llvm/test/Transforms/LoopVectorize/single-early-exit-deref-assumptions.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 4
; RUN: opt -S < %s -p loop-vectorize -enable-early-exit-vectorization -force-vector-width=4 | FileCheck %s

define i64 @early_exit_alignment_and_deref_known_via_assumption_with_constant_size(ptr noalias %p1, ptr noalias %p2) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_known_via_assumption_with_constant_size(
; CHECK-SAME: ptr noalias [[P1:%.*]], ptr noalias [[P2:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P1]], i64 4), "dereferenceable"(ptr [[P1]], i64 1024) ]
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P2]], i64 4), "dereferenceable"(ptr [[P2]], i64 1024) ]
; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT3:%.*]], [[LOOP]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX1]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP0]], align 1
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX1]]
; CHECK-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i8>, ptr [[TMP2]], align 1
; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne <4 x i8> [[WIDE_LOAD]], [[WIDE_LOAD2]]
; CHECK-NEXT:    [[INDEX_NEXT3]] = add nuw i64 [[INDEX1]], 4
; CHECK-NEXT:    [[TMP3:%.*]] = freeze <4 x i1> [[TMP4]]
; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP3]])
; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 1024
; CHECK-NEXT:    [[TMP7:%.*]] = or i1 [[TMP5]], [[TMP6]]
; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_SPLIT:%.*]], label [[LOOP]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       middle.split:
; CHECK-NEXT:    br i1 [[TMP5]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK:       middle.block:
; CHECK-NEXT:    br label [[LOOP_END:%.*]]
; CHECK:       vector.early.exit:
; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP4]], i1 true)
; CHECK-NEXT:    [[TMP9:%.*]] = add i64 [[INDEX1]], [[TMP8]]
; CHECK-NEXT:    br label [[LOOP_END]]
; CHECK:       scalar.ph:
; CHECK-NEXT:    br label [[LOOP1:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 0, [[SCALAR_PH]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT:    [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
; CHECK-NEXT:    br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END]]
; CHECK:       loop.inc:
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP1]], label [[LOOP_END]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       loop.end:
; CHECK-NEXT:    [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP1]] ], [ -1, [[LOOP_INC]] ], [ -1, [[MIDDLE_BLOCK]] ], [ [[TMP9]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT:    ret i64 [[RETVAL]]
;
entry:
  call void @llvm.assume(i1 true) [ "align"(ptr %p1, i64 4), "dereferenceable"(ptr %p1, i64 1024) ]
  call void @llvm.assume(i1 true) [ "align"(ptr %p2, i64 4), "dereferenceable"(ptr %p2, i64 1024) ]
  br label %loop

loop:
  %index = phi i64 [ %index.next, %loop.inc ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i8, ptr %p1, i64 %index
  %ld1 = load i8, ptr %arrayidx, align 1
  %arrayidx1 = getelementptr inbounds i8, ptr %p2, i64 %index
  %ld2 = load i8, ptr %arrayidx1, align 1
  %cmp3 = icmp eq i8 %ld1, %ld2
  br i1 %cmp3, label %loop.inc, label %loop.end

loop.inc:
  %index.next = add i64 %index, 1
  %exitcond = icmp ne i64 %index.next, 1024
  br i1 %exitcond, label %loop, label %loop.end

loop.end:
  %retval = phi i64 [ %index, %loop ], [ -1, %loop.inc ]
  ret i64 %retval
}

