[llvm] [VPlan] Propagate all GEP flags (PR #119899)
Nikita Popov via llvm-commits
llvm-commits at lists.llvm.org
Tue Dec 17 03:09:15 PST 2024
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/119899
>From a56ed2417c8a03591b67b76b65edc3357442b4b5 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Fri, 13 Dec 2024 16:39:09 +0100
Subject: [PATCH 1/2] [VPlan] Propagate all GEP flags
Store GEPNoWrapFlags instead of only InBounds and propagate them.
---
.../Vectorize/LoopVectorizationPlanner.h | 8 ++--
.../Transforms/Vectorize/LoopVectorize.cpp | 7 ++-
llvm/lib/Transforms/Vectorize/VPlan.h | 44 +++++++------------
.../lib/Transforms/Vectorize/VPlanRecipes.cpp | 34 +++++++-------
.../AArch64/low_trip_count_predicates.ll | 22 +++++-----
.../scalable-fp-ext-trunc-illegal-type.ll | 40 ++++++++---------
.../PowerPC/widened-massv-call.ll | 2 +-
.../LoopVectorize/X86/cost-model.ll | 10 ++---
.../LoopVectorize/X86/float-induction-x86.ll | 8 ++--
.../LoopVectorize/X86/interleave-cost.ll | 2 +-
.../LoopVectorize/X86/metadata-enable.ll | 24 +++++-----
...aved-accesses-different-insert-position.ll | 6 +--
.../AArch64/hoist-runtime-checks.ll | 2 +-
...ting-sinking-required-for-vectorization.ll | 8 ++--
.../AArch64/interleavevectorization.ll | 30 ++++++-------
.../AArch64/matrix-extract-insert.ll | 16 +++----
.../AArch64/predicated-reduction.ll | 4 +-
.../PhaseOrdering/X86/pixel-splat.ll | 4 +-
.../Transforms/PhaseOrdering/X86/pr88239.ll | 2 +-
.../X86/preserve-access-group.ll | 8 ++--
.../PhaseOrdering/X86/vdiv-nounroll.ll | 2 +-
.../test/Transforms/PhaseOrdering/X86/vdiv.ll | 8 ++--
.../X86/vector-reduction-known-first-value.ll | 2 +-
23 files changed, 144 insertions(+), 149 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
index fbcf181a45a664..650a4859780da2 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
@@ -222,13 +222,13 @@ class VPBuilder {
VPInstruction *createPtrAdd(VPValue *Ptr, VPValue *Offset, DebugLoc DL = {},
const Twine &Name = "") {
- return tryInsertInstruction(new VPInstruction(
- Ptr, Offset, VPRecipeWithIRFlags::GEPFlagsTy(false), DL, Name));
+ return tryInsertInstruction(
+ new VPInstruction(Ptr, Offset, GEPNoWrapFlags::none(), DL, Name));
}
VPValue *createInBoundsPtrAdd(VPValue *Ptr, VPValue *Offset, DebugLoc DL = {},
const Twine &Name = "") {
- return tryInsertInstruction(new VPInstruction(
- Ptr, Offset, VPRecipeWithIRFlags::GEPFlagsTy(true), DL, Name));
+ return tryInsertInstruction(
+ new VPInstruction(Ptr, Offset, GEPNoWrapFlags::inBounds(), DL, Name));
}
VPDerivedIVRecipe *createDerivedIV(InductionDescriptor::InductionKind Kind,
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 1b78eb27180187..749336939b7109 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -8406,10 +8406,13 @@ VPRecipeBuilder::tryToWidenMemory(Instruction *I, ArrayRef<VPValue *> Operands,
if (Reverse)
VectorPtr = new VPReverseVectorPointerRecipe(
Ptr, &Plan.getVF(), getLoadStoreType(I),
- GEP ? GEP->isInBounds() : false, I->getDebugLoc());
+ GEP && GEP->isInBounds() ? GEPNoWrapFlags::inBounds()
+ : GEPNoWrapFlags::none(),
+ I->getDebugLoc());
else
VectorPtr = new VPVectorPointerRecipe(Ptr, getLoadStoreType(I),
- GEP ? GEP->isInBounds() : false,
+ GEP ? GEP->getNoWrapFlags()
+ : GEPNoWrapFlags::none(),
I->getDebugLoc());
Builder.getInsertBlock()->appendRecipe(VectorPtr);
Ptr = VectorPtr;
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 32db42a4937b43..d26995883cb905 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -952,11 +952,6 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
DisjointFlagsTy(bool IsDisjoint) : IsDisjoint(IsDisjoint) {}
};
- struct GEPFlagsTy {
- char IsInBounds : 1;
- GEPFlagsTy(bool IsInBounds) : IsInBounds(IsInBounds) {}
- };
-
private:
struct ExactFlagsTy {
char IsExact : 1;
@@ -983,7 +978,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
WrapFlagsTy WrapFlags;
DisjointFlagsTy DisjointFlags;
ExactFlagsTy ExactFlags;
- GEPFlagsTy GEPFlags;
+ GEPNoWrapFlags GEPFlags;
NonNegFlagsTy NonNegFlags;
FastMathFlagsTy FMFs;
unsigned AllFlags;
@@ -1020,7 +1015,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
ExactFlags.IsExact = Op->isExact();
} else if (auto *GEP = dyn_cast<GetElementPtrInst>(&I)) {
OpType = OperationType::GEPOp;
- GEPFlags.IsInBounds = GEP->isInBounds();
+ GEPFlags = GEP->getNoWrapFlags();
} else if (auto *PNNI = dyn_cast<PossiblyNonNegInst>(&I)) {
OpType = OperationType::NonNegOp;
NonNegFlags.NonNeg = PNNI->hasNonNeg();
@@ -1060,7 +1055,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
protected:
template <typename IterT>
VPRecipeWithIRFlags(const unsigned char SC, IterT Operands,
- GEPFlagsTy GEPFlags, DebugLoc DL = {})
+ GEPNoWrapFlags GEPFlags, DebugLoc DL = {})
: VPSingleDefRecipe(SC, Operands, DL), OpType(OperationType::GEPOp),
GEPFlags(GEPFlags) {}
@@ -1097,7 +1092,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
ExactFlags.IsExact = false;
break;
case OperationType::GEPOp:
- GEPFlags.IsInBounds = false;
+ GEPFlags = GEPNoWrapFlags::none();
break;
case OperationType::FPMathOp:
FMFs.NoNaNs = false;
@@ -1126,10 +1121,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
I->setIsExact(ExactFlags.IsExact);
break;
case OperationType::GEPOp:
- // TODO(gep_nowrap): Track the full GEPNoWrapFlags in VPlan.
