llvm/test/Transforms/LoopVectorize/RISCV/tail-folding-reverse-load-store.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes=loop-vectorize \
; RUN: -tail-folding-policy=prefer-fold-tail \
; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=IF-EVL

; RUN: opt -passes=loop-vectorize \
; RUN: -tail-folding-policy=dont-fold-tail \
; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=NO-VP

define void @reverse_load_store(i64 %startval, ptr noalias %ptr, ptr noalias %ptr2) vscale_range(2, 1024) {
; IF-EVL-LABEL: @reverse_load_store(
; IF-EVL-NEXT:  entry:
; IF-EVL-NEXT:    br label [[VECTOR_PH:%.*]]
; IF-EVL:       vector.ph:
; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
; IF-EVL:       vector.body:
; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[AVL:%.*]] = phi i64 [ 1024, [[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[TMP5:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; IF-EVL-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL:%.*]], [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[TMP7:%.*]] = add i64 [[OFFSET_IDX]], -1
; IF-EVL-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[TMP7]]
; IF-EVL-NEXT:    [[TMP20:%.*]] = zext i32 [[TMP5]] to i64
; IF-EVL-NEXT:    [[TMP4:%.*]] = sub nuw nsw i64 [[TMP20]], 1
; IF-EVL-NEXT:    [[TMP6:%.*]] = sub i64 0, [[TMP4]]
; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr i32, ptr [[TMP8]], i64 [[TMP6]]
; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP12]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
; IF-EVL-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[PTR2:%.*]], i64 [[TMP7]]
; IF-EVL-NEXT:    [[TMP17:%.*]] = getelementptr i32, ptr [[TMP13]], i64 [[TMP6]]
; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP_LOAD]], ptr align 4 [[TMP17]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add nuw i64 [[TMP20]], [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP20]]
; IF-EVL-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; IF-EVL-NEXT:    br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; IF-EVL:       middle.block:
; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
; IF-EVL:       loopend:
; IF-EVL-NEXT:    ret void
;
; NO-VP-LABEL: @reverse_load_store(
; NO-VP-NEXT:  entry:
; NO-VP-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; NO-VP-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 2
; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; NO-VP:       vector.ph:
; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; NO-VP-NEXT:    [[TMP6:%.*]] = sub i64 [[STARTVAL:%.*]], [[N_VEC]]
; NO-VP-NEXT:    [[TMP7:%.*]] = trunc i64 [[N_VEC]] to i32
; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
; NO-VP:       vector.body:
; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY]] ]
; NO-VP-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL]], [[INDEX]]
; NO-VP-NEXT:    [[TMP8:%.*]] = add i64 [[OFFSET_IDX]], -1
; NO-VP-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[TMP8]]
; NO-VP-NEXT:    [[TMP9:%.*]] = sub nuw nsw i64 [[TMP1]], 1
; NO-VP-NEXT:    [[TMP12:%.*]] = sub i64 0, [[TMP9]]
; NO-VP-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP13]], i64 [[TMP12]]
; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP14]], align 4
; NO-VP-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[PTR2:%.*]], i64 [[TMP8]]
; NO-VP-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP15]], i64 [[TMP12]]
; NO-VP-NEXT:    store <vscale x 4 x i32> [[WIDE_LOAD]], ptr [[TMP20]], align 4
; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; NO-VP-NEXT:    [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; NO-VP-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; NO-VP:       middle.block:
; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; NO-VP-NEXT:    br i1 [[CMP_N]], label [[LOOPEND:%.*]], label [[SCALAR_PH]]
; NO-VP:       scalar.ph:
; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP6]], [[MIDDLE_BLOCK]] ], [ [[STARTVAL]], [[ENTRY:%.*]] ]
; NO-VP-NEXT:    [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[TMP7]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; NO-VP-NEXT:    br label [[FOR_BODY1:%.*]]
; NO-VP:       for.body:
; NO-VP-NEXT:    [[ADD_PHI:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY1]] ]
; NO-VP-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY1]] ]
; NO-VP-NEXT:    [[ADD]] = add i64 [[ADD_PHI]], -1
; NO-VP-NEXT:    [[GEPL:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[ADD]]
; NO-VP-NEXT:    [[TMP:%.*]] = load i32, ptr [[GEPL]], align 4
; NO-VP-NEXT:    [[GEPS:%.*]] = getelementptr inbounds i32, ptr [[PTR2]], i64 [[ADD]]
; NO-VP-NEXT:    store i32 [[TMP]], ptr [[GEPS]], align 4
; NO-VP-NEXT:    [[INC]] = add i32 [[I]], 1
; NO-VP-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[INC]], 1024
; NO-VP-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY1]], label [[LOOPEND]], !llvm.loop [[LOOP3:![0-9]+]]
; NO-VP:       loopend:
; NO-VP-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %add.phi = phi i64 [ %startval, %entry ], [ %add, %for.body ]
  %i = phi i32 [ 0, %entry ], [ %inc, %for.body ]
  %add = add i64 %add.phi, -1
  %gepl = getelementptr inbounds i32, ptr %ptr, i64 %add
  %tmp = load i32, ptr %gepl, align 4
  %geps = getelementptr inbounds i32, ptr %ptr2, i64 %add
  store i32 %tmp, ptr %geps, align 4
  %inc = add i32 %i, 1
  %exitcond = icmp ne i32 %inc, 1024
  br i1 %exitcond, label %for.body, label %loopend

