1 ; RUN: opt < %s -basicaa -loop-interchange -S | FileCheck %s
2 ;; We test the complete .ll for adjustment in outer loop header/latch and inner loop header/latch.
4 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
5 target triple = "x86_64-unknown-linux-gnu"
7 @A = common global [100 x [100 x i32]] zeroinitializer
8 @B = common global [100 x i32] zeroinitializer
10 ;; for(int i=0;i<N;i++)
11 ;; for(int j=1;j<N;j++)
12 ;; A[j][i] = A[j][i]+k;
14 define void @interchange_01(i32 %k, i32 %N) {
16 %cmp21 = icmp sgt i32 %N, 0
17 br i1 %cmp21, label %for.cond1.preheader.lr.ph, label %for.end12
19 for.cond1.preheader.lr.ph:
20 %cmp219 = icmp sgt i32 %N, 1
22 br label %for.cond1.preheader
25 %indvars.iv23 = phi i64 [ 0, %for.cond1.preheader.lr.ph ], [ %indvars.iv.next24, %for.inc10 ]
26 br i1 %cmp219, label %for.body3, label %for.inc10
29 %indvars.iv = phi i64 [ %indvars.iv.next, %for.body3 ], [ 1, %for.cond1.preheader ]
30 %arrayidx5 = getelementptr inbounds [100 x [100 x i32]], [100 x [100 x i32]]* @A, i64 0, i64 %indvars.iv, i64 %indvars.iv23
31 %1 = load i32, i32* %arrayidx5
32 %add = add nsw i32 %1, %k
33 store i32 %add, i32* %arrayidx5
34 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
35 %lftr.wideiv = trunc i64 %indvars.iv to i32
36 %exitcond = icmp eq i32 %lftr.wideiv, %0
37 br i1 %exitcond, label %for.inc10, label %for.body3
40 %indvars.iv.next24 = add nuw nsw i64 %indvars.iv23, 1
41 %lftr.wideiv25 = trunc i64 %indvars.iv23 to i32
42 %exitcond26 = icmp eq i32 %lftr.wideiv25, %0
43 br i1 %exitcond26, label %for.end12, label %for.cond1.preheader
49 ; CHECK-LABEL: @interchange_01
51 ; CHECK: %cmp21 = icmp sgt i32 %N, 0
52 ; CHECK: br i1 %cmp21, label %for.body3.preheader, label %for.end12
53 ; CHECK: for.cond1.preheader.lr.ph:
54 ; CHECK: br label %for.cond1.preheader
55 ; CHECK: for.cond1.preheader:
56 ; CHECK: %indvars.iv23 = phi i64 [ 0, %for.cond1.preheader.lr.ph ], [ %indvars.iv.next24, %for.inc10 ]
57 ; CHECK: br i1 %cmp219, label %for.body3.split1, label %for.end12.loopexit
58 ; CHECK: for.body3.preheader:
59 ; CHECK: %cmp219 = icmp sgt i32 %N, 1
60 ; CHECK: %0 = add i32 %N, -1
61 ; CHECK: br label %for.body3
63 ; CHECK: %indvars.iv = phi i64 [ %indvars.iv.next, %for.body3.split ], [ 1, %for.body3.preheader ]
64 ; CHECK: br label %for.cond1.preheader.lr.ph
65 ; CHECK: for.body3.split1:
66 ; CHECK: %arrayidx5 = getelementptr inbounds [100 x [100 x i32]], [100 x [100 x i32]]* @A, i64 0, i64 %indvars.iv, i64 %indvars.iv23
67 ; CHECK: %1 = load i32, i32* %arrayidx5
68 ; CHECK: %add = add nsw i32 %1, %k
69 ; CHECK: store i32 %add, i32* %arrayidx5
70 ; CHECK: br label %for.inc10.loopexit
71 ; CHECK: for.body3.split:
72 ; CHECK: %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
73 ; CHECK: %lftr.wideiv = trunc i64 %indvars.iv to i32
74 ; CHECK: %exitcond = icmp eq i32 %lftr.wideiv, %0
75 ; CHECK: br i1 %exitcond, label %for.end12.loopexit, label %for.body3
76 ; CHECK: for.inc10.loopexit:
77 ; CHECK: br label %for.inc10
79 ; CHECK: %indvars.iv.next24 = add nuw nsw i64 %indvars.iv23, 1
80 ; CHECK: %lftr.wideiv25 = trunc i64 %indvars.iv23 to i32
81 ; CHECK: %exitcond26 = icmp eq i32 %lftr.wideiv25, %0
82 ; CHECK: br i1 %exitcond26, label %for.body3.split, label %for.cond1.preheader
83 ; CHECK: for.end12.loopexit:
84 ; CHECK: br label %for.end12