1 # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
2 # RUN: llc -mtriple=thumbv8.1m.main-none-eabi -mcpu=cortex-m85 -mattr=+use-mipipeliner -run-pass=pipeliner --pipeliner-force-issue-width=10 -o - %s | FileCheck %s --check-prefix=CHECK
5 define hidden float @dot(ptr nocapture noundef readonly %a, ptr nocapture noundef readonly %b, i32 noundef %sz) local_unnamed_addr #0 {
7 %cmp8 = icmp sgt i32 %sz, 0
8 br i1 %cmp8, label %for.body.preheader, label %for.end
10 for.body.preheader: ; preds = %entry
11 %scevgep = getelementptr float, ptr %b, i32 -1
12 %scevgep4 = getelementptr float, ptr %a, i32 -1
15 for.body: ; preds = %for.body.preheader, %for.body
16 %lsr.iv5 = phi ptr [ %scevgep4, %for.body.preheader ], [ %scevgep6, %for.body ]
17 %lsr.iv1 = phi ptr [ %scevgep, %for.body.preheader ], [ %scevgep2, %for.body ]
18 %lsr.iv = phi i32 [ %sz, %for.body.preheader ], [ %lsr.iv.next, %for.body ]
19 %sum.010 = phi float [ %add, %for.body ], [ 0.000000e+00, %for.body.preheader ]
20 %scevgep7 = getelementptr float, ptr %lsr.iv5, i32 1
21 %0 = load float, ptr %scevgep7, align 4
22 %scevgep3 = getelementptr float, ptr %lsr.iv1, i32 1
23 %1 = load float, ptr %scevgep3, align 4
24 %mul = fmul fast float %1, %0
25 %add = fadd fast float %mul, %sum.010
26 %lsr.iv.next = add i32 %lsr.iv, -1
27 %scevgep2 = getelementptr float, ptr %lsr.iv1, i32 1
28 %scevgep6 = getelementptr float, ptr %lsr.iv5, i32 1
29 %exitcond.not = icmp eq i32 %lsr.iv.next, 0
30 br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
32 for.end: ; preds = %for.body, %entry
33 %sum.0.lcssa = phi float [ 0.000000e+00, %entry ], [ %add, %for.body ]
34 ret float %sum.0.lcssa
37 !0 = distinct !{!0, !1, !2, !3}
38 !1 = !{!"llvm.loop.mustprogress"}
39 !2 = !{!"llvm.loop.unroll.disable"}
40 !3 = !{!"llvm.loop.pipeline.initiationinterval", i32 3}
46 tracksRegLiveness: true
49 value: 'float 0.000000e+00'
51 isTargetSpecific: false
53 ; CHECK-LABEL: name: dot
55 ; CHECK-NEXT: successors: %bb.2(0x50000000), %bb.1(0x30000000)
56 ; CHECK-NEXT: liveins: $r0, $r1, $r2
58 ; CHECK-NEXT: [[COPY:%[0-9]+]]:gprlr = COPY $r2
59 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gprnopc = COPY $r1
60 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gprnopc = COPY $r0
61 ; CHECK-NEXT: t2CMPri [[COPY]], 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
62 ; CHECK-NEXT: t2Bcc %bb.2, 10 /* CC::ge */, $cpsr
65 ; CHECK-NEXT: successors: %bb.4(0x80000000)
67 ; CHECK-NEXT: [[VLDRS:%[0-9]+]]:spr = VLDRS %const.0, 0, 14 /* CC::al */, $noreg :: (load (s32) from constant-pool)
68 ; CHECK-NEXT: t2B %bb.4, 14 /* CC::al */, $noreg
70 ; CHECK-NEXT: bb.2.for.body.preheader:
71 ; CHECK-NEXT: successors: %bb.5(0x80000000)
73 ; CHECK-NEXT: [[t2SUBri:%[0-9]+]]:rgpr = t2SUBri [[COPY1]], 4, 14 /* CC::al */, $noreg, $noreg
74 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:gprnopc = COPY [[t2SUBri]]
