explicit tail calls: support indirect arguments by folkertdev · Pull Request #151143 · rust-lang/rust

folkertdev · 2026-01-14T21:26:52Z

tracking issue: #112788

After discussion in #144855, I was wrong and it is actually possible to support tail calls with PassMode::Indirect { on_stack: false, .. } arguments.

Normally an indirect argument with on_stack: false would be passed as a pointer into the caller's stack frame. For tail calls, that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame.

Therefore we store the argument for the callee in the corresponding caller's slot. Because guaranteed tail calls demand that the caller's signature matches the callee's, the corresponding slot has the correct type.

To handle cases like the one below, the tail call arguments must first be copied to a temporary, and can only then be copied to the caller's argument slots.

// A struct big enough that it is not passed via registers.
pub struct Big([u64; 4]);

fn swapper(a: Big, b: Big) -> (Big, Big) {
    become swapper_helper(b, a);
}

I'm not really familiar with MIR and what tricks/helpers we have, so I'm just cobbling this together. Hopefully we can arrive at something more elegant.

folkertdev · 2026-01-14T21:36:17Z

+                        PassMode::Indirect { on_stack: true, .. } => {
+                            // FIXME: some LLVM backends (notably x86) do not correctly pass byval
+                            // arguments to tail calls (as of LLVM 21). See also:
+                            //
+                            // - https://github.com/rust-lang/rust/pull/144232#discussion_r2218543841
+                            // - https://github.com/rust-lang/rust/issues/144855
+                            span_bug!(
+                                fn_span,
+                                "arguments using PassMode::Indirect {{ on_stack: true, .. }} are currently not supported for tail calls"
+                            )


I've kept the span_bug! for now.

It looks like a fix for x86 might get cherry-picked into LLVM 22. If so, I think that is enough support to allow this variant too. At that point riscv would be the next most commonly used target that would miscompile.

folkertdev · 2026-01-15T17:32:28Z

@bors try=x86_64-msvc-1,x86_64-msvc-2

rust-bors · 2026-01-15T17:32:31Z

Unknown command "try". Run @bors help to see available commands.

folkertdev · 2026-01-15T17:32:48Z

@bors try job=x86_64-msvc-1,x86_64-msvc-2

explicit tail calls: support indirect arguments try-job: x86_64-msvc-1 try-job: x86_64-msvc-2

rust-bors · 2026-01-15T19:20:43Z

💔 Test for e8cb9bb failed: CI. Failed job:

try - x86_64-msvc-1 (web logs, enhanced plaintext logs)

folkertdev · 2026-01-15T22:22:55Z

@bors try job=x86_64-msvc-1,x86_64-msvc-2

explicit tail calls: support indirect arguments try-job: x86_64-msvc-1 try-job: x86_64-msvc-2

rust-bors · 2026-01-16T01:03:26Z

☀️ Try build successful (CI)
Build commit: 282e2c8 (282e2c837faf975d875d5007dcbbdb5567a62f41, parent: b5c2a0fc0ac851e83ef943e3a2b90c1abab06baa)

folkertdev · 2026-01-16T17:09:12Z

This seems to work, though we should probably run a bunch of try jobs once the code looks good.

r? @WaffleLapkin

rustbot · 2026-01-16T17:09:16Z

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

rustbot · 2026-02-27T09:24:54Z

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.

folkertdev · 2026-02-27T09:25:36Z

@bors try job=test-various

explicit tail calls: support indirect arguments try-job: test-various

folkertdev · 2026-02-27T09:52:10Z

@bors try job=test-various

explicit tail calls: support indirect arguments try-job: test-various

rust-bors · 2026-02-27T11:44:58Z

☀️ Try build successful (CI)
Build commit: b8a6c0d (b8a6c0d1a7f881896a4e68b3a59e5a419c4bade6, parent: 0ee5907d597b02afadb5daa26a60fedb72f098d1)

folkertdev · 2026-02-27T11:45:53Z

@bors r=WaffleLapkin

rust-bors · 2026-02-27T11:45:56Z

📌 Commit e6cf5a2 has been approved by WaffleLapkin

It is now in the queue for this repository.

