[llvm] [VPlan] Reassociate header masks and simplify (x && y) || (x && z) -> x && (y || z) (PR #155383)
Luke Lau via llvm-commits
llvm-commits at lists.llvm.org
Mon Sep 1 07:06:16 PDT 2025
https://github.com/lukel97 updated https://github.com/llvm/llvm-project/pull/155383
>From f9f7969f5a8cb5c59536da1a4d90a1b5fec02970 Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Tue, 26 Aug 2025 16:31:06 +0800
Subject: [PATCH] Reassociate header mask
---
.../Transforms/Vectorize/VPlanTransforms.cpp | 105 ++++++++++--------
.../RISCV/blocks-with-dead-instructions.ll | 12 +-
...ruction-or-drop-poison-generating-flags.ll | 11 +-
.../LoopVectorize/X86/constant-fold.ll | 40 +++----
4 files changed, 87 insertions(+), 81 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index a942d52cbca94..4ad054a438b70 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -996,7 +996,8 @@ static Value *tryToFoldLiveIns(const VPRecipeBase &R, unsigned Opcode,
}
/// Try to simplify recipe \p R.
-static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
+static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo,
+ VPValue *HeaderMask) {
VPlan *Plan = R.getParent()->getPlan();
auto *Def = dyn_cast<VPSingleDefRecipe>(&R);
@@ -1119,6 +1120,14 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
return;
}
+ // Reassociate the header mask so it has more opportunities to be simplified.
+ // (headermask && x) && y -> headermask && (x && y)
+ if (HeaderMask && match(Def, m_LogicalAnd(m_LogicalAnd(m_Specific(HeaderMask),
+ m_VPValue(X)),
+ m_VPValue(Y))))
+ return Def->replaceAllUsesWith(
+ Builder.createLogicalAnd(HeaderMask, Builder.createLogicalAnd(X, Y)));
+
if (match(Def, m_c_Mul(m_VPValue(A), m_SpecificInt(1))))
return Def->replaceAllUsesWith(A);
@@ -1263,13 +1272,61 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
}
}
+/// Collect the header mask with the pattern:
+/// (ICMP_ULE, WideCanonicalIV, backedge-taken-count)
+/// TODO: Introduce explicit recipe for header-mask instead of searching
+/// for the header-mask pattern manually.
+static VPSingleDefRecipe *findHeaderMask(VPlan &Plan) {
+ SmallVector<VPValue *> WideCanonicalIVs;
+ auto *FoundWidenCanonicalIVUser = find_if(Plan.getCanonicalIV()->users(),
+ IsaPred<VPWidenCanonicalIVRecipe>);
+ assert(count_if(Plan.getCanonicalIV()->users(),
+ IsaPred<VPWidenCanonicalIVRecipe>) <= 1 &&
+ "Must have at most one VPWideCanonicalIVRecipe");
+ if (FoundWidenCanonicalIVUser != Plan.getCanonicalIV()->users().end()) {
+ auto *WideCanonicalIV =
+ cast<VPWidenCanonicalIVRecipe>(*FoundWidenCanonicalIVUser);
+ WideCanonicalIVs.push_back(WideCanonicalIV);
+ }
+
+ // Also include VPWidenIntOrFpInductionRecipes that represent a widened
+ // version of the canonical induction.
+ VPBasicBlock *HeaderVPBB = Plan.getVectorLoopRegion()->getEntryBasicBlock();
+ for (VPRecipeBase &Phi : HeaderVPBB->phis()) {
+ auto *WidenOriginalIV = dyn_cast<VPWidenIntOrFpInductionRecipe>(&Phi);
+ if (WidenOriginalIV && WidenOriginalIV->isCanonical())
+ WideCanonicalIVs.push_back(WidenOriginalIV);
+ }
+
+ // Walk users of wide canonical IVs and find the single compare of the form
+ // (ICMP_ULE, WideCanonicalIV, backedge-taken-count).