- cast<GetElementPtrInst>(I)->setNoWrapFlags(
- GEPFlags.IsInBounds ? GEPNoWrapFlags::inBounds()
- : GEPNoWrapFlags::none());
+ cast<GetElementPtrInst>(I)->setNoWrapFlags(GEPFlags);
break;
case OperationType::FPMathOp:
I->setHasAllowReassoc(FMFs.AllowReassoc);
@@ -1155,11 +1147,7 @@ class VPRecipeWithIRFlags : public VPSingleDefRecipe {
return CmpPredicate;
}
- bool isInBounds() const {
- assert(OpType == OperationType::GEPOp &&
- "recipe doesn't have inbounds flag");
- return GEPFlags.IsInBounds;
- }
+ GEPNoWrapFlags getGEPNoWrapFlags() const { return GEPFlags; }
/// Returns true if the recipe has fast-math flags.
bool hasFastMathFlags() const { return OpType == OperationType::FPMathOp; }
@@ -1306,7 +1294,7 @@ class VPInstruction : public VPRecipeWithIRFlags,
assert(Opcode == Instruction::Or && "only OR opcodes can be disjoint");
}
- VPInstruction(VPValue *Ptr, VPValue *Offset, GEPFlagsTy Flags,
+ VPInstruction(VPValue *Ptr, VPValue *Offset, GEPNoWrapFlags Flags,
DebugLoc DL = {}, const Twine &Name = "")
: VPRecipeWithIRFlags(VPDef::VPInstructionSC,
ArrayRef<VPValue *>({Ptr, Offset}), Flags, DL),
@@ -1914,10 +1902,9 @@ class VPReverseVectorPointerRecipe : public VPRecipeWithIRFlags,
public:
VPReverseVectorPointerRecipe(VPValue *Ptr, VPValue *VF, Type *IndexedTy,
- bool IsInBounds, DebugLoc DL)
+ GEPNoWrapFlags GEPFlags, DebugLoc DL)
: VPRecipeWithIRFlags(VPDef::VPReverseVectorPointerSC,
- ArrayRef<VPValue *>({Ptr, VF}),
- GEPFlagsTy(IsInBounds), DL),
+ ArrayRef<VPValue *>({Ptr, VF}), GEPFlags, DL),
IndexedTy(IndexedTy) {}
VP_CLASSOF_IMPL(VPDef::VPReverseVectorPointerSC)
@@ -1949,8 +1936,9 @@ class VPReverseVectorPointerRecipe : public VPRecipeWithIRFlags,
}
VPReverseVectorPointerRecipe *clone() override {
- return new VPReverseVectorPointerRecipe(
- getOperand(0), getVFValue(), IndexedTy, isInBounds(), getDebugLoc());
+ return new VPReverseVectorPointerRecipe(getOperand(0), getVFValue(),
+ IndexedTy, getGEPNoWrapFlags(),
+ getDebugLoc());
}
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
@@ -1966,10 +1954,10 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags,
Type *IndexedTy;
public:
- VPVectorPointerRecipe(VPValue *Ptr, Type *IndexedTy, bool IsInBounds,
+ VPVectorPointerRecipe(VPValue *Ptr, Type *IndexedTy, GEPNoWrapFlags GEPFlags,
DebugLoc DL)
: VPRecipeWithIRFlags(VPDef::VPVectorPointerSC, ArrayRef<VPValue *>(Ptr),
- GEPFlagsTy(IsInBounds), DL),
+ GEPFlags, DL),
IndexedTy(IndexedTy) {}
VP_CLASSOF_IMPL(VPDef::VPVectorPointerSC)
@@ -1991,8 +1979,8 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags,
}
VPVectorPointerRecipe *clone() override {
- return new VPVectorPointerRecipe(getOperand(0), IndexedTy, isInBounds(),
- getDebugLoc());
+ return new VPVectorPointerRecipe(getOperand(0), IndexedTy,
+ getGEPNoWrapFlags(), getDebugLoc());
}
/// Return the cost of this VPHeaderPHIRecipe.
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index 1052ce588ad81a..5122eb01fe7a8f 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -621,8 +621,7 @@ Value *VPInstruction::generate(VPTransformState &State) {
"can only generate first lane for PtrAdd");
Value *Ptr = State.get(getOperand(0), VPLane(0));
Value *Addend = State.get(getOperand(1), VPLane(0));
- return isInBounds() ? Builder.CreateInBoundsPtrAdd(Ptr, Addend, Name)
- : Builder.CreatePtrAdd(Ptr, Addend, Name);
+ return Builder.CreatePtrAdd(Ptr, Addend, Name, getGEPNoWrapFlags());
}
case VPInstruction::ResumePhi: {
Value *IncomingFromVPlanPred =
@@ -1276,8 +1275,12 @@ void VPRecipeWithIRFlags::printFlags(raw_ostream &O) const {
getFastMathFlags().print(O);
break;
case OperationType::GEPOp:
- if (GEPFlags.IsInBounds)
+ if (GEPFlags.isInBounds())
O << " inbounds";
+ else if (GEPFlags.hasNoUnsignedSignedWrap())
+ O << " nusw";
+ if (GEPFlags.hasNoUnsignedWrap())
+ O << " nuw";
break;
case OperationType::NonNegOp:
if (NonNegFlags.NonNeg)
@@ -1906,9 +1909,9 @@ void VPWidenGEPRecipe::execute(VPTransformState &State) {
for (unsigned I = 0, E = getNumOperands(); I != E; I++)
Ops.push_back(State.get(getOperand(I), VPLane(0)));
- auto *NewGEP =
- State.Builder.CreateGEP(GEP->getSourceElementType(), Ops[0],
- ArrayRef(Ops).drop_front(), "", isInBounds());
+ auto *NewGEP = State.Builder.CreateGEP(GEP->getSourceElementType(), Ops[0],
+ ArrayRef(Ops).drop_front(), "",
+ getGEPNoWrapFlags());
Value *Splat = State.Builder.CreateVectorSplat(State.VF, NewGEP);
State.set(this, Splat);
State.addMetadata(Splat, GEP);
@@ -1934,7 +1937,7 @@ void VPWidenGEPRecipe::execute(VPTransformState &State) {
// Create the new GEP. Note that this GEP may be a scalar if VF == 1,
// but it should be a vector, otherwise.
