Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing by darkraider01 · Pull Request #163673 · rust-lang/rust

darkraider01 · 2026-10-02T18:21:30Z

This migrates #[doc(inline)] and #[doc(no_inline)] validation from rustc_passes into DocParser in rustc_attr_parsing. It checks for conflicting inlining attributes and non-use targets, and tweaks rustdoc's early abort so lowering errors don't suppress other attribute diagnostics, also added new UI tests in doc-inline-validation.rs.

rustbot · 2026-10-02T18:21:34Z

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

rustbot · 2026-10-02T18:21:36Z

r? @oli-obk

rustbot has assigned @oli-obk.
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 77 candidates
Random selection from 19 candidates

darkraider01 · 2026-10-02T18:27:37Z

@@ -367,13 +367,14 @@ pub(crate) fn run_global_ctxt(

    // NOTE: These are copy/pasted from typeck/lib.rs and should be kept in sync with those changes.
    tcx.sess.time("wf_checking", || tcx.ensure_ok().check_type_wf(()));
+    tcx.sess.time("check_mod_attrs", || {


Had to move this before abort_if_errors. Since conflicting attributes are caught during lowering now, rustdoc was exiting early and swallowing the other errors in invalid-doc-attr.rs.

View changes since the review

@GuillaumeGomez could you confirm this is ok? I'm not familiar with why this abort_if_errors is there

I'm surprised this change is needed. If the rustdoc tests pass and the perf check is happy, then so am I.

Just double-checked and yes, seems ok. We'll wait for tests to confirm but seems fine.

darkraider01 · 2026-10-02T18:28:17Z

        }
    }
+
+    fn deferred_finalize_check(&self) -> Option<(FinalizeCheckFn, Span)> {


Deferred this check since we need the resolved target from FinalizeCheckContext to check if it's actually a use item.

View changes since the review

Is the resolved target needed? I don't see where you're using it

Sorry, I confused the item’s syntactic target with the resolved import target. This only checks Use or ExternCrate, which FinalizeContext already provides. I’ll move it into finalize.

You can also consider putting it in parse_inline

Moved the conflict check into parse_inline and made inline an Option. The target check is still deferred since finalize takes an immutable context. That also avoids an extra target lint when the attributes already conflict.

I think you should also just be able to put the target check in the inline parser right?

Yes, AcceptContext already has the target and can emit the lint. I’ll move that check into parse_inline and remove the deferred callback.

JonathanBrouwer · 2026-10-03T08:37:43Z

r? me

JonathanBrouwer · 2026-10-05T13:47:21Z

        }
    }
+
+    fn finalize_check(cx: &mut FinalizeCheckContext<'_, '_>, _attr_span: Span) {


Can we move this logic to the inline parser?
Then we can solve this comment and make inline an Option

rust/compiler/rustc_attr_ir/src/data_structures.rs

Line 493 in d731eda

// at the same time as the other doc attributes, we store a list instead of using `Option`.

View changes since the review

JonathanBrouwer · 2026-10-05T13:50:42Z

@@ -367,13 +367,14 @@ pub(crate) fn run_global_ctxt(

    // NOTE: These are copy/pasted from typeck/lib.rs and should be kept in sync with those changes.
    tcx.sess.time("wf_checking", || tcx.ensure_ok().check_type_wf(()));
+    tcx.sess.time("check_mod_attrs", || {


@GuillaumeGomez could you confirm this is ok? I'm not familiar with why this abort_if_errors is there

rustbot · 2026-10-05T13:50:51Z

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

GuillaumeGomez · 2026-10-05T13:54:56Z

@bors try @rust-timer queue

Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing

GuillaumeGomez · 2026-10-05T13:55:52Z

Actually it's completely unneeded. Sorry for the noise. ^^'

@bors try-

rust-bors · 2026-10-05T13:56:00Z

Unknown command "try-". Run @bors help or go to https://bors.rust-lang.org/help to see available commands.