loopend:
  ret void
}

define void @reverse_load_store_masked(i64 %startval, ptr noalias %ptr, ptr noalias %ptr1, ptr noalias %ptr2) vscale_range(2, 1024) {
; IF-EVL-LABEL: @reverse_load_store_masked(
; IF-EVL-NEXT:  entry:
; IF-EVL-NEXT:    br label [[VECTOR_PH:%.*]]
; IF-EVL:       vector.ph:
; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
; IF-EVL:       vector.body:
; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[AVL:%.*]] = phi i64 [ 1024, [[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[TMP5:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; IF-EVL-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL:%.*]], [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[OFFSET_IDX3:%.*]] = trunc i64 [[EVL_BASED_IV]] to i32
; IF-EVL-NEXT:    [[TMP11:%.*]] = add i64 [[OFFSET_IDX]], -1
; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i32 [[OFFSET_IDX3]]
; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP12]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
; IF-EVL-NEXT:    [[TMP14:%.*]] = icmp slt <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 100)
; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[PTR1:%.*]], i64 [[TMP11]]
; IF-EVL-NEXT:    [[VP_REVERSE_MASK:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
; IF-EVL-NEXT:    [[TMP26:%.*]] = zext i32 [[TMP5]] to i64
; IF-EVL-NEXT:    [[TMP15:%.*]] = sub nuw nsw i64 [[TMP26]], 1
; IF-EVL-NEXT:    [[TMP7:%.*]] = sub i64 0, [[TMP15]]
; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[TMP16]], i64 [[TMP7]]
; IF-EVL-NEXT:    [[VP_OP_LOAD4:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[VP_REVERSE_MASK]], i32 [[TMP5]])
; IF-EVL-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[PTR2:%.*]], i64 [[TMP11]]
; IF-EVL-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP21]], i64 [[TMP7]]
; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP_LOAD4]], ptr align 4 [[TMP25]], <vscale x 4 x i1> [[VP_REVERSE_MASK]], i32 [[TMP5]])
; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add nuw i64 [[TMP26]], [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP26]]
; IF-EVL-NEXT:    [[TMP29:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; IF-EVL-NEXT:    br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; IF-EVL:       middle.block:
; IF-EVL-NEXT:    br label [[FOR_INC:%.*]]
; IF-EVL:       loopend:
; IF-EVL-NEXT:    ret void
;
; NO-VP-LABEL: @reverse_load_store_masked(
; NO-VP-NEXT:  entry:
; NO-VP-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; NO-VP-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 2
; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[ENTRY:%.*]], label [[VECTOR_PH:%.*]]
; NO-VP:       vector.ph:
; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; NO-VP-NEXT:    [[TMP6:%.*]] = sub i64 [[STARTVAL1:%.*]], [[N_VEC]]
; NO-VP-NEXT:    [[TMP7:%.*]] = trunc i64 [[N_VEC]] to i32
; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
; NO-VP:       vector.body:
; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY]] ]
; NO-VP-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL1]], [[INDEX]]
; NO-VP-NEXT:    [[OFFSET_IDX1:%.*]] = trunc i64 [[INDEX]] to i32
; NO-VP-NEXT:    [[TMP8:%.*]] = add i64 [[OFFSET_IDX]], -1
; NO-VP-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i32 [[OFFSET_IDX1]]
; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 4
; NO-VP-NEXT:    [[TMP10:%.*]] = icmp slt <vscale x 4 x i32> [[WIDE_LOAD]], splat (i32 100)
; NO-VP-NEXT:    [[TMP15:%.*]] = getelementptr i32, ptr [[PTR1:%.*]], i64 [[TMP8]]
; NO-VP-NEXT:    [[TMP11:%.*]] = sub nuw nsw i64 [[TMP1]], 1
; NO-VP-NEXT:    [[TMP14:%.*]] = sub i64 0, [[TMP11]]
; NO-VP-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[TMP15]], i64 [[TMP14]]
; NO-VP-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x i1> @llvm.vector.reverse.nxv4i1(<vscale x 4 x i1> [[TMP10]])
; NO-VP-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP16]], <vscale x 4 x i1> [[REVERSE]], <vscale x 4 x i32> poison)
; NO-VP-NEXT:    [[TMP17:%.*]] = getelementptr i32, ptr [[PTR2:%.*]], i64 [[TMP8]]
; NO-VP-NEXT:    [[TMP22:%.*]] = getelementptr i32, ptr [[TMP17]], i64 [[TMP14]]
; NO-VP-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], ptr align 4 [[TMP22]], <vscale x 4 x i1> [[REVERSE]])
; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; NO-VP-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; NO-VP-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; NO-VP:       middle.block:
; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; NO-VP-NEXT:    br i1 [[CMP_N]], label [[LOOPEND:%.*]], label [[ENTRY]]
; NO-VP:       scalar.ph:
; NO-VP-NEXT:    [[STARTVAL:%.*]] = phi i64 [ [[TMP6]], [[MIDDLE_BLOCK]] ], [ [[STARTVAL1]], [[ENTRY1:%.*]] ]
; NO-VP-NEXT:    [[BC_RESUME_VAL5:%.*]] = phi i32 [ [[TMP7]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY1]] ]
; NO-VP-NEXT:    br label [[FOR_BODY1:%.*]]
; NO-VP:       for.body:
; NO-VP-NEXT:    [[ADD_PHI:%.*]] = phi i64 [ [[STARTVAL]], [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_INC:%.*]] ]
; NO-VP-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL5]], [[ENTRY]] ], [ [[INC:%.*]], [[FOR_INC]] ]
; NO-VP-NEXT:    [[ADD]] = add i64 [[ADD_PHI]], -1
; NO-VP-NEXT:    [[GEPL:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i32 [[I]]
; NO-VP-NEXT:    [[TMP:%.*]] = load i32, ptr [[GEPL]], align 4
; NO-VP-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[TMP]], 100
; NO-VP-NEXT:    br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; NO-VP:       if.then:
; NO-VP-NEXT:    [[GEPL1:%.*]] = getelementptr inbounds i32, ptr [[PTR1]], i64 [[ADD]]
; NO-VP-NEXT:    [[V:%.*]] = load i32, ptr [[GEPL1]], align 4
; NO-VP-NEXT:    [[GEPS:%.*]] = getelementptr inbounds i32, ptr [[PTR2]], i64 [[ADD]]
; NO-VP-NEXT:    store i32 [[V]], ptr [[GEPS]], align 4
; NO-VP-NEXT:    br label [[FOR_INC]]
; NO-VP:       for.inc:
; NO-VP-NEXT:    [[INC]] = add i32 [[I]], 1
; NO-VP-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[INC]], 1024
; NO-VP-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY1]], label [[LOOPEND]], !llvm.loop [[LOOP5:![0-9]+]]
; NO-VP:       loopend:
; NO-VP-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %add.phi = phi i64 [ %startval, %entry ], [ %add, %for.inc ]
  %i = phi i32 [ 0, %entry ], [ %inc, %for.inc ]
  %add = add i64 %add.phi, -1
  %gepl = getelementptr inbounds i32, ptr %ptr, i32 %i
  %tmp = load i32, ptr %gepl, align 4
  %cmp1 = icmp slt i32 %tmp, 100
  br i1 %cmp1, label %if.then, label %for.inc