auto *NewGEP = State.Builder.CreateGEP(GEP->getSourceElementType(), Ptr,
- Indices, "", isInBounds());
+ Indices, "", getGEPNoWrapFlags());
assert((State.VF.isScalar() || NewGEP->getType()->isVectorTy()) &&
"NewGEP is not a pointer vector");
State.set(this, NewGEP);
@@ -1985,9 +1988,10 @@ void VPReverseVectorPointerRecipe::execute(VPTransformState &State) {
// LastLane = 1 - RunTimeVF
Value *LastLane = Builder.CreateSub(ConstantInt::get(IndexTy, 1), RunTimeVF);
Value *Ptr = State.get(getOperand(0), VPLane(0));
- bool InBounds = isInBounds();
- Value *ResultPtr = Builder.CreateGEP(IndexedTy, Ptr, NumElt, "", InBounds);
- ResultPtr = Builder.CreateGEP(IndexedTy, ResultPtr, LastLane, "", InBounds);
+ Value *ResultPtr =
+ Builder.CreateGEP(IndexedTy, Ptr, NumElt, "", getGEPNoWrapFlags());
+ ResultPtr = Builder.CreateGEP(IndexedTy, ResultPtr, LastLane, "",
+ getGEPNoWrapFlags());
State.set(this, ResultPtr, /*IsScalar*/ true);
}
@@ -1997,9 +2001,9 @@ void VPReverseVectorPointerRecipe::print(raw_ostream &O, const Twine &Indent,
VPSlotTracker &SlotTracker) const {
O << Indent;
printAsOperand(O, SlotTracker);
- O << " = reverse-vector-pointer ";
- if (isInBounds())
- O << "inbounds ";
+ O << " = reverse-vector-pointer";
+ printFlags(O);
+ O << " ";
printOperands(O, SlotTracker);
}
#endif
@@ -2011,10 +2015,10 @@ void VPVectorPointerRecipe::execute(VPTransformState &State) {
Type *IndexTy = getGEPIndexTy(State.VF.isScalable(), /*IsReverse*/ false,
CurrentPart, Builder);
Value *Ptr = State.get(getOperand(0), VPLane(0));
- bool InBounds = isInBounds();
Value *Increment = createStepForVF(Builder, IndexTy, State.VF, CurrentPart);
- Value *ResultPtr = Builder.CreateGEP(IndexedTy, Ptr, Increment, "", InBounds);
+ Value *ResultPtr =
+ Builder.CreateGEP(IndexedTy, Ptr, Increment, "", getGEPNoWrapFlags());
State.set(this, ResultPtr, /*IsScalar*/ true);
}
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/low_trip_count_predicates.ll b/llvm/test/Transforms/LoopVectorize/AArch64/low_trip_count_predicates.ll
index a02323ab2108f9..7a7eb71ea2f23c 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/low_trip_count_predicates.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/low_trip_count_predicates.ll
@@ -81,8 +81,8 @@ define void @low_vf_ic_is_better(ptr nocapture noundef %p, i32 %tc, i16 noundef
; CHECK-VS1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-VS1-NEXT: [[TMP20:%.*]] = add i64 [[TMP0]], [[INDEX]]
; CHECK-VS1-NEXT: [[TMP21:%.*]] = add i64 [[TMP20]], 0
-; CHECK-VS1-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 [[TMP21]]
-; CHECK-VS1-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[TMP22]], i32 0
+; CHECK-VS1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[V]], i64 [[TMP21]]
+; CHECK-VS1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP22]], i32 0
; CHECK-VS1-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 16 x i8>, ptr [[TMP23]], align 1
; CHECK-VS1-NEXT: [[TMP24:%.*]] = add <vscale x 16 x i8> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-VS1-NEXT: store <vscale x 16 x i8> [[TMP24]], ptr [[TMP23]], align 1
@@ -115,8 +115,8 @@ define void @low_vf_ic_is_better(ptr nocapture noundef %p, i32 %tc, i16 noundef
; CHECK-VS1-NEXT: [[INDEX5:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT9:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-VS1-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[TMP0]], [[INDEX5]]
; CHECK-VS1-NEXT: [[TMP32:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-VS1-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 [[TMP32]]
-; CHECK-VS1-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[TMP33]], i32 0
+; CHECK-VS1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw i8, ptr [[V]], i64 [[TMP32]]
+; CHECK-VS1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP33]], i32 0
; CHECK-VS1-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x i8>, ptr [[TMP34]], align 1
; CHECK-VS1-NEXT: [[TMP35:%.*]] = add <vscale x 8 x i8> [[WIDE_LOAD6]], [[BROADCAST_SPLAT8]]
; CHECK-VS1-NEXT: store <vscale x 8 x i8> [[TMP35]], ptr [[TMP34]], align 1
@@ -189,8 +189,8 @@ define void @low_vf_ic_is_better(ptr nocapture noundef %p, i32 %tc, i16 noundef
; CHECK-VS2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-VS2-NEXT: [[TMP20:%.*]] = add i64 [[TMP0]], [[INDEX]]
; CHECK-VS2-NEXT: [[TMP21:%.*]] = add i64 [[TMP20]], 0
-; CHECK-VS2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 [[TMP21]]
-; CHECK-VS2-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[TMP22]], i32 0
+; CHECK-VS2-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[V]], i64 [[TMP21]]
+; CHECK-VS2-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP22]], i32 0
; CHECK-VS2-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x i8>, ptr [[TMP23]], align 1
; CHECK-VS2-NEXT: [[TMP24:%.*]] = add <vscale x 8 x i8> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-VS2-NEXT: store <vscale x 8 x i8> [[TMP24]], ptr [[TMP23]], align 1
@@ -223,8 +223,8 @@ define void @low_vf_ic_is_better(ptr nocapture noundef %p, i32 %tc, i16 noundef
; CHECK-VS2-NEXT: [[INDEX5:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT9:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-VS2-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[TMP0]], [[INDEX5]]
; CHECK-VS2-NEXT: [[TMP32:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-VS2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 [[TMP32]]
-; CHECK-VS2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[TMP33]], i32 0
+; CHECK-VS2-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw i8, ptr [[V]], i64 [[TMP32]]
+; CHECK-VS2-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP33]], i32 0
; CHECK-VS2-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i8>, ptr [[TMP34]], align 1
; CHECK-VS2-NEXT: [[TMP35:%.*]] = add <vscale x 4 x i8> [[WIDE_LOAD6]], [[BROADCAST_SPLAT8]]
; CHECK-VS2-NEXT: store <vscale x 4 x i8> [[TMP35]], ptr [[TMP34]], align 1
@@ -279,7 +279,7 @@ while.end:
define void @trip_count_too_small(ptr nocapture noundef %p, i32 noundef %tc, i16 noundef %val) {
; CHECK-LABEL: define void @trip_count_too_small(
-; CHECK-SAME: ptr nocapture noundef [[P:%.*]], i32 noundef [[TC:%.*]], i16 noundef [[VAL:%.*]]) #[[ATTR0]] {
+; CHECK-SAME: ptr nocapture noundef [[P:%.*]], i32 noundef [[TC:%.*]], i16 noundef [[VAL:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[CMP7:%.*]] = icmp ult i32 [[TC]], 3
; CHECK-NEXT: br i1 [[CMP7]], label %[[WHILE_PREHEADER:.*]], label %[[WHILE_END:.*]]
@@ -440,8 +440,8 @@ define void @overflow_indvar_known_false(ptr nocapture noundef %p, i32 noundef %
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 16 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], %[[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[TMP0]], [[INDEX]]
; CHECK-NEXT: [[TMP12:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 [[TMP12]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP13]], i32 0
