1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt -S -loop-vectorize -mtriple=s390x-linux-gnu -tiny-trip-count-interleave-threshold=4 -vectorizer-min-trip-count=8 < %s | FileCheck %s
3; RUN: opt -S -passes=loop-vectorize -mtriple=s390x-linux-gnu -tiny-trip-count-interleave-threshold=4 -vectorizer-min-trip-count=8 < %s | FileCheck %s
4
5define i32 @main(i32 %arg, i8** nocapture readnone %arg1) #0 {
6; CHECK-LABEL: @main(
7; CHECK-NEXT:  entry:
8; CHECK-NEXT:    [[TMP0:%.*]] = alloca i8, align 1
9; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
10; CHECK:       vector.ph:
11; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
12; CHECK:       vector.body:
13; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
14; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = trunc i32 [[INDEX]] to i8
15; CHECK-NEXT:    [[TMP1:%.*]] = add i8 [[OFFSET_IDX]], 0
16; CHECK-NEXT:    [[TMP2:%.*]] = add i8 [[OFFSET_IDX]], 1
17; CHECK-NEXT:    [[TMP3:%.*]] = add i8 [[OFFSET_IDX]], 2
18; CHECK-NEXT:    [[TMP4:%.*]] = add i8 [[OFFSET_IDX]], 3
19; CHECK-NEXT:    [[TMP5:%.*]] = add i8 [[OFFSET_IDX]], 4
20; CHECK-NEXT:    [[TMP6:%.*]] = add i8 [[OFFSET_IDX]], 5
21; CHECK-NEXT:    [[TMP7:%.*]] = add i8 [[OFFSET_IDX]], 6
22; CHECK-NEXT:    [[TMP8:%.*]] = add i8 [[OFFSET_IDX]], 7
23; CHECK-NEXT:    store i8 [[TMP1]], i8* [[TMP0]], align 2
24; CHECK-NEXT:    store i8 [[TMP2]], i8* [[TMP0]], align 2
25; CHECK-NEXT:    store i8 [[TMP3]], i8* [[TMP0]], align 2
26; CHECK-NEXT:    store i8 [[TMP4]], i8* [[TMP0]], align 2
27; CHECK-NEXT:    store i8 [[TMP5]], i8* [[TMP0]], align 2
28; CHECK-NEXT:    store i8 [[TMP6]], i8* [[TMP0]], align 2
29; CHECK-NEXT:    store i8 [[TMP7]], i8* [[TMP0]], align 2
30; CHECK-NEXT:    store i8 [[TMP8]], i8* [[TMP0]], align 2
31; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
32; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], 8
33; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
34; CHECK:       middle.block:
35; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i32 9, 8
36; CHECK-NEXT:    br i1 [[CMP_N]], label [[RET:%.*]], label [[SCALAR_PH]]
37; CHECK:       scalar.ph:
38; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i8 [ 8, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
39; CHECK-NEXT:    br label [[LOOP:%.*]]
40; CHECK:       loop:
41; CHECK-NEXT:    [[STOREMERGE_I_I:%.*]] = phi i8 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[TMP12_I_I:%.*]], [[LOOP]] ]
42; CHECK-NEXT:    store i8 [[STOREMERGE_I_I]], i8* [[TMP0]], align 2
43; CHECK-NEXT:    [[TMP8_I_I:%.*]] = icmp ult i8 [[STOREMERGE_I_I]], 8
44; CHECK-NEXT:    [[TMP12_I_I]] = add nuw nsw i8 [[STOREMERGE_I_I]], 1
45; CHECK-NEXT:    br i1 [[TMP8_I_I]], label [[LOOP]], label [[RET]], !llvm.loop [[LOOP2:![0-9]+]]
46; CHECK:       ret:
47; CHECK-NEXT:    ret i32 0
48;
49entry:
50  %0 = alloca i8, align 1
51  br label %loop
52
53loop:
54  %storemerge.i.i = phi i8 [ 0, %entry ], [ %tmp12.i.i, %loop ]
55  store i8 %storemerge.i.i, i8* %0, align 2
56  %tmp8.i.i = icmp ult i8 %storemerge.i.i, 8
57  %tmp12.i.i = add nuw nsw i8 %storemerge.i.i, 1
58  br i1 %tmp8.i.i, label %loop, label %ret
59
60ret:
61  ret i32 0
62}
63
64attributes #0 = { "target-cpu"="z13" }
65
66