Fix splat ICEs and ban it in closures by teor2345 · Pull Request #158645 · rust-lang/rust

teor2345 · 2026-07-01T08:12:36Z

This PR is part of the splat lang experiment #153629.

It fixes one ICE, and cleans up the surrounding error handling to avoid future ICEs:
Close #158482

Improves the message and error handling in another ICE (but doesn't fix it):
cc #158603

And bans splat on closures and rust-call, because they already de-tuple arguments:
Close #158605

In the process I discovered another symbol mangling bug with splatting, which I added a test for (but didn't fix yet):
cc #158644
~~This test might be flaky, I'll drop it from this PR if it fails in CI.~~

rustbot · 2026-07-01T08:12:41Z

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 16 candidates

folkertdev

A first pass. Overall this PR seems kind of large, it would be easier to review if split up into smaller chunks.

View changes since this review

folkertdev · 2026-07-01T08:32:24Z

+                Extern::Explicit(abi_str, span)
+                    if abi_str.symbol_unescaped.as_str() == "rust-call" =>
+                {
+                    SplatSemantic::NoRustCall(span)
+                }
+                Extern::Explicit(_abi_str, _span) => SplatSemantic::Yes,


Suggested change

Extern::Explicit(abi_str, span)

if abi_str.symbol_unescaped.as_str() == "rust-call" =>

{

SplatSemantic::NoRustCall(span)

}

Extern::Explicit(_abi_str, _span) => SplatSemantic::Yes,

Extern::Explicit(abi_str, span) => match abi_str.symbol_unescaped {

sym::rust_dash_call => SplatSemantic::NoRustCall(span),

_ => SplatSemantic::Yes,

},

and add a comment on why here too perhaps?

I think the rust_dash_call symbol only exists in rust-analyzer, not in the compiler, but I did add the comment.

rustbot · 2026-07-01T08:36:48Z

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

teor2345

A first pass. Overall this PR seems kind of large, it would be easier to review if split up into smaller chunks.

Yeah I wondered about that, I'll split it up today, now I know which parts actually fail in CI.

@rustbot label +F-splat +I-ICE +T-compiler +C-bug

View changes since this review

teor2345 · 2026-07-02T02:14:03Z

I split the extra tests into PR #158677, most of them didn't need any compiler changes to work, so they are fine to merge separately.

rustbot · 2026-07-02T03:00:02Z

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.

teor2345 · 2026-07-02T04:32:58Z

I dropped the large refactor because it broke some unrelated UI tests. Later I'll see if I can get it to work, and submit it as a separate PR. Otherwise this should be ready to review, and independent of #158677 merging.

Let me know if you'd like the PR split further?

@rustbot ready

rustbot · 2026-07-02T04:33:02Z

Error: shortcut handler unexpectedly failed in this comment: error sending request

Please file an issue on GitHub at triagebot if there's a problem with this bot, or reach out on #triagebot on Zulip.

teor2345 · 2026-07-02T05:48:09Z

@rustbot ready

@rustbot

…eywiser Add extra splat tests This PR is part of the splat lang experiment rust-lang#153629. It adds a bunch of tests for existing splat functionality, but doesn't make any compiler code changes. Edit: split off from PR rust-lang#158645 @rustbot label +F-splat +T-compiler

teor2345 · 2026-07-03T02:47:20Z

            TyKind::FnPtr(bfty) => {
                self.check_fn_ptr_safety(bfty.decl_span, bfty.safety);
-                self.check_fn_decl(&bfty.decl, SelfSemantic::No);
+                self.check_fn_decl(&bfty.decl, SelfSemantic::No, SplatSemantic::Yes);


This is wrong because FnPtrs can be rust-call, I'll add a test
https://play.rust-lang.org/?version=nightly&mode=debug&edition=2024&gist=511e43db9905a3a38acb86b175535fe2

@rustbot author

View changes since the review

teor2345 · 2026-07-03T06:31:28Z

Ok, now splat + rust-call is banned on both function item declarations and function pointer types, and splat is banned on closures. And there are tests for all of them.

@rustbot ready

folkertdev · 2026-07-04T14:07:27Z

-        // Whether the function is variadic (e.g. from C)
-        c_variadic: bool,


why was this removed exactly?

View changes since the review

It's not used in this function except for debug logging, and it's not particularly interesting in debug logging because splat and c_variadic aren't allowed on the same function at the AST level.

