• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
2# RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve.fp -run-pass=arm-low-overhead-loops --verify-machineinstrs %s -o - | FileCheck %s
3
4--- |
5  define dso_local arm_aapcs_vfpcc void @remove_mov_lr_chain(float* nocapture readonly %pSrc, float* nocapture %pDst, i32 %blockSize) #0 {
6  entry:
7    %cmp5 = icmp eq i32 %blockSize, 0
8    br i1 %cmp5, label %while.end, label %while.body.preheader
9
10  while.body.preheader:                             ; preds = %entry
11    %min.iters.check = icmp ult i32 %blockSize, 4
12    br i1 %min.iters.check, label %while.body.preheader19, label %vector.memcheck
13
14  vector.memcheck:                                  ; preds = %while.body.preheader
15    %scevgep = getelementptr float, float* %pDst, i32 %blockSize
16    %scevgep12 = getelementptr float, float* %pSrc, i32 %blockSize
17    %bound0 = icmp ugt float* %scevgep12, %pDst
18    %bound1 = icmp ugt float* %scevgep, %pSrc
19    %found.conflict = and i1 %bound0, %bound1
20    %0 = lshr i32 %blockSize, 2
21    %1 = shl nuw i32 %0, 2
22    %2 = add i32 %1, -4
23    %3 = lshr i32 %2, 2
24    %4 = add nuw nsw i32 %3, 1
25    br i1 %found.conflict, label %while.body.preheader19, label %vector.ph
26
27  vector.ph:                                        ; preds = %vector.memcheck
28    %n.vec = and i32 %blockSize, -4
29    %ind.end = sub i32 %blockSize, %n.vec
30    %ind.end15 = getelementptr float, float* %pSrc, i32 %n.vec
31    %ind.end17 = getelementptr float, float* %pDst, i32 %n.vec
32    %scevgep9 = getelementptr float, float* %pDst, i32 -4
33    %scevgep14 = getelementptr float, float* %pSrc, i32 -4
34    %start1 = call i32 @llvm.start.loop.iterations.i32(i32 %4)
35    br label %vector.body
36
37  vector.body:                                      ; preds = %vector.body, %vector.ph
38    %lsr.iv15 = phi float* [ %scevgep16, %vector.body ], [ %scevgep14, %vector.ph ]
39    %lsr.iv10 = phi float* [ %scevgep11, %vector.body ], [ %scevgep9, %vector.ph ]
40    %5 = phi i32 [ %start1, %vector.ph ], [ %7, %vector.body ]
41    %lsr.iv1517 = bitcast float* %lsr.iv15 to <4 x float>*
42    %lsr.iv1012 = bitcast float* %lsr.iv10 to <4 x float>*
43    %scevgep18 = getelementptr <4 x float>, <4 x float>* %lsr.iv1517, i32 1
44    %wide.load = load <4 x float>, <4 x float>* %scevgep18, align 4
45    %6 = call fast <4 x float> @llvm.fabs.v4f32(<4 x float> %wide.load)
46    %scevgep13 = getelementptr <4 x float>, <4 x float>* %lsr.iv1012, i32 1
47    store <4 x float> %6, <4 x float>* %scevgep13, align 4
48    %scevgep11 = getelementptr float, float* %lsr.iv10, i32 4
49    %scevgep16 = getelementptr float, float* %lsr.iv15, i32 4
50    %7 = call i32 @llvm.loop.decrement.reg.i32.i32.i32(i32 %5, i32 1)
51    %8 = icmp ne i32 %7, 0
52    br i1 %8, label %vector.body, label %middle.block
53
54  middle.block:                                     ; preds = %vector.body
