`slice.get(i)` should use a slice projection in MIR, like `slice[i]` does #139118

scottmcm · 2025-03-29T19:05:23Z

slice[i] is built-in magic, so ends up being quite different from slice.get(i) in MIR, even though they're both doing nearly identical operations -- checking the length of the slice then getting a ref/ptr to the element if it's in-bounds.

This PR adds a slice_get_unchecked intrinsic for impl SliceIndex for usize to use to fix that, so it no longer needs to do a bunch of lines of pointer math and instead just gets the obvious single statement. (This is not used for the range versions, since slice[i..] and slice[..k] can't use the mir Slice projection as they're using fenceposts, not indices.)

I originally tried to do this with some kind of GVN pattern, but realized that I'm pretty sure it's not legal to optimize BinOp::Offset to PlaceElem::Index without an extremely complicated condition. Basically, the problem is that the Index projection on a dereferenced slice pointer cares about the metadata, since it's UB to PlaceElem::Index outside the range described by the metadata. But then you cast the fat pointer to a thin pointer then offset it, that ignores the slice length metadata, so it's possible to write things that are legal with Offset but would be UB if translated in the obvious way to Index. Checking (or even determining) the necessary conditions for that would be complicated and error-prone, whereas this intrinsic-based approach is quite straight-forward.

Zero backend changes, because it just lowers to MIR, so it's already supported naturally by CTFE/Miri/cg_llvm/cg_clif.

rustbot · 2025-03-29T19:05:28Z

r? @workingjubilee

rustbot has assigned @workingjubilee.
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 · 2025-03-29T19:05:30Z

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter
gets adapted for the changes, if necessary.

cc @rust-lang/miri, @RalfJung, @oli-obk, @lcnr

scottmcm · 2025-03-29T19:10:04Z

tests/mir-opt/pre-codegen/slice_index.slice_get_mut_usize.PreCodegen.after.panic-abort.mir

    bb2: {
        StorageDead(_3);
-        StorageLive(_7);
-        StorageLive(_5);
-        _5 = &raw mut (*_1);
-        StorageLive(_6);
-        _6 = copy _5 as *mut u32 (PtrToPtr);
-        _7 = Offset(copy _6, copy _2);
-        StorageDead(_6);
-        StorageDead(_5);
-        _8 = &mut (*_7);
-        _0 = Option::<&mut u32>::Some(copy _8);
-        StorageDead(_7);
+        _5 = &mut (*_1)[_2];
+        _0 = Option::<&mut u32>::Some(copy _5);
        goto -> bb3;
    }


annot: this pre-codegen test is the clearest demonstration of what the PR does.

RalfJung · 2025-03-31T06:44:51Z

library/core/src/intrinsics/mod.rs

+    index: usize,
+) -> ItemPtr;
+
+pub trait SliceGetUnchecked<T> {}


A brief doc comment would be good, it is a public trait after all. (I am surprised CI does not complain.)

library/core/src/intrinsics/mod.rs

rustbot · 2025-04-01T23:54:29Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

scottmcm · 2025-04-06T04:19:01Z

Ok, updated:

Redid the comment to talk about the projection primarily, just mentioning MIR as an aside that backends don't need to implement it.
Moved the trait to a private module to emphasize that it's not for public use, gave it some comments, and redid it to be more general so it can also replace the AggregateRawPtr trait.

@rustbot ready

workingjubilee · 2025-05-01T04:45:42Z

?

workingjubilee · 2025-05-01T04:46:46Z

uhh, sorry for not noticing this earlier, can you clarify its status?

@rustbot author

RalfJung · 2025-05-01T07:23:32Z

Looks like we got really unlucky in the linux-futex test... and/or the probabilities shifted since this affects a lot of MIR.

Chances are if you rebase and try again, it'll work this time. I can play around with the numbers in that test later.

scottmcm · 2025-05-01T19:05:30Z

Thanks, Ralf! Looks like it worked out with just the rebase, as you predicted.

@rustbot ready

workingjubilee · 2025-05-30T20:30:37Z

Sorry, I noticed while scanning the queue that this is not rollup=never... it was not run through perf, it seems? My fault. The odds this does not affect perf significantly seem... small. Will reapprove assuming it's perf-neutral.

