From 4c3bfdc7f100bc04e61e226e2886c3fc8899c3bc Mon Sep 17 00:00:00 2001 From: David Green Date: Mon, 15 Nov 2021 09:44:41 +0000 Subject: [PATCH] [ARM] Fix GatherScatter AddLikeOr condition --- .../Target/ARM/MVEGatherScatterLowering.cpp | 2 +- .../Thumb2/mve-gather-optimisation-deep.ll | 209 ++++++++++++++++-- 2 files changed, 188 insertions(+), 23 deletions(-) diff --git a/llvm/lib/Target/ARM/MVEGatherScatterLowering.cpp b/llvm/lib/Target/ARM/MVEGatherScatterLowering.cpp index 3c57f8ebe68c..cfd275bc0621 100644 --- a/llvm/lib/Target/ARM/MVEGatherScatterLowering.cpp +++ b/llvm/lib/Target/ARM/MVEGatherScatterLowering.cpp @@ -376,7 +376,7 @@ MVEGatherScatterLowering::getVarAndConst(Value *Inst, int TypeScale) { // add-like-or. Instruction *Add = dyn_cast(Inst); if (Add == nullptr || - (Add->getOpcode() != Instruction::Add && isAddLikeOr(Add, *DL))) + (Add->getOpcode() != Instruction::Add && !isAddLikeOr(Add, *DL))) return ReturnFalse; Value *Summand; diff --git a/llvm/test/CodeGen/Thumb2/mve-gather-optimisation-deep.ll b/llvm/test/CodeGen/Thumb2/mve-gather-optimisation-deep.ll index 1617ce36d4ca..bde97e404c30 100644 --- a/llvm/test/CodeGen/Thumb2/mve-gather-optimisation-deep.ll +++ b/llvm/test/CodeGen/Thumb2/mve-gather-optimisation-deep.ll @@ -1,27 +1,36 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; RUN: opt --arm-mve-gather-scatter-lowering -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp %s -S -o - | FileCheck %s +target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64" + define arm_aapcs_vfpcc void @push_out_add_sub_block(i32* noalias nocapture readonly %data, i32* noalias nocapture %dst, i32 %n.vec) { ; CHECK-LABEL: @push_out_add_sub_block( ; CHECK-NEXT: vector.ph: ; CHECK-NEXT: [[PUSHEDOUTADD:%.*]] = add <4 x i32> , +; CHECK-NEXT: [[SCALEDINDEX:%.*]] = shl <4 x i32> [[PUSHEDOUTADD]], +; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint i32* [[DATA:%.*]] to i32 +; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TMP0]], i32 0 +; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[STARTINDEX:%.*]] = add <4 x i32> [[SCALEDINDEX]], [[DOTSPLAT]] +; CHECK-NEXT: [[PREINCREMENTSTARTINDEX:%.*]] = sub <4 x i32> [[STARTINDEX]], ; CHECK-NEXT: br label [[VECTOR_BODY:%.*]] ; CHECK: vector.body: ; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY_END:%.*]] ] -; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[PUSHEDOUTADD]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY_END]] ] -; CHECK-NEXT: [[TMP0:%.*]] = icmp eq i32 [[INDEX]], 48 -; CHECK-NEXT: br i1 [[TMP0]], label [[LOWER_BLOCK:%.*]], label [[END:%.*]] +; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[PREINCREMENTSTARTINDEX]], [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY_END]] ] +; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[INDEX]], 48 +; CHECK-NEXT: br i1 [[TMP1]], label [[LOWER_BLOCK:%.*]], label [[END:%.*]] ; CHECK: lower.block: -; CHECK-NEXT: [[TMP1:%.*]] = call <4 x i32> @llvm.arm.mve.vldr.gather.offset.v4i32.p0i32.v4i32(i32* [[DATA:%.*]], <4 x i32> [[VEC_IND]], i32 32, i32 2, i32 1) -; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i32 [[INDEX]] -; CHECK-NEXT: [[TMP3:%.*]] = bitcast i32* [[TMP2]] to <4 x i32>* -; CHECK-NEXT: store <4 x i32> [[TMP1]], <4 x i32>* [[TMP3]], align 4 +; CHECK-NEXT: [[TMP2:%.*]] = call { <4 x i32>, <4 x i32> } @llvm.arm.mve.vldr.gather.base.wb.v4i32.v4i32(<4 x i32> [[VEC_IND]], i32 32) +; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <4 x i32>, <4 x i32> } [[TMP2]], 0 +; CHECK-NEXT: [[TMP4]] = extractvalue { <4 x i32>, <4 x i32> } [[TMP2]], 1 +; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i32 [[INDEX]] +; CHECK-NEXT: [[TMP6:%.*]] = bitcast i32* [[TMP5]] to <4 x i32>* +; CHECK-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* [[TMP6]], align 4 ; CHECK-NEXT: [[INDEX_NEXT]] = add i32 [[INDEX]], 4 -; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], ; CHECK-NEXT: br label [[VECTOR_BODY_END]] ; CHECK: vector.body.end: -; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC:%.*]] -; CHECK-NEXT: br i1 [[TMP4]], label [[END]], label [[VECTOR_BODY]] +; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC:%.*]] +; CHECK-NEXT: br i1 [[TMP7]], label [[END]], label [[VECTOR_BODY]] ; CHECK: end: ; CHECK-NEXT: ret void ; @@ -60,23 +69,30 @@ define arm_aapcs_vfpcc void @push_out_mul_sub_block(i32* noalias nocapture reado ; CHECK-NEXT: [[PUSHEDOUTMUL:%.*]] = mul <4 x i32> , ; CHECK-NEXT: [[PRODUCT:%.*]] = mul <4 x i32> , ; CHECK-NEXT: [[PUSHEDOUTADD:%.*]] = add <4 x i32> [[PUSHEDOUTMUL]], +; CHECK-NEXT: [[SCALEDINDEX:%.*]] = shl <4 x i32> [[PUSHEDOUTADD]], +; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint i32* [[DATA:%.*]] to i32 +; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TMP0]], i32 0 +; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[STARTINDEX:%.*]] = add <4 x i32> [[SCALEDINDEX]], [[DOTSPLAT]] +; CHECK-NEXT: [[PREINCREMENTSTARTINDEX:%.*]] = sub <4 x i32> [[STARTINDEX]], ; CHECK-NEXT: br label [[VECTOR_BODY:%.*]] ; CHECK: vector.body: ; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY_END:%.*]] ] -; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[PUSHEDOUTADD]], [[VECTOR_PH]] ], [ [[INCREMENTPUSHEDOUTMUL:%.*]], [[VECTOR_BODY_END]] ] -; CHECK-NEXT: [[TMP0:%.*]] = icmp eq i32 [[INDEX]], 48 -; CHECK-NEXT: br i1 [[TMP0]], label [[LOWER_BLOCK:%.*]], label [[END:%.*]] +; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[PREINCREMENTSTARTINDEX]], [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY_END]] ] +; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[INDEX]], 48 +; CHECK-NEXT: br i1 [[TMP1]], label [[LOWER_BLOCK:%.*]], label [[END:%.*]] ; CHECK: lower.block: -; CHECK-NEXT: [[TMP1:%.*]] = call <4 x i32> @llvm.arm.mve.vldr.gather.offset.v4i32.p0i32.v4i32(i32* [[DATA:%.*]], <4 x i32> [[VEC_IND]], i32 32, i32 2, i32 1) -; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i32 [[INDEX]] -; CHECK-NEXT: [[TMP3:%.*]] = bitcast i32* [[TMP2]] to <4 x i32>* -; CHECK-NEXT: store <4 x i32> [[TMP1]], <4 x i32>* [[TMP3]], align 4 +; CHECK-NEXT: [[TMP2:%.*]] = call { <4 x i32>, <4 x i32> } @llvm.arm.mve.vldr.gather.base.wb.v4i32.v4i32(<4 x i32> [[VEC_IND]], i32 96) +; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <4 x i32>, <4 x i32> } [[TMP2]], 0 +; CHECK-NEXT: [[TMP4]] = extractvalue { <4 x i32>, <4 x i32> } [[TMP2]], 1 +; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i32 [[INDEX]] +; CHECK-NEXT: [[TMP6:%.*]] = bitcast i32* [[TMP5]] to <4 x i32>* +; CHECK-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* [[TMP6]], align 4 ; CHECK-NEXT: [[INDEX_NEXT]] = add i32 [[INDEX]], 4 ; CHECK-NEXT: br label [[VECTOR_BODY_END]] ; CHECK: vector.body.end: -; CHECK-NEXT: [[INCREMENTPUSHEDOUTMUL]] = add <4 x i32> [[VEC_IND]], [[PRODUCT]] -; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC:%.