55    %cmp.n = icmp eq i32 %n.vec, %blockSize
56    br i1 %cmp.n, label %while.end, label %while.body.preheader19
57
58  while.body.preheader19:                           ; preds = %middle.block, %vector.memcheck, %while.body.preheader
59    %blkCnt.08.ph = phi i32 [ %blockSize, %vector.memcheck ], [ %blockSize, %while.body.preheader ], [ %ind.end, %middle.block ]
60    %pSrc.addr.07.ph = phi float* [ %pSrc, %vector.memcheck ], [ %pSrc, %while.body.preheader ], [ %ind.end15, %middle.block ]
61    %pDst.addr.06.ph = phi float* [ %pDst, %vector.memcheck ], [ %pDst, %while.body.preheader ], [ %ind.end17, %middle.block ]
62    %scevgep1 = getelementptr float, float* %pSrc.addr.07.ph, i32 -1
63    %scevgep4 = getelementptr float, float* %pDst.addr.06.ph, i32 -1
64    %start2 = call i32 @llvm.start.loop.iterations.i32(i32 %blkCnt.08.ph)
65    br label %while.body
66
67  while.body:                                       ; preds = %while.body, %while.body.preheader19
68    %lsr.iv5 = phi float* [ %scevgep6, %while.body ], [ %scevgep4, %while.body.preheader19 ]
69    %lsr.iv = phi float* [ %scevgep2, %while.body ], [ %scevgep1, %while.body.preheader19 ]
70    %9 = phi i32 [ %start2, %while.body.preheader19 ], [ %12, %while.body ]
71    %scevgep3 = getelementptr float, float* %lsr.iv, i32 1
72    %scevgep7 = getelementptr float, float* %lsr.iv5, i32 1
73    %10 = load float, float* %scevgep3, align 4
74    %11 = tail call fast float @llvm.fabs.f32(float %10)
75    store float %11, float* %scevgep7, align 4
76    %scevgep2 = getelementptr float, float* %lsr.iv, i32 1
77    %scevgep6 = getelementptr float, float* %lsr.iv5, i32 1
78    %12 = call i32 @llvm.loop.decrement.reg.i32.i32.i32(i32 %9, i32 1)
79    %13 = icmp ne i32 %12, 0
80    br i1 %13, label %while.body, label %while.end
81
82  while.end:                                        ; preds = %while.body, %middle.block, %entry
83    ret void
84  }
85  declare float @llvm.fabs.f32(float)
86  declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
87  declare i32 @llvm.start.loop.iterations.i32(i32)
88  declare i32 @llvm.loop.decrement.reg.i32.i32.i32(i32, i32)
89
90...
91---
92name:            remove_mov_lr_chain
93alignment:       2
94exposesReturnsTwice: false
95legalized:       false
96regBankSelected: false
97selected:        false
98failedISel:      false
99tracksRegLiveness: true
100hasWinCFI:       false
101registers:       []
102liveins:
103  - { reg: '$r0', virtual-reg: '' }
104  - { reg: '$r1', virtual-reg: '' }
105  - { reg: '$r2', virtual-reg: '' }
106frameInfo:
107  isFrameAddressTaken: false
108  isReturnAddressTaken: false
109  hasStackMap:     false
110  hasPatchPoint:   false
111  stackSize:       16
112  offsetAdjustment: 0
113  maxAlignment:    4
114  adjustsStack:    false
115  hasCalls:        false
116  stackProtector:  ''
117  maxCallFrameSize: 0
118  cvBytesOfCalleeSavedRegisters: 0
119  hasOpaqueSPAdjustment: false
