Less unsafe in `dangling`/`without_provenance` #135344

scottmcm · 2025-01-10T21:10:05Z

This PR was inspired by the new NonNull::without_provenance (cc #135243 (comment)) since it made me realize that we could write NonNull::dangling in completely-safe code using other existing things.

Then doing that led me to a few more places that could be simplified, like now that GVN will optimize Transmute-then-PtrToPtr, we can just implement ptr::without_provenance by calling ptr::without_provenance_mut since the shipped rlib of core ends up with the same single statement as the implementation (thanks to GVN merging the steps) and thus there's no need to duplicate the transmute -- and more importantly, no need to repeat a long safety comment.

There did end up being a couple of other changes needed to avoid exploding certain bits of MIR, though -- like <Box<[i32]>>::default()'s MIR originally got way worse as certain things didn't inline, or had a bunch of extraneous UbChecks -- so there's a couple of other changes to solve that.

rustbot · 2025-01-10T21:10:14Z

r? @ibraheemdev

rustbot has assigned @ibraheemdev.
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

scottmcm · 2025-01-10T21:14:03Z

...r-opt/dataflow-const-prop/default_boxed_slice.main.DataflowConstProp.64bit.panic-unwind.diff

-      bb3: {
-          StorageLive(_10);
-          _10 = const {0x1 as *mut ()};
-          _9 = NonNull::<T>::new_unchecked::precondition_check(const {0x1 as *mut ()}) -> [return: bb4, unwind unreachable];


annot: it was always impossible to trigger this UbCheck -- there's no user value passed to Box<[_]>::default() -- so having it here was just making rustc do more work for no reason.

(Similarly for <&[_]>::default() and friends as well.)

joboet · 2025-01-14T12:28:21Z

This looks very good, great idea!
r=me if perf is green.

@bors try @rust-timer queue

bors · 2025-01-14T12:29:37Z

⌛ Trying commit c0d5988 with merge f97ac68...

Less unsafe in `dangling`/`without_provenance` This PR was inspired by the new `NonNull::without_provenance` (cc rust-lang#135243 (comment)) since it made me realize that we could write `NonNull::dangling` in completely-safe code using other existing things. Then doing that led me to a few more places that could be simplified, like now that GVN will optimize Transmute-then-PtrToPtr, we can just implement `ptr::without_provenance` by calling `ptr::without_provenance_mut` since the shipped rlib of `core` ends up with the same single statement as the implementation (thanks to GVN merging the steps) and thus there's no need to duplicate the `transmute` -- and more importantly, no need to repeat a long safety comment. There did end up being a couple of other changes needed to avoid exploding certain bits of MIR, though -- like `<Box<[i32]>>::default()`'s MIR originally got way worse as certain things didn't inline, or had a bunch of extraneous UbChecks -- so there's a couple of other changes to solve that.

bors · 2025-01-14T14:18:35Z

☀️ Try build successful - checks-actions
Build commit: f97ac68 (f97ac682281f313dccb58a15ed0c3b90972f9d4b)

rust-timer · 2025-01-14T16:47:21Z

Finished benchmarking commit (f97ac68): comparison URL.

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

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 the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	2.2%	[2.2%, 2.2%]	1
Regressions ❌ (secondary)	0.2%	[0.2%, 0.2%]	5
Improvements ✅ (primary)	-0.5%	[-0.8%, -0.3%]	2
Improvements ✅ (secondary)	-1.1%	[-1.6%, -0.5%]	4
All ❌✅ (primary)	0.4%	[-0.8%, 2.2%]	3

Max RSS (memory usage)

Results (primary 0.7%, secondary 2.4%)

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)	5.1%	[4.3%, 6.5%]	3
Regressions ❌ (secondary)	2.4%	[2.4%, 2.5%]	3
Improvements ✅ (primary)	-3.7%	[-4.5%, -2.8%]	3
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.7%	[-4.5%, 6.5%]	6

Cycles

Results (primary 1.6%, secondary -2.6%)

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)	1.6%	[1.6%, 1.6%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.6%	[-2.6%, -2.6%]	1
All ❌✅ (primary)	1.6%	[1.6%, 1.6%]	1

Binary size

Results (primary 0.0%, secondary -1.1%)

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)	0.1%	[0.0%, 0.5%]	19
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.1%	[-0.2%, -0.0%]	31
Improvements ✅ (secondary)	-1.1%	[-2.4%, -0.0%]	19
All ❌✅ (primary)	0.0%	[-0.2%, 0.5%]	50

Bootstrap: 763.089s -> 764.045s (0.13%)
Artifact size: 326.08 MiB -> 326.03 MiB (-0.01%)

scottmcm · 2025-01-15T00:48:58Z

The perf change here looks like inlining churn to me, more than any substantial move one way or the other. Some losses, some wins on icounts, but not a ton of them. Both wins and losses for binary size in both debug and opt.

