diff options
author | Nikita Popov <npopov@redhat.com> | 2024-02-16 14:50:14 +0100 |
---|---|---|
committer | GitHub <noreply@github.com> | 2024-02-16 05:50:14 -0800 |
commit | 1a69056c899a74c311d700bd0f5618cbfee23518 (patch) | |
tree | df27821a80f9f0891b5495e8ed4f6f1903165894 /llvm | |
parent | d01a4ab21044ceb20e39b783a5983a8d4cc93cb6 (diff) | |
download | llvm-1a69056c899a74c311d700bd0f5618cbfee23518.zip llvm-1a69056c899a74c311d700bd0f5618cbfee23518.tar.gz llvm-1a69056c899a74c311d700bd0f5618cbfee23518.tar.bz2 |
Backport [DAGCombine] Fix multi-use miscompile in load combine (#81586) (#81633)
(cherry picked from commit 25b9ed6e4964344e3710359bec4c831e5a8448b9)
Diffstat (limited to 'llvm')
-rw-r--r-- | llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp | 2 | ||||
-rw-r--r-- | llvm/test/CodeGen/X86/load-combine.ll | 32 |
2 files changed, 33 insertions, 1 deletions
diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp index 98d8a6d..3135ec7 100644 --- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp +++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp @@ -9253,7 +9253,7 @@ SDValue DAGCombiner::MatchLoadCombine(SDNode *N) { // Transfer chain users from old loads to the new load. for (LoadSDNode *L : Loads) - DAG.ReplaceAllUsesOfValueWith(SDValue(L, 1), SDValue(NewLoad.getNode(), 1)); + DAG.makeEquivalentMemoryOrdering(L, NewLoad); if (!NeedsBswap) return NewLoad; diff --git a/llvm/test/CodeGen/X86/load-combine.ll b/llvm/test/CodeGen/X86/load-combine.ll index 7f8115d..b5f3e78 100644 --- a/llvm/test/CodeGen/X86/load-combine.ll +++ b/llvm/test/CodeGen/X86/load-combine.ll @@ -1282,3 +1282,35 @@ define i32 @zext_load_i32_by_i8_bswap_shl_16(ptr %arg) { %tmp8 = or i32 %tmp7, %tmp30 ret i32 %tmp8 } + +define i32 @pr80911_vector_load_multiuse(ptr %ptr, ptr %clobber) nounwind { +; CHECK-LABEL: pr80911_vector_load_multiuse: +; CHECK: # %bb.0: +; CHECK-NEXT: pushl %esi +; CHECK-NEXT: movl {{[0-9]+}}(%esp), %ecx +; CHECK-NEXT: movl {{[0-9]+}}(%esp), %edx +; CHECK-NEXT: movl (%edx), %esi +; CHECK-NEXT: movzwl (%edx), %eax +; CHECK-NEXT: movl $0, (%ecx) +; CHECK-NEXT: movl %esi, (%edx) +; CHECK-NEXT: popl %esi +; CHECK-NEXT: retl +; +; CHECK64-LABEL: pr80911_vector_load_multiuse: +; CHECK64: # %bb.0: +; CHECK64-NEXT: movl (%rdi), %ecx +; CHECK64-NEXT: movzwl (%rdi), %eax +; CHECK64-NEXT: movl $0, (%rsi) +; CHECK64-NEXT: movl %ecx, (%rdi) +; CHECK64-NEXT: retq + %load = load <4 x i8>, ptr %ptr, align 16 + store i32 0, ptr %clobber + store <4 x i8> %load, ptr %ptr, align 16 + %e1 = extractelement <4 x i8> %load, i64 1 + %e1.ext = zext i8 %e1 to i32 + %e1.ext.shift = shl nuw nsw i32 %e1.ext, 8 + %e0 = extractelement <4 x i8> %load, i64 0 + %e0.ext = zext i8 %e0 to i32 + %res = or i32 %e1.ext.shift, %e0.ext + ret i32 %res +} |