120  hasVAStart:      false
121  hasMustTailInVarArgFunc: false
122  localFrameSize:  0
123  savePoint:       ''
124  restorePoint:    ''
125fixedStack:      []
126stack:
127  - { id: 0, name: '', type: spill-slot, offset: -4, size: 4, alignment: 4,
128      stack-id: default, callee-saved-register: '$lr', callee-saved-restored: false,
129      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
130  - { id: 1, name: '', type: spill-slot, offset: -8, size: 4, alignment: 4,
131      stack-id: default, callee-saved-register: '$r7', callee-saved-restored: true,
132      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
133  - { id: 2, name: '', type: spill-slot, offset: -12, size: 4, alignment: 4,
134      stack-id: default, callee-saved-register: '$r5', callee-saved-restored: true,
135      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
136  - { id: 3, name: '', type: spill-slot, offset: -16, size: 4, alignment: 4,
137      stack-id: default, callee-saved-register: '$r4', callee-saved-restored: true,
138      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
139callSites:       []
140constants:       []
141machineFunctionInfo: {}
142body:             |
143  ; CHECK-LABEL: name: remove_mov_lr_chain
144  ; CHECK: bb.0.entry:
145  ; CHECK:   successors: %bb.9(0x30000000), %bb.1(0x50000000)
146  ; CHECK:   liveins: $lr, $r0, $r1, $r2, $r4, $r5, $r7
147  ; CHECK:   frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r4, killed $r5, killed $r7, killed $lr, implicit-def $sp, implicit $sp
148  ; CHECK:   frame-setup CFI_INSTRUCTION def_cfa_offset 16
149  ; CHECK:   frame-setup CFI_INSTRUCTION offset $lr, -4
150  ; CHECK:   frame-setup CFI_INSTRUCTION offset $r7, -8
151  ; CHECK:   frame-setup CFI_INSTRUCTION offset $r5, -12
152  ; CHECK:   frame-setup CFI_INSTRUCTION offset $r4, -16
153  ; CHECK:   tCMPi8 renamable $r2, 0, 14 /* CC::al */, $noreg, implicit-def $cpsr
154  ; CHECK:   tBcc %bb.9, 0 /* CC::eq */, killed $cpsr
155  ; CHECK: bb.1.while.body.preheader:
156  ; CHECK:   successors: %bb.6(0x40000000), %bb.2(0x40000000)
157  ; CHECK:   liveins: $r0, $r1, $r2
158  ; CHECK:   tCMPi8 renamable $r2, 4, 14 /* CC::al */, $noreg, implicit-def $cpsr
159  ; CHECK:   tBcc %bb.6, 3 /* CC::lo */, killed $cpsr
160  ; CHECK: bb.2.vector.memcheck:
161  ; CHECK:   successors: %bb.3(0x40000000), %bb.6(0x40000000)
162  ; CHECK:   liveins: $r0, $r1, $r2
163  ; CHECK:   renamable $r3 = t2ADDrs renamable $r0, renamable $r2, 18, 14 /* CC::al */, $noreg, $noreg
164  ; CHECK:   tCMPr killed renamable $r3, renamable $r1, 14 /* CC::al */, $noreg, implicit-def $cpsr
165  ; CHECK:   t2IT 8, 4, implicit-def $itstate
166  ; CHECK:   renamable $r3 = t2ADDrs renamable $r1, renamable $r2, 18, 8 /* CC::hi */, $cpsr, $noreg, implicit $itstate
167  ; CHECK:   tCMPr killed renamable $r3, renamable $r0, 8 /* CC::hi */, killed $cpsr, implicit-def $cpsr, implicit killed $itstate
168  ; CHECK:   tBcc %bb.6, 8 /* CC::hi */, killed $cpsr