…affleLapkin explicit tail calls: support indirect arguments tracking issue: rust-lang#112788 After discussion in rust-lang#144855, I was wrong and it is actually possible to support tail calls with `PassMode::Indirect { on_stack: false, .. }` arguments. Normally an indirect argument with `on_stack: false` would be passed as a pointer into the caller's stack frame. For tail calls, that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame. Therefore we store the argument for the callee in the corresponding caller's slot. Because guaranteed tail calls demand that the caller's signature matches the callee's, the corresponding slot has the correct type. To handle cases like the one below, the tail call arguments must first be copied to a temporary, and can only then be copied to the caller's argument slots. ```rust // A struct big enough that it is not passed via registers. pub struct Big([u64; 4]); fn swapper(a: Big, b: Big) -> (Big, Big) { become swapper_helper(b, a); } ``` --- I'm not really familiar with MIR and what tricks/helpers we have, so I'm just cobbling this together. Hopefully we can arrive at something more elegant.

…uwer Rollup of 12 pull requests Successful merges: - #151143 (explicit tail calls: support indirect arguments) - #153012 (Stop using `LinkedGraph` in `lexical_region_resolve`) - #153175 (Clarify a confusing green-path function) - #153179 (Force a CI LLVM stamp bump) - #150828 (Improved security section in rustdoc for `current_exe`) - #152673 (rustc_public: rewrite `bridge_impl` to reduce boilerplate) - #152674 (rustc_public: remove the `CrateDefItems` trait) - #153073 (Fix mem::conjure_zst panic message to use any::type_name instead) - #153117 (Remove mutation from macro path URL construction) - #153128 (Recover feature lang_items for emscripten) - #153138 (Print path root when printing path) - #153159 (Work around a false `err.emit();` type error in rust-analyzer)

Rollup merge of #151143 - folkertdev:tail-call-indirect, r=WaffleLapkin explicit tail calls: support indirect arguments tracking issue: #112788 After discussion in #144855, I was wrong and it is actually possible to support tail calls with `PassMode::Indirect { on_stack: false, .. }` arguments. Normally an indirect argument with `on_stack: false` would be passed as a pointer into the caller's stack frame. For tail calls, that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame. Therefore we store the argument for the callee in the corresponding caller's slot. Because guaranteed tail calls demand that the caller's signature matches the callee's, the corresponding slot has the correct type. To handle cases like the one below, the tail call arguments must first be copied to a temporary, and can only then be copied to the caller's argument slots. ```rust // A struct big enough that it is not passed via registers. pub struct Big([u64; 4]); fn swapper(a: Big, b: Big) -> (Big, Big) { become swapper_helper(b, a); } ``` --- I'm not really familiar with MIR and what tricks/helpers we have, so I'm just cobbling this together. Hopefully we can arrive at something more elegant.

…uwer Rollup of 12 pull requests Successful merges: - rust-lang/rust#151143 (explicit tail calls: support indirect arguments) - rust-lang/rust#153012 (Stop using `LinkedGraph` in `lexical_region_resolve`) - rust-lang/rust#153175 (Clarify a confusing green-path function) - rust-lang/rust#153179 (Force a CI LLVM stamp bump) - rust-lang/rust#150828 (Improved security section in rustdoc for `current_exe`) - rust-lang/rust#152673 (rustc_public: rewrite `bridge_impl` to reduce boilerplate) - rust-lang/rust#152674 (rustc_public: remove the `CrateDefItems` trait) - rust-lang/rust#153073 (Fix mem::conjure_zst panic message to use any::type_name instead) - rust-lang/rust#153117 (Remove mutation from macro path URL construction) - rust-lang/rust#153128 (Recover feature lang_items for emscripten) - rust-lang/rust#153138 (Print path root when printing path) - rust-lang/rust#153159 (Work around a false `err.emit();` type error in rust-analyzer)

@bar

codegen: Copy to an alloca when the argument is neither by-val nor by-move for indirect pointer. Fixes #155241. When a value is passed via an indirect pointer, the value needs to be copied to a new alloca. For x86_64-unknown-linux-gnu, `Thing` is the case: ```rust #[derive(Clone, Copy)] struct Thing(usize, usize, usize); pub fn foo() { let thing = Thing(0, 0, 0); bar(thing); assert_eq!(thing.0, 0); } #[inline(never)] #[unsafe(no_mangle)] pub fn bar(mut thing: Thing) { thing.0 = 1; } ``` Before passing the thing to the bar function, the thing needs to be copied to an alloca that is passed to bar. ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) call void @bar(ptr %0) ``` This patch applies the rule to the untupled arguments as well. ```rust #![feature(fn_traits)] #[derive(Clone, Copy)] struct Thing(usize, usize, usize); #[inline(never)] #[unsafe(no_mangle)] pub fn foo() { let thing = (Thing(0, 0, 0),); (|mut thing: Thing| { thing.0 = 1; }).call(thing); assert_eq!(thing.0.0, 0); } ``` For this case, this patch changes from ```llvm ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %thing) ``` to ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %0) ``` However, the same rule cannot be applied to tail calls that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame. Fortunately, #151143 has already handled the special case. We must not copy again. No copy is needed for by-move arguments, because the argument is passed to the called "in-place". No copy is also needed for by-val arguments, because the attribute implies that a hidden copy of the pointee is made between the caller and the callee. NOTE: The patch has a trick for tail calls that we pass by-move. We can choose to copy an alloca even for by-move arguments, but tail calls require MUST-by-move.