define i64 @early_exit_alignment_and_deref_via_assumption_with_constant_size_too_small(ptr noalias %p1, ptr noalias %p2) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_via_assumption_with_constant_size_too_small(
; CHECK-SAME: ptr noalias [[P1:%.*]], ptr noalias [[P2:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P1]], i64 4), "dereferenceable"(ptr [[P1]], i64 1024) ]
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P2]], i64 4), "dereferenceable"(ptr [[P2]], i64 1024) ]
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT:    [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
; CHECK-NEXT:    br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END:%.*]]
; CHECK:       loop.inc:
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 1025
; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]]
; CHECK:       loop.end:
; CHECK-NEXT:    [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ -1, [[LOOP_INC]] ]
; CHECK-NEXT:    ret i64 [[RETVAL]]
;
entry:
  call void @llvm.assume(i1 true) [ "align"(ptr %p1, i64 4), "dereferenceable"(ptr %p1, i64 1024) ]
  call void @llvm.assume(i1 true) [ "align"(ptr %p2, i64 4), "dereferenceable"(ptr %p2, i64 1024) ]
  br label %loop

loop:
  %index = phi i64 [ %index.next, %loop.inc ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i8, ptr %p1, i64 %index
  %ld1 = load i8, ptr %arrayidx, align 1
  %arrayidx1 = getelementptr inbounds i8, ptr %p2, i64 %index
  %ld2 = load i8, ptr %arrayidx1, align 1
  %cmp3 = icmp eq i8 %ld1, %ld2
  br i1 %cmp3, label %loop.inc, label %loop.end

loop.inc:
  %index.next = add i64 %index, 1
  %exitcond = icmp ne i64 %index.next, 1025
  br i1 %exitcond, label %loop, label %loop.end

loop.end:
  %retval = phi i64 [ %index, %loop ], [ -1, %loop.inc ]
  ret i64 %retval
}

define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero(ptr noalias %p1, ptr noalias %p2, i64 %n) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero(
; CHECK-SAME: ptr noalias [[P1:%.*]], ptr noalias [[P2:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P1]], i64 4), "dereferenceable"(ptr [[P1]], i64 [[N]]) ]
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P2]], i64 4), "dereferenceable"(ptr [[P2]], i64 [[N]]) ]
; CHECK-NEXT:    [[C:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT:    br i1 [[C]], label [[LOOP_PREHEADER:%.*]], label [[LOOP_END:%.*]]
; CHECK:       loop.preheader:
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT3:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX1]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP0]], align 1
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX1]]
; CHECK-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i8>, ptr [[TMP1]], align 1
; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne <4 x i8> [[WIDE_LOAD]], [[WIDE_LOAD2]]
; CHECK-NEXT:    [[INDEX_NEXT3]] = add nuw i64 [[INDEX1]], 4
; CHECK-NEXT:    [[TMP3:%.*]] = freeze <4 x i1> [[TMP2]]
; CHECK-NEXT:    [[TMP4:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP3]])
; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT3]], [[N_VEC]]
; CHECK-NEXT:    [[TMP6:%.*]] = or i1 [[TMP4]], [[TMP5]]
; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK:       middle.split:
; CHECK-NEXT:    br i1 [[TMP4]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], label [[LOOP_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK:       vector.early.exit:
; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP2]], i1 true)
; CHECK-NEXT:    [[TMP8:%.*]] = add i64 [[INDEX1]], [[TMP7]]
; CHECK-NEXT:    br label [[LOOP_END_LOOPEXIT]]
; CHECK:       scalar.ph:
; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ]
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT:    [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
; CHECK-NEXT:    br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END_LOOPEXIT]]
; CHECK:       loop.inc:
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], [[N]]
; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END_LOOPEXIT]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:       loop.end.loopexit:
; CHECK-NEXT:    [[RETVAL_PH:%.*]] = phi i64 [ -1, [[LOOP_INC]] ], [ [[INDEX]], [[LOOP]] ], [ -1, [[MIDDLE_BLOCK]] ], [ [[TMP8]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT:    br label [[LOOP_END]]
; CHECK:       loop.end:
; CHECK-NEXT:    [[RETVAL:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[RETVAL_PH]], [[LOOP_END_LOOPEXIT]] ]
; CHECK-NEXT:    ret i64 [[RETVAL]]
;
entry:
  call void @llvm.assume(i1 true) [ "align"(ptr %p1, i64 4), "dereferenceable"(ptr %p1, i64 %n) ]
  call void @llvm.assume(i1 true) [ "align"(ptr %p2, i64 4), "dereferenceable"(ptr %p2, i64 %n) ]
  %c = icmp ne i64 %n, 0
  br i1 %c, label %loop, label %loop.end

