destringlyfy lint infrastructure by mejrs · Pull Request #164607 · rust-lang/rust

mejrs · 2026-10-19T22:09:47Z

Currently, lints are passed as strings and parsed/split/reformatted/compared/etc in many many places. It's not always clear whether that string includes the tool or not.

Instead, they are passed around as:

pub struct LintName {
    pub tool: Option<Symbol>,
    pub name: Symbol,
}

r? @JonathanBrouwer

rustbot · 2026-10-19T22:09:50Z

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

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_attr_ir

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann, @JonathanBrouwer

rustbot · 2026-10-19T22:09:54Z

JonathanBrouwer is currently at their maximum review capacity.
They may take a while to respond.

mejrs · 2026-10-19T22:14:16Z

haven't updated clippy yet. But I think I can do a perf run.

@bors try @rust-timer queue

destringlyfy lint infrastructure

rust-bors · 2026-10-19T23:13:08Z

☀️ Try build successful (CI)
Build commit: 3dcd667 (3dcd66770271913d28347dd2820933e94b0f53a4)
Base parent: 07ea2af (07ea2af746e09a0e9e3ac9b39ee866bd9bf702a2)

rust-timer · 2026-10-19T23:59:10Z

Finished benchmarking commit (3dcd667): comparison URL.

Overall result: ❌✅ regressions and improvements - BENCHMARK(S) FAILED

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

❗ ❗ ❗ ❗ ❗
Warning ⚠️: The following benchmark(s) failed to build:

libc-0.2.172
Job failure

❗ ❗ ❗ ❗ ❗

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
Regressions ❌ (secondary)	0.3%	[0.2%, 0.5%]	5
Improvements ✅ (primary)	-0.4%	[-1.3%, -0.1%]	45
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.1%]	2
All ❌✅ (primary)	-0.4%	[-1.3%, -0.1%]	45

Max RSS (memory usage)

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

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)	4.1%	[0.5%, 9.5%]	7
Improvements ✅ (primary)	-1.0%	[-1.0%, -1.0%]	1
Improvements ✅ (secondary)	-5.5%	[-8.3%, -3.2%]	8
All ❌✅ (primary)	-1.0%	[-1.0%, -1.0%]	1

Cycles

Results (primary 2.2%, 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)	2.2%	[2.2%, 2.2%]	1
Regressions ❌ (secondary)	3.1%	[0.5%, 12.1%]	8
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.7%	[-4.2%, -0.6%]	6
All ❌✅ (primary)	2.2%	[2.2%, 2.2%]	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
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.0%	[-0.0%, -0.0%]	1
Improvements ✅ (secondary)	-0.0%	[-0.0%, -0.0%]	1
All ❌✅ (primary)	-0.0%	[-0.0%, -0.0%]	1

Bootstrap: 489.513s -> 489.49s (-0.00%)
Artifact size: 409.13 MiB -> 409.09 MiB (-0.01%)

Kivooeo · 2026-10-20T00:38:41Z

huh, any ideas why this might happenened in libc?

   thread 'rustc' (1970916) panicked at /rustc-dev/3dcd66770271913d28347dd2820933e94b0f53a4/compiler/rustc_lint/src/levels.rs:839:25:
    renamed lint does not exist: rustdoc::bare_urls

mejrs · 2026-10-20T10:38:56Z

huh, any ideas why this might happenened in libc?

   thread 'rustc' (1970916) panicked at /rustc-dev/3dcd66770271913d28347dd2820933e94b0f53a4/compiler/rustc_lint/src/levels.rs:839:25:
    renamed lint does not exist: rustdoc::bare_urls

it's one of the cases where a function takes (lint, tool) but in this case it's passed (rustdoc::bare_urls, None). i just didnt catch this one, it happened for clippy lints too.

rustbot · 2026-10-20T11:00:08Z

clippy is developed in its own repository. If possible, consider making this change to rust-lang/rust-clippy instead.

cc @rust-lang/clippy

ada4a

Please apologize the drive-by review. The idea seems very sensible, just some minor questions 👀

View changes since this review

ada4a · 2026-10-20T11:10:57Z

+#[derive(Copy, Clone, Debug, StableHash, Encodable, Decodable, PrintAttribute)]
+pub struct LintName {
+    /// The lint's tool name, if present.
+    pub tool: Option<Symbol>,


Imo it would make sense to leave this as Option<Ident>, so that you don't need to ads a separate LintCheck.tool_span field?

ada4a · 2026-10-20T11:11:27Z

-    /// produces multiple `LintCheck`s, one with a span pointing to `foo`
-    /// and another pointing to `bar`.
-    pub lint_span: Span,
+    pub lint_name: LintName,


Maybe put this inside LintName as well, as it seems pretty related?

mejrs · 2026-10-20T12:41:50Z

Please apologize the drive-by review. The idea seems very sensible, just some minor questions 👀

No need to apologize, all reviews are welcome :)

To answer your question w/r to spans: it would make comparing different LintNames complicated, because lint names with equal values but different spans will compare unequal.

