1# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py 2# RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve,+lob -run-pass=arm-low-overhead-loops --verify-machineinstrs %s -o - | FileCheck %s 3 4--- | 5 define dso_local void @vctp_tsubi3(i32* noalias nocapture %A, i32* noalias nocapture readonly %B, i32* noalias nocapture readonly %C, i32 %N) local_unnamed_addr #0 { 6 entry: 7 %cmp8 = icmp sgt i32 %N, 0 8 %0 = add i32 %N, 3 9 %1 = lshr i32 %0, 2 10 %2 = shl nuw i32 %1, 2 11 %3 = add i32 %2, -4 12 %4 = lshr i32 %3, 2 13 %5 = add nuw nsw i32 %4, 1 14 br i1 %cmp8, label %vector.ph, label %for.cond.cleanup 15 16 vector.ph: ; preds = %entry 17 %start = call i32 @llvm.start.loop.iterations.i32(i32 %5) 18 br label %vector.body 19 20 vector.body: ; preds = %vector.body, %vector.ph 21 %lsr.iv17 = phi i32* [ %scevgep18, %vector.body ], [ %A, %vector.ph ] 22 %lsr.iv14 = phi i32* [ %scevgep15, %vector.body ], [ %C, %vector.ph ] 23 %lsr.iv = phi i32* [ %scevgep, %vector.body ], [ %B, %vector.ph ] 24 %6 = phi i32 [ %start, %vector.ph ], [ %11, %vector.body ] 25 %7 = phi i32 [ %N, %vector.ph ], [ %9, %vector.body ] 26 %lsr.iv13 = bitcast i32* %lsr.iv to <4 x i32>* 27 %lsr.iv1416 = bitcast i32* %lsr.iv14 to <4 x i32>* 28 %lsr.iv1719 = bitcast i32* %lsr.iv17 to <4 x i32>* 29 %8 = call <4 x i1> @llvm.arm.mve.vctp32(i32 %7) 30 %9 = sub i32 %7, 5 31 %wide.masked.load = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %lsr.iv13, i32 4, <4 x i1> %8, <4 x i32> undef) 32 %wide.masked.load12 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %lsr.iv1416, i32 4, <4 x i1> %8, <4 x i32> undef) 33 %10 = add nsw <4 x i32> %wide.masked.load12, %wide.masked.load 34 call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %10, <4 x i32>* %lsr.iv1719, i32 4, <4 x i1> %8) 35 %scevgep = getelementptr i32, i32* %lsr.iv, i32 4 36 %scevgep15 = getelementptr i32, i32* %lsr.iv14, i32 4 37 %scevgep18 = getelementptr i32, i32* %lsr.iv17, i32 4 38 %11 = call i32 @llvm.loop.decrement.reg.i32.i32.i32(i32 %6, i32 1) 39 %12 = icmp ne i32 %11, 0 40 br i1 %12, label %vector.body, label %for.cond.cleanup 41 42 for.cond.cleanup: ; preds = %vector.body, %entry 43 ret void 44 } 45 declare i32 @llvm.start.loop.iterations.i32(i32) 46 declare <4 x i1> @llvm.arm.mve.vctp32(i32) 47 declare i32 @llvm.loop.decrement.reg.i32.i32.i32(i32, i32) 48 declare <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>*, i32 immarg, <4 x i1>, <4 x i32>) 49 declare void @llvm.masked.store.v4i32.p0v4i32(<4 x i32>, <4 x i32>*, i32 immarg, <4 x i1>) 50 51... 52--- 53name: vctp_tsubi3 54alignment: 2 55exposesReturnsTwice: false 56legalized: false 57regBankSelected: false 58selected: false 59failedISel: false 60tracksRegLiveness: true 61hasWinCFI: false 62registers: [] 63liveins: 64 - { reg: '$r0', virtual-reg: '' } 65 - { reg: '$r1', virtual-reg: '' } 66 - { reg: '$r2', virtual-reg: '' } 67 - { reg: '$r3', virtual-reg: '' } 68frameInfo: 69 isFrameAddressTaken: false 70 isReturnAddressTaken: false 71 hasStackMap: false 72 hasPatchPoint: false 73 stackSize: 8 74 offsetAdjustment: 0 75 maxAlignment: 4 76 adjustsStack: false 77 hasCalls: false 78 stackProtector: '' 79 maxCallFrameSize: 0 80 cvBytesOfCalleeSavedRegisters: 0 81 hasOpaqueSPAdjustment: false 82 hasVAStart: false 83 hasMustTailInVarArgFunc: false 84 localFrameSize: 0 85 savePoint: '' 86 restorePoint: '' 87fixedStack: [] 88stack: 89 - { id: 0, name: '', type: spill-slot, offset: -4, size: 4, alignment: 4, 90 stack-id: default, callee-saved-register: '$lr', callee-saved-restored: false, 91 debug-info-variable: '', debug-info-expression: '', debug-info-location: '' } 92 - { id: 1, name: '', type: spill-slot, offset: -8, size: 4, alignment: 4, 93 stack-id: default, callee-saved-register: '$r7', callee-saved-restored: true, 94 debug-info-variable: '', debug-info-expression: '', debug-info-location: '' } 95callSites: [] 96constants: [] 97machineFunctionInfo: {} 98body: | 99 ; CHECK-LABEL: name: vctp_tsubi3 100 ; CHECK: bb.0.entry: 101 ; CHECK: successors: %bb.1(0x80000000) 102 ; CHECK: liveins: $lr, $r0, $r1, $r2, $r3, $r7 103 ; CHECK: frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r7, killed $lr, implicit-def $sp, implicit $sp 104 ; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8 105 ; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -4 106 ; CHECK: frame-setup CFI_INSTRUCTION offset $r7, -8 107 ; CHECK: tCMPi8 renamable $r3, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr 108 ; CHECK: t2IT 11, 8, implicit-def $itstate 109 ; CHECK: tPOP_RET 11 /* CC::lt */, killed $cpsr, def $r7, def $pc, implicit killed $itstate 110 ; CHECK: bb.1.vector.ph: 111 ; CHECK: successors: %bb.2(0x80000000) 112 ; CHECK: liveins: $r0, $r1, $r2, $r3 113 ; CHECK: $lr = MVE_DLSTP_32 killed renamable $r3 114 ; CHECK: bb.2.vector.body: 115 ; CHECK: successors: %bb.2(0x7c000000), %bb.3(0x04000000) 116 ; CHECK: liveins: $lr, $r0, $r1, $r2 117 ; CHECK: renamable $r1, renamable $q0 = MVE_VLDRWU32_post killed renamable $r1, 16, 0, $noreg :: (load 16 from %ir.lsr.iv13, align 4) 118 ; CHECK: renamable $r2, renamable $q1 = MVE_VLDRWU32_post killed renamable $r2, 16, 0, $noreg :: (load 16 from %ir.lsr.iv1416, align 4) 119 ; CHECK: renamable $q0 = nsw MVE_VADDi32 killed renamable $q1, killed renamable $q0, 0, $noreg, undef renamable $q0 120 ; CHECK: renamable $r0 = MVE_VSTRWU32_post killed renamable $q0, killed renamable $r0, 16, 0, killed $noreg :: (store 16 into %ir.lsr.iv1719, align 4) 121 ; CHECK: $lr = MVE_LETP killed renamable $lr, %bb.2 122 ; CHECK: bb.3.for.cond.cleanup: 123 ; CHECK: tPOP_RET 14 /* CC::al */, $noreg, def $r7, def $pc 124 bb.0.entry: 125 successors: %bb.1(0x80000000) 126 liveins: $r0, $r1, $r2, $r3, $r7, $lr 127 128 frame-setup tPUSH 14, $noreg, killed $r7, killed $lr, implicit-def $sp, implicit $sp 129 frame-setup CFI_INSTRUCTION def_cfa_offset 8 130 frame-setup CFI_INSTRUCTION offset $lr, -4 131 frame-setup CFI_INSTRUCTION offset $r7, -8 132 tCMPi8 renamable $r3, 1, 14, $noreg, implicit-def $cpsr 133 t2IT 11, 8, implicit-def $itstate 134 tPOP_RET 11, killed $cpsr, def $r7, def $pc, implicit killed $itstate 135 136 bb.1.vector.ph: 137 successors: %bb.2(0x80000000) 138 liveins: $r0, $r1, $r2, $r3, $r7, $lr 139 140 renamable $r12 = t2ADDri renamable $r3, 3, 14, $noreg, $noreg 141 renamable $lr = t2MOVi 1, 14, $noreg, $noreg 142 renamable $r12 = t2BICri killed renamable $r12, 3, 14, $noreg, $noreg 143 renamable $r12 = t2SUBri killed renamable $r12, 4, 14, $noreg, $noreg 144 renamable $lr = nuw nsw t2ADDrs killed renamable $lr, killed renamable $r12, 19, 14, $noreg, $noreg 145 $lr = t2DoLoopStart renamable $lr 146 147 bb.2.vector.body: 148 successors: %bb.2(0x7c000000), %bb.3(0x04000000) 149 liveins: $lr, $r0, $r1, $r2, $r3 150 151 renamable $vpr = MVE_VCTP32 renamable $r3, 0, $noreg 152 MVE_VPST 4, implicit $vpr 153 renamable $r1, renamable $q0 = MVE_VLDRWU32_post killed renamable $r1, 16, 1, renamable $vpr :: (load 16 from %ir.lsr.iv13, align 4) 154 renamable $r2, renamable $q1 = MVE_VLDRWU32_post killed renamable $r2, 16, 1, renamable $vpr :: (load 16 from %ir.lsr.iv1416, align 4) 155 renamable $r3 = t2SUBri killed renamable $r3, 4, 14, $noreg, $noreg 156 renamable $q0 = nsw MVE_VADDi32 killed renamable $q1, killed renamable $q0, 0, $noreg, undef renamable $q0 157 MVE_VPST 8, implicit $vpr 158 renamable $r0 = MVE_VSTRWU32_post killed renamable $q0, killed renamable $r0, 16, 1, killed renamable $vpr :: (store 16 into %ir.lsr.iv1719, align 4) 159 renamable $lr = t2LoopDec killed renamable $lr, 1 160 t2LoopEnd renamable $lr, %bb.2, implicit-def dead $cpsr 161 tB %bb.3, 14, $noreg 162 163 bb.3.for.cond.cleanup: 164 tPOP_RET 14, $noreg, def $r7, def $pc 165 166... 167