loop:
  %index = phi i64 [ %index.next, %loop.inc ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i8, ptr %p1, i64 %index
  %ld1 = load i8, ptr %arrayidx, align 1
  %arrayidx1 = getelementptr inbounds i8, ptr %p2, i64 %index
  %ld2 = load i8, ptr %arrayidx1, align 1
  %cmp3 = icmp eq i8 %ld1, %ld2
  br i1 %cmp3, label %loop.inc, label %loop.end

loop.inc:
  %index.next = add i64 %index, 1
  %exitcond = icmp ne i64 %index.next, %n
  br i1 %exitcond, label %loop, label %loop.end

loop.end:
  %retval = phi i64 [ 0, %entry ], [ %index, %loop ], [ -1, %loop.inc ]
  ret i64 %retval
}

define i64 @early_exit_alignment_and_deref_known_via_assumption_n_may_be_zero(ptr noalias %p1, ptr noalias %p2, i64 %n) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_known_via_assumption_n_may_be_zero(
; CHECK-SAME: ptr noalias [[P1:%.*]], ptr noalias [[P2:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P1]], i64 4), "dereferenceable"(ptr [[P1]], i64 [[N]]) ]
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P2]], i64 4), "dereferenceable"(ptr [[P2]], i64 [[N]]) ]
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT:    [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
; CHECK-NEXT:    br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END:%.*]]
; CHECK:       loop.inc:
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], [[N]]
; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]]
; CHECK:       loop.end:
; CHECK-NEXT:    [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ -1, [[LOOP_INC]] ]
; CHECK-NEXT:    ret i64 [[RETVAL]]
;
entry:
  call void @llvm.assume(i1 true) [ "align"(ptr %p1, i64 4), "dereferenceable"(ptr %p1, i64 %n) ]
  call void @llvm.assume(i1 true) [ "align"(ptr %p2, i64 4), "dereferenceable"(ptr %p2, i64 %n) ]
  br label %loop

loop:
  %index = phi i64 [ %index.next, %loop.inc ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i8, ptr %p1, i64 %index
  %ld1 = load i8, ptr %arrayidx, align 1
  %arrayidx1 = getelementptr inbounds i8, ptr %p2, i64 %index
  %ld2 = load i8, ptr %arrayidx1, align 1
  %cmp3 = icmp eq i8 %ld1, %ld2
  br i1 %cmp3, label %loop.inc, label %loop.end

loop.inc:
  %index.next = add i64 %index, 1
  %exitcond = icmp ne i64 %index.next, %n
  br i1 %exitcond, label %loop, label %loop.end

loop.end:
  %retval = phi i64 [ %index, %loop ], [ -1, %loop.inc ]
  ret i64 %retval
}