169  ; CHECK: bb.3.vector.ph:
170  ; CHECK:   successors: %bb.4(0x80000000)
171  ; CHECK:   liveins: $r0, $r1, $r2
172  ; CHECK:   renamable $r4 = t2BICri renamable $r2, 3, 14 /* CC::al */, $noreg, $noreg
173  ; CHECK:   renamable $r3, dead $cpsr = tMOVi8 1, 14 /* CC::al */, $noreg
174  ; CHECK:   renamable $r12 = t2SUBri renamable $r4, 4, 14 /* CC::al */, $noreg, $noreg
175  ; CHECK:   renamable $r7, dead $cpsr = tSUBrr renamable $r2, renamable $r4, 14 /* CC::al */, $noreg
176  ; CHECK:   renamable $r3 = nuw nsw t2ADDrs killed renamable $r3, killed renamable $r12, 19, 14 /* CC::al */, $noreg, $noreg
177  ; CHECK:   renamable $r12 = t2ADDrs renamable $r0, renamable $r4, 18, 14 /* CC::al */, $noreg, $noreg
178  ; CHECK:   renamable $r0, dead $cpsr = tSUBi8 killed renamable $r0, 16, 14 /* CC::al */, $noreg
179  ; CHECK:   $r5 = tMOVr killed $r3, 14 /* CC::al */, $noreg
180  ; CHECK:   renamable $r3 = t2ADDrs renamable $r1, renamable $r4, 18, 14 /* CC::al */, $noreg, $noreg
181  ; CHECK:   renamable $r1, dead $cpsr = tSUBi8 killed renamable $r1, 16, 14 /* CC::al */, $noreg
182  ; CHECK: bb.4.vector.body:
183  ; CHECK:   successors: %bb.4(0x7c000000), %bb.5(0x04000000)
184  ; CHECK:   liveins: $r0, $r1, $r2, $r3, $r4, $r5, $r7, $r12
185  ; CHECK:   renamable $r0, renamable $q0 = MVE_VLDRWU32_pre killed renamable $r0, 16, 0, $noreg :: (load 16 from %ir.scevgep18, align 4)
186  ; CHECK:   $lr = tMOVr killed $r5, 14 /* CC::al */, $noreg
187  ; CHECK:   renamable $q0 = nnan ninf nsz arcp contract afn reassoc MVE_VABSf32 killed renamable $q0, 0, $noreg, undef renamable $q0
188  ; CHECK:   renamable $r1 = MVE_VSTRBU8_pre killed renamable $q0, killed renamable $r1, 16, 0, $noreg :: (store 16 into %ir.scevgep13, align 4)
189  ; CHECK:   $lr = t2SUBri killed renamable $lr, 1, 14 /* CC::al */, $noreg, def $cpsr
190  ; CHECK:   $r5 = tMOVr killed $lr, 14 /* CC::al */, $noreg
191  ; CHECK:   tBcc %bb.4, 1 /* CC::ne */, killed $cpsr
192  ; CHECK:   tB %bb.5, 14 /* CC::al */, $noreg
193  ; CHECK: bb.5.middle.block:
194  ; CHECK:   successors: %bb.7(0x80000000)
195  ; CHECK:   liveins: $r2, $r3, $r4, $r7, $r12
196  ; CHECK:   tCMPr killed renamable $r4, killed renamable $r2, 14 /* CC::al */, $noreg, implicit-def $cpsr
197  ; CHECK:   $lr = tMOVr killed $r7, 14 /* CC::al */, $noreg
198  ; CHECK:   t2IT 0, 8, implicit-def $itstate
199  ; CHECK:   tPOP_RET 0 /* CC::eq */, killed $cpsr, def $r4, def $r5, def $r7, def $pc, implicit killed $itstate
200  ; CHECK:   tB %bb.7, 14 /* CC::al */, $noreg
201  ; CHECK: bb.6:
202  ; CHECK:   successors: %bb.7(0x80000000)
203  ; CHECK:   liveins: $r0, $r1, $r2
204  ; CHECK:   $lr = tMOVr killed $r2, 14 /* CC::al */, $noreg
205  ; CHECK:   $r12 = tMOVr killed $r0, 14 /* CC::al */, $noreg
206  ; CHECK:   $r3 = tMOVr killed $r1, 14 /* CC::al */, $noreg
207  ; CHECK: bb.7.while.body.preheader19:
208  ; CHECK:   successors: %bb.8(0x80000000)
209  ; CHECK:   liveins: $lr, $r3, $r12