*]] -; CHECK-NEXT: br i1 [[TMP4]], label [[END]], label [[VECTOR_BODY]] +; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC:%.*]] +; CHECK-NEXT: br i1 [[TMP7]], label [[END]], label [[VECTOR_BODY]] ; CHECK: end: ; CHECK-NEXT: ret void ; @@ -123,8 +139,12 @@ define arm_aapcs_vfpcc void @push_out_mul_sub_loop(i32* noalias nocapture readon ; CHECK-NEXT: br label [[VECTOR_2_BODY:%.*]] ; CHECK: vector.2.body: ; CHECK-NEXT: [[TMP0:%.*]] = mul <4 x i32> [[VEC_IND]], -; CHECK-NEXT: [[TMP1:%.*]] = add <4 x i32> [[TMP0]], -; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i32> @llvm.arm.mve.vldr.gather.offset.v4i32.p0i32.v4i32(i32* [[DATA:%.*]], <4 x i32> [[TMP1]], i32 32, i32 2, i32 1) +; CHECK-NEXT: [[SCALEDINDEX:%.*]] = shl <4 x i32> [[TMP0]], +; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint i32* [[DATA:%.*]] to i32 +; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TMP1]], i32 0 +; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[STARTINDEX:%.*]] = add <4 x i32> [[SCALEDINDEX]], [[DOTSPLAT]] +; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i32> @llvm.arm.mve.vldr.gather.base.v4i32.v4i32(<4 x i32> [[STARTINDEX]], i32 24) ; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i32 [[INDEX]] ; CHECK-NEXT: [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>* ; CHECK-NEXT: store <4 x i32> [[TMP2]], <4 x i32>* [[TMP4]], align 4 @@ -221,5 +241,150 @@ end: ret void; } +define void @gatherload(i32 %n, i32 %m, i32* nocapture %a, i32* nocapture readonly %b, i32 %call.us.us) { +; CHECK-LABEL: @gatherload( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[A57:%.*]] = bitcast i32* [[A:%.*]] to i8* +; CHECK-NEXT: [[CMP38:%.*]] = icmp sgt i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[CMP38]], label [[FOR_BODY_LR_PH:%.*]], label [[FOR_END16:%.*]] +; CHECK: for.body.lr.ph: +; CHECK-NEXT: [[CMP636:%.*]] = icmp sgt i32 [[M:%.*]], 0 +; CHECK-NEXT: br i1 [[CMP636]], label [[FOR_BODY_US_US_PREHEADER:%.*]], label [[FOR_BODY:%.*]] +; CHECK: for.body.us.us.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = shl nuw i32 [[M]], 2 +; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i32, i32* [[A]], i32 [[M]] +; CHECK-NEXT: [[SCEVGEP64:%.*]] = getelementptr i32, i32* [[B:%.*]], i32 [[M]] +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[M]], 4 +; CHECK-NEXT: [[BOUND0:%.*]] = icmp ugt i32* [[SCEVGEP64]], [[A]] +; CHECK-NEXT: [[BOUND1:%.*]] = icmp ugt i32* [[SCEVGEP]], [[B]] +; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] +; CHECK-NEXT: [[N_VEC:%.*]] = and i32 [[M]], -4 +; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[M]] +; CHECK-NEXT: br label [[FOR_BODY_US_US:%.*]] +; CHECK: for.body.us.us: +; CHECK-NEXT: [[I_039_US_US:%.*]] = phi i32 [ [[INC15_US_US:%.*]], [[FOR_COND5_FOR_END13_CRIT_EDGE_US_US:%.*]] ], [ 0, [[FOR_BODY_US_US_PREHEADER]] ] +; CHECK-NEXT: [[VLA_US_US:%.*]] = alloca i32, i32 [[CALL_US_US:%.*]], align 4 +; CHECK-NEXT: [[VLA_US_US56:%.*]] = bitcast i32* [[VLA_US_US]] to i8* +; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 4 [[VLA_US_US56]], i8* align 4 [[A57]], i32 [[TMP0]], i1 false) +; CHECK-NEXT: [[BRMERGE:%.*]] = select i1 [[MIN_ITERS_CHECK]], i1 true, i1 [[FOUND_CONFLICT]] +; CHECK-NEXT: br i1 [[BRMERGE]], label [[FOR_BODY7_US_US_PREHEADER:%.