[LSV] Change the default value of InstertElement to poison

This patch is changing the InsertElement's placeholder to poison without changing the LSV's behavior.

Regardless of whether `StoreTy` is FixedVectorType or not, the poison value will be overwritten with a different value.
Therefore, whether the InsertElement's placeholder is poison or undef will not affect the result of the program.

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D111005

diff --git a/llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp b/llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
index 68da163..1b52088 100644 (file)
--- a/llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
@@ -1114,7 +1114,7 @@ bool Vectorizer::vectorizeStoreChain(
   std::tie(First, Last) = getBoundaryInstrs(Chain);
   Builder.SetInsertPoint(&*Last);
 
-  Value *Vec = UndefValue::get(VecTy);
+  Value *Vec = PoisonValue::get(VecTy);
 
   if (VecStoreTy) {
     unsigned VecWidth = VecStoreTy->getNumElements();

diff --git a/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll b/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll
index c1889cd..03a55b3 100644 (file)
--- a/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll
+++ b/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll
@@ -221,7 +221,7 @@ define amdgpu_kernel void @merge_global_store_4_constants_i64(i64 addrspace(1)*
 ; CHECK: [[LOAD:%[^ ]+]] = load <2 x i32>
 ; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 0
 ; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 1
-; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> undef, i32 [[ELT0]], i32 0
+; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> poison, i32 [[ELT0]], i32 0
 ; CHECK: [[INSERT1:%[^ ]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT1]], i32 1
 ; CHECK: store <2 x i32> [[INSERT1]]
 define amdgpu_kernel void @merge_global_store_2_adjacent_loads_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {
@@ -260,7 +260,7 @@ define amdgpu_kernel void @merge_global_store_2_adjacent_loads_i32_nonzero_base(
 ; CHECK: [[LOAD:%[^ ]+]] = load <2 x i32>
 ; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 0
 ; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 1
-; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> undef, i32 [[ELT1]], i32 0
+; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> poison, i32 [[ELT1]], i32 0
 ; CHECK: [[INSERT1:%[^ ]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT0]], i32 1
 ; CHECK: store <2 x i32> [[INSERT1]]
 define amdgpu_kernel void @merge_global_store_2_adjacent_loads_shuffle_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {

diff --git a/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll b/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll
index 8a6dafe..ef6ee2e 100644 (file)
--- a/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll
+++ b/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll
@@ -74,7 +74,7 @@ entry:
 
 ; CHECK-LABEL: @merge_store_ptr64_i64(
 ; CHECK: [[ELT0:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
-; CHECK: insertelement <2 x i64> undef, i64 [[ELT0]], i32 0
+; CHECK: insertelement <2 x i64> poison, i64 [[ELT0]], i32 0
 ; CHECK: store <2 x i64>
 define amdgpu_kernel void @merge_store_ptr64_i64(i64 addrspace(1)* nocapture %a, i8 addrspace(1)* %ptr0, i64 %val1) #0 {
 entry:
@@ -135,7 +135,7 @@ entry:
 
 ; CHECK-LABEL: @merge_store_ptr32_i32(
 ; CHECK: [[ELT0:%[^ ]+]] = ptrtoint i8 addrspace(3)* %ptr0 to i32
-; CHECK: insertelement <2 x i32> undef, i32 [[ELT0]], i32 0
+; CHECK: insertelement <2 x i32> poison, i32 [[ELT0]], i32 0
 ; CHECK: store <2 x i32>
 define amdgpu_kernel void @merge_store_ptr32_i32(i32 addrspace(3)* nocapture %a, i8 addrspace(3)* %ptr0, i32 %val1) #0 {
 entry:
@@ -275,7 +275,7 @@ entry:
 
 ; CHECK-LABEL: @merge_store_ptr64_f64(
 ; CHECK: [[ELT0_INT:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
-; CHECK: insertelement <2 x i64> undef, i64 [[ELT0_INT]], i32 0
+; CHECK: insertelement <2 x i64> poison, i64 [[ELT0_INT]], i32 0
 ; CHECK: [[ELT1_INT:%[^ ]+]] = bitcast double %val1 to i64
 ; CHECK: insertelement <2 x i64> %{{[^ ]+}}, i64 [[ELT1_INT]], i32 1
 ; CHECK: store <2 x i64>
@@ -292,7 +292,7 @@ entry:
 
 ; CHECK-LABEL: @merge_store_f64_ptr64(
 ; CHECK: [[ELT0_INT:%[^ ]+]] = bitcast double %val0 to i64
-; CHECK: insertelement <2 x i64> undef, i64 [[ELT0_INT]], i32 0
+; CHECK: insertelement <2 x i64> poison, i64 [[ELT0_INT]], i32 0
 ; CHECK: [[ELT1_INT:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr1 to i64
 ; CHECK: insertelement <2 x i64> %{{[^ ]+}}, i64 [[ELT1_INT]], i32 1
 ; CHECK: store <2 x i64>

diff --git a/llvm/test/Transforms/LoadStoreVectorizer/int_sideeffect.ll b/llvm/test/Transforms/LoadStoreVectorizer/int_sideeffect.ll
index 5c6f4ce..23108e3 100644 (file)
--- a/llvm/test/Transforms/LoadStoreVectorizer/int_sideeffect.ll
+++ b/llvm/test/Transforms/LoadStoreVectorizer/int_sideeffect.ll
@@ -17,7 +17,7 @@ define void @test_sideeffect(float* %p) {
 ; CHECK-NEXT:    [[L34:%.*]] = extractelement <4 x float> [[TMP2]], i32 3
 ; CHECK-NEXT:    call void @llvm.sideeffect()
 ; CHECK-NEXT:    call void @llvm.sideeffect()
-; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x float> undef, float [[L01]], i32 0
+; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x float> poison, float [[L01]], i32 0
 ; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <4 x float> [[TMP3]], float [[L12]], i32 1
 ; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <4 x float> [[TMP4]], float [[L23]], i32 2
 ; CHECK-NEXT:    [[TMP6:%.*]] = insertelement <4 x float> [[TMP5]], float [[L34]], i32 3
@@ -55,7 +55,7 @@ define void @test_inaccessiblememonly(float* %p) {
 ; CHECK-NEXT:    [[L34:%.*]] = extractelement <4 x float> [[TMP2]], i32 3
 ; CHECK-NEXT:    call void @foo() #[[ATTR1:[0-9]+]]
 ; CHECK-NEXT:    call void @foo() #[[ATTR1]]
-; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x float> undef, float [[L01]], i32 0
+; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x float> poison, float [[L01]], i32 0
 ; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <4 x float> [[TMP3]], float [[L12]], i32 1
 ; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <4 x float> [[TMP4]], float [[L23]], i32 2
 ; CHECK-NEXT:    [[TMP6:%.*]] = insertelement <4 x float> [[TMP5]], float [[L34]], i32 3