[llvm] [VPlan] Preserve nusw on VectorEndPointer (PR #151558)
Ramkumar Ramachandra via llvm-commits
llvm-commits at lists.llvm.org
Mon Aug 4 04:20:56 PDT 2025
https://github.com/artagnon updated https://github.com/llvm/llvm-project/pull/151558
>From 3ddfd273469eaf0bfa23ce564cd62e8611a1b078 Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Thu, 31 Jul 2025 17:43:18 +0100
Subject: [PATCH 1/2] [VPlan] Create VectorEndPointer with full NW flags
In createInterleaveGroups, get the no-wrap flags other than inbounds
from the existing GEP, and set them on the VPVectorEndPointerRecipe.
---
.../Transforms/Vectorize/VPlanTransforms.cpp | 13 +-
.../LoopVectorize/interleaved-accesses.ll | 117 ++++++++++++++----
2 files changed, 96 insertions(+), 34 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 3ecffc7593d49..aaab9ba3fe105 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -2568,10 +2568,10 @@ void VPlanTransforms::createInterleaveGroups(
auto *InsertPos =
cast<VPWidenMemoryRecipe>(RecipeBuilder.getRecipe(IRInsertPos));
- bool InBounds = false;
+ GEPNoWrapFlags NW = GEPNoWrapFlags::none();
if (auto *Gep = dyn_cast<GetElementPtrInst>(
getLoadStorePointerOperand(IRInsertPos)->stripPointerCasts()))
- InBounds = Gep->isInBounds();
+ NW = Gep->getNoWrapFlags();
// Get or create the start address for the interleave group.
auto *Start =
@@ -2595,8 +2595,9 @@ void VPlanTransforms::createInterleaveGroups(
VPValue *OffsetVPV =
Plan.getOrAddLiveIn(ConstantInt::get(Plan.getContext(), -Offset));
VPBuilder B(InsertPos);
- Addr = InBounds ? B.createInBoundsPtrAdd(InsertPos->getAddr(), OffsetVPV)
- : B.createPtrAdd(InsertPos->getAddr(), OffsetVPV);
+ Addr = NW.isInBounds()
+ ? B.createInBoundsPtrAdd(InsertPos->getAddr(), OffsetVPV)
+ : B.createPtrAdd(InsertPos->getAddr(), OffsetVPV);
}
// If the group is reverse, adjust the index to refer to the last vector
// lane instead of the first. We adjust the index from the first vector
@@ -2605,9 +2606,7 @@ void VPlanTransforms::createInterleaveGroups(
if (IG->isReverse()) {
auto *ReversePtr = new VPVectorEndPointerRecipe(
Addr, &Plan.getVF(), getLoadStoreType(IRInsertPos),
- -(int64_t)IG->getFactor(),
- InBounds ? GEPNoWrapFlags::inBounds() : GEPNoWrapFlags::none(),
- InsertPos->getDebugLoc());
+ -(int64_t)IG->getFactor(), NW, InsertPos->getDebugLoc());
ReversePtr->insertBefore(InsertPos);
Addr = ReversePtr;
}
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
index add58758788f9..953d106e3d390 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
@@ -398,6 +398,69 @@ for.body: ; preds = %for.body, %entry
br i1 %cmp, label %for.body, label %for.cond.cleanup
}
+; Variant of the above with nusw flag on the GEP.