75 ; CHECK-NEXT: [[t2SUBri1:%[0-9]+]]:rgpr = t2SUBri [[COPY2]], 4, 14 /* CC::al */, $noreg, $noreg
76 ; CHECK-NEXT: [[VLDRS1:%[0-9]+]]:spr = VLDRS %const.0, 0, 14 /* CC::al */, $noreg :: (load (s32) from constant-pool)
77 ; CHECK-NEXT: [[t2DoLoopStart:%[0-9]+]]:gprlr = t2DoLoopStart [[COPY]]
78 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:gprnopc = COPY [[t2SUBri1]]
80 ; CHECK-NEXT: bb.5.for.body:
81 ; CHECK-NEXT: successors: %bb.6(0x80000000), %bb.7(0x00000000)
83 ; CHECK-NEXT: [[t2ADDri:%[0-9]+]]:rgpr = t2ADDri [[COPY4]], 4, 14 /* CC::al */, $noreg, $noreg
84 ; CHECK-NEXT: [[VLDRS2:%[0-9]+]]:spr = VLDRS [[COPY4]], 1, 14 /* CC::al */, $noreg :: (load (s32) from %ir.scevgep7)
85 ; CHECK-NEXT: [[t2ADDri1:%[0-9]+]]:rgpr = t2ADDri [[COPY3]], 4, 14 /* CC::al */, $noreg, $noreg
86 ; CHECK-NEXT: [[VLDRS3:%[0-9]+]]:spr = VLDRS [[COPY3]], 1, 14 /* CC::al */, $noreg :: (load (s32) from %ir.scevgep3)
87 ; CHECK-NEXT: [[VMULS:%[0-9]+]]:spr = nnan ninf nsz arcp contract afn reassoc VMULS [[VLDRS3]], [[VLDRS2]], 14 /* CC::al */, $noreg
88 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:gprlr = COPY [[t2DoLoopStart]]
89 ; CHECK-NEXT: [[t2LoopDec:%[0-9]+]]:gprlr = t2LoopDec [[COPY5]], 1
90 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:gpr = COPY [[t2LoopDec]]
91 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:gpr = COPY [[t2ADDri1]]
92 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:gpr = COPY [[t2ADDri]]
93 ; CHECK-NEXT: t2CMPri [[t2LoopDec]], 0, 14 /* CC::al */, $noreg, implicit-def $cpsr
94 ; CHECK-NEXT: t2Bcc %bb.7, 0 /* CC::eq */, $cpsr
95 ; CHECK-NEXT: t2B %bb.6, 14 /* CC::al */, $noreg
97 ; CHECK-NEXT: bb.6.for.body:
98 ; CHECK-NEXT: successors: %bb.7(0x04000000), %bb.6(0x7c000000)
100 ; CHECK-NEXT: [[PHI:%[0-9]+]]:gprnopc = PHI [[COPY8]], %bb.5, %43, %bb.6
101 ; CHECK-NEXT: [[PHI1:%[0-9]+]]:gprnopc = PHI [[COPY7]], %bb.5, %44, %bb.6
102 ; CHECK-NEXT: [[PHI2:%[0-9]+]]:gpr = PHI [[COPY6]], %bb.5, %47, %bb.6
103 ; CHECK-NEXT: [[PHI3:%[0-9]+]]:spr = PHI [[VLDRS1]], %bb.5, %46, %bb.6
104 ; CHECK-NEXT: [[PHI4:%[0-9]+]]:spr = PHI [[VMULS]], %bb.5, %45, %bb.6
105 ; CHECK-NEXT: [[VLDRS4:%[0-9]+]]:spr = VLDRS [[PHI1]], 1, 14 /* CC::al */, $noreg :: (load unknown-size from %ir.scevgep3, align 4)
106 ; CHECK-NEXT: [[VLDRS5:%[0-9]+]]:spr = VLDRS [[PHI]], 1, 14 /* CC::al */, $noreg :: (load unknown-size from %ir.scevgep7, align 4)
107 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:gprlr = COPY [[PHI2]]
108 ; CHECK-NEXT: [[t2LoopDec1:%[0-9]+]]:gprlr = t2LoopDec [[COPY9]], 1
109 ; CHECK-NEXT: [[t2ADDri2:%[0-9]+]]:rgpr = t2ADDri [[PHI]], 4, 14 /* CC::al */, $noreg, $noreg
110 ; CHECK-NEXT: [[t2ADDri3:%[0-9]+]]:rgpr = t2ADDri [[PHI1]], 4, 14 /* CC::al */, $noreg, $noreg
111 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:gpr = COPY [[t2ADDri2]]
112 ; CHECK-NEXT: [[COPY11:%[0-9]+]]:gpr = COPY [[t2ADDri3]]