+ VPSingleDefRecipe *HeaderMask = nullptr;
+ for (auto *Wide : WideCanonicalIVs) {
+ for (VPUser *U : SmallVector<VPUser *>(Wide->users())) {
+ auto *VPI = dyn_cast<VPInstruction>(U);
+ if (!VPI || !vputils::isHeaderMask(VPI, Plan))
+ continue;
+
+ assert(VPI->getOperand(0) == Wide &&
+ "WidenCanonicalIV must be the first operand of the compare");
+ assert(!HeaderMask && "Multiple header masks found?");
+ HeaderMask = VPI;
+ }
+ }
+ return HeaderMask;
+}
+
void VPlanTransforms::simplifyRecipes(VPlan &Plan) {
+ VPValue *HeaderMask = nullptr;
+ // Ignore post-unrolling as there can be multiple header masks.
+ if (!Plan.isUnrolled())
+ HeaderMask = findHeaderMask(Plan);
ReversePostOrderTraversal<VPBlockDeepTraversalWrapper<VPBlockBase *>> RPOT(
Plan.getEntry());
VPTypeAnalysis TypeInfo(Plan);
for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly<VPBasicBlock>(RPOT)) {
for (VPRecipeBase &R : make_early_inc_range(*VPBB)) {
- simplifyRecipe(R, TypeInfo);
+ simplifyRecipe(R, TypeInfo, HeaderMask);
}
}
}
@@ -2192,50 +2249,6 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch(
return LaneMaskPhi;
}
-/// Collect the header mask with the pattern:
-/// (ICMP_ULE, WideCanonicalIV, backedge-taken-count)
-/// TODO: Introduce explicit recipe for header-mask instead of searching
-/// for the header-mask pattern manually.
-static VPSingleDefRecipe *findHeaderMask(VPlan &Plan) {
- SmallVector<VPValue *> WideCanonicalIVs;
- auto *FoundWidenCanonicalIVUser = find_if(Plan.getCanonicalIV()->users(),
- IsaPred<VPWidenCanonicalIVRecipe>);
- assert(count_if(Plan.getCanonicalIV()->users(),
- IsaPred<VPWidenCanonicalIVRecipe>) <= 1 &&
- "Must have at most one VPWideCanonicalIVRecipe");
- if (FoundWidenCanonicalIVUser != Plan.getCanonicalIV()->users().end()) {
- auto *WideCanonicalIV =
- cast<VPWidenCanonicalIVRecipe>(*FoundWidenCanonicalIVUser);
- WideCanonicalIVs.push_back(WideCanonicalIV);
- }
-
- // Also include VPWidenIntOrFpInductionRecipes that represent a widened
- // version of the canonical induction.
- VPBasicBlock *HeaderVPBB = Plan.getVectorLoopRegion()->getEntryBasicBlock();
- for (VPRecipeBase &Phi : HeaderVPBB->phis()) {
- auto *WidenOriginalIV = dyn_cast<VPWidenIntOrFpInductionRecipe>(&Phi);
- if (WidenOriginalIV && WidenOriginalIV->isCanonical())
- WideCanonicalIVs.push_back(WidenOriginalIV);
- }
-
- // Walk users of wide canonical IVs and find the single compare of the form
- // (ICMP_ULE, WideCanonicalIV, backedge-taken-count).