+define void @nowrap_flag_preservation(ptr noalias %A, ptr %B) {
+; CHECK-LABEL: @nowrap_flag_preservation(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nusw [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr nusw i8, ptr [[TMP0]], i64 -24
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP3:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr nusw [[STRUCT_ST2]], ptr [[B:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nusw i8, ptr [[TMP4]], i64 -24
+; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
+; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP5]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 -4)
+; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.cond.cleanup:
+; CHECK-NEXT: ret void
+; CHECK: for.body:
+; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP13:![0-9]+]]
+;
+entry:
+ br label %for.body
+
+for.cond.cleanup: ; preds = %for.body
+ ret void
+
+for.body: ; preds = %for.body, %entry
+ %iv = phi i64 [ 1023, %entry ], [ %iv.next, %for.body ]
+ %x = getelementptr nusw %struct.ST2, ptr %A, i64 %iv, i32 0
+ %tmp = load i32, ptr %x, align 4
+ %tmp1 = trunc i64 %iv to i32
+ %add = add nsw i32 %tmp, %tmp1
+ %y = getelementptr nusw %struct.ST2, ptr %A, i64 %iv, i32 1
+ %tmp2 = load i32, ptr %y, align 4
+ %sub = sub nsw i32 %tmp2, %tmp1
+ %x5 = getelementptr nusw %struct.ST2, ptr %B, i64 %iv, i32 0
+ store i32 %add, ptr %x5, align 4
+ %y8 = getelementptr nusw %struct.ST2, ptr %B, i64 %iv, i32 1
+ store i32 %sub, ptr %y8, align 4
+ %iv.next = add nsw i64 %iv, -1
+ %cmp = icmp sgt i64 %iv, 0
+ br i1 %cmp, label %for.body, label %for.cond.cleanup
+}
+
; Check vectorization on an interleaved load group of factor 2 with 1 gap
; (missing the load of odd elements). Because the vectorized loop would
; speculatively access memory out-of-bounds, we must execute at least one
@@ -427,7 +490,7 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store <4 x i32> [[TMP1]], ptr [[TMP3]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 508
-; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -444,7 +507,7 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1022
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP15:![0-9]+]]
;
entry:
br label %for.body
@@ -503,7 +566,7 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -521,7 +584,7 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP15:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP17:![0-9]+]]
;
entry:
br label %for.body
@@ -614,13 +677,13 @@ define void @load_gap_reverse(ptr noalias nocapture %P1, ptr noalias nocapture %
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 -4)
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_EXIT:%.*]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_EXIT]], !llvm.loop [[LOOP17:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_EXIT]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK: for.exit:
; CHECK-NEXT: ret void
;
@@ -676,7 +739,7 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
-; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
; CHECK: scalar.ph:
@@ -684,7 +747,7 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP19:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP21:![0-9]+]]
;
entry:
br label %for.body
@@ -748,7 +811,7 @@ define void @mixed_load3_store3(ptr nocapture %A) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4)
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
; CHECK: scalar.ph:
@@ -756,7 +819,7 @@ define void @mixed_load3_store3(ptr nocapture %A) {
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]]
;
entry:
br label %for.body
@@ -831,7 +894,7 @@ define void @int_float_struct(ptr nocapture readonly %A) #0 {
; CHECK-NEXT: [[TMP3]] = fadd fast <4 x float> [[VEC_PHI]], [[TMP1]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP5:%.*]] = call fast float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> [[TMP3]])
; CHECK-NEXT: [[TMP6:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP2]])
@@ -845,7 +908,7 @@ define void @int_float_struct(ptr nocapture readonly %A) #0 {
; CHECK-NEXT: store float [[ADD3_LCSSA]], ptr @SB, align 4
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP25:![0-9]+]]
;
entry:
br label %for.body
