[llvm] 27d3528 - [SLP]Fix vectorization of insertelements with multiple uses.
Alexey Bataev via llvm-commits
llvm-commits at lists.llvm.org
Wed May 26 09:43:14 PDT 2021
Author: Alexey Bataev
Date: 2021-05-26T09:42:18-07:00
New Revision: 27d3528acf8aacc62a955dc13b0f08d4167b5b48
URL: https://github.com/llvm/llvm-project/commit/27d3528acf8aacc62a955dc13b0f08d4167b5b48
DIFF: https://github.com/llvm/llvm-project/commit/27d3528acf8aacc62a955dc13b0f08d4167b5b48.diff
LOG: [SLP]Fix vectorization of insertelements with multiple uses.
SLP vectorizer should not consider in sertelements with multiple uses as
a part of high level build vector, it must be considered as
a terminating insertelement in the vector build, otherwise it may
produce incorrect code.
Differential Revision: https://reviews.llvm.org/D103164
Added:
Modified:
llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
llvm/test/Transforms/SLPVectorizer/X86/insert-element-multiple-uses.ll
Removed:
################################################################################
diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index 8de5188de814..0f8fb09b6f6c 100644
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -7722,7 +7722,8 @@ static bool findBuildAggregate_rec(Instruction *LastInsertInst,
LastInsertInst = dyn_cast<Instruction>(LastInsertInst->getOperand(0));
} while (LastInsertInst != nullptr &&
(isa<InsertValueInst>(LastInsertInst) ||
- isa<InsertElementInst>(LastInsertInst)));
+ isa<InsertElementInst>(LastInsertInst)) &&
+ LastInsertInst->hasOneUse());
return true;
}
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/insert-element-multiple-uses.ll b/llvm/test/Transforms/SLPVectorizer/X86/insert-element-multiple-uses.ll
index 713e0dad5ee2..c04217e4913f 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/insert-element-multiple-uses.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/insert-element-multiple-uses.ll
@@ -5,10 +5,14 @@ define void @main() {
; CHECK-LABEL: @main(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, <2 x i64>* undef, align 16
-; CHECK-NEXT: [[TMP1:%.*]] = add <2 x i64> [[TMP0]], <i64 1, i64 1>
-; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
-; CHECK-NEXT: [[CMP_I:%.*]] = icmp eq i64 [[TMP2]], 0
-; CHECK-NEXT: [[VEC_0_I:%.*]] = select i1 [[CMP_I]], <2 x i64> [[TMP1]], <2 x i64> [[TMP1]]
+; CHECK-NEXT: [[VEC_0_VEC_EXTRACT_I:%.*]] = extractelement <2 x i64> [[TMP0]], i32 0
+; CHECK-NEXT: [[ADD_I:%.*]] = add i64 [[VEC_0_VEC_EXTRACT_I]], 1
+; CHECK-NEXT: [[VEC_0_VEC_INSERT_I:%.*]] = insertelement <2 x i64> [[TMP0]], i64 [[ADD_I]], i32 0
+; CHECK-NEXT: [[CMP_I:%.*]] = icmp eq i64 [[ADD_I]], 0
+; CHECK-NEXT: [[VEC_8_VEC_EXTRACT_I:%.*]] = extractelement <2 x i64> [[TMP0]], i32 1
+; CHECK-NEXT: [[INC_I:%.*]] = add i64 [[VEC_8_VEC_EXTRACT_I]], 1
+; CHECK-NEXT: [[VEC_8_VEC_INSERT_I:%.*]] = insertelement <2 x i64> [[VEC_0_VEC_INSERT_I]], i64 [[INC_I]], i32 1
+; CHECK-NEXT: [[VEC_0_I:%.*]] = select i1 [[CMP_I]], <2 x i64> [[VEC_8_VEC_INSERT_I]], <2 x i64> [[VEC_0_VEC_INSERT_I]]
; CHECK-NEXT: ret void
;
entry:
More information about the llvm-commits
mailing list