1; RUN: opt < %s -basic-aa -loop-interchange -pass-remarks-missed='loop-interchange' -pass-remarks-output=%t -S \
2; RUN:     -verify-dom-info -verify-loop-info -verify-loop-lcssa 2>&1 | FileCheck -check-prefix=IR %s
3; RUN: FileCheck --input-file=%t %s
4
5; Inner loop only reductions are not supported currently. See discussion at
6; D53027 for more information on the required checks.
7
8@A = common global [500 x [500 x i32]] zeroinitializer
9@X = common global i32 0
10@B = common global [500 x [500 x i32]] zeroinitializer
11@Y = common global i32 0
12
13;; global X
14
15;;  for( int i=1;i<N;i++)
16;;    for( int j=1;j<N;j++)
17;;      X+=A[j][i];
18
19; CHECK: --- !Missed
20; CHECK-NEXT: Pass:            loop-interchange
21; CHECK-NEXT: Name:            UnsupportedPHI
22; CHECK-NEXT: Function:        reduction_01
23
24; IR-LABEL: @reduction_01(
25; IR-NOT: split
26
27define void @reduction_01(i32 %N) {
28entry:
29  %cmp16 = icmp sgt i32 %N, 1
30  br i1 %cmp16, label %for.body3.lr.ph, label %for.end8
31
32for.body3.lr.ph:                                  ; preds = %for.cond1.for.inc6_crit_edge, %entry
33  %indvars.iv18 = phi i64 [ %indvars.iv.next19, %for.cond1.for.inc6_crit_edge ], [ 1, %entry ]
34  %X.promoted = load i32, i32* @X
35  br label %for.body3
36
37for.body3:                                        ; preds = %for.body3, %for.body3.lr.ph
38  %indvars.iv = phi i64 [ 1, %for.body3.lr.ph ], [ %indvars.iv.next, %for.body3 ]
39  %add15 = phi i32 [ %X.promoted, %for.body3.lr.ph ], [ %add, %for.body3 ]
40  %arrayidx5 = getelementptr inbounds [500 x [500 x i32]], [500 x [500 x i32]]* @A, i64 0, i64 %indvars.iv, i64 %indvars.iv18
41  %0 = load i32, i32* %arrayidx5
42  %add = add nsw i32 %add15, %0
43  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
44  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
45  %exitcond = icmp eq i32 %lftr.wideiv, %N
46  br i1 %exitcond, label %for.cond1.for.inc6_crit_edge, label %for.body3
47
48for.cond1.for.inc6_crit_edge:                     ; preds = %for.body3
49  %add.lcssa = phi i32 [ %add, %for.body3 ]
50  store i32 %add.lcssa, i32* @X
51  %indvars.iv.next19 = add nuw nsw i64 %indvars.iv18, 1
52  %lftr.wideiv20 = trunc i64 %indvars.iv.next19 to i32
53  %exitcond21 = icmp eq i32 %lftr.wideiv20, %N
54  br i1 %exitcond21, label %for.end8, label %for.body3.lr.ph
55
56for.end8:                                         ; preds = %for.cond1.for.inc6_crit_edge, %entry
57  ret void
58}
59
60;; Not tightly nested. Do not interchange.
61;;  for( int i=1;i<N;i++)
62;;    for( int j=1;j<N;j++) {
63;;      for( int k=1;k<N;k++) {
64;;        X+=A[k][j];
65;;      }
66;;      Y+=B[j][i];
67;;    }
68
69;; Not tightly nested. Do not interchange.
70;; Not interchanged hence the phi's in the inner loop will not be split.
71
72; CHECK: --- !Missed
73; CHECK-NEXT: Pass:            loop-interchange
74; CHECK-NEXT: Name:            UnsupportedPHIOuter
75; CHECK-NEXT: Function:        reduction_03
76
77; IR-LABEL: @reduction_03(
78; IR-NOT: split
79
80define void @reduction_03(i32 %N) {
81entry:
82  %cmp35 = icmp sgt i32 %N, 1
83  br i1 %cmp35, label %for.cond4.preheader.lr.ph, label %for.end19
84
85for.cond4.preheader.lr.ph:                        ; preds = %for.cond1.for.inc17_crit_edge, %entry
86  %indvars.iv41 = phi i64 [ %indvars.iv.next42, %for.cond1.for.inc17_crit_edge ], [ 1, %entry ]
87  %Y.promoted = load i32, i32* @Y
88  br label %for.body6.lr.ph
89
90for.body6.lr.ph:                                  ; preds = %for.cond4.for.end_crit_edge, %for.cond4.preheader.lr.ph
91  %indvars.iv37 = phi i64 [ 1, %for.cond4.preheader.lr.ph ], [ %indvars.iv.next38, %for.cond4.for.end_crit_edge ]
92  %add1334 = phi i32 [ %Y.promoted, %for.cond4.preheader.lr.ph ], [ %add13, %for.cond4.for.end_crit_edge ]
93  %X.promoted = load i32, i32* @X
94  br label %for.body6
95
96for.body6:                                        ; preds = %for.body6, %for.body6.lr.ph
97  %indvars.iv = phi i64 [ 1, %for.body6.lr.ph ], [ %indvars.iv.next, %for.body6 ]
98  %arrayidx8 = getelementptr inbounds [500 x [500 x i32]], [500 x [500 x i32]]* @A, i64 0, i64 %indvars.iv, i64 %indvars.iv37
99  %0 = load i32, i32* %arrayidx8
100  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
101  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
102  %exitcond = icmp eq i32 %lftr.wideiv, %N
103  br i1 %exitcond, label %for.cond4.for.end_crit_edge, label %for.body6
104
105for.cond4.for.end_crit_edge:                      ; preds = %for.body6
106  %arrayidx12 = getelementptr inbounds [500 x [500 x i32]], [500 x [500 x i32]]* @B, i64 0, i64 %indvars.iv37, i64 %indvars.iv41
107  %1 = load i32, i32* %arrayidx12
108  %add13 = add nsw i32 %add1334, %1
109  %indvars.iv.next38 = add nuw nsw i64 %indvars.iv37, 1
110  %lftr.wideiv39 = trunc i64 %indvars.iv.next38 to i32
111  %exitcond40 = icmp eq i32 %lftr.wideiv39, %N
112  br i1 %exitcond40, label %for.cond1.for.inc17_crit_edge, label %for.body6.lr.ph
113
114for.cond1.for.inc17_crit_edge:                    ; preds = %for.cond4.for.end_crit_edge
115  %add13.lcssa = phi i32 [ %add13, %for.cond4.for.end_crit_edge ]
116  store i32 %add13.lcssa, i32* @Y
117  %indvars.iv.next42 = add nuw nsw i64 %indvars.iv41, 1
118  %lftr.wideiv43 = trunc i64 %indvars.iv.next42 to i32
119  %exitcond44 = icmp eq i32 %lftr.wideiv43, %N
120  br i1 %exitcond44, label %for.end19, label %for.cond4.preheader.lr.ph
121
122for.end19:                                        ; preds = %for.cond1.for.inc17_crit_edge, %entry
123  ret void
124}
125