@@ -928,7 +991,7 @@ define void @PR27626_0(ptr %p, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[TMP16]], ptr [[TMP12]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -942,7 +1005,7 @@ define void @PR27626_0(ptr %p, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_Y]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP25:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP27:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1012,7 +1075,7 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK-NEXT: [[TMP14]] = add <4 x i32> [[STRIDED_VEC2]], [[VEC_PHI]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP16:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP14]])
; CHECK-NEXT: br label [[SCALAR_PH]]
@@ -1030,7 +1093,7 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK-NEXT: [[TMP18]] = add nsw i32 [[TMP17]], [[S]]
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP27:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP29:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret i32 [[TMP18]]
;
@@ -1106,7 +1169,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: store i32 [[TMP17]], ptr [[TMP13]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -1122,7 +1185,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: store i32 [[TMP19]], ptr [[P_I_Y]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP29:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP31:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1198,7 +1261,7 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP17]])
; CHECK-NEXT: br label [[SCALAR_PH]]
@@ -1219,7 +1282,7 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[TMP22]] = add nsw i32 [[TMP21]], [[S]]
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP31:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP33:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret i32 [[TMP22]]
;
@@ -1294,7 +1357,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP14]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -1310,7 +1373,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I_PLUS_1]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP33:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP35:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1404,7 +1467,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 8)
; CHECK-NEXT: [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -1423,7 +1486,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP35:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP37:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1499,10 +1562,10 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[TMP12:%.*]] = mul nsw <4 x i32> [[TMP9]], [[TMP10]]
; CHECK-NEXT: [[TMP13:%.*]] = mul nsw <4 x i32> [[TMP12]], [[TMP11]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP14]], align 4, !alias.scope [[META36:![0-9]+]], !noalias [[META39:![0-9]+]]
+; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP14]], align 4, !alias.scope [[META38:![0-9]+]], !noalias [[META41:![0-9]+]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP41:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP43:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <8 x i16> [[WIDE_VEC]], i64 7
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
@@ -1531,7 +1594,7 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I]]
; CHECK-NEXT: store i32 [[MUL012]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[IV]], [[N]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[END]], label [[LOOP]], !llvm.loop [[LOOP42:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[END]], label [[LOOP]], !llvm.loop [[LOOP44:![0-9]+]]
; CHECK: end:
; CHECK-NEXT: ret void
;
>From b319dd889753a1064d5e92d12537ed151ebde06d Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Mon, 4 Aug 2025 12:05:29 +0100
Subject: [PATCH 2/2] [VPlan] Drop nuw: thinko
---
.../Transforms/Vectorize/VPlanTransforms.cpp | 2 +-
.../interleaved-accesses-gep-nowrap-flags.ll | 187 ++++++++++++++++++
.../LoopVectorize/interleaved-accesses.ll | 117 +++--------
3 files changed, 215 insertions(+), 91 deletions(-)
create mode 100644 llvm/test/Transforms/LoopVectorize/interleaved-accesses-gep-nowrap-flags.ll
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index aaab9ba3fe105..8b6dc7dff2634 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -2571,7 +2571,7 @@ void VPlanTransforms::createInterleaveGroups(
GEPNoWrapFlags NW = GEPNoWrapFlags::none();
if (auto *Gep = dyn_cast<GetElementPtrInst>(
getLoadStorePointerOperand(IRInsertPos)->stripPointerCasts()))
- NW = Gep->getNoWrapFlags();
+ NW = Gep->getNoWrapFlags().withoutNoUnsignedWrap();
// Get or create the start address for the interleave group.
