Skip to content

[SPIRV_LLVM_BACKEND] carry patch for atomic_cmpxchg issue #11881

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 20, 2025
Merged

[SPIRV_LLVM_BACKEND] carry patch for atomic_cmpxchg issue #11881

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion S/SPIRV_LLVM_Backend/build_tarballs.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ sources = [

# Bash recipe for building across all platforms
script = raw"""
mv llvm-* llvm
mv llvm-*.src llvm
mv cmake-* cmake
mv third-party-* third-party

Expand All @@ -27,6 +27,7 @@ LLVM_SRCDIR=$(pwd)

atomic_patch -p1 $WORKSPACE/srcdir/patches/avoid_builtin_available.patch
atomic_patch -p1 $WORKSPACE/srcdir/patches/fix_insertvalue.patch
atomic_patch -p1 $WORKSPACE/srcdir/patches/atomic_cmpxchg_64bit.patch

install_license LICENSE.TXT

Expand Down
74 changes: 74 additions & 0 deletions S/SPIRV_LLVM_Backend/bundled/patches/atomic_cmpxchg_64bit.patch
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,74 @@
From 2ffa8bb80d2806aaa913dcd2a35a8f298d511458 Mon Sep 17 00:00:00 2001
From: Simeon David Schaub <[email protected]>
Date: Tue, 19 Aug 2025 11:14:53 +0200
Subject: [PATCH] [SPIR-V] fix return type for OpAtomicCompareExchange

fixes #152863

Tests were written with some help from Copilot
---
lib/Target/SPIRV/SPIRVBuiltins.cpp | 3 +-
.../AtomicCompareExchange64BitPhiNode.ll | 35 +++++++++++++++++++
2 files changed, 36 insertions(+), 2 deletions(-)
create mode 100644 llvm/test/CodeGen/SPIRV/AtomicCompareExchange64BitPhiNode.ll

diff --git a/lib/Target/SPIRV/SPIRVBuiltins.cpp b/lib/Target/SPIRV/SPIRVBuiltins.cpp
index 95fa7bc3894f..4bc5c4ab377b 100644
--- a/lib/Target/SPIRV/SPIRVBuiltins.cpp
+++ b/lib/Target/SPIRV/SPIRVBuiltins.cpp
@@ -791,10 +791,9 @@ static bool buildAtomicCompareExchangeInst(
MRI->setRegClass(Tmp, GR->getRegClass(SpvDesiredTy));
GR->assignSPIRVTypeToVReg(SpvDesiredTy, Tmp, MIRBuilder.getMF());

- SPIRVType *IntTy = GR->getOrCreateSPIRVIntegerType(32, MIRBuilder);
MIRBuilder.buildInstr(Opcode)
.addDef(Tmp)
- .addUse(GR->getSPIRVTypeID(IntTy))
+ .addUse(GR->getSPIRVTypeID(SpvDesiredTy))
.addUse(ObjectPtr)
.addUse(ScopeReg)
.addUse(MemSemEqualReg)
diff --git a/test/CodeGen/SPIRV/AtomicCompareExchange64BitPhiNode.ll b/test/CodeGen/SPIRV/AtomicCompareExchange64BitPhiNode.ll
new file mode 100644
index 000000000000..63023c63958d
--- /dev/null
+++ b/test/CodeGen/SPIRV/AtomicCompareExchange64BitPhiNode.ll
@@ -0,0 +1,35 @@
+; RUN: llc -O0 -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s --check-prefix=CHECK-SPIRV
+; RUN: llc -O0 -mtriple=spirv64-unknown-unknown %s -o %t.spv -filetype=obj
+; RUN: spirv-val %t.spv
+
+; Regression test for issue https://github.com/llvm/llvm-project/issues/152863
+; Ensure OpAtomicCompareExchange returns the correct i64 type when used in phi nodes.
+; Previously, this would generate invalid SPIR-V where the atomic operation returned
+; uint (32-bit) but the phi node expected ulong (64-bit), causing validation errors.
+
+; CHECK-SPIRV-DAG: %[[#Long:]] = OpTypeInt 64 0
+; CHECK-SPIRV-DAG: %[[#Ptr:]] = OpTypePointer CrossWorkgroup %[[#Long]]
+; CHECK-SPIRV-DAG: %[[#Void:]] = OpTypeVoid
+; CHECK-SPIRV-DAG: %[[#Int:]] = OpTypeInt 32 0
+; CHECK-SPIRV-DAG: %[[#Zero64:]] = OpConstantNull %[[#Long]]
+; CHECK-SPIRV-DAG: %[[#Scope:]] = OpConstant %[[#Int]] 2
+; CHECK-SPIRV-DAG: %[[#MemSem:]] = OpConstant %[[#Int]] 0
+
+; Verify that both the phi node and atomic operation use the same i64 type
+; CHECK-SPIRV: %[[#ValuePhi:]] = OpPhi %[[#Long]] %[[#Zero64]] %[[#]] %[[#AtomicResult:]] %[[#]]
+; CHECK-SPIRV: %[[#AtomicResult]] = OpAtomicCompareExchange %[[#Long]] %[[#]] %[[#Scope]] %[[#MemSem]] %[[#MemSem]] %[[#Zero64]] %[[#ValuePhi]]
+
+target datalayout = "e-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-G1"
+target triple = "spirv64-unknown-unknown"
+
+declare i64 @_Z14atomic_cmpxchgPU8CLglobalVlll(ptr addrspace(1), i64, i64)
+
+define spir_kernel void @test_atomic_cmpxchg_phi(ptr addrspace(1) %ptr) {
+conversion:
+ br label %L6
+
+L6: ; preds = %L6, %conversion
+ %value_phi = phi i64 [ 0, %conversion ], [ %1, %L6 ]
+ %1 = call i64 @_Z14atomic_cmpxchgPU8CLglobalVlll(ptr addrspace(1) %ptr, i64 %value_phi, i64 0)
+ br label %L6
+}
--
2.50.1