And the top of the function early exits if splat / Fn* traits / rust-call isn't being used, so c_variadic should always be false.

folkertdev · 2026-07-04T14:11:32Z

-        )
+            }
+        } else {
+            let fn_ty = self.typeck_results.expr_ty_adjusted(expr);


maybe comment here that this case is for function pointers?

View changes since the review

Perhaps a custom enum even makes sense? "None" meaning we're dealing with function pointers is a but subtle. On the other hand, the ergonomics of that might be worse.

Happy to add a comment.

I expect to populate the side-table in an upcoming PR, as part of that change I might need a custom enum with different fields for FnPtrs and FnDefs. I was holding off on that to avoid churn.

folkertdev · 2026-07-04T14:14:11Z

+error: `#[splat]` is not allowed on functions with the `rust-call` ABI
+  --> $DIR/reject-closure-issue-158605.rs:10:5
+   |
+LL |     extern "rust-call" fn method(#[splat] self: Self);
+   |     ^^^^^^^^^^^^^^^^^^                    ^^^^
+   |
+   = help: remove `#[splat]` or change the ABI


is the span right here? I'd expect it on the #[splat] really?

also the #[splat] is not really "on" the function, it is on an argument. Not a native speaker, but "is not allowed in/for functions with ..." sounds more natural?

View changes since the review

is the span right here? I'd expect it on the #[splat] really?

Ideally on splat and the ABI, as one of them needs to be removed. ~~I'll add a FIXME if that's not possible right now.~~

I think "splat is not allowed in the arguments of functions with the rust-call ABI" would be clearer, as "in" could mean the body as well.

folkertdev · 2026-07-04T14:18:49Z

+            // FIXME(splat): should splatting extern "C" or other ABIs be allowed?
+            Extern::Implicit(_) => SplatSemantic::Yes,
+            // For now, splatting rust-call is banned, because it already de-tuples args.
+            Extern::Explicit(abi_str, span) if abi_str.symbol_unescaped.as_str() == "rust-call" => {


it would be fine to add a rust_dash_call to rustc too.

View changes since the review

@rustbot

…eywiser Add extra splat tests This PR is part of the splat lang experiment rust-lang#153629. It adds a bunch of tests for existing splat functionality, but doesn't make any compiler code changes. Edit: split off from PR rust-lang#158645 @rustbot label +F-splat +T-compiler

@rustbot

Rollup merge of #158677 - teor2345:extra-splat-tests, r=wesleywiser Add extra splat tests This PR is part of the splat lang experiment #153629. It adds a bunch of tests for existing splat functionality, but doesn't make any compiler code changes. Edit: split off from PR #158645 @rustbot label +F-splat +T-compiler

teor2345

I'll have a look at these changes later today or tomorrow

View changes since this review

teor2345 · 2026-07-04T23:20:30Z

-        // Whether the function is variadic (e.g. from C)
-        c_variadic: bool,


It's not used in this function except for debug logging, and it's not particularly interesting in debug logging because splat and c_variadic aren't allowed on the same function at the AST level.

And the top of the function early exits if splat / Fn* traits / rust-call isn't being used, so c_variadic should always be false.

teor2345 · 2026-07-04T23:25:21Z

-        )
+            }
+        } else {
+            let fn_ty = self.typeck_results.expr_ty_adjusted(expr);


Happy to add a comment.

I expect to populate the side-table in an upcoming PR, as part of that change I might need a custom enum with different fields for FnPtrs and FnDefs. I was holding off on that to avoid churn.

teor2345 · 2026-07-04T23:28:30Z

+error: `#[splat]` is not allowed on functions with the `rust-call` ABI
+  --> $DIR/reject-closure-issue-158605.rs:10:5
+   |
+LL |     extern "rust-call" fn method(#[splat] self: Self);
+   |     ^^^^^^^^^^^^^^^^^^                    ^^^^
+   |
+   = help: remove `#[splat]` or change the ABI


is the span right here? I'd expect it on the #[splat] really?

Ideally on splat and the ABI, as one of them needs to be removed. ~~I'll add a FIXME if that's not possible right now.~~

I think "splat is not allowed in the arguments of functions with the rust-call ABI" would be clearer, as "in" could mean the body as well.

teor2345 · 2026-07-06T02:24:54Z

Alright, all of those should be fixed up now. This ended up a bit more complicated than I thought, because we were missing a test (and AST handling) for multiple splats on the same argument.

teor2345 · 2026-07-06T02:25:19Z

@rustbot ready

rustbot assigned folkertdev Jul 1, 2026

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. labels Jul 1, 2026

This was referenced Jul 1, 2026

Fix ICE in label_fn_like for splatted args and add UI tests #157434

Draft

[ICE] due to #[splat] in a function pointer type #158603

Open

programmerjake reviewed Jul 1, 2026

View reviewed changes

Comment thread tests/ui/splat/splat-fn-ptr-tuple.rs Outdated

folkertdev requested changes Jul 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 Jul 1, 2026

This comment has been minimized.

Sign in to view

Ajay-singh1 reviewed Jul 1, 2026

View reviewed changes

Comment thread tests/ui/splat/splat-dyn-asref-tuple-fail.stderr Outdated

This comment has been minimized.

Sign in to view

teor2345 commented Jul 1, 2026

View reviewed changes

Comment thread compiler/rustc_ast_passes/src/ast_validation.rs Outdated

Comment thread tests/ui/splat/splat-fn-ptr-tuple.rs Outdated

Comment thread tests/ui/splat/splat-dyn-asref-tuple-fail.stderr Outdated

rustbot added C-bug Category: This is a bug. F-splat `#![feature(splat)]` I-ICE Issue: The compiler panicked, giving an Internal Compilation Error (ICE) ❄️ labels Jul 1, 2026

teor2345 mentioned this pull request Jul 2, 2026

Add extra splat tests #158677

Merged

teor2345 added 4 commits July 2, 2026 12:56

Split unsafe fn ptr accesses into their own test

ea1849e

Check arg errors then write splatted call info

3a0cd1d

Avoid an unwrap using a u16 splat index

b1cfe90

Remove a splat check arg only used for debugging

42a0767

teor2345 force-pushed the splat-fn-fixes branch from 81fcc4e to babf77b Compare July 2, 2026 02:59

rustbot removed the S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. label Jul 2, 2026

teor2345 mentioned this pull request Jul 2, 2026

Tracking Issue for argument splatting lang experiment #153629

Open

25 tasks

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jul 2, 2026

teor2345 commented Jul 3, 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 Jul 3, 2026

teor2345 force-pushed the splat-fn-fixes branch from babf77b to 83fe45c Compare July 3, 2026 06:30

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 Jul 3, 2026

folkertdev requested changes Jul 4, 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 Jul 4, 2026

teor2345 commented Jul 6, 2026

View reviewed changes

teor2345 added 2 commits July 6, 2026 11:34

Improve ICE diag for splatted FnPtrs

77a2384

Add splat mangling bug tests (flaky?)

6c1a5b2

teor2345 force-pushed the splat-fn-fixes branch from 83fe45c to 5f1feac Compare July 6, 2026 02:19

Ban splat on closures and rust-call

488662a

teor2345 force-pushed the splat-fn-fixes branch from 5f1feac to 488662a Compare July 6, 2026 02:24

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 Jul 6, 2026

teor2345 mentioned this pull request Jul 6, 2026

Avoid underflow for splatted rust-call receivers #158850

Open

		// Whether the function is variadic (e.g. from C)
		c_variadic: bool,

Uh oh!

Uh oh!

Conversation

teor2345 commented Jul 1, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Jul 1, 2026

Uh oh!

Uh oh!

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!

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!

Uh oh!

rustbot commented Jul 1, 2026

Uh oh!

This comment has been minimized.

Uh oh!

This comment has been minimized.

teor2345 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!

teor2345 commented Jul 2, 2026

Uh oh!

rustbot commented Jul 2, 2026

Uh oh!

teor2345 commented Jul 2, 2026

Uh oh!

rustbot commented Jul 2, 2026

Uh oh!

teor2345 commented Jul 2, 2026

Uh oh!

teor2345 Jul 3, 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!

teor2345 commented Jul 3, 2026

Uh oh!

folkertdev Jul 4, 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!

folkertdev Jul 4, 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!

folkertdev Jul 4, 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!

teor2345 Jul 4, 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!

folkertdev Jul 4, 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!

teor2345 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

teor2345 commented Jul 1, 2026 •

edited by rustbot

Loading

folkertdev left a comment •

edited by rustbot

Loading

teor2345 left a comment •

edited by rustbot

Loading

teor2345 Jul 3, 2026 •

edited by rustbot

Loading

folkertdev Jul 4, 2026 •

edited by rustbot

Loading

folkertdev Jul 4, 2026 •

edited by rustbot

Loading

folkertdev Jul 4, 2026 •

edited by rustbot

Loading

teor2345 Jul 4, 2026 •

edited

Loading

folkertdev Jul 4, 2026 •

edited by rustbot

Loading

teor2345 left a comment •

edited by rustbot

Loading

teor2345 Jul 4, 2026 •

edited

Loading