And Span: !Ord, so this would require manual Ord impls, and I'd rather not go there...

mejrs · 2026-10-20T13:02:04Z

Also, possible followups:

clippy has it's own LintName, it could use this one instead
store all the builtin lints as symbols rather than string literals. just pre-interning them may also matter for perf.

mejrs · 2026-10-20T13:17:35Z

@bors try @rust-timer queue

ada4a · 2026-10-20T13:55:16Z

To answer your question w/r to spans: it would make comparing different LintNames complicated, because lint names with equal values but different spans will compare unequal.

And Span: !Ord, so this would require manual Ord impls, and I'd rather not go there...

I guess it'd be nice to have a wrapper like

struct SpannedLintName {
    tool: Option<Ident>,
    name: Symbol,
    /// The span of the complete lint name
    span: Span,
}

and use that inside LintCheck,
but converting it to/from LintName would probably not be zero-cost, so idk...

rust-bors · 2026-10-20T14:16:50Z

☀️ Try build successful (CI)
Build commit: 767dd6c (767dd6cf7268381167f2af26e795d4de6291f3fc)
Base parent: a7c37f9 (a7c37f9f0ee5e93396106e9de20f4a86b4154e63)

rust-timer · 2026-10-20T15:01:59Z

Finished benchmarking commit (767dd6c): 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
Regressions ❌ (secondary)	0.4%	[0.4%, 0.5%]	3
Improvements ✅ (primary)	-0.4%	[-1.3%, -0.1%]	44
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.1%]	3
All ❌✅ (primary)	-0.4%	[-1.3%, -0.1%]	44

Max RSS (memory usage)

Results (secondary -1.1%)

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)	4.0%	[0.5%, 9.7%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-5.4%	[-9.4%, -0.7%]	7
All ❌✅ (primary)	-	-	0

Cycles

Results (primary 2.6%, secondary 1.3%)

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

	mean	range	count
Regressions ❌ (primary)	2.6%	[2.1%, 3.1%]	2
Regressions ❌ (secondary)	3.5%	[0.4%, 12.8%]	10
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.1%	[-6.7%, -1.2%]	5
All ❌✅ (primary)	2.6%	[2.1%, 3.1%]	2

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 491.717s -> 488.128s (-0.73%)
Artifact size: 408.99 MiB -> 408.98 MiB (-0.00%)

mejrs · 2026-10-20T16:56:41Z

I guess it'd be nice to have a wrapper like

I don't think it'd help much at this time, we also won't need this if we move lint registration into the attribute parser.

JonathanBrouwer · 2026-10-25T11:18:13Z

+
+impl LintName {
+    pub fn new(segments: impl Iterator<Item = Symbol>) -> LintName {
+        let segments = segments.collect::<Vec<_>>();


Follow-up (unless you want to fix this immediately): Are we losing performance in this collect?

View changes since the review

its a bit vestigial (and it's easier to match on it), but i'll take a look

I was thinking do it as a follow-up so we can independently measure a performance effect, and see if the change is worth it

JonathanBrouwer · 2026-10-25T11:19:07Z

+        LintName { tool, name }
+    }
+
+    /// Panics if `lint` is empty.


I feel this comment is misleading if we only do this with debug_assertions

View changes since the review

JonathanBrouwer · 2026-10-25T11:27:18Z

-LL - #[warn(clippy::dead_cod)]
-LL + #[warn(dead_code)]
-   |
+   |        ^^^^^^^^^^^^^^^^ help: did you mean: `dead_code`


I kinda prefer the old diagnostic here? Why did this change and how hard is this to undo?

View changes since the review

JonathanBrouwer · 2026-10-25T11:52:16Z

-    /// produces multiple `LintCheck`s, one with a span pointing to `foo`
-    /// and another pointing to `bar`.
-    pub lint_span: Span,
+    pub lint_name: LintName,


My dream solution would be to get anLintId here instead of syms.
I'm gonna do a quick experiment to see how realistic that is

View changes since the review

First exploration seems to show potential here, I think this should be possible! (with a few minor hacks)
I'm going to try this in depth tomorrow

how realistic that is

there are some reasons to not do this:

how do you represent a lint that doesn't exist, because e.g. the input is wrong or it belongs to a tool that isn't currently running

you'd need to move Lint and LintId far down in the crate graph to do this

it would be incompatible with implementing lint resolution for proc/decl macro lints, because we'd likely need to run queries to retrieve the lint information

how do you represent a lint that doesn't exist, because e.g. the input is wrong or it belongs to a tool that isn't currently running

I think that could either stay in the HIR by name if it's useful, and if it's not useful removed entirely.

you'd need to move Lint and LintId far down in the crate graph to do this

rustc_attr_ir can depend directly on rustc_lint_defs. It already does transitively so it doesn't even worsen bootstrap times.

it would be incompatible with implementing lint resolution for proc/decl macro lints, because we'd likely need to run queries to retrieve the lint information

The only information we'd need is from the LintStore, which is available very early through the DynLintStore in Session

The only significant problem I've encountered so far is that LintIds are not encodable, so can't be part of the HIR. I was thinking to encode the index of the lint in LintStore::lints (field) instead.