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[V]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i32 0
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8.p0(ptr [[TMP14]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]], <vscale x 16 x i8> poison)
; CHECK-NEXT: [[TMP15:%.*]] = add <vscale x 16 x i8> [[WIDE_MASKED_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP15]], ptr [[TMP14]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-fp-ext-trunc-illegal-type.ll b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-fp-ext-trunc-illegal-type.ll
index fb5ff66989a679..7d71ff87ec8be2 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-fp-ext-trunc-illegal-type.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-fp-ext-trunc-illegal-type.ll
@@ -14,36 +14,36 @@ target triple = "aarch64-unknown-linux-gnu"
define void @load_ext_trunc_store(ptr readonly %in, ptr noalias %out, i64 %N) {
; CHECK-LABEL: define void @load_ext_trunc_store(
; CHECK-SAME: ptr readonly [[IN:%.*]], ptr noalias [[OUT:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
-; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: entry:
; CHECK-NEXT: [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[N]], i64 1)
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[UMAX]], 4
-; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
-; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[UMAX]], 4
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[UMAX]], [[N_MOD_VF]]
-; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
-; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[IN]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[TMP2]], i32 0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw double, ptr [[IN]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw double, ptr [[TMP2]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = fpext <4 x double> [[WIDE_LOAD]] to <4 x fp128>
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw float, ptr [[OUT]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP5:%.*]] = fptrunc <4 x fp128> [[TMP3]] to <4 x float>
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP8]], i32 0
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw float, ptr [[TMP8]], i32 0
; CHECK-NEXT: store <4 x float> [[TMP5]], ptr [[TMP12]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP14]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[UMAX]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_EXIT:.*]], label %[[SCALAR_PH]]
-; CHECK: [[SCALAR_PH]]:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: br label %[[FOR_BODY:.*]]
-; CHECK: [[FOR_BODY]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[GEP_IN:%.*]] = getelementptr inbounds nuw double, ptr [[IN]], i64 [[IV]]
; CHECK-NEXT: [[LOAD_IN:%.*]] = load double, ptr [[GEP_IN]], align 8
; CHECK-NEXT: [[LOAD_EXT:%.*]] = fpext double [[LOAD_IN]] to fp128
@@ -52,8 +52,8 @@ define void @load_ext_trunc_store(ptr readonly %in, ptr noalias %out, i64 %N) {
; CHECK-NEXT: store float [[TRUNC_OUT]], ptr [[GEP_OUT]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ult i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label %[[FOR_BODY]], label %[[FOR_EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK: [[FOR_EXIT]]:
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_BODY]], label [[FOR_EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK: for.exit:
; CHECK-NEXT: ret void
;
entry:
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
index 9f13a461b2057c..12b1d24945c62a 100644
--- a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
@@ -10,7 +10,7 @@ define dso_local double @test(ptr %Arr) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, [[ENTRY]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[ARR:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw float, ptr [[ARR:%.*]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fpext <2 x float> [[WIDE_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP2:%.*]] = tail call fast <2 x double> @__sind2(<2 x double> [[TMP1]])
diff --git a/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll b/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
index 4ca988a2d9b531..af1e78cc65bbe8 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
@@ -612,10 +612,10 @@ define void @cost_duplicate_recipe_for_sinking(ptr %A, i64 %N) #2 {
; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP4]], 2
; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[TMP5]], 2
; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[TMP6]], 2
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr double, ptr [[A:%.*]], i64 [[TMP7]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP9]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr nusw double, ptr [[A:%.*]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr nusw double, ptr [[A]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr nusw double, ptr [[A]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr nusw double, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <16 x double> [[WIDE_VEC]], <16 x double> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <16 x double>, ptr [[TMP12]], align 8
@@ -788,7 +788,7 @@ define void @cost_duplicate_recipe_for_sinking(ptr %A, i64 %N) #2 {
; CHECK-NEXT: [[INDEX40:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL1]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT51:%.*]], [[PRED_STORE_CONTINUE50:%.*]] ]
; CHECK-NEXT: [[TMP86:%.*]] = add i64 [[INDEX40]], 0
; CHECK-NEXT: [[TMP87:%.*]] = shl nsw i64 [[TMP86]], 2
-; CHECK-NEXT: [[TMP89:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP87]]
+; CHECK-NEXT: [[TMP89:%.*]] = getelementptr nusw double, ptr [[A]], i64 [[TMP87]]
; CHECK-NEXT: [[WIDE_VEC41:%.*]] = load <16 x double>, ptr [[TMP89]], align 8
; CHECK-NEXT: [[STRIDED_VEC42:%.*]] = shufflevector <16 x double> [[WIDE_VEC41]], <16 x double> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT: [[TMP90:%.*]] = fcmp oeq <4 x double> [[STRIDED_VEC42]], zeroinitializer
diff --git a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
index 12eb3bd03348a4..1fab786f7c74a3 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
@@ -38,7 +38,7 @@ define void @fp_iv_loop1(ptr noalias nocapture %A, i32 %N) #0 {
; AUTO_VEC-NEXT: [[STEP_ADD:%.*]] = fadd fast <8 x float> [[VEC_IND]], splat (float 4.000000e+00)
; AUTO_VEC-NEXT: [[STEP_ADD2:%.*]] = fadd fast <8 x float> [[VEC_IND]], splat (float 8.000000e+00)
; AUTO_VEC-NEXT: [[STEP_ADD3:%.*]] = fadd fast <8 x float> [[VEC_IND]], splat (float 1.200000e+01)
-; AUTO_VEC-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; AUTO_VEC-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[A:%.*]], i64 [[INDEX]]
; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 32
; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 64
; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 96
@@ -74,7 +74,7 @@ define void @fp_iv_loop1(ptr noalias nocapture %A, i32 %N) #0 {
; AUTO_VEC: vec.epilog.vector.body:
; AUTO_VEC-NEXT: [[INDEX10:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT13:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; AUTO_VEC-NEXT: [[VEC_IND11:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VEC_EPILOG_PH]] ], [ [[VEC_IND_NEXT12:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; AUTO_VEC-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX10]]
+; AUTO_VEC-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDEX10]]
; AUTO_VEC-NEXT: store <4 x float> [[VEC_IND11]], ptr [[TMP8]], align 4
; AUTO_VEC-NEXT: [[INDEX_NEXT13]] = add nuw i64 [[INDEX10]], 4
; AUTO_VEC-NEXT: [[VEC_IND_NEXT12]] = fadd fast <4 x float> [[VEC_IND11]], splat (float 2.000000e+00)
@@ -410,7 +410,7 @@ define void @fadd_reassoc_FMF(ptr nocapture %p, i32 %N) {
; AUTO_VEC-NEXT: [[STEP_ADD:%.*]] = fadd reassoc <8 x float> [[VEC_IND]], splat (float 3.360000e+02)
; AUTO_VEC-NEXT: [[STEP_ADD2:%.*]] = fadd reassoc <8 x float> [[STEP_ADD]], splat (float 3.360000e+02)
; AUTO_VEC-NEXT: [[STEP_ADD3:%.*]] = fadd reassoc <8 x float> [[STEP_ADD2]], splat (float 3.360000e+02)
-; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[P:%.*]], i64 [[INDEX]]
+; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[P:%.*]], i64 [[INDEX]]
; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 32
; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 64
; AUTO_VEC-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 96
@@ -454,7 +454,7 @@ define void @fadd_reassoc_FMF(ptr nocapture %p, i32 %N) {
; AUTO_VEC: vec.epilog.vector.body:
; AUTO_VEC-NEXT: [[INDEX13:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT17:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; AUTO_VEC-NEXT: [[VEC_IND14:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VEC_EPILOG_PH]] ], [ [[VEC_IND_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; AUTO_VEC-NEXT: [[TMP13:%.*]] = getelementptr inbounds float, ptr [[P]], i64 [[INDEX13]]
+; AUTO_VEC-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw float, ptr [[P]], i64 [[INDEX13]]
; AUTO_VEC-NEXT: [[WIDE_LOAD16:%.*]] = load <4 x float>, ptr [[TMP13]], align 4
; AUTO_VEC-NEXT: [[TMP14:%.*]] = fadd reassoc <4 x float> [[VEC_IND14]], [[WIDE_LOAD16]]
; AUTO_VEC-NEXT: store <4 x float> [[TMP14]], ptr [[TMP13]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/X86/interleave-cost.ll b/llvm/test/Transforms/LoopVectorize/X86/interleave-cost.ll
index fd89c8b0193415..e5dd723be0c20e 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/interleave-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/interleave-cost.ll
@@ -479,7 +479,7 @@ define void @geps_feeding_interleave_groups_with_reuse2(ptr %A, ptr %B, i64 %N)
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 8
; CHECK-NEXT: [[TMP50:%.*]] = add i64 [[OFFSET_IDX]], 0
; CHECK-NEXT: [[TMP51:%.*]] = lshr exact i64 [[TMP50]], 1
-; CHECK-NEXT: [[TMP52:%.*]] = getelementptr i32, ptr [[B]], i64 [[TMP51]]
+; CHECK-NEXT: [[TMP52:%.*]] = getelementptr nusw i32, ptr [[B]], i64 [[TMP51]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x i32>, ptr [[TMP52]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT: [[STRIDED_VEC34:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <4 x i32> <i32 1, i32 5, i32 9, i32 13>
diff --git a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
index 7c467c264843a3..2b96797036814e 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
@@ -1186,16 +1186,16 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O1VEC2: vector.body:
; O1VEC2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; O1VEC2-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP0]]
-; O1VEC2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
-; O1VEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 4
+; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP0]]
+; O1VEC2-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i32 0
+; O1VEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i32 4
; O1VEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; O1VEC2-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; O1VEC2-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O1VEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; O1VEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
-; O1VEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 4
+; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[TMP0]]
+; O1VEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i32 0
+; O1VEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i32 4
; O1VEC2-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP7]], align 4
; O1VEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP8]], align 4
; O1VEC2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -1230,16 +1230,16 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; OzVEC2: vector.body:
; OzVEC2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; OzVEC2-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP0]]
-; OzVEC2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
-; OzVEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 4
+; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP0]]
+; OzVEC2-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i32 0
+; OzVEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i32 4
; OzVEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; OzVEC2-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; OzVEC2-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; OzVEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; OzVEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
-; OzVEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 4
+; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[TMP0]]
+; OzVEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i32 0
+; OzVEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i32 4
; OzVEC2-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP7]], align 4
; OzVEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP8]], align 4
; OzVEC2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-different-insert-position.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-different-insert-position.ll
index 5913bae082f1da..5818c3ae79979f 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-different-insert-position.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-different-insert-position.ll
@@ -32,11 +32,11 @@ define void @gep_for_first_member_does_not_dominate_insert_point(ptr %str, ptr n
; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 100, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i64 [ 200, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i64 [ 200, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: br label %[[LOOP:.*]]
; CHECK: [[LOOP]]:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; CHECK-NEXT: [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], %[[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL2]], %[[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT: [[OR_1:%.*]] = or disjoint i64 [[IV2]], 1