@bors r-

@bors2 try @rust-timer queue

rust-bors · 2025-05-30T20:30:40Z

⌛ Trying commit 4668124 with merge 462bbcc…

`slice.get(i)` should use a slice projection in MIR, like `slice[i]` does `slice[i]` is built-in magic, so ends up being quite different from `slice.get(i)` in MIR, even though they're both doing nearly identical operations -- checking the length of the slice then getting a ref/ptr to the element if it's in-bounds. This PR adds a `slice_get_unchecked` intrinsic for `impl SliceIndex for usize` to use to fix that, so it no longer needs to do a bunch of lines of pointer math and instead just gets the obvious single statement. (This is *not* used for the range versions, since `slice[i..]` and `slice[..k]` can't use the mir Slice projection as they're using fenceposts, not indices.) I originally tried to do this with some kind of GVN pattern, but realized that I'm pretty sure it's not legal to optimize `BinOp::Offset` to `PlaceElem::Index` without an extremely complicated condition. Basically, the problem is that the `Index` projection on a dereferenced slice pointer *cares about the metadata*, since it's UB to `PlaceElem::Index` outside the range described by the metadata. But then you cast the fat pointer to a thin pointer then offset it, that *ignores* the slice length metadata, so it's possible to write things that are legal with `Offset` but would be UB if translated in the obvious way to `Index`. Checking (or even determining) the necessary conditions for that would be complicated and error-prone, whereas this intrinsic-based approach is quite straight-forward. Zero backend changes, because it just lowers to MIR, so it's already supported naturally by CTFE/Miri/cg_llvm/cg_clif.

rust-bors · 2025-05-30T22:52:55Z

☀️ Try build successful

CI ✅
Build commit: 462bbcc (462bbcc7f1ef36956f33c903203abed871d66de3)

rust-timer · 2025-05-31T06:58:57Z

Finished benchmarking commit (462bbcc): 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)	0.6%	[0.6%, 0.6%]	1
Regressions ❌ (secondary)	0.2%	[0.2%, 0.2%]	1
Improvements ✅ (primary)	-1.0%	[-1.0%, -1.0%]	1
Improvements ✅ (secondary)	-0.5%	[-0.5%, -0.5%]	1
All ❌✅ (primary)	-0.2%	[-1.0%, 0.6%]	2

Max RSS (memory usage)

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

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)	4.4%	[2.3%, 6.5%]	2
Regressions ❌ (secondary)	0.7%	[0.4%, 0.8%]	7
Improvements ✅ (primary)	-3.5%	[-5.0%, -2.6%]	3
Improvements ✅ (secondary)	-1.0%	[-2.9%, -0.4%]	8
All ❌✅ (primary)	-0.3%	[-5.0%, 6.5%]	5

Cycles

Results (primary -0.8%, secondary -0.4%)

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)	-	-	0
Regressions ❌ (secondary)	1.1%	[0.5%, 3.1%]	6
Improvements ✅ (primary)	-0.8%	[-0.8%, -0.8%]	1
Improvements ✅ (secondary)	-1.3%	[-4.1%, -0.5%]	10
All ❌✅ (primary)	-0.8%	[-0.8%, -0.8%]	1

Binary size

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

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)	0.0%	[0.0%, 0.0%]	1
Regressions ❌ (secondary)	7.8%	[7.8%, 7.8%]	1
Improvements ✅ (primary)	-0.2%	[-0.7%, -0.0%]	51
Improvements ✅ (secondary)	-0.4%	[-0.5%, -0.0%]	41
All ❌✅ (primary)	-0.2%	[-0.7%, 0.0%]	52

Bootstrap: 778.613s -> 778.03s (-0.07%)
Artifact size: 370.16 MiB -> 370.29 MiB (0.03%)

scottmcm · 2025-05-31T19:29:15Z

icount changes are neutral (slightly green overall, but too few to really say that matters), bootstrap is slightly faster, and there's a small but pervasive binary size improvement (not just for debug either, with cargo & image improving in opt-full). Seems good to go to me.

@bors r=workingjubilee

bors · 2025-05-31T19:29:17Z

📌 Commit 4668124 has been approved by workingjubilee

It is now in the queue for this repository.

bors · 2025-05-31T21:38:24Z

⌛ Testing commit 4668124 with merge f0999ff...

bors · 2025-06-01T01:01:45Z

☀️ Test successful - checks-actions
Approved by: workingjubilee
Pushing f0999ff to master...

github-actions · 2025-06-01T01:04:53Z

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 4d08223 (parent) -> f0999ff (this PR)

