Add unstable loop unrolling hint attributes by saethlin · Pull Request #156816 · rust-lang/rust

saethlin · 2026-05-22T08:56:02Z

Tracking issue: #156874

This adds as new attribute #[unroll]/#[unroll(full)]/#[unroll(never)]/#[unroll(16)] (or any u32).

#[unroll] is behind a new feature gate #![feature(loop_hints)] because I intend to add an attribute for loop vectorization as well. If a user wants to turn off loop unrolling to locally minimize code size, LLVM may vectorize the loop even though it isn't unrolled which can produce a similar code size explosion.

rustbot · 2026-05-31T20:58:49Z

Some changes occurred in compiler/rustc_passes/src/check_attr.rs

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_hir/src/attrs

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

Some changes occurred in match lowering

cc @Nadrieril

Some changes occurred in coverage instrumentation.

cc @Zalathar

rustbot · 2026-05-31T20:58:51Z

r? @folkertdev

rustbot has assigned @folkertdev.
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

Why was this reviewer chosen?

The reviewer was selected based on:

Owners of files modified in this PR: compiler
compiler expanded to 73 candidates
Random selection from 17 candidates

JonathanBrouwer · 2026-05-31T21:01:32Z

(would like to take a look at this as well, should have time in the next few days)

folkertdev

Looks reasonable, I'll let Jonathan take a closer look at the attribute stuff.

Maybe I'm missing something or am just used to different parts of the code base, but some style things seemed off to me. Feel free to disregard that though, I guess.

View changes since this review

folkertdev · 2026-05-31T21:40:15Z

+#[no_mangle]
+pub fn unroll_count() {
+    // CHECK-LABEL: @unroll_count
+    // CHECK: !llvm.loop ![[COUNT:[0-9]+]]


is checking the actual number tricky? or does it not map one-to-one?

I am checking the number, it's at the very bottom of this file. Loop metadata looks like this:

bb7: ; preds = %bb3 call void @maybe_has_side_effect() #5 br label %bb1, !llvm.loop !11 ... !11 = distinct !{!11, !12} !12 = !{!"llvm.loop.unroll.count", i32 5}

JonathanBrouwer · 2026-06-01T09:17:13Z

Wow wonder what causes that insane improvement in image, this only adds a new attribute so that's quite unexpected right?

folkertdev · 2026-06-01T09:28:44Z

It might be due to layout changes because some types got bigger so now something is aligned that wasn't before? Still that is a massive change.

JonathanBrouwer

I've not reviewed everything yet, but I think these are the biggest points.
PR looks good in general and happy to see this new feature :)

View changes since this review

JonathanBrouwer · 2026-06-01T09:20:17Z

    MacroCall,
    Crate,
    Delegation { mac: bool },
+    ForLoop,


Could you add this information as a field of Target::Expr, rather than its own target type?
I think this is confusing because now not all expressions produce Target::Expr

Furthermore, does it make sense to combine these three to just Loop, or do they need to be separate targets?

For #[loop_match] it's kind of nice to just have a Loop one corresponding to loop {}, but that's validated separately further down the line too.

Ah right if that attribute is only valid on Loop then keeping the seperate targets is perfectly reasonable

JonathanBrouwer · 2026-06-01T09:21:30Z

    use super::*;
    // tidy-alphabetical-start
-    static_assert_size!(BasicBlockData<'_>, 152);
+    static_assert_size!(BasicBlockData<'_>, 160);


The perf result of (probably) this change is a bad sad, can we improve that?

I looked through the cachegrind diffs and I'm pretty sure the perf impact is mostly caused by adding a new field to an encoded struct (MIR Terminators), not by the size increase of the struct.

I did think about where to stash the data for a while. In THIR I was able to create a single collection for each Body, which means that the effect when not used is a single empty collection (or single zero byte when encoding/decoding). But in MIR, I want this to have a chance of surviving MIR optimizations, so putting attributes on something like a FxHashMap<BasicBlock, Vec<Attribute>> would mean that we'd need to repair that mapping any time a basic block was added or removed by a MIR transform. That sounds hard to maintain. So I think the only viable approach is to attach this to the Goto Terminators somehow, and there are often many terminators per basic block, so it's not shocking that the overhead surfaces here.

