ (secondary)	0.4%	[0.1%, 0.8%]	27
Improvements ✅ (primary)	-0.5%	[-0.8%, -0.2%]	14
Improvements ✅ (secondary)	-0.5%	[-0.9%, -0.2%]	21
All ❌✅ (primary)	1.7%	[-0.8%, 25.9%]	45

Max RSS (memory usage)

Results (primary -1.0%, secondary -0.4%)

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

	mean	range	count
Regressions ❌ (primary)	3.3%	[2.3%, 4.2%]	2
Regressions ❌ (secondary)	6.4%	[6.4%, 6.4%]	1
Improvements ✅ (primary)	-2.0%	[-4.2%, -1.1%]	9
Improvements ✅ (secondary)	-2.7%	[-4.3%, -1.5%]	3
All ❌✅ (primary)	-1.0%	[-4.2%, 4.2%]	11

Cycles

Results (primary 11.4%, secondary -2.7%)

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

	mean	range	count
Regressions ❌ (primary)	11.4%	[2.5%, 26.1%]	7
Regressions ❌ (secondary)	2.3%	[2.3%, 2.3%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-7.7%	[-7.7%, -7.7%]	1
All ❌✅ (primary)	11.4%	[2.5%, 26.1%]	7

Binary size

Results (secondary -0.6%)

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
Improvements ✅ (secondary)	-0.6%	[-1.1%, -0.0%]	12
All ❌✅ (primary)	-	-	0

Bootstrap: 489.468s -> 490.649s (0.24%)
Artifact size: 394.17 MiB -> 394.27 MiB (0.03%)

tmandry added 4 commits April 8, 2026 21:47

Add mir_post_borrowck_cleanup and use it from mir_coroutine_witnesses

c9bcdce

Use mir_post_borrowck_cleanup from everywhere else (coroutines only)

3c2bae3

Run RemoveDeadDrops everywhere

5b6aa44

remove_dead_drops: run simplify_cfg if anything changed

26833b6

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 Apr 9, 2026

tmandry changed the title ~~Remove dead drops~~ Remove redundant drop terminators ahead of mir_coroutine_witnesses Apr 9, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Apr 9, 2026

Auto merge of #155022 - tmandry:remove-dead-drops, r=<try>

5cfde26

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 9, 2026

This comment was marked as off-topic.

Sign in to view

This comment has been minimized.

Sign in to view

rustbot added perf-regression Performance regression. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Apr 9, 2026

This comment was marked as off-topic.

Sign in to view

restrict to coroutines again for perf

11e9fef

tmandry force-pushed the remove-dead-drops branch from 4d04916 to 11e9fef Compare April 9, 2026 06:13

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Apr 9, 2026

Auto merge of #155022 - tmandry:remove-dead-drops, r=<try>

dc4f33d

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 9, 2026

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 10, 2026

rust-bors Bot pushed a commit that referenced this pull request Apr 10, 2026

Auto merge of #155022 - tmandry:remove-dead-drops, r=<try>

0af2e44

Remove redundant drop terminators ahead of mir_coroutine_witnesses

This comment was marked as off-topic.

Sign in to view

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 10, 2026

Disable simplify_cfg

619036b

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Apr 10, 2026

Auto merge of #155022 - tmandry:remove-dead-drops, r=<try>

a0e885e

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 10, 2026

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 11, 2026

tmandry added 2 commits April 13, 2026 15:21

Revert "Disable simplify_cfg"

fe8b3d3

This reverts commit 619036b.

Run RemoveDeadDrops ahead of borrowck

42549e3

This should fix performance issues, but includes at least one test change that is likely unacceptable (pass => fail).

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Apr 13, 2026

Auto merge of #155022 - tmandry:remove-dead-drops, r=<try>

764a690

Remove redundant drop terminators ahead of mir_coroutine_witnesses

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 13, 2026

This comment was marked as off-topic.

Sign in to view

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Apr 14, 2026

Uh oh!

Conversation

tmandry commented Apr 9, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tmandry commented Apr 9, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment was marked as off-topic.

rust-bors Bot commented Apr 9, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Apr 9, 2026

Overall result: ❌ regressions - please read:

Uh oh!

This comment was marked as off-topic.

tmandry commented Apr 9, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented Apr 9, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Apr 9, 2026

Overall result: ❌ regressions - please read:

Uh oh!

This comment was marked as off-topic.

rust-bors Bot commented Apr 10, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Apr 10, 2026

Overall result: ❌ regressions - please read:

Uh oh!

tmandry commented Apr 10, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented Apr 11, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Apr 11, 2026

Overall result: ❌ regressions - please read:

Uh oh!

tmandry commented Apr 13, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment was marked as off-topic.

rust-bors Bot commented Apr 14, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Apr 14, 2026

Overall result: ❌✅ regressions and improvements - please read:

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

tmandry commented Apr 9, 2026 •

edited by rustbot

Loading