llvm/test/Transforms/LoopVectorize/early-exit-calls.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -use-dereferenceable-at-point-semantics -S %s | FileCheck %s

define i32 @early_exit_with_extractvalue(ptr dereferenceable(1024) align 8 %src, i32 noundef %x) {
; CHECK-LABEL: define i32 @early_exit_with_extractvalue(
; CHECK-SAME: ptr align 8 dereferenceable(1024) [[SRC:%.*]], i32 noundef [[X:%.*]]) {
; CHECK-NEXT:  [[LOOP_PREHEADER:.*:]]
; CHECK-NEXT:    br label %[[LOOP:.*]]
; CHECK:       [[LOOP]]:
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[X]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    br label %[[LATCH:.*]]
; CHECK:       [[LATCH]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[LOOP]] ], [ [[IV_NEXT:%.*]], %[[EARLY_EXIT:.*]] ]
; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[LOOP]] ], [ [[VEC_IND_NEXT:%.*]], %[[EARLY_EXIT]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> [[VEC_IND]], <4 x i32> [[BROADCAST_SPLAT]])
; CHECK-NEXT:    [[TMP2:%.*]] = extractvalue { <4 x i32>, <4 x i1> } [[TMP1]], 1
; CHECK-NEXT:    [[TMP3:%.*]] = freeze <4 x i1> [[TMP2]]
; CHECK-NEXT:    [[TMP4:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP3]])
; CHECK-NEXT:    [[IV_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[IV_NEXT]], 60
; CHECK-NEXT:    [[VEC_IND_NEXT]] = add nuw nsw <4 x i32> [[VEC_IND]], splat (i32 4)
; CHECK-NEXT:    br i1 [[TMP4]], label %[[VECTOR_EARLY_EXIT:.*]], label %[[EARLY_EXIT]]
; CHECK:       [[EARLY_EXIT]]:
; CHECK-NEXT:    br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[LATCH]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT_LOOPEXIT:.*]]
; CHECK:       [[VECTOR_EARLY_EXIT]]:
; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP3]], i1 false)
; CHECK-NEXT:    [[TMP8:%.*]] = trunc i64 [[TMP7]] to i32
; CHECK-NEXT:    [[IV_LCSSA1:%.*]] = add i32 [[INDEX]], [[TMP8]]
; CHECK-NEXT:    br label %[[EARLY_EXIT1:.*]]
; CHECK:       [[EXIT_LOOPEXIT]]:
; CHECK-NEXT:    br label %[[LOOP_HEADER:.*]]
; CHECK:       [[LOOP_HEADER]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ 60, %[[EXIT_LOOPEXIT]] ], [ [[IV_NEXT1:%.*]], %[[LOOP_LATCH:.*]] ]
; CHECK-NEXT:    [[MUL_OV:%.*]] = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 [[IV]], i32 [[X]])
; CHECK-NEXT:    [[MUL_OV_OVERFLOW:%.*]] = extractvalue { i32, i1 } [[MUL_OV]], 1
; CHECK-NEXT:    br i1 [[MUL_OV_OVERFLOW]], label %[[EARLY_EXIT1]], label %[[LOOP_LATCH]]
; CHECK:       [[LOOP_LATCH]]:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[IV]]
; CHECK-NEXT:    [[VAL:%.*]] = load i32, ptr [[GEP]], align 4
; CHECK-NEXT:    [[IV_NEXT1]] = add nuw nsw i32 [[IV]], 1
; CHECK-NEXT:    [[EC:%.*]] = icmp eq i32 [[IV_NEXT1]], 63
; CHECK-NEXT:    br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[EARLY_EXIT1]]:
; CHECK-NEXT:    [[IV_LCSSA2:%.*]] = phi i32 [ [[IV]], %[[LOOP_HEADER]] ], [ [[IV_LCSSA1]], %[[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT:    ret i32 [[IV_LCSSA2]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    [[IV_LCSSA:%.*]] = phi i32 [ [[VAL]], %[[LOOP_LATCH]] ]
; CHECK-NEXT:    ret i32 [[IV_LCSSA]]
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %mul.ov = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 %iv, i32 %x)
  %mul.ov.overflow = extractvalue { i32, i1 } %mul.ov, 1
  br i1 %mul.ov.overflow, label %early.exit, label %loop.latch

loop.latch:
  %gep = getelementptr inbounds i32, ptr %src, i32 %iv
  %val = load i32, ptr %gep, align 4
  %iv.next = add nuw nsw i32 %iv, 1
  %ec = icmp eq i32 %iv.next, 63
  br i1 %ec, label %exit, label %loop.header

early.exit:
  ret i32 %iv

exit:
  ret i32 %val
}

define void @readnone_call_site_non_deref_load(ptr dereferenceable(1024) %a, ptr %b) {
; CHECK-LABEL: define void @readnone_call_site_non_deref_load(
; CHECK-SAME: ptr dereferenceable(1024) [[A:%.*]], ptr [[B:%.*]]) {
; CHECK-NEXT:  [[SCALAR_PH:.*]]:
; CHECK-NEXT:    br label %[[LOOP:.*]]
; CHECK:       [[LOOP]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LATCH:.*]] ]
; CHECK-NEXT:    [[GEP_A:%.*]] = getelementptr i8, ptr [[A]], i64 [[IV]]
; CHECK-NEXT:    [[L_A:%.*]] = load i8, ptr [[GEP_A]], align 1
; CHECK-NEXT:    [[GEP_B:%.*]] = getelementptr float, ptr [[B]], i64 [[IV]]
; CHECK-NEXT:    [[L_B:%.*]] = load float, ptr [[GEP_B]], align 4
; CHECK-NEXT:    [[C:%.*]] = call float @fabsf(float [[L_B]]) #[[ATTR3:[0-9]+]]
; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i8 [[L_A]], 0
; CHECK-NEXT:    br i1 [[CMP]], label %[[EXIT:.*]], label %[[LATCH]]
; CHECK:       [[LATCH]]:
; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
; CHECK-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
  %gep.a = getelementptr i8, ptr %a, i64 %iv
  %l.a = load i8, ptr %gep.a
  %gep.b = getelementptr float, ptr %b, i64 %iv
  %l.b = load float, ptr %gep.b
  %c = call float @fabsf(float %l.b) nounwind memory(none)
  %cmp = icmp eq i8 %l.a, 0
  br i1 %cmp, label %exit, label %latch

latch:
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 1024
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

declare float @fabsf(float) speculatable