There are things I could do here so I'll try one, but you might not like it 😛

JonathanBrouwer · 2026-06-01T11:15:17Z

+                }
+            }
+            ArgParser::NameValue(_) => {
+                cx.adcx().warn_ill_formed_attribute_input(ILL_FORMED_ATTRIBUTE_INPUT);


This can be .expected_list_or_no_args

We should start linting against weird parsing practices :3 we totally can detect these. @JonathanBrouwer

JonathanBrouwer · 2026-06-01T11:17:28Z

+                    }
+                }
+
+                match l.meta_item().and_then(|i| i.path().word_sym()) {


This forgets to check whether l has any arguments, use the new meta_item_no_args method introduced in #155193 (which is in the queue atm)

JonathanBrouwer · 2026-06-01T11:20:15Z

 print_tup!(A B C D E F G H);
 print_skip!(Span, (), ErrorGuaranteed, AttrId);
-print_disp!(u8, u16, u32, u128, usize, bool, NonZero<u32>, Limit);
+print_disp!(u8, u16, u32, u64, u128, usize, bool, NonZero<u32>, Limit);


Is the u64 used anywhere?

Whoops. Previously I was accepting any u64 in the unroll count, but I changed to u32 because that's what clang does.

saethlin · 2026-06-01T15:34:05Z

I think the image perf result is just something broken in collection. https://rust-lang.zulipchat.com/#narrow/channel/247081-t-compiler.2Fperformance/topic/sus.20perf.20results/near/599190868

Kobzol · 2026-06-01T15:35:46Z

Let's try again to see if it persists.

@bors try @rust-timer queue

Add unstable loop unrolling hint attributes

rust-bors · 2026-06-01T17:47:40Z

☀️ Try build successful (CI)
Build commit: 57c302e (57c302e0d23431b9eb5bf77a5403230fb921e506, parent: 968d50ad35115bc2c8c19cb9039f7ed3dfe56a81)

rust-timer · 2026-06-01T18:28:08Z

Finished benchmarking commit (57c302e): 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)	0.2%	[0.1%, 0.3%]	21
Regressions ❌ (secondary)	0.2%	[0.0%, 0.3%]	21
Improvements ✅ (primary)	-36.8%	[-77.8%, -11.1%]	4
Improvements ✅ (secondary)	-0.4%	[-0.6%, -0.1%]	3
All ❌✅ (primary)	-5.7%	[-77.8%, 0.3%]	25

Max RSS (memory usage)

Results (primary -2.1%, 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)	1.7%	[0.6%, 4.6%]	10
Regressions ❌ (secondary)	1.6%	[0.5%, 2.8%]	12
Improvements ✅ (primary)	-14.8%	[-24.2%, -1.9%]	3
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.1%	[-24.2%, 4.6%]	13

Cycles

Results (primary -37.5%)

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)	-37.5%	[-78.5%, -11.0%]	4
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-37.5%	[-78.5%, -11.0%]	4

Binary size

Results (primary 0.3%, 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)	0.3%	[0.0%, 0.7%]	87
Regressions ❌ (secondary)	0.4%	[0.0%, 1.4%]	70
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.3%	[0.0%, 0.7%]	87

Bootstrap: 511.125s -> 514.465s (0.65%)
Artifact size: 400.79 MiB -> 401.10 MiB (0.08%)

rustbot · 2026-06-02T02:39:36Z

This PR was rebased onto a different main 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.

saethlin · 2026-06-02T02:47:10Z

@bors try @rust-timer queue

Add unstable loop unrolling hint attributes

rust-log-analyzer · 2026-06-02T02:52:04Z

The job x86_64-gnu-tools failed! Check out the build log: (web) (plain enhanced) (plain)

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

[TIMING:end] compile::StdLink { compiler: Compiler { stage: 0, host: x86_64-unknown-linux-gnu, forced_compiler: false }, target_compiler: Compiler { stage: 0, host: x86_64-unknown-linux-gnu, forced_compiler: false }, target: x86_64-unknown-linux-gnu, crates: [], force_recompile: false } -- 0.001
##[group]Building stage1 compiler artifacts (stage0 -> stage1, x86_64-unknown-linux-gnu)
error: process didn't exit successfully: `sccache /checkout/obj/build/bootstrap/debug/rustc -vV` (exit status: 2)
--- stderr
sccache: error: Timed out waiting for server startup. Maybe the remote service is unreachable?
Run with SCCACHE_LOG=debug SCCACHE_NO_DAEMON=1 to get more information

Bootstrap failed while executing `build --stage 2 compiler rustdoc`
Build completed unsuccessfully in 0:00:30
  local time: Tue Jun  2 02:51:46 UTC 2026
  network time: Tue, 02 Jun 2026 02:51:46 GMT

rust-bors · 2026-06-02T04:57:44Z

☀️ Try build successful (CI)
Build commit: c1e5e0f (c1e5e0ffb239a664235b939ffad67af5426f4d9f, parent: 4a31759ad18b3c29c5ec99ca23c4764a8bedcf52)

rust-timer · 2026-06-02T05:38:38Z

Finished benchmarking commit (c1e5e0f): 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)	0.2%	[0.1%, 0.3%]	17
Regressions ❌ (secondary)	0.2%	[0.0%, 0.6%]	13
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.1%]	1
All ❌✅ (primary)	0.2%	[0.1%, 0.3%]	17

