[X86] Fix missing ByValTemporaries update in CopyViaTemp path for musttail calls by xiongzile · Pull Request #190540 · llvm/llvm-project

xiongzile · 2026-04-05T17:18:34Z

This fixes a miscompilation in musttail calls with byval arguments on X86.

In the CopyViaTemp path, a temporary stack object is created and the argument is copied into it.
However, the temporary is not recorded in ByValTemporaries,
so the final lowering phase does not emit the copy to the real outgoing argument slot.

As a result, the callee may read incorrect values from the stack.

Fix this by recording the temporary in ByValTemporaries so that the final lowering step correctly copies the argument to the expected stack location.

Reproducer: #190429

github-actions · 2026-04-05T17:18:53Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2026-04-05T17:19:25Z

@llvm/pr-subscribers-backend-x86

Author: Zile Xiong (xiongzile)

Changes

This fixes a miscompilation in musttail calls with byval arguments on X86.

In the CopyViaTemp path, a temporary stack object is created and the argument is copied into it.
However, the temporary is not recorded in ByValTemporaries,
so the final lowering phase does not emit the copy to the real outgoing argument slot.

As a result, the callee may read incorrect values from the stack.

Fix this by recording the temporary in ByValTemporaries so that the final lowering step correctly copies the argument to the expected stack location.

Reproducer: #190429

Full diff: https://github.com/llvm/llvm-project/pull/190540.diff

1 Files Affected:

(modified) llvm/lib/Target/X86/X86ISelLoweringCall.cpp (+1)

diff --git a/llvm/lib/Target/X86/X86ISelLoweringCall.cpp b/llvm/lib/Target/X86/X86ISelLoweringCall.cpp
index 37c80e27f4bd2..65d77769b3c45 100644
--- a/llvm/lib/Target/X86/X86ISelLoweringCall.cpp
+++ b/llvm/lib/Target/X86/X86ISelLoweringCall.cpp
@@ -2237,6 +2237,7 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
         SDValue CopyChain =
             CreateCopyOfByValArgument(Src, Temp, Chain, Flags, DAG, dl);
         ByValCopyChains.push_back(CopyChain);
+        ByValTemporaries[ArgIdx] = Temp;
       }
     }
     if (!ByValCopyChains.empty())

xiongzile · 2026-04-06T15:19:50Z

Hi @RKSimon , could you please take a look at this PR when you have a moment?

This fixes a miscompilation in musttail calls with byval arguments on X86. All checks are passing.

Thanks!

efriedma-quic

CC @folkertdev @rnk

efriedma-quic · 2026-04-06T16:23:21Z

@@ -0,0 +1,90 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6


Does this new test cover any constructs that are significantly different from the existing test coverage in musttail-struct.ll?

You're right — the existing test already covers this case after the update.
I've removed the additional test in the latest commit. Thanks!

…ttail calls The CopyViaTemp path creates a temporary for byval arguments but does not record it in ByValTemporaries. As a result, the final lowering phase does not emit the copy to the outgoing argument slot, leading to incorrect values being observed by the callee. Record the temporary in ByValTemporaries to ensure correct lowering. Fixes: llvm#190429

folkertdev

LGTM

folkertdev · 2026-04-06T16:49:35Z

        SDValue CopyChain =
            CreateCopyOfByValArgument(Src, Temp, Chain, Flags, DAG, dl);
        ByValCopyChains.push_back(CopyChain);
+        ByValTemporaries[ArgIdx] = Temp;


The fix here is consistent with the arm implementation that the x86 implementation is based on

llvm-project/llvm/lib/Target/ARM/ARMISelLowering.cpp

Lines 2193 to 2201 in e7ac60c

SDVTList VTs = DAG.getVTList(MVT::Other, MVT::Glue);

SDValue Ops[] = {Chain, Temp, Src, SizeNode, AlignNode};

ByValCopyChains.push_back(

DAG.getNode(ARMISD::COPY_STRUCT_BYVAL, dl, VTs, Ops));

ByValTemporaries[ArgIdx] = Temp;

}

}

if (!ByValCopyChains.empty())

ByValTempChain =

That line must have gotten lost at some point...

efriedma-quic

LGTM

Let's cherry-pick this to 22.

It would also be nice if anyone has suggestions for better testing, to make sure there aren't any further issues...