if.then:
  %gepl1 = getelementptr inbounds i32, ptr %ptr1, i64 %add
  %v = load i32, ptr %gepl1, align 4
  %geps = getelementptr inbounds i32, ptr %ptr2, i64 %add
  store i32 %v, ptr %geps, align 4
  br label %for.inc

for.inc:
  %inc = add i32 %i, 1
  %exitcond = icmp ne i32 %inc, 1024
  br i1 %exitcond, label %for.body, label %loopend

loopend:
  ret void
}

; From a miscompile originally reported at
; https://github.com/llvm/llvm-project/issues/122681

define void @multiple_reverse_vector_pointer(ptr noalias %a, ptr noalias %b, ptr noalias %c, ptr noalias %d) vscale_range(2, 1024) {
; IF-EVL-LABEL: @multiple_reverse_vector_pointer(
; IF-EVL-NEXT:  entry:
; IF-EVL-NEXT:    br label [[VECTOR_PH:%.*]]
; IF-EVL:       vector.ph:
; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
; IF-EVL:       vector.body:
; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[AVL:%.*]] = phi i64 [ 1025, [[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], [[VECTOR_BODY]] ]
; IF-EVL-NEXT:    [[TMP6:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 16, i1 true)
; IF-EVL-NEXT:    [[TMP10:%.*]] = sub i64 1024, [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[TMP8:%.*]], i64 [[TMP10]]
; IF-EVL-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP6]] to i64
; IF-EVL-NEXT:    [[TMP3:%.*]] = sub nuw nsw i64 [[TMP9]], 1
; IF-EVL-NEXT:    [[TMP11:%.*]] = sub i64 0, [[TMP3]]
; IF-EVL-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[TMP12]], i64 [[TMP11]]
; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 16 x i8> @llvm.vp.load.nxv16i8.p0(ptr align 1 [[TMP13]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    [[VP_REVERSE:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> [[VP_OP_LOAD]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    [[TMP14:%.*]] = getelementptr i8, ptr [[B:%.*]], <vscale x 16 x i8> [[VP_REVERSE]]
; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 16 x i8> @llvm.vp.gather.nxv16i8.nxv16p0(<vscale x 16 x ptr> align 1 [[TMP14]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    [[TMP15:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[TMP10]]
; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP11]]
; IF-EVL-NEXT:    [[VP_REVERSE1:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> [[WIDE_MASKED_GATHER]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    call void @llvm.vp.store.nxv16i8.p0(<vscale x 16 x i8> [[VP_REVERSE1]], ptr align 1 [[TMP20]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr i8, ptr [[D:%.*]], i64 [[TMP10]]
; IF-EVL-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[TMP16]], i64 [[TMP11]]
; IF-EVL-NEXT:    call void @llvm.vp.store.nxv16i8.p0(<vscale x 16 x i8> [[VP_REVERSE1]], ptr align 1 [[TMP26]], <vscale x 16 x i1> splat (i1 true), i32 [[TMP6]])
; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add nuw i64 [[TMP9]], [[EVL_BASED_IV]]
; IF-EVL-NEXT:    [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; IF-EVL-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; IF-EVL-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; IF-EVL:       middle.block:
; IF-EVL-NEXT:    br label [[LOOP:%.*]]
; IF-EVL:       exit:
; IF-EVL-NEXT:    ret void
;
; NO-VP-LABEL: @multiple_reverse_vector_pointer(
; NO-VP-NEXT:  entry:
; NO-VP-NEXT:    [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; NO-VP-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP5]], 4
; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1025, [[TMP1]]
; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; NO-VP:       vector.ph:
; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1025, [[TMP1]]
; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 1025, [[N_MOD_VF]]
; NO-VP-NEXT:    [[TMP4:%.*]] = sub i64 1024, [[N_VEC]]
; NO-VP-NEXT:    br label [[LOOP:%.*]]
; NO-VP:       vector.body:
; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[LOOP]] ]
; NO-VP-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1024, [[INDEX]]
; NO-VP-NEXT:    [[TMP0:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
; NO-VP-NEXT:    [[TMP11:%.*]] = sub nuw nsw i64 [[TMP1]], 1
; NO-VP-NEXT:    [[TMP8:%.*]] = sub i64 0, [[TMP11]]
; NO-VP-NEXT:    [[TMP9:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[TMP8]]
; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 16 x i8>, ptr [[TMP9]], align 1
; NO-VP-NEXT:    [[REVERSE:%.*]] = call <vscale x 16 x i8> @llvm.vector.reverse.nxv16i8(<vscale x 16 x i8> [[WIDE_LOAD]])
; NO-VP-NEXT:    [[TMP10:%.*]] = getelementptr i8, ptr [[B:%.*]], <vscale x 16 x i8> [[REVERSE]]
; NO-VP-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 16 x i8> @llvm.masked.gather.nxv16i8.nxv16p0(<vscale x 16 x ptr> align 1 [[TMP10]], <vscale x 16 x i1> splat (i1 true), <vscale x 16 x i8> poison)
; NO-VP-NEXT:    [[TMP7:%.*]] = getelementptr i8, ptr [[D:%.*]], i64 [[OFFSET_IDX]]
; NO-VP-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[TMP7]], i64 [[TMP8]]
; NO-VP-NEXT:    [[REVERSE1:%.*]] = call <vscale x 16 x i8> @llvm.vector.reverse.nxv16i8(<vscale x 16 x i8> [[WIDE_MASKED_GATHER]])
; NO-VP-NEXT:    store <vscale x 16 x i8> [[REVERSE1]], ptr [[TMP12]], align 1
; NO-VP-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[D1:%.*]], i64 [[OFFSET_IDX]]
; NO-VP-NEXT:    [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 [[TMP8]]
; NO-VP-NEXT:    store <vscale x 16 x i8> [[REVERSE1]], ptr [[TMP14]], align 1
; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; NO-VP-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; NO-VP-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[LOOP]], !llvm.loop [[LOOP6:![0-9]+]]
; NO-VP:       middle.block:
; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1025, [[N_VEC]]
; NO-VP-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
; NO-VP:       scalar.ph:
; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP4]], [[MIDDLE_BLOCK]] ], [ 1024, [[ENTRY:%.*]] ]
; NO-VP-NEXT:    br label [[LOOP1:%.*]]
; NO-VP:       loop:
; NO-VP-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP1]] ]
; NO-VP-NEXT:    [[GEP_A:%.*]] = getelementptr i8, ptr [[A]], i64 [[IV]]
; NO-VP-NEXT:    [[X:%.*]] = load i8, ptr [[GEP_A]], align 1
; NO-VP-NEXT:    [[GEP_B:%.*]] = getelementptr i8, ptr [[B]], i8 [[X]]
; NO-VP-NEXT:    [[Y:%.*]] = load i8, ptr [[GEP_B]], align 1
; NO-VP-NEXT:    [[GEP_C:%.*]] = getelementptr i8, ptr [[D]], i64 [[IV]]
; NO-VP-NEXT:    store i8 [[Y]], ptr [[GEP_C]], align 1
; NO-VP-NEXT:    [[GEP_D:%.*]] = getelementptr i8, ptr [[D1]], i64 [[IV]]
; NO-VP-NEXT:    store i8 [[Y]], ptr [[GEP_D]], align 1
; NO-VP-NEXT:    [[IV_NEXT]] = add i64 [[IV]], -1
; NO-VP-NEXT:    [[CMP_NOT:%.*]] = icmp eq i64 [[IV]], 0
; NO-VP-NEXT:    br i1 [[CMP_NOT]], label [[EXIT]], label [[LOOP1]], !llvm.loop [[LOOP7:![0-9]+]]
; NO-VP:       exit:
; NO-VP-NEXT:    ret void
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 1024, %entry ], [ %iv.next, %loop ]

  %gep.a = getelementptr i8, ptr %a, i64 %iv
  %x = load i8, ptr %gep.a

  %gep.b = getelementptr i8, ptr %b, i8 %x
  %y = load i8, ptr %gep.b

  %gep.c = getelementptr i8, ptr %c, i64 %iv
  store i8 %y, ptr %gep.c

  %gep.d = getelementptr i8, ptr %d, i64 %iv
  store i8 %y, ptr %gep.d

  %iv.next = add i64 %iv, -1
  %cmp.not = icmp eq i64 %iv, 0
  br i1 %cmp.not, label %exit, label %loop

exit:
  ret void
}