Max RSS (memory usage)

Results (primary 0.1%, secondary 2.9%)

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

	mean	range	count
Regressions ❌ (primary)	1.3%	[0.5%, 2.2%]	3
Regressions ❌ (secondary)	4.7%	[1.4%, 6.6%]	3
Improvements ✅ (primary)	-3.7%	[-3.7%, -3.7%]	1
Improvements ✅ (secondary)	-2.7%	[-2.7%, -2.7%]	1
All ❌✅ (primary)	0.1%	[-3.7%, 2.2%]	4

Cycles

Results (primary -2.1%, 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)	-	-	0
Regressions ❌ (secondary)	5.1%	[2.9%, 7.3%]	2
Improvements ✅ (primary)	-2.1%	[-2.1%, -2.1%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.1%	[-2.1%, -2.1%]	1

Binary size

Results (primary -0.0%, 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)	0.0%	[0.0%, 0.0%]	12
Regressions ❌ (secondary)	0.0%	[0.0%, 0.0%]	11
Improvements ✅ (primary)	-0.0%	[-0.1%, -0.0%]	26
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.0%]	20
All ❌✅ (primary)	-0.0%	[-0.1%, 0.0%]	38

Bootstrap: 511.17s -> 513.02s (0.36%)
Artifact size: 400.72 MiB -> 400.98 MiB (0.06%)

This comment has been minimized.

Sign in to view

saethlin force-pushed the loop-attributes branch from 523f17c to ec9cf15 Compare May 23, 2026 12:32

This comment has been minimized.

Sign in to view

saethlin mentioned this pull request May 24, 2026

Tracking Issue for loop optimization hint attributes #156874

Open

8 tasks

This comment has been minimized.

Sign in to view

saethlin force-pushed the loop-attributes branch 2 times, most recently from 9c8f21c to 22381f6 Compare May 24, 2026 21:15

This comment has been minimized.

Sign in to view

saethlin force-pushed the loop-attributes branch from 22381f6 to bf74a8e Compare May 25, 2026 11:16

This comment has been minimized.

Sign in to view

saethlin force-pushed the loop-attributes branch 2 times, most recently from 4d232fd to 9c8493b Compare May 31, 2026 20:58

saethlin marked this pull request as ready for review May 31, 2026 20:58

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label May 31, 2026

rustbot removed the S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. label May 31, 2026

rustbot assigned folkertdev May 31, 2026

saethlin changed the title ~~Prototype loop unrolling hint attributes~~ Add unstable loop unrolling hint attributes May 31, 2026

JonathanBrouwer self-assigned this May 31, 2026

This comment has been minimized.

Sign in to view

folkertdev reviewed May 31, 2026

View reviewed changes

JonathanBrouwer requested changes Jun 1, 2026

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 Jun 1, 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 Jun 1, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Jun 1, 2026

Auto merge of #156816 - saethlin:loop-attributes, r=<try>

57c302e

Add unstable loop unrolling hint attributes

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 Jun 1, 2026

saethlin added 2 commits June 1, 2026 22:39

Add #![feature(loop_hints)] and #[unroll]

e36eb95

Add a sus encoding bitpacking trick

5b32bf8

saethlin force-pushed the loop-attributes branch from fe5d3c8 to 5b32bf8 Compare June 2, 2026 02:39

Use meta_item_no_args

d8e3738

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 Jun 2, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Jun 2, 2026

Auto merge of #156816 - saethlin:loop-attributes, r=<try>

c1e5e0f

Add unstable loop unrolling hint attributes

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 Jun 2, 2026

Uh oh!

Conversation

saethlin commented May 22, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

rustbot commented May 31, 2026

Uh oh!

rustbot commented May 31, 2026

Uh oh!

JonathanBrouwer commented May 31, 2026

Uh oh!

This comment has been minimized.

folkertdev left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

JonathanBrouwer commented Jun 1, 2026

Uh oh!

folkertdev commented Jun 1, 2026

Uh oh!

JonathanBrouwer left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

JonathanBrouwer Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

saethlin commented Jun 1, 2026

Uh oh!

Kobzol commented Jun 1, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented Jun 1, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Jun 1, 2026

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

Uh oh!

saethlin commented May 22, 2026 •

edited by rustbot

Loading

folkertdev left a comment •

edited by rustbot

Loading

JonathanBrouwer left a comment •

edited by rustbot

Loading

JonathanBrouwer Jun 1, 2026 •

edited

Loading