The one primary regression has a very different codegen schedule:

Which makes it look more to me like it got unlucky on codegen unit splitting more than this change being a fundamental pessimization. Interestingly, while its opt-full is +2.21% on icounts, its opt-incr-full is only +0.07%.

@bors r=joboet

bors · 2025-01-15T00:49:00Z

📌 Commit c0d5988 has been approved by joboet

It is now in the queue for this repository.

Less unsafe in `dangling`/`without_provenance` This PR was inspired by the new `NonNull::without_provenance` (cc rust-lang#135243 (comment)) since it made me realize that we could write `NonNull::dangling` in completely-safe code using other existing things. Then doing that led me to a few more places that could be simplified, like now that GVN will optimize Transmute-then-PtrToPtr, we can just implement `ptr::without_provenance` by calling `ptr::without_provenance_mut` since the shipped rlib of `core` ends up with the same single statement as the implementation (thanks to GVN merging the steps) and thus there's no need to duplicate the `transmute` -- and more importantly, no need to repeat a long safety comment. There did end up being a couple of other changes needed to avoid exploding certain bits of MIR, though -- like `<Box<[i32]>>::default()`'s MIR originally got way worse as certain things didn't inline, or had a bunch of extraneous UbChecks -- so there's a couple of other changes to solve that.

bors · 2025-01-15T03:19:36Z

⌛ Testing commit c0d5988 with merge cfa4bd0...

bors · 2025-01-15T03:53:13Z

💔 Test failed - checks-actions

scottmcm · 2025-01-15T04:29:13Z

Doh, conflicted with myself. Rebased atop #135236 and fixed the MIR.

@bors r=joboet

bors · 2025-01-15T04:29:16Z

📌 Commit 27be78c has been approved by joboet

It is now in the queue for this repository.

scottmcm · 2025-01-15T04:56:46Z

Oops, CI failed
@bors r-

scottmcm · 2025-01-16T07:10:08Z

Didn't change anything in the rebase but CI's passing now 🤷

@bors r=joboet

bors · 2025-01-16T07:10:10Z

📌 Commit c18718c has been approved by joboet

It is now in the queue for this repository.

bors · 2025-01-16T10:52:14Z

⌛ Testing commit c18718c with merge d8a64098c9d0fb25699f657c6efff0bb418f7e18...

bors · 2025-01-16T13:34:19Z

☀️ Test successful - checks-actions
Approved by: joboet
Pushing d8a6409 to master...

rust-timer · 2025-01-16T14:51:19Z

Finished benchmarking commit (d8a6409): 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

This is the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	1.2%	[0.4%, 2.1%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.5%	[-0.9%, -0.3%]	3
Improvements ✅ (secondary)	-1.1%	[-1.6%, -0.6%]	4
All ❌✅ (primary)	0.2%	[-0.9%, 2.1%]	5

Max RSS (memory usage)

Results (primary 0.3%, secondary 1.2%)

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%	[1.3%, 4.8%]	5
Regressions ❌ (secondary)	2.4%	[2.1%, 2.7%]	2
Improvements ✅ (primary)	-4.7%	[-7.2%, -2.7%]	3
Improvements ✅ (secondary)	-1.1%	[-1.1%, -1.1%]	1
All ❌✅ (primary)	0.3%	[-7.2%, 4.8%]	8

Cycles

Results (primary -0.0%, secondary -2.9%)

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)	1.5%	[1.5%, 1.5%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.8%	[-0.8%, -0.8%]	2
Improvements ✅ (secondary)	-2.9%	[-3.4%, -2.1%]	3
All ❌✅ (primary)	-0.0%	[-0.8%, 1.5%]	3

Binary size

Results (primary -0.1%, secondary -1.1%)

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)	0.1%	[0.0%, 0.1%]	15
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.1%	[-0.5%, -0.0%]	33
Improvements ✅ (secondary)	-1.1%	[-2.4%, -0.0%]	19
All ❌✅ (primary)	-0.1%	[-0.5%, 0.1%]	48