113 ; CHECK-NEXT: [[VMULS1:%[0-9]+]]:spr = nnan ninf nsz arcp contract afn reassoc VMULS [[VLDRS4]], [[VLDRS5]], 14 /* CC::al */, $noreg
114 ; CHECK-NEXT: [[VADDS:%[0-9]+]]:spr = nnan ninf nsz arcp contract afn reassoc VADDS [[PHI4]], [[PHI3]], 14 /* CC::al */, $noreg
115 ; CHECK-NEXT: [[COPY12:%[0-9]+]]:gpr = COPY [[t2LoopDec1]]
116 ; CHECK-NEXT: t2LoopEnd [[t2LoopDec1]], %bb.6, implicit-def $cpsr
117 ; CHECK-NEXT: t2B %bb.7, 14 /* CC::al */, $noreg
120 ; CHECK-NEXT: successors: %bb.4(0x80000000)
122 ; CHECK-NEXT: [[PHI5:%[0-9]+]]:spr = PHI [[VLDRS1]], %bb.5, [[VADDS]], %bb.6
123 ; CHECK-NEXT: [[PHI6:%[0-9]+]]:spr = PHI [[VMULS]], %bb.5, [[VMULS1]], %bb.6
124 ; CHECK-NEXT: [[VADDS1:%[0-9]+]]:spr = nnan ninf nsz arcp contract afn reassoc VADDS [[PHI6]], [[PHI5]], 14 /* CC::al */, $noreg
125 ; CHECK-NEXT: t2B %bb.4, 14 /* CC::al */, $noreg
127 ; CHECK-NEXT: bb.4.for.end:
128 ; CHECK-NEXT: [[PHI7:%[0-9]+]]:spr = PHI [[VLDRS]], %bb.1, [[VADDS1]], %bb.7
129 ; CHECK-NEXT: [[VMOVRS:%[0-9]+]]:gpr = VMOVRS [[PHI7]], 14 /* CC::al */, $noreg
130 ; CHECK-NEXT: $r0 = COPY [[VMOVRS]]
131 ; CHECK-NEXT: tBX_RET 14 /* CC::al */, $noreg, implicit $r0
133 successors: %bb.1(0x50000000), %bb.4(0x30000000)
134 liveins: $r0, $r1, $r2
137 %12:gprnopc = COPY $r1
138 %11:gprnopc = COPY $r0
139 t2CMPri %13, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
140 t2Bcc %bb.1, 10 /* CC::ge */, $cpsr
143 successors: %bb.3(0x80000000)
145 %14:spr = VLDRS %const.0, 0, 14 /* CC::al */, $noreg :: (load (s32) from constant-pool)
146 t2B %bb.3, 14 /* CC::al */, $noreg
148 bb.1.for.body.preheader:
149 successors: %bb.2(0x80000000)
151 %16:rgpr = t2SUBri %12, 4, 14 /* CC::al */, $noreg, $noreg
153 %17:rgpr = t2SUBri %11, 4, 14 /* CC::al */, $noreg, $noreg
154 %15:spr = VLDRS %const.0, 0, 14 /* CC::al */, $noreg :: (load (s32) from constant-pool)
155 %44:gprlr = t2DoLoopStart %13
159 successors: %bb.3(0x04000000), %bb.2(0x7c000000)
161 %2:gprnopc = PHI %1, %bb.1, %9, %bb.2
162 %3:gprnopc = PHI %0, %bb.1, %8, %bb.2
163 %4:gpr = PHI %44, %bb.1, %7, %bb.2
164 %5:spr = PHI %15, %bb.1, %6, %bb.2
165 %18:rgpr = t2ADDri %2, 4, 14 /* CC::al */, $noreg, $noreg
166 %19:spr = VLDRS %2, 1, 14 /* CC::al */, $noreg :: (load (s32) from %ir.scevgep7)
167 %20:rgpr = t2ADDri %3, 4, 14 /* CC::al */, $noreg, $noreg
168 %21:spr = VLDRS %3, 1, 14 /* CC::al */, $noreg :: (load (s32) from %ir.scevgep3)
169 %22:spr = nnan ninf nsz arcp contract afn reassoc VMULS killed %21, killed %19, 14 /* CC::al */, $noreg
170 %6:spr = nnan ninf nsz arcp contract afn reassoc VADDS killed %22, %5, 14 /* CC::al */, $noreg
172 %23:gprlr = t2LoopDec %42:gprlr, 1
176 t2LoopEnd %23:gprlr, %bb.2, implicit-def dead $cpsr
177 t2B %bb.3, 14 /* CC::al */, $noreg
180 %10:spr = PHI %14, %bb.4, %6, %bb.2
181 %24:gpr = VMOVRS %10, 14 /* CC::al */, $noreg
183 tBX_RET 14 /* CC::al */, $noreg, implicit $r0