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[STR]], i64 [[OR_1]]
; CHECK-NEXT: [[TMP9:%.*]] = load i8, ptr [[GEP1]], align 1
@@ -91,7 +91,7 @@ define void @test_ig_insert_pos_at_end_of_vpbb(ptr noalias %dst, ptr noalias %sr
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr { i16, i16, i16, i16 }, ptr [[SRC]], i64 [[TMP3]], i32 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr nusw { i16, i16, i16, i16 }, ptr [[SRC]], i64 [[TMP3]], i32 2
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[TMP4]], i32 -4
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x i16>, ptr [[TMP5]], align 2
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <16 x i16> [[WIDE_VEC]], <16 x i16> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
index 7097171ab78c5f..b2d6455a82944b 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
@@ -25,7 +25,7 @@ define i32 @read_only_loop_with_runtime_check(ptr noundef %array, i32 noundef %c
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI11:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[ARRAY]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
index 89b2f039c827d9..82b1cf90b47207 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
@@ -48,7 +48,7 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
@@ -60,7 +60,7 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: [[TMP8:%.*]] = select <2 x i1> [[TMP6]], <2 x double> splat (double 6.000000e+00), <2 x double> [[WIDE_LOAD8]]
; CHECK-NEXT: [[TMP9:%.*]] = select <2 x i1> [[TMP3]], <2 x double> zeroinitializer, <2 x double> [[TMP7]]
; CHECK-NEXT: [[TMP10:%.*]] = select <2 x i1> [[TMP4]], <2 x double> zeroinitializer, <2 x double> [[TMP8]]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP11]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr [[TMP11]], align 8
; CHECK-NEXT: store <2 x double> [[TMP10]], ptr [[TMP12]], align 8
@@ -148,13 +148,13 @@ define void @loop2(ptr %A, ptr %B, ptr %C, float %x) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i32, ptr [[C]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4, !alias.scope [[META4:![0-9]+]]
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !alias.scope [[META4]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], splat (i32 20)
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD7]], splat (i32 20)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP4]], i64 16
; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <4 x float>, ptr [[TMP4]], align 4, !alias.scope [[META7:![0-9]+]]
; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <4 x float>, ptr [[TMP5]], align 4, !alias.scope [[META7]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
index 69adf2365140a9..be57809132d3fa 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
@@ -17,9 +17,9 @@ define void @add4(ptr noalias noundef %x, ptr noalias noundef %y, i32 noundef %n
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: store <32 x i16> [[INTERLEAVED_VEC]], ptr [[TMP1]], align 2
@@ -137,9 +137,9 @@ define void @addsubs(ptr noalias noundef %x, ptr noundef %y, i32 noundef %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: [[TMP3:%.*]] = sub <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
@@ -263,9 +263,9 @@ define void @add2sub2(ptr noalias noundef %x, ptr noundef %y, i32 noundef %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: [[TMP3:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
@@ -389,11 +389,11 @@ define void @addmul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z, i32
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP2]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = add <32 x i16> [[TMP4]], [[WIDE_VEC36]]
@@ -545,12 +545,12 @@ define void @addsubsmul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z,
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = add <32 x i16> [[TMP2]], [[WIDE_VEC36]]
; CHECK-NEXT: [[TMP5:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
@@ -710,12 +710,12 @@ define void @add2sub2mul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z,
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = add <32 x i16> [[TMP2]], [[WIDE_VEC36]]
; CHECK-NEXT: [[TMP5:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
index ff9ad9ffa14080..e22e4a4b31ebe6 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
@@ -123,7 +123,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP15]])
; CHECK-NEXT: [[TMP16:%.*]] = extractelement <2 x i1> [[TMP12]], i64 1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP16]])
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds <225 x double>, ptr [[A]], i64 0, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[A]], i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP17]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP17]], align 8, !alias.scope [[META0:![0-9]+]]
; CHECK-NEXT: [[WIDE_LOAD21:%.*]] = load <2 x double>, ptr [[TMP18]], align 8, !alias.scope [[META0]]
@@ -132,7 +132,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP20:%.*]] = fmul <2 x double> [[WIDE_LOAD]], [[BROADCAST_SPLAT23]]
; CHECK-NEXT: [[TMP21:%.*]] = fmul <2 x double> [[WIDE_LOAD21]], [[BROADCAST_SPLAT23]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds <225 x double>, ptr [[B]], i64 0, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[B]], i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP22]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24:%.*]] = load <2 x double>, ptr [[TMP22]], align 8, !alias.scope [[META5:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25:%.*]] = load <2 x double>, ptr [[TMP23]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -199,7 +199,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP45]])
; CHECK-NEXT: [[TMP46:%.*]] = extractelement <2 x i1> [[TMP42]], i64 1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP46]])
-; CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds <225 x double>, ptr [[A]], i64 0, i64 [[TMP33]]
+; CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[A]], i64 0, i64 [[TMP33]]
; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP47]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_1:%.*]] = load <2 x double>, ptr [[TMP47]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_1:%.*]] = load <2 x double>, ptr [[TMP48]], align 8, !alias.scope [[META0]]
@@ -208,7 +208,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_1:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP50:%.*]] = fmul <2 x double> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT23_1]]
; CHECK-NEXT: [[TMP51:%.*]] = fmul <2 x double> [[WIDE_LOAD21_1]], [[BROADCAST_SPLAT23_1]]
-; CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds <225 x double>, ptr [[B]], i64 0, i64 [[TMP33]]
+; CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[B]], i64 0, i64 [[TMP33]]
; CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP52]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_1:%.*]] = load <2 x double>, ptr [[TMP52]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_1:%.*]] = load <2 x double>, ptr [[TMP53]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -276,7 +276,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP76]])
; CHECK-NEXT: [[TMP77:%.*]] = extractelement <2 x i1> [[TMP73]], i64 1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP77]])
-; CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds <225 x double>, ptr [[A]], i64 0, i64 [[TMP64]]
+; CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[A]], i64 0, i64 [[TMP64]]
; CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP78]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_2:%.*]] = load <2 x double>, ptr [[TMP78]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_2:%.*]] = load <2 x double>, ptr [[TMP79]], align 8, !alias.scope [[META0]]
@@ -285,7 +285,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_2:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP81:%.*]] = fmul <2 x double> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT23_2]]
; CHECK-NEXT: [[TMP82:%.*]] = fmul <2 x double> [[WIDE_LOAD21_2]], [[BROADCAST_SPLAT23_2]]
-; CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds <225 x double>, ptr [[B]], i64 0, i64 [[TMP64]]
+; CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[B]], i64 0, i64 [[TMP64]]
; CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP83]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_2:%.*]] = load <2 x double>, ptr [[TMP83]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_2:%.*]] = load <2 x double>, ptr [[TMP84]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -353,7 +353,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP107]])
; CHECK-NEXT: [[TMP108:%.*]] = extractelement <2 x i1> [[TMP104]], i64 1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP108]])
-; CHECK-NEXT: [[TMP109:%.*]] = getelementptr inbounds <225 x double>, ptr [[A]], i64 0, i64 [[TMP95]]
+; CHECK-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[A]], i64 0, i64 [[TMP95]]
; CHECK-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP109]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_3:%.*]] = load <2 x double>, ptr [[TMP109]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_3:%.*]] = load <2 x double>, ptr [[TMP110]], align 8, !alias.scope [[META0]]
@@ -362,7 +362,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_3:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_3]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP112:%.*]] = fmul <2 x double> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT23_3]]
; CHECK-NEXT: [[TMP113:%.*]] = fmul <2 x double> [[WIDE_LOAD21_3]], [[BROADCAST_SPLAT23_3]]
-; CHECK-NEXT: [[TMP114:%.*]] = getelementptr inbounds <225 x double>, ptr [[B]], i64 0, i64 [[TMP95]]
+; CHECK-NEXT: [[TMP114:%.*]] = getelementptr inbounds nuw <225 x double>, ptr [[B]], i64 0, i64 [[TMP95]]
; CHECK-NEXT: [[TMP115:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP114]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_3:%.*]] = load <2 x double>, ptr [[TMP114]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_3:%.*]] = load <2 x double>, ptr [[TMP115]], align 8, !alias.scope [[META5]], !noalias [[META0]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
index 68b64a5dad4fae..260567723aad09 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
@@ -27,7 +27,7 @@ define nofpclass(nan inf) double @monte_simple(i32 noundef %nblocks, i32 noundef
; CHECK-NEXT: [[VEC_PHI15:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP19:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI16:%.*]] = phi <4 x double> [ <double 0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %[[VECTOR_PH]] ], [ [[TMP14:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI17:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP15:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX1]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[WIDE_LOAD18:%.*]] = load <4 x float>, ptr [[TMP23]], align 4
@@ -213,7 +213,7 @@ define nofpclass(nan inf) double @monte_exp(i32 noundef %nblocks, i32 noundef %R
; CHECK-NEXT: [[VEC_PHI31:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP23:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI32:%.*]] = phi <4 x double> [ [[TMP27]], %[[VECTOR_PH]] ], [ [[TMP18:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI33:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP19:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX_US1:%.*]] = getelementptr inbounds float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
+; CHECK-NEXT: [[ARRAYIDX_US1:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX_US1]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[ARRAYIDX_US1]], align 4
; CHECK-NEXT: [[WIDE_LOAD34:%.*]] = load <4 x float>, ptr [[TMP3]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
index 8994afcdc629d0..9177f279217fa7 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
@@ -32,7 +32,7 @@ define void @loop_or(ptr noalias %pIn, ptr noalias %pOut, i32 %s) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[PIN:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[PIN:%.*]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 4
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP0]], align 1
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i8>, ptr [[TMP1]], align 1
@@ -42,7 +42,7 @@ define void @loop_or(ptr noalias %pIn, ptr noalias %pOut, i32 %s) {
; CHECK-NEXT: [[TMP5:%.*]] = mul nuw nsw <4 x i32> [[TMP3]], splat (i32 65793)
; CHECK-NEXT: [[TMP6:%.*]] = or disjoint <4 x i32> [[TMP4]], splat (i32 -16777216)
; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <4 x i32> [[TMP5]], splat (i32 -16777216)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[POUT:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[POUT:%.*]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP8]], i64 16
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: store <4 x i32> [[TMP7]], ptr [[TMP9]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
index fb7a4a3df74ef2..b3625094f07ea1 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
@@ -17,7 +17,7 @@ define void @foo(ptr noalias noundef %0, ptr noalias noundef %1) optsize {
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = load <8 x i32>, ptr [[GEP]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = add nsw <8 x i32> [[WIDE_MASKED_GATHER]], splat (i32 5)
; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <8 x i32> [[TMP5]], <8 x i32> poison, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 [[INDVARS_IV]]
; CHECK-NEXT: store <8 x i32> [[TMP6]], ptr [[TMP10]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw i64 [[INDVARS_IV]], 8
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 256
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll b/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
index 1e062041b12869..fe5bba1ae97766 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