- VPSingleDefRecipe *HeaderMask = nullptr;
- for (auto *Wide : WideCanonicalIVs) {
- for (VPUser *U : SmallVector<VPUser *>(Wide->users())) {
- auto *VPI = dyn_cast<VPInstruction>(U);
- if (!VPI || !vputils::isHeaderMask(VPI, Plan))
- continue;
-
- assert(VPI->getOperand(0) == Wide &&
- "WidenCanonicalIV must be the first operand of the compare");
- assert(!HeaderMask && "Multiple header masks found?");
- HeaderMask = VPI;
- }
- }
- return HeaderMask;
-}
-
void VPlanTransforms::addActiveLaneMask(
VPlan &Plan, bool UseActiveLaneMaskForControlFlow,
bool DataAndControlFlowWithoutRuntimeCheck) {
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/blocks-with-dead-instructions.ll b/llvm/test/Transforms/LoopVectorize/RISCV/blocks-with-dead-instructions.ll
index 631328a9a0964..c06b06ed4aee5 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/blocks-with-dead-instructions.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/blocks-with-dead-instructions.ll
@@ -436,23 +436,17 @@ define void @multiple_blocks_with_dead_inst_multiple_successors_6(ptr %src, i1 %
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 8 x i64> [ [[INDUCTION]], %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[TMP2]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP27:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 8, i1 true)
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT3:%.*]] = insertelement <vscale x 8 x i32> poison, i32 [[TMP27]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT4:%.*]] = shufflevector <vscale x 8 x i32> [[BROADCAST_SPLATINSERT3]], <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
; CHECK-NEXT: [[TMP12:%.*]] = zext i32 [[TMP27]] to i64
; CHECK-NEXT: [[TMP16:%.*]] = mul i64 3, [[TMP12]]
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i64> poison, i64 [[TMP16]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i64> [[DOTSPLATINSERT]], <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP14:%.*]] = call <vscale x 8 x i32> @llvm.stepvector.nxv8i32()
-; CHECK-NEXT: [[TMP15:%.*]] = icmp ult <vscale x 8 x i32> [[TMP14]], [[BROADCAST_SPLAT4]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i16, ptr [[SRC]], <vscale x 8 x i64> [[VEC_IND]]
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 8 x i16> @llvm.vp.gather.nxv8i16.nxv8p0(<vscale x 8 x ptr> align 2 [[TMP20]], <vscale x 8 x i1> splat (i1 true), i32 [[TMP27]])
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq <vscale x 8 x i16> [[WIDE_MASKED_GATHER]], zeroinitializer
-; CHECK-NEXT: [[TMP18:%.*]] = select <vscale x 8 x i1> [[TMP15]], <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP19:%.*]] = select <vscale x 8 x i1> [[TMP18]], <vscale x 8 x i1> [[TMP8]], <vscale x 8 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP29:%.*]] = select <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i1> [[TMP8]], <vscale x 8 x i1> zeroinitializer
; CHECK-NEXT: [[TMP28:%.*]] = xor <vscale x 8 x i1> [[TMP17]], splat (i1 true)
-; CHECK-NEXT: [[TMP21:%.*]] = select <vscale x 8 x i1> [[TMP15]], <vscale x 8 x i1> [[TMP28]], <vscale x 8 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP22:%.*]] = or <vscale x 8 x i1> [[TMP19]], [[TMP21]]
-; CHECK-NEXT: [[TMP23:%.*]] = select <vscale x 8 x i1> [[TMP18]], <vscale x 8 x i1> [[BROADCAST_SPLAT]], <vscale x 8 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP22:%.*]] = or <vscale x 8 x i1> [[TMP29]], [[TMP28]]
+; CHECK-NEXT: [[TMP23:%.*]] = select <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i1> [[BROADCAST_SPLAT]], <vscale x 8 x i1> zeroinitializer
; CHECK-NEXT: [[TMP24:%.*]] = or <vscale x 8 x i1> [[TMP22]], [[TMP23]]
; CHECK-NEXT: call void @llvm.vp.scatter.nxv8i16.nxv8p0(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x ptr> align 2 [[TMP20]], <vscale x 8 x i1> [[TMP24]], i32 [[TMP27]])
; CHECK-NEXT: [[TMP25:%.*]] = zext i32 [[TMP27]] to i64
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/pr87378-vpinstruction-or-drop-poison-generating-flags.ll b/llvm/test/Transforms/LoopVectorize/RISCV/pr87378-vpinstruction-or-drop-poison-generating-flags.ll
index db6185087bac5..a4b90c658cd6e 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/pr87378-vpinstruction-or-drop-poison-generating-flags.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/pr87378-vpinstruction-or-drop-poison-generating-flags.ll
@@ -34,17 +34,16 @@ define void @pr87378_vpinstruction_or_drop_poison_generating_flags(ptr %arg, i64
; CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 8 x i32> @llvm.stepvector.nxv8i32()
; CHECK-NEXT: [[TMP11:%.*]] = icmp ult <vscale x 8 x i32> [[TMP10]], [[BROADCAST_SPLAT8]]
; CHECK-NEXT: [[TMP13:%.*]] = icmp ule <vscale x 8 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP28:%.*]] = select <vscale x 8 x i1> [[TMP11]], <vscale x 8 x i1> [[TMP13]], <vscale x 8 x i1> zeroinitializer
; CHECK-NEXT: [[TMP14:%.*]] = icmp ule <vscale x 8 x i64> [[VEC_IND]], [[BROADCAST_SPLAT2]]
-; CHECK-NEXT: [[TMP15:%.*]] = select <vscale x 8 x i1> [[TMP28]], <vscale x 8 x i1> [[TMP14]], <vscale x 8 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP9:%.*]] = select <vscale x 8 x i1> [[TMP13]], <vscale x 8 x i1> [[TMP14]], <vscale x 8 x i1> zeroinitializer
; CHECK-NEXT: [[TMP16:%.*]] = xor <vscale x 8 x i1> [[TMP13]], splat (i1 true)
-; CHECK-NEXT: [[TMP29:%.*]] = select <vscale x 8 x i1> [[TMP11]], <vscale x 8 x i1> [[TMP16]], <vscale x 8 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP17:%.*]] = or <vscale x 8 x i1> [[TMP15]], [[TMP29]]
+; CHECK-NEXT: [[TMP17:%.*]] = or <vscale x 8 x i1> [[TMP9]], [[TMP16]]
; CHECK-NEXT: [[TMP18:%.*]] = icmp ule <vscale x 8 x i64> [[VEC_IND]], [[BROADCAST_SPLAT4]]
; CHECK-NEXT: [[TMP19:%.*]] = select <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i1> [[TMP18]], <vscale x 8 x i1> zeroinitializer
; CHECK-NEXT: [[TMP20:%.*]] = xor <vscale x 8 x i1> [[TMP14]], splat (i1 true)
-; CHECK-NEXT: [[TMP21:%.*]] = select <vscale x 8 x i1> [[TMP28]], <vscale x 8 x i1> [[TMP20]], <vscale x 8 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP22:%.*]] = or <vscale x 8 x i1> [[TMP19]], [[TMP21]]
+; CHECK-NEXT: [[TMP28:%.*]] = select <vscale x 8 x i1> [[TMP13]], <vscale x 8 x i1> [[TMP20]], <vscale x 8 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP21:%.*]] = select <vscale x 8 x i1> [[TMP11]], <vscale x 8 x i1> [[TMP28]], <vscale x 8 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP22:%.*]] = or <vscale x 8 x i1> [[TMP19]], [[TMP28]]
; CHECK-NEXT: [[TMP23:%.*]] = extractelement <vscale x 8 x i1> [[TMP21]], i32 0
; CHECK-NEXT: [[PREDPHI:%.*]] = select i1 [[TMP23]], i64 poison, i64 [[INDEX]]
; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i16, ptr [[ARG]], i64 [[PREDPHI]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll b/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll
index db54ca61f715b..54d738388ea73 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll
@@ -65,35 +65,35 @@ define void @redundant_or_1(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK: vector.ph:
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C_0:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT]], splat (i1 true)
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i1> poison, i1 [[C_1:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT2]], splat (i1 true)
+; CHECK-NEXT: [[TMP0:%.*]] = select <4 x i1> [[TMP1]], <4 x i1> [[BROADCAST_SPLAT]], <4 x i1> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[TMP2:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = select <4 x i1> [[TMP2]], <4 x i1> [[BROADCAST_SPLAT2]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP5]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP2]], i32 0
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i32 0
; CHECK-NEXT: store i32 0, ptr [[TMP8]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE]]
; CHECK: pred.store.continue:
-; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP5]], i32 1
+; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP2]], i32 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; CHECK: pred.store.if3:
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 1
; CHECK-NEXT: store i32 0, ptr [[TMP11]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE4]]
; CHECK: pred.store.continue4:
-; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP5]], i32 2
+; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP2]], i32 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]]
; CHECK: pred.store.if5:
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 2
; CHECK-NEXT: store i32 0, ptr [[TMP14]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE6]]
; CHECK: pred.store.continue6:
-; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP5]], i32 3
+; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP2]], i32 3
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_STORE_IF7:%.*]], label [[PRED_STORE_CONTINUE8:%.*]]
; CHECK: pred.store.if7:
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 3
@@ -107,11 +107,11 @@ define void @redundant_or_1(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 0, [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
-; CHECK-NEXT: br i1 [[C_0]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
+; CHECK-NEXT: br i1 [[C_1]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
; CHECK: then.1:
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[IV]], 2
; CHECK-NEXT: [[OR:%.*]] = or i1 [[CMP]], true
-; CHECK-NEXT: [[COND:%.*]] = select i1 [[OR]], i1 [[C_1]], i1 false
+; CHECK-NEXT: [[COND:%.*]] = select i1 [[OR]], i1 [[C_0]], i1 false
; CHECK-NEXT: br i1 [[COND]], label [[THEN_2:%.*]], label [[LOOP_LATCH]]
; CHECK: then.2:
; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]]
@@ -158,35 +158,35 @@ define void @redundant_or_2(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK: vector.ph:
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C_1:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT]], splat (i1 true)
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i1> poison, i1 [[C_0:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT2]], splat (i1 true)
+; CHECK-NEXT: [[TMP0:%.*]] = select <4 x i1> [[TMP1]], <4 x i1> [[BROADCAST_SPLAT]], <4 x i1> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP3]], <4 x i1> [[BROADCAST_SPLAT2]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i32 0
+; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP3]], i32 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i32 0
; CHECK-NEXT: store i32 0, ptr [[TMP7]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE]]
; CHECK: pred.store.continue:
-; CHECK-NEXT: [[TMP8:%.*]] = extractelement <4 x i1> [[TMP4]], i32 1
+; CHECK-NEXT: [[TMP8:%.*]] = extractelement <4 x i1> [[TMP3]], i32 1
; CHECK-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; CHECK: pred.store.if3:
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 1
; CHECK-NEXT: store i32 0, ptr [[TMP10]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE4]]
; CHECK: pred.store.continue4:
-; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x i1> [[TMP4]], i32 2
+; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x i1> [[TMP3]], i32 2
; CHECK-NEXT: br i1 [[TMP11]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]]
; CHECK: pred.store.if5:
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 2
; CHECK-NEXT: store i32 0, ptr [[TMP13]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE6]]
; CHECK: pred.store.continue6:
-; CHECK-NEXT: [[TMP14:%.*]] = extractelement <4 x i1> [[TMP4]], i32 3
+; CHECK-NEXT: [[TMP14:%.*]] = extractelement <4 x i1> [[TMP3]], i32 3
; CHECK-NEXT: br i1 [[TMP14]], label [[PRED_STORE_IF7:%.*]], label [[PRED_STORE_CONTINUE8:%.*]]
; CHECK: pred.store.if7:
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 3
@@ -200,11 +200,11 @@ define void @redundant_or_2(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 0, [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
-; CHECK-NEXT: br i1 [[C_1]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
+; CHECK-NEXT: br i1 [[C_0]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
; CHECK: then.1:
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[IV]], 2
; CHECK-NEXT: [[OR:%.*]] = or i1 true, [[CMP]]
-; CHECK-NEXT: [[COND:%.*]] = select i1 [[OR]], i1 [[C_0]], i1 false
+; CHECK-NEXT: [[COND:%.*]] = select i1 [[OR]], i1 [[C_1]], i1 false
; CHECK-NEXT: br i1 [[COND]], label [[THEN_2:%.*]], label [[LOOP_LATCH]]
; CHECK: then.2:
; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]]
@@ -256,9 +256,9 @@ define void @redundant_and_1(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[TMP2:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = select <4 x i1> <i1 false, i1 false, i1 true, i1 false>, <4 x i1> [[BROADCAST_SPLAT2]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = select <4 x i1> [[TMP2]], <4 x i1> [[TMP5]], <4 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP2:%.*]] = select <4 x i1> [[TMP0]], <4 x i1> [[TMP5]], <4 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP6:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP2]], <4 x i1> zeroinitializer
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP6]], i32 0
; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
@@ -346,10 +346,10 @@ define void @redundant_and_2(ptr %dst, i1 %c.0, i1 %c.1) {
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i1> poison, i1 [[C_0:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT2]], splat (i1 true)
+; CHECK-NEXT: [[TMP1:%.*]] = select <4 x i1> [[TMP0]], <4 x i1> zeroinitializer, <4 x i1> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP3]], <4 x i1> zeroinitializer, <4 x i1> zeroinitializer
+; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> <i1 true, i1 true, i1 true, i1 false>, <4 x i1> [[TMP1]], <4 x i1> zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i32 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
More information about the llvm-commits
mailing list