Bootstrap: 764.032s -> 762.749s (-0.17%)
Artifact size: 326.06 MiB -> 326.07 MiB (0.00%)

scottmcm · 2025-01-16T21:17:02Z

tests/mir-opt/pre-codegen/slice_iter.enumerated_loop.PreCodegen.after.panic-abort.mir

        }
    }
-    scope 17 (inlined <Enumerate<std::slice::Iter<'_, T>> as IntoIterator>::into_iter) {
+    scope 18 (inlined <Enumerate<std::slice::Iter<'_, T>> as IntoIterator>::into_iter) {


annot: note that there's no changes to the actual MIR blocks (statements/terminators) in any of these pre-codegen tests, just changes to which things ended up inlined along the way.

rustbot assigned ibraheemdev Jan 10, 2025

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Jan 10, 2025

scottmcm commented Jan 10, 2025

View reviewed changes

scottmcm force-pushed the safe-dangling branch from 0c7c016 to c0d5988 Compare January 10, 2025 21:14

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 Jan 14, 2025

joboet assigned joboet and unassigned ibraheemdev Jan 14, 2025

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 Jan 14, 2025

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 15, 2025

This comment has been minimized.

Sign in to view

bors added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Jan 15, 2025

scottmcm force-pushed the safe-dangling branch from c0d5988 to 27be78c Compare January 15, 2025 04:24

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 15, 2025

This comment has been minimized.

Sign in to view

bors added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Jan 15, 2025

scottmcm closed this Jan 15, 2025

scottmcm reopened this Jan 15, 2025

This comment has been minimized.

Sign in to view

Less unsafe in dangling/without_provenance

c18718c

scottmcm force-pushed the safe-dangling branch from 27be78c to c18718c Compare January 16, 2025 06:18

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Jan 16, 2025

bors added the merged-by-bors This PR was explicitly merged by bors. label Jan 16, 2025

bors merged commit d8a6409 into rust-lang:master Jan 16, 2025
7 checks passed

rustbot added this to the 1.86.0 milestone Jan 16, 2025

This was referenced Jan 16, 2025

Add some track_caller info to precondition panics #129658

Draft

Add post-mono MIR passes to make mono-reachable analysis more accurate #131650

Draft

scottmcm commented Jan 16, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Less unsafe in `dangling`/`without_provenance` #135344

Less unsafe in `dangling`/`without_provenance` #135344

scottmcm commented Jan 10, 2025

rustbot commented Jan 10, 2025

scottmcm Jan 10, 2025

joboet commented Jan 14, 2025

This comment has been minimized.

bors commented Jan 14, 2025

bors commented Jan 14, 2025

This comment has been minimized.

rust-timer commented Jan 14, 2025

scottmcm commented Jan 15, 2025

bors commented Jan 15, 2025

bors commented Jan 15, 2025

This comment has been minimized.

bors commented Jan 15, 2025

scottmcm commented Jan 15, 2025

bors commented Jan 15, 2025

This comment has been minimized.

scottmcm commented Jan 15, 2025

This comment has been minimized.

scottmcm commented Jan 16, 2025

bors commented Jan 16, 2025

bors commented Jan 16, 2025

bors commented Jan 16, 2025

rust-timer commented Jan 16, 2025

scottmcm Jan 16, 2025

Less unsafe in dangling/without_provenance #135344

Less unsafe in dangling/without_provenance #135344

Conversation

scottmcm commented Jan 10, 2025

rustbot commented Jan 10, 2025

Choose a reason for hiding this comment

joboet commented Jan 14, 2025

This comment has been minimized.

bors commented Jan 14, 2025

bors commented Jan 14, 2025

This comment has been minimized.

rust-timer commented Jan 14, 2025

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

scottmcm commented Jan 15, 2025

bors commented Jan 15, 2025

bors commented Jan 15, 2025

This comment has been minimized.

bors commented Jan 15, 2025

scottmcm commented Jan 15, 2025

bors commented Jan 15, 2025

This comment has been minimized.

scottmcm commented Jan 15, 2025

This comment has been minimized.

scottmcm commented Jan 16, 2025

bors commented Jan 16, 2025

bors commented Jan 16, 2025

bors commented Jan 16, 2025

rust-timer commented Jan 16, 2025

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

Choose a reason for hiding this comment

Less unsafe in `dangling`/`without_provenance` #135344

Less unsafe in `dangling`/`without_provenance` #135344