@@ -22,10 +22,10 @@ define void @test(i32 noundef %nface, i32 noundef %ncell, ptr noalias noundef %f
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV_EPIL:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i32, ptr [[FACE_CELL]], i64 [[INDVARS_IV_EPIL]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_EPIL]], align 4, !tbaa [[TBAA0:![0-9]+]], !llvm.access.group [[ACC_GRP4:![0-9]+]]
-; CHECK-NEXT: [[GEP_EPIL:%.*]] = getelementptr inbounds i32, ptr [[INVARIANT_GEP]], i64 [[INDVARS_IV_EPIL]]
-; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[GEP_EPIL]], align 4, !tbaa [[TBAA0]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i32, ptr [[FACE_CELL]], i64 [[INDVARS_IV_EPIL]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4, !tbaa [[TBAA0:![0-9]+]], !llvm.access.group [[ACC_GRP4:![0-9]+]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[INVARIANT_GEP]], i64 [[INDVARS_IV_EPIL]]
+; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[TBAA0]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[TMP3:%.*]] = sext <4 x i32> [[WIDE_LOAD]] to <4 x i64>
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[Y]], <4 x i64> [[TMP3]]
; CHECK-NEXT: [[TMP5:%.*]] = sext <4 x i32> [[WIDE_LOAD12]] to <4 x i64>
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll b/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
index 993e949d6d2a0c..f60bc26c5111c6 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
@@ -23,7 +23,7 @@ define void @vdiv(ptr %a, float %b) #0 {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4, !tbaa [[TBAA3:![0-9]+]]
; CHECK-NEXT: [[TMP3:%.*]] = fmul fast <4 x float> [[WIDE_LOAD]], [[TMP0]]
; CHECK-NEXT: store <4 x float> [[TMP3]], ptr [[TMP1]], align 4, !tbaa [[TBAA3]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
index 7349f603822cd7..7817c23e6a3ec9 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
@@ -39,7 +39,7 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 32
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 64
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 96
@@ -51,7 +51,7 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: [[TMP10:%.*]] = fmul fast <4 x double> [[WIDE_LOAD6]], [[TMP2]]
; CHECK-NEXT: [[TMP11:%.*]] = fmul fast <4 x double> [[WIDE_LOAD7]], [[TMP3]]
; CHECK-NEXT: [[TMP12:%.*]] = fmul fast <4 x double> [[WIDE_LOAD8]], [[TMP4]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 32
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 64
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 96
@@ -78,10 +78,10 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT16:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX12]]
+; CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX12]]
; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <4 x double>, ptr [[TMP39]], align 8, !tbaa [[TBAA3]]
; CHECK-NEXT: [[TMP40:%.*]] = fmul fast <4 x double> [[WIDE_LOAD13]], [[TMP38]]
-; CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX12]]
+; CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDEX12]]
; CHECK-NEXT: store <4 x double> [[TMP40]], ptr [[TMP41]], align 8, !tbaa [[TBAA3]]
; CHECK-NEXT: [[INDEX_NEXT16]] = add nuw i64 [[INDEX12]], 4
; CHECK-NEXT: [[TMP42:%.*]] = icmp eq i64 [[INDEX_NEXT16]], [[N_VEC11]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vector-reduction-known-first-value.ll b/llvm/test/Transforms/PhaseOrdering/X86/vector-reduction-known-first-value.ll
index e2c85bb5153698..3bd911b8ac48fa 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vector-reduction-known-first-value.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vector-reduction-known-first-value.ll
@@ -16,7 +16,7 @@ define i16 @test(ptr %ptr) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <8 x i16> [ zeroinitializer, [[ENTRY]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <8 x i16> [ zeroinitializer, [[ENTRY]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 8
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP0]], align 1
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
>From 41d53e2fcfe3b538d5b9a1bfc6495b9bb8bc2980 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Tue, 17 Dec 2024 12:08:20 +0100
Subject: [PATCH 2/2] Add gep nuw/nusw test variants to
drop-poison-generating-flags
---
.../X86/drop-poison-generating-flags.ll | 80 +++++++++++++++++++
1 file changed, 80 insertions(+)
diff --git a/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll b/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
index f6f77e274273e3..eaea69735fcbe9 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
@@ -57,6 +57,86 @@ loop.exit:
ret void
}
+; Variant with getelementptr nusw.
+define void @drop_scalar_gep_nusw(ptr noalias nocapture readonly %input,
+ ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: @drop_scalar_gep_nusw(
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
+; CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[TMP0]], 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+entry:
+ br label %loop.header
+
+loop.header:
+ %iv = phi i64 [ 0, %entry ], [ %iv.inc, %if.end ]
+ %i23 = icmp eq i64 %iv, 0
+ br i1 %i23, label %if.end, label %if.then
+
+if.then:
+ %i27 = sub nuw nsw i64 %iv, 1
+ %i29 = getelementptr nusw float, ptr %input, i64 %i27
+ %i30 = load float, ptr %i29, align 4, !invariant.load !0
+ br label %if.end
+
+if.end:
+ %i34 = phi float [ 0.000000e+00, %loop.header ], [ %i30, %if.then ]
+ %i35 = getelementptr nusw float, ptr %output, i64 %iv
+ store float %i34, ptr %i35, align 4
+ %iv.inc = add nuw nsw i64 %iv, 1
+ %exitcond = icmp eq i64 %iv.inc, 4
+ br i1 %exitcond, label %loop.exit, label %loop.header
+
+loop.exit:
+ ret void
+}
+
+; Variant with getelementptr nuw.
+define void @drop_scalar_gep_nuw(ptr noalias nocapture readonly %input,
+ ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: @drop_scalar_gep_nuw(
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
+; CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[TMP0]], 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+entry:
+ br label %loop.header
+
+loop.header:
+ %iv = phi i64 [ 0, %entry ], [ %iv.inc, %if.end ]
+ %i23 = icmp eq i64 %iv, 0
+ br i1 %i23, label %if.end, label %if.then
+
+if.then:
+ %i27 = sub nuw nsw i64 %iv, 1
+ %i29 = getelementptr nuw float, ptr %input, i64 %i27
+ %i30 = load float, ptr %i29, align 4, !invariant.load !0
+ br label %if.end
+
+if.end:
+ %i34 = phi float [ 0.000000e+00, %loop.header ], [ %i30, %if.then ]
+ %i35 = getelementptr nuw float, ptr %output, i64 %iv
+ store float %i34, ptr %i35, align 4
+ %iv.inc = add nuw nsw i64 %iv, 1
+ %exitcond = icmp eq i64 %iv.inc, 4
+ br i1 %exitcond, label %loop.exit, label %loop.header
+
+loop.exit:
+ ret void
+}
+
; Drop poison-generating flags from 'sub' and 'getelementptr' feeding a masked load.
; In this case, 'sub' and 'getelementptr' are not guarded by the predicate.
define void @drop_nonpred_scalar_nuw_nsw(ptr noalias nocapture readonly %input,
More information about the llvm-commits
mailing list