define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero_i16(ptr noalias %p1, ptr noalias %p2, i32 %n) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero_i16(
; CHECK-SAME: ptr noalias [[P1:%.*]], ptr noalias [[P2:%.*]], i32 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[N_EXT:%.*]] = zext i32 [[N]] to i64
; CHECK-NEXT:    [[N_SCALED:%.*]] = shl nuw nsw i64 [[N_EXT]], 1
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P1]], i64 4), "dereferenceable"(ptr [[P1]], i64 [[N_SCALED]]) ]
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[P2]], i64 4), "dereferenceable"(ptr [[P2]], i64 [[N_SCALED]]) ]
; CHECK-NEXT:    [[C:%.*]] = icmp ne i64 [[N_EXT]], 0
; CHECK-NEXT:    br i1 [[C]], label [[LOOP_PREHEADER:%.*]], label [[LOOP_END:%.*]]
; CHECK:       loop.preheader:
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 0, [[LOOP_PREHEADER]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT:    [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT:    [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
; CHECK-NEXT:    br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END_LOOPEXIT:%.*]]
; CHECK:       loop.inc:
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], [[N_EXT]]
; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END_LOOPEXIT]]
; CHECK:       loop.end.loopexit:
; CHECK-NEXT:    [[RETVAL_PH:%.*]] = phi i64 [ -1, [[LOOP_INC]] ], [ [[INDEX]], [[LOOP]] ]
; CHECK-NEXT:    br label [[LOOP_END]]
; CHECK:       loop.end:
; CHECK-NEXT:    [[RETVAL:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[RETVAL_PH]], [[LOOP_END_LOOPEXIT]] ]
; CHECK-NEXT:    ret i64 [[RETVAL]]
;
entry:
  %n.ext = zext i32 %n to i64
  %n.scaled = shl nuw nsw i64 %n.ext, 1
  call void @llvm.assume(i1 true) [ "align"(ptr %p1, i64 4), "dereferenceable"(ptr %p1, i64 %n.scaled) ]
  call void @llvm.assume(i1 true) [ "align"(ptr %p2, i64 4), "dereferenceable"(ptr %p2, i64 %n.scaled) ]
  %c = icmp ne i64 %n.ext, 0
  br i1 %c, label %loop, label %loop.end

loop:
  %index = phi i64 [ 0, %entry ], [ %index.next, %loop.inc ]
  %gep.p1 = getelementptr inbounds i8, ptr %p1, i64 %index
  %ld1 = load i8, ptr %gep.p1, align 1
  %gep.p2 = getelementptr inbounds i8, ptr %p2, i64 %index
  %ld2 = load i8, ptr %gep.p2, align 1
  %c.0 = icmp eq i8 %ld1, %ld2
  br i1 %c.0, label %loop.inc, label %loop.end

loop.inc:
  %index.next = add i64 %index, 1
  %exitcond = icmp ne i64 %index.next, %n.ext
  br i1 %exitcond, label %loop, label %loop.end

loop.end:
  %retval = phi i64 [ 0, %entry ], [ %index, %loop ], [ -1, %loop.inc ]
  ret i64 %retval
}