210  ; CHECK:   renamable $r0, dead $cpsr = tSUBi3 killed renamable $r3, 4, 14 /* CC::al */, $noreg
211  ; CHECK:   renamable $r1 = t2SUBri killed renamable $r12, 4, 14 /* CC::al */, $noreg, $noreg
212  ; CHECK:   $lr = t2DLS killed renamable $lr
213  ; CHECK: bb.8.while.body:
214  ; CHECK:   successors: %bb.8(0x7c000000), %bb.9(0x04000000)
215  ; CHECK:   liveins: $lr, $r0, $r1
216  ; CHECK:   renamable $s0 = VLDRS renamable $r1, 1, 14 /* CC::al */, $noreg :: (load 4 from %ir.scevgep3)
217  ; CHECK:   renamable $r1, dead $cpsr = tADDi8 killed renamable $r1, 4, 14 /* CC::al */, $noreg
218  ; CHECK:   renamable $s0 = nnan ninf nsz arcp contract afn reassoc VABSS killed renamable $s0, 14 /* CC::al */, $noreg
219  ; CHECK:   VSTRS killed renamable $s0, renamable $r0, 1, 14 /* CC::al */, $noreg :: (store 4 into %ir.scevgep7)
220  ; CHECK:   renamable $r0, dead $cpsr = tADDi8 killed renamable $r0, 4, 14 /* CC::al */, $noreg
221  ; CHECK:   $lr = t2LEUpdate killed renamable $lr, %bb.8
222  ; CHECK: bb.9.while.end:
223  ; CHECK:   tPOP_RET 14 /* CC::al */, $noreg, def $r4, def $r5, def $r7, def $pc
224  bb.0.entry:
225    successors: %bb.9(0x30000000), %bb.1(0x50000000)
226    liveins: $r0, $r1, $r2, $r4, $r5, $r7, $lr
227
228    frame-setup tPUSH 14, $noreg, killed $r4, killed $r5, killed $r7, killed $lr, implicit-def $sp, implicit $sp
229    frame-setup CFI_INSTRUCTION def_cfa_offset 16
230    frame-setup CFI_INSTRUCTION offset $lr, -4
231    frame-setup CFI_INSTRUCTION offset $r7, -8
232    frame-setup CFI_INSTRUCTION offset $r5, -12
233    frame-setup CFI_INSTRUCTION offset $r4, -16
234    tCMPi8 renamable $r2, 0, 14, $noreg, implicit-def $cpsr
235    tBcc %bb.9, 0, killed $cpsr
236
237  bb.1.while.body.preheader:
238    successors: %bb.6(0x40000000), %bb.2(0x40000000)
239    liveins: $r0, $r1, $r2
240
241    tCMPi8 renamable $r2, 4, 14, $noreg, implicit-def $cpsr
242    tBcc %bb.6, 3, killed $cpsr
243
244  bb.2.vector.memcheck:
245    successors: %bb.3(0x40000000), %bb.6(0x40000000)
246    liveins: $r0, $r1, $r2
247
248    renamable $r3 = t2ADDrs renamable $r0, renamable $r2, 18, 14, $noreg, $noreg
249    tCMPr killed renamable $r3, renamable $r1, 14, $noreg, implicit-def $cpsr
250    t2IT 8, 4, implicit-def $itstate
251    renamable $r3 = t2ADDrs renamable $r1, renamable $r2, 18, 8, $cpsr, $noreg, implicit $itstate
252    tCMPr killed renamable $r3, renamable $r0, 8, killed $cpsr, implicit-def $cpsr, implicit killed $itstate
253    tBcc %bb.6, 8, killed $cpsr
254
255  bb.3.vector.ph:
256    successors: %bb.4(0x80000000)
257    liveins: $r0, $r1, $r2
258
259    renamable $r4 = t2BICri renamable $r2, 3, 14, $noreg, $noreg
260    renamable $r3, dead $cpsr = tMOVi8 1, 14, $noreg
261    renamable $r12 = t2SUBri renamable $r4, 4, 14, $noreg, $noreg
262    renamable $r7, dead $cpsr = tSUBrr renamable $r2, renamable $r4, 14, $noreg
263    renamable $r3 = nuw nsw t2ADDrs killed renamable $r3, killed renamable $r12, 19, 14, $noreg, $noreg