RKSimon

LGTM

RKSimon · 2026-04-08T14:44:11Z

LGTM

Let's cherry-pick this to 22.

I'll let this stew in trunk for a few days and then get it cherry picked (probably next week during EuroLLVM)

github-actions · 2026-04-08T15:12:31Z

@xiongzile Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

RKSimon · 2026-04-16T18:24:13Z

/cherry-pick abd502a

llvmbot · 2026-04-16T18:31:50Z

/pull-request #192507

…ttail calls (llvm#190540) This fixes a miscompilation in musttail calls with byval arguments on X86. In the CopyViaTemp path, a temporary stack object is created and the argument is copied into it. However, the temporary is not recorded in ByValTemporaries, so the final lowering phase does not emit the copy to the real outgoing argument slot. As a result, the callee may read incorrect values from the stack. Fix this by recording the temporary in ByValTemporaries so that the final lowering step correctly copies the argument to the expected stack location. Reproducer: llvm#190429 (cherry picked from commit abd502a)

…ttail calls (llvm#190540) This fixes a miscompilation in musttail calls with byval arguments on X86. In the CopyViaTemp path, a temporary stack object is created and the argument is copied into it. However, the temporary is not recorded in ByValTemporaries, so the final lowering phase does not emit the copy to the real outgoing argument slot. As a result, the callee may read incorrect values from the stack. Fix this by recording the temporary in ByValTemporaries so that the final lowering step correctly copies the argument to the expected stack location. Reproducer: llvm#190429

llvmbot added the backend:X86 label Apr 5, 2026

xiongzile force-pushed the musttail-fix branch 2 times, most recently from b0f1215 to 03109e8 Compare April 5, 2026 18:23

RKSimon requested review from RKSimon and efriedma-quic April 6, 2026 15:31

efriedma-quic reviewed Apr 6, 2026

View reviewed changes

xiongzile force-pushed the musttail-fix branch from 03109e8 to 0ccfded Compare April 6, 2026 16:42

xiongzile requested a review from efriedma-quic April 6, 2026 16:47

folkertdev approved these changes Apr 6, 2026

View reviewed changes

efriedma-quic approved these changes Apr 6, 2026

View reviewed changes

RKSimon added this to the LLVM 22.x Release milestone Apr 8, 2026

github-project-automation Bot moved this to Needs Triage in LLVM Release Status Apr 8, 2026

github-project-automation Bot added this to LLVM Release Status Apr 8, 2026

RKSimon approved these changes Apr 8, 2026

View reviewed changes

Merge branch 'main' into musttail-fix

990c47c

RKSimon enabled auto-merge (squash) April 8, 2026 14:42

RKSimon merged commit abd502a into llvm:main Apr 8, 2026
9 of 10 checks passed

github-project-automation Bot moved this from Needs Triage to Done in LLVM Release Status Apr 8, 2026

xroche mentioned this pull request May 23, 2026

[Clang] Forward incoming Indirect parameters across musttail calls #199351

Open

		@@ -0,0 +1,90 @@
		; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6

	SDVTList VTs = DAG.getVTList(MVT::Other, MVT::Glue);
	SDValue Ops[] = {Chain, Temp, Src, SizeNode, AlignNode};
	ByValCopyChains.push_back(
	DAG.getNode(ARMISD::COPY_STRUCT_BYVAL, dl, VTs, Ops));
	ByValTemporaries[ArgIdx] = Temp;
	}
	}
	if (!ByValCopyChains.empty())
	ByValTempChain =

Uh oh!

Conversation

xiongzile commented Apr 5, 2026

Uh oh!

github-actions Bot commented Apr 5, 2026

Uh oh!

llvmbot commented Apr 5, 2026

Uh oh!

xiongzile commented Apr 6, 2026

Uh oh!

efriedma-quic left a comment

Choose a reason for hiding this comment

Uh oh!

efriedma-quic Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

xiongzile Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

folkertdev left a comment

Choose a reason for hiding this comment

Uh oh!

folkertdev Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

efriedma-quic left a comment

Choose a reason for hiding this comment

Uh oh!

RKSimon left a comment

Choose a reason for hiding this comment

Uh oh!

RKSimon commented Apr 8, 2026

Uh oh!

Uh oh!

github-actions Bot commented Apr 8, 2026

Uh oh!

RKSimon commented Apr 16, 2026

Uh oh!

llvmbot commented Apr 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants