1; RUN: opt -S -scalable-vectorization=on -force-target-supports-scalable-vectors=true -loop-vectorize -force-vector-width=2 -force-vector-interleave=1 < %s | FileCheck %s
2
3target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
4
5; Make sure we can vectorize loops which contain lifetime markers.
6
7define void @test(i32 *%d) {
8; CHECK-LABEL: @test(
9; CHECK:      entry:
10; CHECK:        [[ALLOCA:%.*]] = alloca [1024 x i32], align 16
11; CHECK-NEXT:   [[BC:%.*]] = bitcast [1024 x i32]* [[ALLOCA]] to i8*
12; CHECK:      vector.body:
13; CHECK:        call void @llvm.lifetime.end.p0i8(i64 4096, i8* [[BC]])
14; CHECK:        store <vscale x 2 x i32>
15; CHECK:        call void @llvm.lifetime.start.p0i8(i64 4096, i8* [[BC]])
16
17entry:
18  %arr = alloca [1024 x i32], align 16
19  %0 = bitcast [1024 x i32]* %arr to i8*
20  call void @llvm.lifetime.start.p0i8(i64 4096, i8* %0) #1
21  br label %for.body
22
23for.body:
24  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
25  call void @llvm.lifetime.end.p0i8(i64 4096, i8* %0) #1
26  %arrayidx = getelementptr inbounds i32, i32* %d, i64 %indvars.iv
27  %1 = load i32, i32* %arrayidx, align 8
28  store i32 100, i32* %arrayidx, align 8
29  call void @llvm.lifetime.start.p0i8(i64 4096, i8* %0) #1
30  %indvars.iv.next = add i64 %indvars.iv, 1
31  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
32  %exitcond = icmp ne i32 %lftr.wideiv, 128
33  br i1 %exitcond, label %for.body, label %for.end, !llvm.loop !0
34
35for.end:
36  call void @llvm.lifetime.end.p0i8(i64 4096, i8* %0) #1
37  ret void
38}
39
40; CHECK-LABEL: @testloopvariant(
41; CHECK:      entry:
42; CHECK:        [[ALLOCA:%.*]] = alloca [1024 x i32], align 16
43; CHECK:      vector.ph:
44; CHECK:        [[TMP1:%.*]] = insertelement <vscale x 2 x [1024 x i32]*> poison, [1024 x i32]* %arr, i32 0
45; CHECK-NEXT:   [[SPLAT_ALLOCA:%.*]] = shufflevector <vscale x 2 x [1024 x i32]*> [[TMP1]], <vscale x 2 x [1024 x i32]*> poison, <vscale x 2 x i32> zeroinitializer
46; CHECK:      vector.body:
47; CHECK:        [[BC_ALLOCA:%.*]] = bitcast <vscale x 2 x [1024 x i32]*> [[SPLAT_ALLOCA]] to <vscale x 2 x i8*>
48; CHECK-NEXT:   [[ONE_LIFETIME:%.*]] = extractelement <vscale x 2 x i8*> [[BC_ALLOCA]], i32 0
49; CHECK-NEXT:   call void @llvm.lifetime.end.p0i8(i64 4096, i8* [[ONE_LIFETIME]])
50; CHECK:        store <vscale x 2 x i32>
51; CHECK-NEXT:   call void @llvm.lifetime.start.p0i8(i64 4096, i8* [[ONE_LIFETIME]])
52
53define void @testloopvariant(i32 *%d) {
54entry:
55  %arr = alloca [1024 x i32], align 16
56  br label %for.body
57
58for.body:
59  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
60  %0 = getelementptr [1024 x i32], [1024 x i32]* %arr, i32 0, i64 %indvars.iv
61  %1 = bitcast [1024 x i32]* %arr to i8*
62  call void @llvm.lifetime.end.p0i8(i64 4096, i8* %1) #1
63  %arrayidx = getelementptr inbounds i32, i32* %d, i64 %indvars.iv
64  %2 = load i32, i32* %arrayidx, align 8
65  store i32 100, i32* %arrayidx, align 8
66  call void @llvm.lifetime.start.p0i8(i64 4096, i8* %1) #1
67  %indvars.iv.next = add i64 %indvars.iv, 1
68  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
69  %exitcond = icmp ne i32 %lftr.wideiv, 128
70  br i1 %exitcond, label %for.body, label %for.end, !llvm.loop !0
71
72for.end:
73  ret void
74}
75
76declare void @llvm.lifetime.start.p0i8(i64, i8* nocapture) #1
77
78declare void @llvm.lifetime.end.p0i8(i64, i8* nocapture) #1
79
80!0 = distinct !{!0, !1}
81!1 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
82