define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero_i16_ptr_iv(ptr %A, i32 noundef %n) nofree nosync {
; CHECK-LABEL: define i64 @early_exit_alignment_and_deref_known_via_assumption_n_not_zero_i16_ptr_iv(
; CHECK-SAME: ptr [[A:%.*]], i32 noundef [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "align"(ptr [[A]], i64 2) ]
; CHECK-NEXT:    [[N_EXT:%.*]] = zext i32 [[N]] to i64
; CHECK-NEXT:    [[MUL:%.*]] = shl i64 [[N_EXT]], 1
; CHECK-NEXT:    call void @llvm.assume(i1 true) [ "dereferenceable"(ptr [[A]], i64 [[MUL]]) ]
; CHECK-NEXT:    [[A_END:%.*]] = getelementptr i8, ptr [[A]], i64 [[MUL]]
; CHECK-NEXT:    [[PRE:%.*]] = icmp eq i32 [[N]], 0
; CHECK-NEXT:    br i1 [[PRE]], label [[EXIT:%.*]], label [[LOOP_HEADER_PREHEADER:%.*]]
; CHECK:       loop.header.preheader:
; CHECK-NEXT:    [[TMP0:%.*]] = add nsw i64 [[MUL]], -2
; CHECK-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
; CHECK-NEXT:    [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP2]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP2]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP2]], [[N_MOD_VF]]
; CHECK-NEXT:    [[TMP3:%.*]] = mul i64 [[N_VEC]], 2
; CHECK-NEXT:    [[IV_NEXT1:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP3]]
; CHECK-NEXT:    br label [[LOOP_HEADER1:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[LOOP_HEADER1]] ]
; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
; CHECK-NEXT:    [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 [[OFFSET_IDX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[NEXT_GEP]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq <4 x i16> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP6:%.*]] = freeze <4 x i1> [[TMP5]]
; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[LOOP_HEADER1]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK:       middle.split:
; CHECK-NEXT:    br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[LOOP_LATCH1:%.*]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK:       vector.early.exit:
; CHECK-NEXT:    [[TMP10:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP5]], i1 true)
; CHECK-NEXT:    [[TMP11:%.*]] = add i64 [[INDEX]], [[TMP10]]
; CHECK-NEXT:    [[TMP12:%.*]] = mul i64 [[TMP11]], 2
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP12]]
; CHECK-NEXT:    br label [[EXIT_LOOPEXIT]]
; CHECK:       scalar.ph:
; CHECK-NEXT:    [[IV1:%.*]] = phi ptr [ [[IV_NEXT1]], [[LOOP_LATCH1]] ], [ [[A]], [[LOOP_HEADER_PREHEADER]] ]
; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
; CHECK:       loop.header:
; CHECK-NEXT:    [[IV:%.*]] = phi ptr [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ [[IV1]], [[SCALAR_PH]] ]
; CHECK-NEXT:    [[L:%.*]] = load i16, ptr [[IV]], align 2
; CHECK-NEXT:    [[C_0:%.*]] = icmp eq i16 [[L]], 0
; CHECK-NEXT:    br i1 [[C_0]], label [[EXIT_LOOPEXIT]], label [[LOOP_LATCH]]
; CHECK:       loop.latch:
; CHECK-NEXT:    [[IV_NEXT]] = getelementptr inbounds nuw i8, ptr [[IV]], i64 2
; CHECK-NEXT:    [[EC:%.*]] = icmp eq ptr [[IV_NEXT]], [[A_END]]
; CHECK-NEXT:    br i1 [[EC]], label [[EXIT_LOOPEXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK:       exit.loopexit:
; CHECK-NEXT:    [[P_PH:%.*]] = phi ptr [ [[A_END]], [[LOOP_LATCH]] ], [ [[IV]], [[LOOP_HEADER]] ], [ [[A_END]], [[LOOP_LATCH1]] ], [ [[TMP13]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT:    br label [[EXIT]]
; CHECK:       exit:
; CHECK-NEXT:    [[P:%.*]] = phi ptr [ [[A]], [[ENTRY:%.*]] ], [ [[P_PH]], [[EXIT_LOOPEXIT]] ]
; CHECK-NEXT:    [[RES:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT:    ret i64 [[RES]]
;
entry:
  call void @llvm.assume(i1 true) [ "align"(ptr %A, i64 2) ]
  %n.ext = zext i32 %n to i64
  %mul = shl nuw nsw i64 %n.ext, 1
  call void @llvm.assume(i1 true) [ "dereferenceable"(ptr %A, i64 %mul) ]
  %A.end = getelementptr i8, ptr %A, i64 %mul
  %pre = icmp eq i32 %n, 0
  br i1 %pre, label %exit, label %loop.header

loop.header:
  %iv = phi ptr [ %iv.next, %loop.latch ], [ %A, %entry ]
  %l = load i16, ptr %iv, align 2
  %c.0 = icmp eq i16 %l, 0
  br i1 %c.0, label %exit, label %loop.latch

loop.latch:
  %iv.next = getelementptr inbounds nuw i8, ptr %iv, i64 2
  %ec = icmp eq ptr %iv.next, %A.end
  br i1 %ec, label %exit, label %loop.header

exit:
  %p = phi ptr [ %A, %entry ], [ %iv, %loop.header ], [ %A.end, %loop.latch ]
  %res = ptrtoint ptr %p to i64
  ret i64 %res
}