llvm/test/Transforms/LoopVectorize/RISCV/conditional-scalar-assignment-fold-tail.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
; RUN: opt -passes=loop-vectorize -mtriple riscv64 -mattr=+v -S < %s | FileCheck %s

define i32 @simple_find_last_reduction(i64 %N, ptr %data, i32 %a) {
; CHECK-LABEL: define i32 @simple_find_last_reduction(
; CHECK-SAME: i64 [[N:%.*]], ptr [[DATA:%.*]], i32 [[A:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  [[SCALAR_PH:.*:]]
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[A]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, %[[EXIT]] ], [ [[CURRENT_ITERATION_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ splat (i32 -1), %[[EXIT]] ], [ [[TMP10:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[EXIT]] ], [ [[TMP9:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[AVL:%.*]] = phi i64 [ [[N]], %[[EXIT]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT:    [[LD_ADDR:%.*]] = getelementptr inbounds i32, ptr [[DATA]], i64 [[IV]]
; CHECK-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[LD_ADDR]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP1]])
; CHECK-NEXT:    [[TMP5:%.*]] = icmp slt <vscale x 4 x i32> [[BROADCAST_SPLAT]], [[VP_OP_LOAD]]
; CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i1> zeroinitializer, i32 [[TMP1]])
; CHECK-NEXT:    [[TMP7:%.*]] = freeze <vscale x 4 x i1> [[TMP6]]
; CHECK-NEXT:    [[TMP8:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP7]])
; CHECK-NEXT:    [[TMP9]] = select i1 [[TMP8]], <vscale x 4 x i1> [[TMP6]], <vscale x 4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP10]] = select i1 [[TMP8]], <vscale x 4 x i32> [[VP_OP_LOAD]], <vscale x 4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP1]] to i64
; CHECK-NEXT:    [[CURRENT_ITERATION_NEXT]] = add i64 [[TMP11]], [[IV]]
; CHECK-NEXT:    [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP11]]
; CHECK-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT:    br i1 [[TMP12]], label %[[EXIT1:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[EXIT1]]:
; CHECK-NEXT:    [[SELECT_DATA_LCSSA:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> [[TMP10]], <vscale x 4 x i1> [[TMP9]], i32 -1)
; CHECK-NEXT:    br label %[[EXIT2:.*]]
; CHECK:       [[EXIT2]]:
; CHECK-NEXT:    ret i32 [[SELECT_DATA_LCSSA]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %data.phi = phi i32 [ -1, %entry ], [ %select.data, %loop ]
  %ld.addr = getelementptr inbounds i32, ptr %data, i64 %iv
  %ld = load i32, ptr %ld.addr, align 4
  %select.cmp = icmp slt i32 %a, %ld
  %select.data = select i1 %select.cmp, i32 %ld, i32 %data.phi
  %iv.next = add nuw nsw i64 %iv, 1
  %exit.cmp = icmp eq i64 %iv.next, %N
  br i1 %exit.cmp, label %exit, label %loop

exit:
  ret i32 %select.data
}

; This test is derived from the following C program:
; int non_speculatable_find_last_reduction(
;   int* a, int* b, int default_val, int N, int threshold)
; {
;   int result = default_val;
;   for (int i = 0; i < N; ++i)
;     if (a[i] > threshold)
;       result = b[i];
;   return result;
; }
define i32 @non_speculatable_find_last_reduction(ptr noalias %a, ptr noalias %b, i32 %default_val, i64 %N, i32 %threshold) {
; CHECK-LABEL: define i32 @non_speculatable_find_last_reduction(
; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], i32 [[DEFAULT_VAL:%.*]], i64 [[N:%.*]], i32 [[THRESHOLD:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[SCALAR_PH:.*:]]
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[THRESHOLD]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[DEFAULT_VAL]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, %[[EXIT]] ], [ [[CURRENT_ITERATION_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ [[BROADCAST_SPLAT2]], %[[EXIT]] ], [ [[TMP12:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[EXIT]] ], [ [[TMP11:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[AVL:%.*]] = phi i64 [ [[N]], %[[EXIT]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT:    [[A_ADDR:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[IV]]
; CHECK-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[A_ADDR]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP1]])
; CHECK-NEXT:    [[TMP5:%.*]] = icmp sgt <vscale x 4 x i32> [[VP_OP_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i1> zeroinitializer, i32 [[TMP1]])
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr i32, ptr [[B]], i64 [[IV]]
; CHECK-NEXT:    [[VP_OP_LOAD5:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP7]], <vscale x 4 x i1> [[TMP5]], i32 [[TMP1]])
; CHECK-NEXT:    [[TMP9:%.*]] = freeze <vscale x 4 x i1> [[TMP8]]
; CHECK-NEXT:    [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
; CHECK-NEXT:    [[TMP11]] = select i1 [[TMP10]], <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP12]] = select i1 [[TMP10]], <vscale x 4 x i32> [[VP_OP_LOAD5]], <vscale x 4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[TMP13:%.*]] = zext i32 [[TMP1]] to i64
; CHECK-NEXT:    [[CURRENT_ITERATION_NEXT]] = add i64 [[TMP13]], [[IV]]
; CHECK-NEXT:    [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP13]]
; CHECK-NEXT:    [[TMP14:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT:    br i1 [[TMP14]], label %[[IF_THEN:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[IF_THEN]]:
; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 4 x i32> [[BROADCAST_SPLAT2]], i64 0
; CHECK-NEXT:    [[SELECT_DATA_LCSSA:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> [[TMP12]], <vscale x 4 x i1> [[TMP11]], i32 [[TMP15]])
; CHECK-NEXT:    br label %[[LATCH:.*]]
; CHECK:       [[LATCH]]:
; CHECK-NEXT:    ret i32 [[SELECT_DATA_LCSSA]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
  %data.phi = phi i32 [ %default_val, %entry ], [ %select.data, %latch ]
  %a.addr = getelementptr inbounds nuw i32, ptr %a, i64 %iv
  %ld.a = load i32, ptr %a.addr, align 4
  %if.cond = icmp sgt i32 %ld.a, %threshold
  br i1 %if.cond, label %if.then, label %latch

if.then:
  %b.addr = getelementptr inbounds nuw i32, ptr %b, i64 %iv
  %ld.b = load i32, ptr %b.addr, align 4
  br label %latch

latch:
  %select.data = phi i32 [ %ld.b, %if.then ], [ %data.phi, %loop ]
  %iv.next = add nuw nsw i64 %iv, 1
  %exit.cmp = icmp eq i64 %iv.next, %N
  br i1 %exit.cmp, label %exit, label %loop

exit:
  ret i32 %select.data
}