Test differences

Show 104 test diffs

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

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard f0999ffdc4818e498949d3b1f2a0ce6be02a0436 --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

x86_64-apple-2: 3869.9s -> 5396.2s (39.4%)
dist-aarch64-linux: 7977.2s -> 5419.4s (-32.1%)
aarch64-apple: 4583.6s -> 5280.1s (15.2%)
x86_64-apple-1: 7436.7s -> 6663.1s (-10.4%)
x86_64-gnu-stable: 6844.3s -> 7519.1s (9.9%)
mingw-check: 1256.5s -> 1347.6s (7.3%)
dist-x86_64-netbsd: 4935.4s -> 5256.0s (6.5%)
dist-aarch64-apple: 5459.8s -> 5167.6s (-5.4%)
dist-x86_64-apple: 7510.2s -> 7862.0s (4.7%)
dist-apple-various: 6066.1s -> 6334.0s (4.4%)

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 · 2025-06-01T02:33:31Z

Finished benchmarking commit (f0999ff): 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)	-	-	0
Regressions ❌ (secondary)	0.3%	[0.2%, 0.4%]	3
Improvements ✅ (primary)	-1.2%	[-1.3%, -1.1%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.2%	[-1.3%, -1.1%]	2

Max RSS (memory usage)

Results (primary -0.0%, secondary -0.8%)

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)	4.4%	[1.5%, 7.3%]	3
Regressions ❌ (secondary)	1.1%	[0.6%, 2.6%]	4
Improvements ✅ (primary)	-3.3%	[-4.2%, -1.3%]	4
Improvements ✅ (secondary)	-1.8%	[-5.6%, -0.5%]	7
All ❌✅ (primary)	-0.0%	[-4.2%, 7.3%]	7

Cycles

Results (secondary -0.5%)

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
Regressions ❌ (secondary)	1.5%	[0.4%, 2.6%]	5
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.4%	[-6.0%, -0.4%]	12
All ❌✅ (primary)	-	-	0

Binary size

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

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)	0.3%	[0.1%, 0.5%]	2
Regressions ❌ (secondary)	3.9%	[0.1%, 7.8%]	2
Improvements ✅ (primary)	-0.2%	[-1.0%, -0.0%]	54
Improvements ✅ (secondary)	-0.4%	[-0.5%, -0.0%]	40
All ❌✅ (primary)	-0.2%	[-1.0%, 0.5%]	56

Bootstrap: 776.152s -> 774.682s (-0.19%)
Artifact size: 372.23 MiB -> 372.26 MiB (0.01%)

Mark-Simulacrum · 2025-06-02T12:52:46Z

Within noise threshold for (new) regressed benchmark, otherwise see #139118 (comment)

…ic, r=workingjubilee `slice.get(i)` should use a slice projection in MIR, like `slice[i]` does `slice[i]` is built-in magic, so ends up being quite different from `slice.get(i)` in MIR, even though they're both doing nearly identical operations -- checking the length of the slice then getting a ref/ptr to the element if it's in-bounds. This PR adds a `slice_get_unchecked` intrinsic for `impl SliceIndex for usize` to use to fix that, so it no longer needs to do a bunch of lines of pointer math and instead just gets the obvious single statement. (This is *not* used for the range versions, since `slice[i..]` and `slice[..k]` can't use the mir Slice projection as they're using fenceposts, not indices.) I originally tried to do this with some kind of GVN pattern, but realized that I'm pretty sure it's not legal to optimize `BinOp::Offset` to `PlaceElem::Index` without an extremely complicated condition. Basically, the problem is that the `Index` projection on a dereferenced slice pointer *cares about the metadata*, since it's UB to `PlaceElem::Index` outside the range described by the metadata. But then you cast the fat pointer to a thin pointer then offset it, that *ignores* the slice length metadata, so it's possible to write things that are legal with `Offset` but would be UB if translated in the obvious way to `Index`. Checking (or even determining) the necessary conditions for that would be complicated and error-prone, whereas this intrinsic-based approach is quite straight-forward. Zero backend changes, because it just lowers to MIR, so it's already supported naturally by CTFE/Miri/cg_llvm/cg_clif.

rustbot assigned workingjubilee Mar 29, 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 Mar 29, 2025

scottmcm commented Mar 29, 2025

View reviewed changes

RalfJung reviewed Mar 31, 2025

View reviewed changes

rustbot 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-review Status: Awaiting review from the assignee but also interested parties. labels Apr 1, 2025

scottmcm force-pushed the slice-get-unchecked-intrinsic branch from 4cf8aa1 to bb5dd9b Compare April 6, 2025 04:10

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Apr 6, 2025