1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; RUN: opt -p loop-vectorize -S %s | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
; Test case for https://github.com/llvm/llvm-project/issues/106248.
define i64 @test_foldable_live_in_via_scev() {
; CHECK-LABEL: define i64 @test_foldable_live_in_via_scev() {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: [[CONV:%.*]] = zext i16 -6 to i64
; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[CONV]], -65528
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ splat (i64 1), %[[VECTOR_PH]] ], [ [[TMP0:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i64> [ splat (i64 1), %[[VECTOR_PH]] ], [ [[TMP1:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0]] = mul <2 x i64> [[VEC_PHI]], splat (i64 2)
; CHECK-NEXT: [[TMP1]] = mul <2 x i64> [[VEC_PHI1]], splat (i64 2)
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 96
; CHECK-NEXT: br i1 [[TMP2]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[BIN_RDX:%.*]] = mul <2 x i64> [[TMP1]], [[TMP0]]
; CHECK-NEXT: [[TMP3:%.*]] = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> [[BIN_RDX]])
; CHECK-NEXT: br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ 97, %[[MIDDLE_BLOCK]] ], [ 1, %[[ENTRY]] ]
; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i64 [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ 1, %[[ENTRY]] ]
; CHECK-NEXT: br label %[[LOOP:.*]]
; CHECK: [[LOOP]]:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT: [[RED:%.*]] = phi i64 [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ], [ [[MUL:%.*]], %[[LOOP]] ]
; CHECK-NEXT: [[MUL]] = mul nsw i64 [[RED]], [[ADD]]
; CHECK-NEXT: [[IV_NEXT]] = add nsw i32 [[IV]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i32 [[IV_NEXT]], 100
; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: [[RET:%.*]] = phi i64 [ [[MUL]], %[[LOOP]] ], [ [[TMP3]], %[[MIDDLE_BLOCK]] ]
; CHECK-NEXT: ret i64 [[RET]]
;
entry:
%conv = zext i16 -6 to i64
%add = add nsw i64 %conv, -65528
br label %loop
loop:
%iv = phi i32 [ 1, %entry ], [ %iv.next, %loop ]
%red = phi i64 [ 1, %entry ], [ %mul, %loop ]
%mul = mul nsw i64 %red, %add
%iv.next = add nsw i32 %iv, 1
%ec = icmp eq i32 %iv.next, 100
br i1 %ec, label %exit, label %loop
exit:
%ret = phi i64 [ %mul, %loop ]
ret i64 %ret
}
; Test case for https://github.com/llvm/llvm-project/issues/109528.
define i64 @second_lshr_operand_zero_via_scev() {
; CHECK-LABEL: define i64 @second_lshr_operand_zero_via_scev() {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: [[EXT_0:%.*]] = sext i8 0 to i32
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP10:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i64> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP11:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND2:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[STEP_ADD:%.*]] = add <2 x i64> [[VEC_IND]], splat (i64 2)
; CHECK-NEXT: [[STEP_ADD4:%.*]] = add <2 x i32> [[VEC_IND2]], splat (i32 2)
; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <2 x i64> [[VEC_IND]], zeroinitializer
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <2 x i64> [[STEP_ADD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = and <2 x i64> [[VEC_IND]], zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = and <2 x i64> [[STEP_ADD]], zeroinitializer
; CHECK-NEXT: [[TMP4:%.*]] = lshr <2 x i32> [[VEC_IND2]], zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = lshr <2 x i32> [[STEP_ADD4]], zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = zext <2 x i32> [[TMP4]] to <2 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = zext <2 x i32> [[TMP5]] to <2 x i64>
; CHECK-NEXT: [[TMP8:%.*]] = select <2 x i1> [[TMP0]], <2 x i64> [[TMP2]], <2 x i64> [[TMP6]]
; CHECK-NEXT: [[TMP9:%.*]] = select <2 x i1> [[TMP1]], <2 x i64> [[TMP3]], <2 x i64> [[TMP7]]
; CHECK-NEXT: [[TMP10]] = or <2 x i64> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT: [[TMP11]] = or <2 x i64> [[TMP9]], [[VEC_PHI1]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i64> [[STEP_ADD]], splat (i64 2)
; CHECK-NEXT: [[VEC_IND_NEXT3]] = add <2 x i32> [[STEP_ADD4]], splat (i32 2)
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[BIN_RDX:%.*]] = or <2 x i64> [[TMP11]], [[TMP10]]
; CHECK-NEXT: [[TMP13:%.*]] = call i64 @llvm.vector.reduce.or.v2i64(<2 x i64> [[BIN_RDX]])
; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i64 [ [[TMP13]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: br label %[[LOOPS:.*]]
; CHECK: [[LOOPS]]:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOPS]] ]
; CHECK-NEXT: [[RED:%.*]] = phi i64 [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ], [ [[RED_NEXT:%.*]], %[[LOOPS]] ]
; CHECK-NEXT: [[C:%.*]] = icmp eq i64 [[IV]], 0
; CHECK-NEXT: [[AND:%.*]] = and i64 [[IV]], 0
; CHECK-NEXT: [[TMP14:%.*]] = trunc i64 [[IV]] to i32
; CHECK-NEXT: [[SHR:%.*]] = lshr i32 [[TMP14]], [[EXT_0]]
; CHECK-NEXT: [[CONV_1:%.*]] = zext i32 [[SHR]] to i64
; CHECK-NEXT: [[RED_NEXT_V:%.*]] = select i1 [[C]], i64 [[AND]], i64 [[CONV_1]]
; CHECK-NEXT: [[RED_NEXT]] = or i64 [[RED_NEXT_V]], [[RED]]
; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOPS]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: [[RES:%.*]] = phi i64 [ [[RED_NEXT]], %[[LOOPS]] ], [ [[TMP13]], %[[MIDDLE_BLOCK]] ]
; CHECK-NEXT: ret i64 [[RES]]
;
entry:
%ext.0 = sext i8 0 to i32
br label %loops
loops:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loops ]
%red = phi i64 [ 0, %entry ], [ %red.next, %loops ]
%c = icmp eq i64 %iv, 0
%and = and i64 %iv, 0
%0 = trunc i64 %iv to i32
%shr = lshr i32 %0, %ext.0
%conv.1 = zext i32 %shr to i64
%red.next.v = select i1 %c, i64 %and, i64 %conv.1
%red.next = or i64 %red.next.v, %red
%iv.next = add i64 %iv, 1
%ec = icmp eq i64 %iv.next, 1000
br i1 %ec, label %exit, label %loops
exit:
%res = phi i64 [ %red.next, %loops ]
ret i64 %res
}
;.
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
;.
|