*]], label [[VECTOR_BODY:%.*]] +; CHECK: vector.body: +; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[FOR_BODY_US_US]] ] +; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, i32* [[B]], i32 [[INDEX]] +; CHECK-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP1]] to <4 x i32>* +; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, <4 x i32>* [[TMP2]], align 4 +; CHECK-NEXT: [[TMP3:%.*]] = call <4 x i32> @llvm.arm.mve.vldr.gather.offset.v4i32.p0i32.v4i32(i32* [[VLA_US_US]], <4 x i32> [[WIDE_LOAD]], i32 32, i32 2, i32 1) +; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, i32* [[A]], i32 [[INDEX]] +; CHECK-NEXT: [[TMP5:%.*]] = bitcast i32* [[TMP4]] to <4 x i32>* +; CHECK-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* [[TMP5]], align 4 +; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4 +; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] +; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]] +; CHECK: middle.block: +; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_COND5_FOR_END13_CRIT_EDGE_US_US]], label [[FOR_BODY7_US_US_PREHEADER]] +; CHECK: for.body7.us.us.preheader: +; CHECK-NEXT: [[J_137_US_US_PH:%.*]] = phi i32 [ 0, [[FOR_BODY_US_US]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ] +; CHECK-NEXT: br label [[FOR_BODY7_US_US:%.*]] +; CHECK: for.body7.us.us: +; CHECK-NEXT: [[J_137_US_US:%.*]] = phi i32 [ [[INC12_US_US:%.*]], [[FOR_BODY7_US_US]] ], [ [[J_137_US_US_PH]], [[FOR_BODY7_US_US_PREHEADER]] ] +; CHECK-NEXT: [[ARRAYIDX8_US_US:%.*]] = getelementptr inbounds i32, i32* [[B]], i32 [[J_137_US_US]] +; CHECK-NEXT: [[TMP7:%.*]] = load i32, i32* [[ARRAYIDX8_US_US]], align 4 +; CHECK-NEXT: [[ARRAYIDX9_US_US:%.*]] = getelementptr inbounds i32, i32* [[VLA_US_US]], i32 [[TMP7]] +; CHECK-NEXT: [[TMP8:%.*]] = load i32, i32* [[ARRAYIDX9_US_US]], align 4 +; CHECK-NEXT: [[ARRAYIDX10_US_US:%.*]] = getelementptr inbounds i32, i32* [[A]], i32 [[J_137_US_US]] +; CHECK-NEXT: store i32 [[TMP8]], i32* [[ARRAYIDX10_US_US]], align 4 +; CHECK-NEXT: [[INC12_US_US]] = add nuw nsw i32 [[J_137_US_US]], 1 +; CHECK-NEXT: [[EXITCOND58_NOT:%.*]] = icmp eq i32 [[INC12_US_US]], [[M]] +; CHECK-NEXT: br i1 [[EXITCOND58_NOT]], label [[FOR_COND5_FOR_END13_CRIT_EDGE_US_US]], label [[FOR_BODY7_US_US]] +; CHECK: for.cond5.for.end13_crit_edge.us.us: +; CHECK-NEXT: [[INC15_US_US]] = add nuw nsw i32 [[I_039_US_US]], 1 +; CHECK-NEXT: [[EXITCOND59_NOT:%.*]] = icmp eq i32 [[INC15_US_US]], [[N]] +; CHECK-NEXT: br i1 [[EXITCOND59_NOT]], label [[FOR_END16]], label [[FOR_BODY_US_US]] +; CHECK: for.body: +; CHECK-NEXT: [[I_039:%.*]] = phi i32 [ [[INC15:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_LR_PH]] ] +; CHECK-NEXT: [[INC15]] = add nuw nsw i32 [[I_039]], 1 +; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC15]], [[N]] +; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END16]], label [[FOR_BODY]] +; CHECK: for.end16: +; CHECK-NEXT: ret void +; +entry: + %a57 = bitcast i32* %a to i8* + %cmp38 = icmp sgt i32 %n, 0 + br i1 %cmp38, label %for.body.lr.ph, label %for.end16 + +for.body.lr.ph: ; preds = %entry + %cmp636 = icmp sgt i32 %m, 0 + br i1 %cmp636, label %for.body.us.us.preheader, label %for.body + +for.body.us.us.preheader: ; preds = %for.body.lr.ph + %0 = shl nuw i32 %m, 2 + %scevgep = getelementptr i32, i32* %a, i32 %m + %scevgep64 = getelementptr i32, i32* %b, i32 %m + %min.iters.check = icmp ult i32 %m, 4 + %bound0 = icmp ugt i32* %scevgep64, %a + %bound1 = icmp ugt i32* %scevgep, %b + %found.conflict = and i1 %bound0, %bound1 + %n.vec = and i32 %m, -4 + %cmp.n = icmp eq i32 %n.vec, %m + br label %for.body.us.us + +for.body.us.us: ; preds = %for.body.us.us.preheader, %for.cond5.for.end13_crit_edge.us.us + %i.039.us.us = phi i32 [ %inc15.us.us, %for.cond5.for.end13_crit_edge.us.us ], [ 0, %for.body.us.us.preheader ] + %1 = add i32 0, 0 + %vla.us.us = alloca i32, i32 %call.us.us, align 4 + %vla.us.us56 = bitcast i32* %vla.us.us to i8* + call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 4 %vla.us.us56, i8* align 4 %a57, i32 %0, i1 false) + %brmerge = select i1 %min.iters.check, i1 true, i1 %found.conflict + br i1 %brmerge, label %for.body7.us.us.preheader, label %vector.body + +vector.body: ; preds = %for.body.us.us, %vector.body + %index = phi i32 [ %index.next, %vector.body ], [ 0, %for.body.us.us ] + %2 = getelementptr inbounds i32, i32* %b, i32 %index + %3 = bitcast i32* %2 to <4 x i32>* + %wide.load = load <4 x i32>, <4 x i32>* %3, align 4 + %4 = getelementptr inbounds i32, i32* %vla.us.us, <4 x i32> %wide.load + %wide.masked.gather = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %4, i32 4, <4 x i1> , <4 x i32> undef) + %5 = getelementptr inbounds i32, i32* %a, i32 %index + %6 = bitcast i32* %5 to <4 x i32>* + store <4 x i32> %wide.masked.gather, <4 x i32>* %6, align 4 + %index.next = add nuw i32 %index, 4 + %7 = icmp eq i32 %index.next, %n.vec + br i1 %7, label %middle.block, label %vector.body + +middle.block: ; preds = %vector.body + br i1 %cmp.n, label %for.cond5.for.end13_crit_edge.us.us, label %for.body7.us.us.preheader + +for.body7.us.us.preheader: ; preds = %for.body.us.us, %middle.block + %j.137.us.us.ph = phi i32 [ 0, %for.body.us.us ], [ %n.vec, %middle.block ] + br label %for.body7.us.us + +for.body7.us.us: ; preds = %for.body7.us.us.preheader, %for.body7.us.us + %j.137.us.us = phi i32 [ %inc12.us.us, %for.body7.us.us ], [ %j.137.us.us.ph, %for.body7.us.us.preheader ] + %arrayidx8.us.us = getelementptr inbounds i32, i32* %b, i32 %j.137.us.us + %8 = load i32, i32* %arrayidx8.us.us, align 4 + %arrayidx9.us.us = getelementptr inbounds i32, i32* %vla.us.us, i32 %8 + %9 = load i32, i32* %arrayidx9.us.us, align 4 + %arrayidx10.us.us = getelementptr inbounds i32, i32* %a, i32 %j.137.us.us + store i32 %9, i32* %arrayidx10.us.us, align 4 + %inc12.us.us = add nuw nsw i32 %j.137.us.us, 1 + %exitcond58.not = icmp eq i32 %inc12.us.us, %m + br i1 %exitcond58.not, label %for.cond5.for.end13_crit_edge.us.us, label %for.body7.us.us + +for.cond5.for.end13_crit_edge.us.us: ; preds = %for.body7.us.us, %middle.block + %inc15.us.us = add nuw nsw i32 %i.039.us.us, 1 + %exitcond59.not = icmp eq i32 %inc15.us.us, %n + br i1 %exitcond59.not, label %for.end16, label %for.body.us.us + +for.body: ; preds = %for.body.lr.ph, %for.body + %i.039 = phi i32 [ %inc15, %for.body ], [ 0, %for.body.lr.ph ] + %inc15 = add nuw nsw i32 %i.039, 1 + %exitcond.not = icmp eq i32 %inc15, %n + br i1 %exitcond.not, label %for.end16, label %for.body + +for.end16: ; preds = %for.body, %for.cond5.for.end13_crit_edge.us.us, %entry + ret void +} declare <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*>, i32, <4 x i1>, <4 x i32>) +declare void @llvm.memcpy.p0i8.p0i8.i32(i8*, i8*, i32, i1)