auto *Start =
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-gep-nowrap-flags.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-gep-nowrap-flags.ll
new file mode 100644
index 0000000000000..552f6a4ec62d9
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-gep-nowrap-flags.ll
@@ -0,0 +1,187 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph" --version 5
+; RUN: opt -S -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -enable-interleaved-mem-accesses=true %s | FileCheck %s
+
+%struct.i32.pair = type { i32, i32 }
+
+define void @nusw_preservation(ptr noalias %A, ptr %B) {
+; CHECK-LABEL: define void @nusw_preservation(
+; CHECK-SAME: ptr noalias [[A:%.*]], ptr [[B:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nusw [[STRUCT_I32_PAIR:%.*]], ptr [[A]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr nusw i32, ptr [[TMP0]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr nusw i32, ptr [[TMP1]], i64 -6
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP4:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nusw [[STRUCT_I32_PAIR]], ptr [[B]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr nusw i32, ptr [[TMP5]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr nusw i32, ptr [[TMP6]], i64 -6
+; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP4]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i32> [[TMP8]], <8 x i32> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
+; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 -4)
+; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br [[EXIT:label %.*]]
+; CHECK: [[SCALAR_PH]]:
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 1023, %entry ], [ %iv.next, %loop ]
+ %x = getelementptr nusw %struct.i32.pair, ptr %A, i64 %iv, i32 0
+ %load.x = load i32, ptr %x, align 4
+ %trunc = trunc i64 %iv to i32
+ %add = add nsw i32 %load.x, %trunc
+ %y = getelementptr nusw %struct.i32.pair, ptr %A, i64 %iv, i32 1
+ %load.y = load i32, ptr %y, align 4
+ %sub = sub nsw i32 %load.y, %trunc
+ %gep.B.iv.0 = getelementptr nusw %struct.i32.pair, ptr %B, i64 %iv, i32 0
+ store i32 %add, ptr %gep.B.iv.0, align 4
+ %gep.B.iv.1 = getelementptr nusw %struct.i32.pair, ptr %B, i64 %iv, i32 1
+ store i32 %sub, ptr %gep.B.iv.1, align 4
+ %iv.next = add nsw i64 %iv, -1
+ %exit.cond = icmp sgt i64 %iv, 0
+ br i1 %exit.cond, label %loop, label %exit
+
+exit:
+ ret void
+}
+
+define void @inbounds_preservation(ptr noalias %A, ptr %B) {
+; CHECK-LABEL: define void @inbounds_preservation(
+; CHECK-SAME: ptr noalias [[A:%.*]], ptr [[B:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_I32_PAIR:%.*]], ptr [[A]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 -6
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP4:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_I32_PAIR]], ptr [[B]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i64 -6
+; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP4]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i32> [[TMP8]], <8 x i32> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
+; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 -4)
+; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br [[EXIT:label %.*]]
+; CHECK: [[SCALAR_PH]]:
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 1023, %entry ], [ %iv.next, %loop ]
+ %x = getelementptr inbounds %struct.i32.pair, ptr %A, i64 %iv, i32 0
+ %load.x = load i32, ptr %x, align 4
+ %trunc = trunc i64 %iv to i32
+ %add = add nsw i32 %load.x, %trunc
+ %y = getelementptr inbounds %struct.i32.pair, ptr %A, i64 %iv, i32 1
+ %load.y = load i32, ptr %y, align 4
+ %sub = sub nsw i32 %load.y, %trunc
+ %gep.B.iv.0 = getelementptr inbounds %struct.i32.pair, ptr %B, i64 %iv, i32 0
+ store i32 %add, ptr %gep.B.iv.0, align 4
+ %gep.B.iv.1 = getelementptr inbounds %struct.i32.pair, ptr %B, i64 %iv, i32 1
+ store i32 %sub, ptr %gep.B.iv.1, align 4
+ %iv.next = add nsw i64 %iv, -1
+ %exit.cond = icmp sgt i64 %iv, 0
+ br i1 %exit.cond, label %loop, label %exit
+
+exit:
+ ret void
+}
+
+define void @nuw_drop(ptr noalias %A, ptr %B) {
+; CHECK-LABEL: define void @nuw_drop(
+; CHECK-SAME: ptr noalias [[A:%.*]], ptr [[B:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw [[STRUCT_I32_PAIR:%.*]], ptr [[A]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[TMP0]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[TMP1]], i64 -6
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP4:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nuw [[STRUCT_I32_PAIR]], ptr [[B]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[TMP5]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP6]], i64 -6
+; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP4]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i32> [[TMP8]], <8 x i32> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
+; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 -4)
+; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br [[EXIT:label %.*]]
+; CHECK: [[SCALAR_PH]]:
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 1023, %entry ], [ %iv.next, %loop ]
+ %x = getelementptr nuw %struct.i32.pair, ptr %A, i64 %iv, i32 0
+ %load.x = load i32, ptr %x, align 4
+ %trunc = trunc i64 %iv to i32
+ %add = add nsw i32 %load.x, %trunc
+ %y = getelementptr nuw %struct.i32.pair, ptr %A, i64 %iv, i32 1
+ %load.y = load i32, ptr %y, align 4
+ %sub = sub nsw i32 %load.y, %trunc
+ %gep.B.iv.0 = getelementptr nuw %struct.i32.pair, ptr %B, i64 %iv, i32 0
+ store i32 %add, ptr %gep.B.iv.0, align 4
+ %gep.B.iv.1 = getelementptr nuw %struct.i32.pair, ptr %B, i64 %iv, i32 1
+ store i32 %sub, ptr %gep.B.iv.1, align 4
+ %iv.next = add nsw i64 %iv, -1
+ %exit.cond = icmp sgt i64 %iv, 0
+ br i1 %exit.cond, label %loop, label %exit
+
+exit:
+ ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
index 953d106e3d390..add58758788f9 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
@@ -398,69 +398,6 @@ for.body: ; preds = %for.body, %entry
br i1 %cmp, label %for.body, label %for.cond.cleanup
}
-; Variant of the above with nusw flag on the GEP.
-define void @nowrap_flag_preservation(ptr noalias %A, ptr %B) {
-; CHECK-LABEL: @nowrap_flag_preservation(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK: vector.ph:
-; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nusw [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr nusw i8, ptr [[TMP0]], i64 -24
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
-; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
-; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP3:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr nusw [[STRUCT_ST2]], ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nusw i8, ptr [[TMP4]], i64 -24
-; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
-; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP5]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 -4)
-; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
-; CHECK: middle.block:
-; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
-; CHECK: scalar.ph:
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
-; CHECK: for.cond.cleanup:
-; CHECK-NEXT: ret void
-; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP13:![0-9]+]]
-;
-entry:
- br label %for.body
-
-for.cond.cleanup: ; preds = %for.body
- ret void
-
-for.body: ; preds = %for.body, %entry
- %iv = phi i64 [ 1023, %entry ], [ %iv.next, %for.body ]
- %x = getelementptr nusw %struct.ST2, ptr %A, i64 %iv, i32 0
- %tmp = load i32, ptr %x, align 4
- %tmp1 = trunc i64 %iv to i32
- %add = add nsw i32 %tmp, %tmp1
- %y = getelementptr nusw %struct.ST2, ptr %A, i64 %iv, i32 1
- %tmp2 = load i32, ptr %y, align 4
- %sub = sub nsw i32 %tmp2, %tmp1
- %x5 = getelementptr nusw %struct.ST2, ptr %B, i64 %iv, i32 0
- store i32 %add, ptr %x5, align 4
- %y8 = getelementptr nusw %struct.ST2, ptr %B, i64 %iv, i32 1
- store i32 %sub, ptr %y8, align 4
- %iv.next = add nsw i64 %iv, -1
- %cmp = icmp sgt i64 %iv, 0
- br i1 %cmp, label %for.body, label %for.cond.cleanup
-}
-
; Check vectorization on an interleaved load group of factor 2 with 1 gap
; (missing the load of odd elements). Because the vectorized loop would
; speculatively access memory out-of-bounds, we must execute at least one
@@ -490,7 +427,7 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store <4 x i32> [[TMP1]], ptr [[TMP3]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 508
-; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -507,7 +444,7 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1022
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP15:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP13:![0-9]+]]
;
entry:
br label %for.body
@@ -566,7 +503,7 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -584,7 +521,7 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP17:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP15:![0-9]+]]
;
entry:
br label %for.body
@@ -677,13 +614,13 @@ define void @load_gap_reverse(ptr noalias nocapture %P1, ptr noalias nocapture %
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 -4)
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_EXIT:%.*]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_EXIT]], !llvm.loop [[LOOP19:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_EXIT]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK: for.exit:
; CHECK-NEXT: ret void
;
@@ -739,7 +676,7 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
-; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
; CHECK: scalar.ph:
@@ -747,7 +684,7 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP21:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_BODY]], label [[FOR_COND_CLEANUP]], !llvm.loop [[LOOP19:![0-9]+]]
;
entry:
br label %for.body
@@ -811,7 +748,7 @@ define void @mixed_load3_store3(ptr nocapture %A) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4)
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
; CHECK: scalar.ph:
@@ -819,7 +756,7 @@ define void @mixed_load3_store3(ptr nocapture %A) {
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
;
entry:
br label %for.body
@@ -894,7 +831,7 @@ define void @int_float_struct(ptr nocapture readonly %A) #0 {
; CHECK-NEXT: [[TMP3]] = fadd fast <4 x float> [[VEC_PHI]], [[TMP1]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP5:%.*]] = call fast float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> [[TMP3]])
; CHECK-NEXT: [[TMP6:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP2]])
@@ -908,7 +845,7 @@ define void @int_float_struct(ptr nocapture readonly %A) #0 {
; CHECK-NEXT: store float [[ADD3_LCSSA]], ptr @SB, align 4
; CHECK-NEXT: ret void
; CHECK: for.body:
-; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP25:![0-9]+]]
+; CHECK-NEXT: br i1 poison, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]]
;
entry:
br label %for.body
@@ -991,7 +928,7 @@ define void @PR27626_0(ptr %p, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[TMP16]], ptr [[TMP12]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -1005,7 +942,7 @@ define void @PR27626_0(ptr %p, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_Y]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP27:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP25:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1075,7 +1012,7 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK-NEXT: [[TMP14]] = add <4 x i32> [[STRIDED_VEC2]], [[VEC_PHI]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP16:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP14]])
; CHECK-NEXT: br label [[SCALAR_PH]]
@@ -1093,7 +1030,7 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK-NEXT: [[TMP18]] = add nsw i32 [[TMP17]], [[S]]
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP29:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP27:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret i32 [[TMP18]]
;
@@ -1169,7 +1106,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: store i32 [[TMP17]], ptr [[TMP13]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -1185,7 +1122,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: store i32 [[TMP19]], ptr [[P_I_Y]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP31:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP29:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1261,7 +1198,7 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP17]])
; CHECK-NEXT: br label [[SCALAR_PH]]
@@ -1282,7 +1219,7 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[TMP22]] = add nsw i32 [[TMP21]], [[S]]
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP33:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]], !llvm.loop [[LOOP31:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret i32 [[TMP22]]
;
@@ -1357,7 +1294,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP14]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -1373,7 +1310,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I_PLUS_1]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP35:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP33:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1467,7 +1404,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 8)
; CHECK-NEXT: [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -1486,7 +1423,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP37:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP35:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -1562,10 +1499,10 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[TMP12:%.*]] = mul nsw <4 x i32> [[TMP9]], [[TMP10]]
; CHECK-NEXT: [[TMP13:%.*]] = mul nsw <4 x i32> [[TMP12]], [[TMP11]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP14]], align 4, !alias.scope [[META38:![0-9]+]], !noalias [[META41:![0-9]+]]
+; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP14]], align 4, !alias.scope [[META36:![0-9]+]], !noalias [[META39:![0-9]+]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP43:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP41:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <8 x i16> [[WIDE_VEC]], i64 7
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
@@ -1594,7 +1531,7 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I]]
; CHECK-NEXT: store i32 [[MUL012]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[IV]], [[N]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[END]], label [[LOOP]], !llvm.loop [[LOOP44:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[END]], label [[LOOP]], !llvm.loop [[LOOP42:![0-9]+]]
; CHECK: end:
; CHECK-NEXT: ret void
;
More information about the llvm-commits
mailing list