264    renamable $r12 = t2ADDrs renamable $r0, renamable $r4, 18, 14, $noreg, $noreg
265    $lr = t2DoLoopStart renamable $r3
266    renamable $r0, dead $cpsr = tSUBi8 killed renamable $r0, 16, 14, $noreg
267    $r5 = tMOVr killed $r3, 14, $noreg
268    renamable $r3 = t2ADDrs renamable $r1, renamable $r4, 18, 14, $noreg, $noreg
269    renamable $r1, dead $cpsr = tSUBi8 killed renamable $r1, 16, 14, $noreg
270
271  bb.4.vector.body:
272    successors: %bb.4(0x7c000000), %bb.5(0x04000000)
273    liveins: $r0, $r1, $r2, $r3, $r4, $r5, $r7, $r12
274
275    renamable $r0, renamable $q0 = MVE_VLDRWU32_pre killed renamable $r0, 16, 0, $noreg :: (load 16 from %ir.scevgep18, align 4)
276    $lr = tMOVr killed $r5, 14, $noreg
277    renamable $q0 = nnan ninf nsz arcp contract afn reassoc MVE_VABSf32 killed renamable $q0, 0, $noreg, undef renamable $q0
278    renamable $r1 = MVE_VSTRBU8_pre killed renamable $q0, killed renamable $r1, 16, 0, $noreg :: (store 16 into %ir.scevgep13, align 4)
279    renamable $lr = t2LoopDec killed renamable $lr, 1
280    $r5 = tMOVr $lr, 14, $noreg
281    t2LoopEnd killed renamable $lr, %bb.4, implicit-def dead $cpsr
282    tB %bb.5, 14, $noreg
283
284  bb.5.middle.block:
285    successors: %bb.7(0x80000000)
286    liveins: $r2, $r3, $r4, $r7, $r12
287
288    tCMPr killed renamable $r4, killed renamable $r2, 14, $noreg, implicit-def $cpsr
289    $lr = tMOVr killed $r7, 14, $noreg
290    t2IT 0, 8, implicit-def $itstate
291    tPOP_RET 0, killed $cpsr, def $r4, def $r5, def $r7, def $pc, implicit killed $itstate
292    tB %bb.7, 14, $noreg
293
294  bb.6:
295    successors: %bb.7(0x80000000)
296    liveins: $r0, $r1, $r2
297
298    $lr = tMOVr killed $r2, 14, $noreg
299    $r12 = tMOVr killed $r0, 14, $noreg
300    $r3 = tMOVr killed $r1, 14, $noreg
301
302  bb.7.while.body.preheader19:
303    successors: %bb.8(0x80000000)
304    liveins: $lr, $r3, $r12
305
306    renamable $r0, dead $cpsr = tSUBi3 killed renamable $r3, 4, 14, $noreg
307    renamable $r1 = t2SUBri killed renamable $r12, 4, 14, $noreg, $noreg
308    $lr = t2DoLoopStart renamable $lr
309
310  bb.8.while.body:
311    successors: %bb.8(0x7c000000), %bb.9(0x04000000)
312    liveins: $lr, $r0, $r1
313
314    renamable $s0 = VLDRS renamable $r1, 1, 14, $noreg :: (load 4 from %ir.scevgep3)
315    renamable $r1, dead $cpsr = tADDi8 killed renamable $r1, 4, 14, $noreg
316    renamable $s0 = nnan ninf nsz arcp contract afn reassoc VABSS killed renamable $s0, 14, $noreg
317    VSTRS killed renamable $s0, renamable $r0, 1, 14, $noreg :: (store 4 into %ir.scevgep7)
318    renamable $r0, dead $cpsr = tADDi8 killed renamable $r0, 4, 14, $noreg
319    renamable $lr = t2LoopDec killed renamable $lr, 1
320    t2LoopEnd renamable $lr, %bb.8, implicit-def dead $cpsr
321    tB %bb.9, 14, $noreg
322
323  bb.9.while.end:
324    tPOP_RET 14, $noreg, def $r4, def $r5, def $r7, def $pc
325
326...
327