I got nerd-sniped too hard so this is part 1 of this approach: #164918

Fwiw, I think our approaches are actually mostly compatible. It's still possible to introduce a LintName type and use it in most places that your PR does, and I think that would be good. It's just that it shouldn't be what's stored in the LintCheck type

This PR also has better perf than #164918 (tho both are improvements) so doing both definitely makes sense

mejrs · 2026-10-25T14:36:53Z

fwiw, what I want to experiment with next is to make Lint store a LintName rather than a &'static str

mejrs · 2026-10-25T22:52:14Z

        CheckLintNameResult::NoLint(suggestion)
    }
-
-    fn check_tool_name_for_backwards_compat(


This code made:

legacy_numeric_constants work as clippy::legacy_numeric_constants

#[warn(DEAD_CODE)] would say that "in rustc" there is a lint named dead_code

The fix was simply to delete it, with no other effects.

rustbot · 2026-10-26T16:36:16Z

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.

rustbot added A-attributes Area: Attributes (`#[…]`, `#![…]`) 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. labels Oct 19, 2026

rustbot assigned JonathanBrouwer Oct 19, 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 Oct 19, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Oct 19, 2026

Auto merge of #164607 - mejrs:lint_name, r=<try>

3dcd667

destringlyfy lint infrastructure

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 Oct 19, 2026

mejrs force-pushed the lint_name branch from 641283a to 3b88f0e Compare October 20, 2026 10:27

This comment has been minimized.

Sign in to view

mejrs force-pushed the lint_name branch from 3b88f0e to 90fe486 Compare October 20, 2026 11:00

rustbot added the T-clippy Relevant to the Clippy team. label Oct 20, 2026

ada4a reviewed Oct 20, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

mejrs force-pushed the lint_name branch from 90fe486 to 68d7402 Compare October 20, 2026 12:13

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 Oct 20, 2026

ojeda mentioned this pull request Oct 21, 2026

Rust unstable features needed for the kernel Rust-for-Linux/linux#2

Open

57 tasks

JonathanBrouwer requested changes Oct 25, 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 Oct 25, 2026

JonathanBrouwer mentioned this pull request Oct 25, 2026

Move lint validation to attribute parsing #164918

Open

mejrs force-pushed the lint_name branch from 68d7402 to c114030 Compare October 25, 2026 17:26

This comment has been minimized.

Sign in to view

mejrs force-pushed the lint_name branch from c114030 to a97467a Compare October 25, 2026 22:39

This comment has been minimized.

Sign in to view

mejrs commented Oct 25, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

mejrs force-pushed the lint_name branch from a97467a to 07f6a25 Compare October 26, 2026 08:50

This comment has been minimized.

Sign in to view

mejrs added 7 commits October 26, 2026 17:24

factor out lint name

5afdb7a

edit check_lint_name to accept LintName

123d20d

use Lintname in lint machinery

6a3df09

store Lintname in error types

f6f8683

Use LintName in LintLevelSource

09a5060

Use LintName more

a8a0f42

return tool_name in NoTool error

232a750

mejrs force-pushed the lint_name branch from 07f6a25 to 232a750 Compare October 26, 2026 16:36

Conversation

mejrs commented Oct 19, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Oct 19, 2026

Uh oh!

rustbot commented Oct 19, 2026

Uh oh!

mejrs commented Oct 19, 2026 • edited 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.

rust-bors Bot commented Oct 19, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Oct 19, 2026

Overall result: ❌✅ regressions and improvements - BENCHMARK(S) FAILED

Uh oh!

Kivooeo commented Oct 20, 2026

Uh oh!

mejrs commented Oct 20, 2026

Uh oh!

This comment has been minimized.

rustbot commented Oct 20, 2026

Uh oh!

ada4a 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!

This comment has been minimized.

mejrs commented Oct 20, 2026

Uh oh!

mejrs commented Oct 20, 2026

Uh oh!

mejrs commented Oct 20, 2026

Uh oh!

ada4a commented Oct 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rust-bors Bot commented Oct 20, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Oct 20, 2026

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

Uh oh!

mejrs commented Oct 20, 2026

Uh oh!

JonathanBrouwer Oct 25, 2026 • 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 Oct 25, 2026 • 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!

JonathanBrouwer Oct 25, 2026 • 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!

JonathanBrouwer Oct 25, 2026 • 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!

mejrs Oct 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

mejrs commented Oct 19, 2026 •

edited by rustbot

Loading

mejrs commented Oct 19, 2026 •

edited

Loading

ada4a left a comment •

edited by rustbot

Loading

ada4a commented Oct 20, 2026 •

edited

Loading

JonathanBrouwer Oct 25, 2026 •

edited by rustbot

Loading

JonathanBrouwer Oct 25, 2026 •

edited by rustbot

Loading

JonathanBrouwer Oct 25, 2026 •

edited by rustbot

Loading

JonathanBrouwer Oct 25, 2026 •

edited by rustbot

Loading

mejrs Oct 25, 2026 •

edited

Loading

JonathanBrouwer Oct 25, 2026 •

edited

Loading

JonathanBrouwer Oct 25, 2026 •

edited

Loading

mejrs Oct 25, 2026 •

edited

Loading