@bar

codegen: Copy to an alloca when the argument is neither by-val nor by-move for indirect pointer. Fixes rust-lang/rust#155241. When a value is passed via an indirect pointer, the value needs to be copied to a new alloca. For x86_64-unknown-linux-gnu, `Thing` is the case: ```rust #[derive(Clone, Copy)] struct Thing(usize, usize, usize); pub fn foo() { let thing = Thing(0, 0, 0); bar(thing); assert_eq!(thing.0, 0); } #[inline(never)] #[unsafe(no_mangle)] pub fn bar(mut thing: Thing) { thing.0 = 1; } ``` Before passing the thing to the bar function, the thing needs to be copied to an alloca that is passed to bar. ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) call void @bar(ptr %0) ``` This patch applies the rule to the untupled arguments as well. ```rust #![feature(fn_traits)] #[derive(Clone, Copy)] struct Thing(usize, usize, usize); #[inline(never)] #[unsafe(no_mangle)] pub fn foo() { let thing = (Thing(0, 0, 0),); (|mut thing: Thing| { thing.0 = 1; }).call(thing); assert_eq!(thing.0.0, 0); } ``` For this case, this patch changes from ```llvm ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %thing) ``` to ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %0) ``` However, the same rule cannot be applied to tail calls that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame. Fortunately, rust-lang/rust#151143 has already handled the special case. We must not copy again. No copy is needed for by-move arguments, because the argument is passed to the called "in-place". No copy is also needed for by-val arguments, because the attribute implies that a hidden copy of the pointee is made between the caller and the callee. NOTE: The patch has a trick for tail calls that we pass by-move. We can choose to copy an alloca even for by-move arguments, but tail calls require MUST-by-move.

@bar

codegen: Copy to an alloca when the argument is neither by-val nor by-move for indirect pointer. Fixes rust-lang/rust#155241. When a value is passed via an indirect pointer, the value needs to be copied to a new alloca. For x86_64-unknown-linux-gnu, `Thing` is the case: ```rust #[derive(Clone, Copy)] struct Thing(usize, usize, usize); pub fn foo() { let thing = Thing(0, 0, 0); bar(thing); assert_eq!(thing.0, 0); } #[inline(never)] #[unsafe(no_mangle)] pub fn bar(mut thing: Thing) { thing.0 = 1; } ``` Before passing the thing to the bar function, the thing needs to be copied to an alloca that is passed to bar. ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) call void @bar(ptr %0) ``` This patch applies the rule to the untupled arguments as well. ```rust #![feature(fn_traits)] #[derive(Clone, Copy)] struct Thing(usize, usize, usize); #[inline(never)] #[unsafe(no_mangle)] pub fn foo() { let thing = (Thing(0, 0, 0),); (|mut thing: Thing| { thing.0 = 1; }).call(thing); assert_eq!(thing.0.0, 0); } ``` For this case, this patch changes from ```llvm ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %thing) ``` to ```llvm %0 = alloca [24 x i8], align 8 call void @llvm.memcpy.p0.p0.i64(ptr align 8 %0, ptr align 8 %thing, i64 24, i1 false) ; call example::foo::{closure#0} call void @_RNCNvCs15qdZVLwHPA_7example3foo0B3_(ptr ..., ptr %0) ``` However, the same rule cannot be applied to tail calls that would be unsound, because the caller's stack frame is overwritten by the callee's stack frame. Fortunately, rust-lang/rust#151143 has already handled the special case. We must not copy again. No copy is needed for by-move arguments, because the argument is passed to the called "in-place". No copy is also needed for by-val arguments, because the attribute implies that a hidden copy of the pointee is made between the caller and the callee. NOTE: The patch has a trick for tail calls that we pass by-move. We can choose to copy an alloca even for by-move arguments, but tail calls require MUST-by-move.

folkertdev added the F-explicit_tail_calls `#![feature(explicit_tail_calls)]` label Jan 14, 2026

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Jan 14, 2026

folkertdev force-pushed the tail-call-indirect branch 2 times, most recently from 8a1cd78 to 6f7eede Compare January 14, 2026 21:41

folkertdev commented Jan 14, 2026

View reviewed changes