rust-bors · 2026-10-05T14:56:08Z

☀️ Try build successful (CI)
Build commit: 61e9eef (61e9eefe54c9e382b1ccf541d9f4ae2942d9c4d3)
Base parent: 602727f (602727f26878dcf9eb2999e2f58e5df1269ff47e)

rustbot · 2026-10-05T15:02:17Z

Some changes occurred in compiler/rustc_attr_ir

cc @jdonszelmann, @JonathanBrouwer

darkraider01 · 2026-10-05T15:06:59Z

@rustbot ready

JonathanBrouwer · 2026-10-05T15:21:35Z

 pub(crate) struct DocParser {
    attribute: DocAttribute,
    nb_doc_attrs: usize,
+    inline_conflict: bool,


Is this needed?
I'd lean toward emitting multiple diagnostics if there are multiple wrong attributes being fine

View changes since the review

Moved the target check into parse_inline and removed the conflict flag and deferred callback. It now reports multiple diagnostics when multiple attributes are invalid, and I’ve updated the tests accordingly.
@rustbot ready

rust-timer · 2026-10-05T15:39:42Z

Finished benchmarking commit (61e9eef): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking means the PR may be perf-sensitive. Consider adding rollup=never if this change is not fit for rolling up.

@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

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

Max RSS (memory usage)

Results (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)	-	-	0
Regressions ❌ (secondary)	6.5%	[6.5%, 6.5%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.3%	[-3.3%, -3.3%]	1
All ❌✅ (primary)	-	-	0

Cycles

Results (primary -2.2%, secondary 3.2%)

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)	3.2%	[2.3%, 4.1%]	2
Improvements ✅ (primary)	-2.2%	[-2.2%, -2.2%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.2%	[-2.2%, -2.2%]	1

Binary size

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

Bootstrap: 490.152s -> 492.034s (0.38%)
Artifact size: 408.63 MiB -> 408.69 MiB (0.01%)

JonathanBrouwer

@bors r+ rollup
Thanks a lot for the work!

View changes since this review

rust-bors · 2026-10-05T17:28:37Z

📌 Commit 5d640cb has been approved by JonathanBrouwer

It is now in the queue for this repository.

…, r=JonathanBrouwer Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing Part of rust-lang#153101 This migrates `#[doc(inline)]` and `#[doc(no_inline)]` validation from `rustc_passes` into `DocParser` in `rustc_attr_parsing`. It checks for conflicting inlining attributes and non-use targets, and tweaks rustdoc's early abort so lowering errors don't suppress other attribute diagnostics, also added new UI tests in doc-inline-validation.rs.

…uwer Rollup of 4 pull requests Successful merges: - #160915 ([rustdoc] Fix how `Deref` items is handled.) - #163267 (fix `VisibleForLeakCheck` in `RegionOutlives` fast path) - #163673 (Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing) - #163814 (Add some comments and docs related to `AllocatorNightly`)

…uwer Rollup of 5 pull requests Successful merges: - #160915 ([rustdoc] Fix how `Deref` items is handled.) - #163267 (fix `VisibleForLeakCheck` in `RegionOutlives` fast path) - #163673 (Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing) - #163795 (When documenting without a specified `--edition`, emit a message) - #163814 (Add some comments and docs related to `AllocatorNightly`)

Rollup merge of #163673 - darkraider01:doc-inline-validation, r=JonathanBrouwer Move #[doc(inline)] and #[doc(no_inline)] validation into rustc_attr_parsing Part of #153101 This migrates `#[doc(inline)]` and `#[doc(no_inline)]` validation from `rustc_passes` into `DocParser` in `rustc_attr_parsing`. It checks for conflicting inlining attributes and non-use targets, and tweaks rustdoc's early abort so lowering errors don't suppress other attribute diagnostics, also added new UI tests in doc-inline-validation.rs.

darkraider01 added 3 commits October 2, 2026 22:58

Add doc inline validation tests

d58b146

Move doc inline validation to attribute parsing

99137e3

Run rustdoc attribute checks before early abort

bc8eca0

rustbot assigned oli-obk Oct 2, 2026

darkraider01 commented Oct 2, 2026

View reviewed changes