1; RUN: opt -S -basic-aa -loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -dce -instcombine < %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@uf = common global [100 x i32] zeroinitializer, align 16
6@xi = common global [100 x i32] zeroinitializer, align 16
7@q = common global [100 x i32] zeroinitializer, align 16
8
9; PR16455
10
11
12; Due to a bug in the way we handled reverse induction stores we would generate
13; a shuffle too many.
14
15define void @t()  {
16entry:
17  br label %for.body
18
19; CHECK-LABEL: @t(
20; CHECK: vector.body:
21; CHECK: [[VAR1:%[a-zA-Z0-9.]+]] = load <4 x i32>
22; CHECK: [[VAR2:%[a-zA-Z0-9.]+]] = load <4 x i32>
23; CHECK: [[VAR3:%[a-zA-Z0-9]+]] = add nsw <4 x i32> [[VAR2]], [[VAR1]]
24; CHECK: store <4 x i32> [[VAR3]]
25; CHECK: [[VAR4:%[a-zA-Z0-9.]+]] = load <4 x i32>
26; CHECK: add nsw <4 x i32> [[VAR3]], [[VAR4]]
27; CHECK-NOT: shufflevector
28
29for.body:
30  %indvars.iv = phi i64 [ 93, %entry ], [ %indvars.iv.next, %for.body ]
31  %0 = add i64 %indvars.iv, 1
32  %arrayidx = getelementptr inbounds [100 x i32], [100 x i32]* @uf, i64 0, i64 %0
33  %arrayidx3 = getelementptr inbounds [100 x i32], [100 x i32]* @xi, i64 0, i64 %0
34  %1 = load i32, i32* %arrayidx3, align 4
35  %2 = load i32, i32* %arrayidx, align 4
36  %add4 = add nsw i32 %2, %1
37  store i32 %add4, i32* %arrayidx, align 4
38  %arrayidx7 = getelementptr inbounds [100 x i32], [100 x i32]* @q, i64 0, i64 %0
39  %3 = load i32, i32* %arrayidx7, align 4
40  %add8 = add nsw i32 %add4, %3
41  store i32 %add8, i32* %arrayidx, align 4
42  %indvars.iv.next = add i64 %indvars.iv, -1
43  %4 = trunc i64 %indvars.iv.next to i32
44  %cmp = icmp ugt i32 %4, 2
45  br i1 %cmp, label %for.body, label %for.end
46
47for.end:
48  ret void
49}
50