if (LocalDef && InstIds.lookup(LocalDef) < InstIds.lookup(MI))
return LocalDef;
- SmallPtrSet<MachineBasicBlock*, 4> VisitedBBs;
SmallPtrSet<MachineInstr*, 2> Incoming;
MachineBasicBlock *Parent = MI->getParent();
for (auto *Pred : Parent->predecessors())
return isPredicatedOnVCTP(Insts.front(), Exclusive);
}
- static bool isValid() {
+ // If this block begins with a VPT, we can check whether it's using
+ // at least one predicated input(s), as well as possible loop invariant
+ // which would result in it being implicitly predicated.
+ static bool hasImplicitlyValidVPT(VPTState &Block,
+ ReachingDefAnalysis &RDA) {
+ SmallVectorImpl<MachineInstr *> &Insts = Block.getInsts();
+ MachineInstr *VPT = Insts.front();
+ assert(isVPTOpcode(VPT->getOpcode()) &&
+ "Expected VPT block to begin with VPT/VPST");
+
+ if (VPT->getOpcode() == ARM::MVE_VPST)
+ return false;
+
+ auto IsOperandPredicated = [&](MachineInstr *MI, unsigned Idx) {
+ MachineInstr *Op = RDA.getMIOperand(MI, MI->getOperand(Idx));
+ return Op && PredicatedInsts.count(Op) && isPredicatedOnVCTP(Op);
+ };
+
+ auto IsOperandInvariant = [&](MachineInstr *MI, unsigned Idx) {
+ MachineOperand &MO = MI->getOperand(Idx);
+ if (!MO.isReg() || !MO.getReg())
+ return true;
+
+ SmallPtrSet<MachineInstr *, 2> Defs;
+ RDA.getGlobalReachingDefs(MI, MO.getReg(), Defs);
+ if (Defs.empty())
+ return true;
+
+ for (auto *Def : Defs)
+ if (Def->getParent() == VPT->getParent())
+ return false;
+ return true;
+ };
+
+ // Check that at least one of the operands is directly predicated on a
+ // vctp and allow an invariant value too.
+ return (IsOperandPredicated(VPT, 1) || IsOperandPredicated(VPT, 2)) &&
+ (IsOperandPredicated(VPT, 1) || IsOperandInvariant(VPT, 1)) &&
+ (IsOperandPredicated(VPT, 2) || IsOperandInvariant(VPT, 2));
+ }
+
+ static bool isValid(ReachingDefAnalysis &RDA) {
// All predication within the loop should be based on vctp. If the block
// isn't predicated on entry, check whether the vctp is within the block
// and that all other instructions are then predicated on it.
for (auto &Block : Blocks) {
- if (isEntryPredicatedOnVCTP(Block))
+ if (isEntryPredicatedOnVCTP(Block, false) ||
+ hasImplicitlyValidVPT(Block, RDA))
continue;
SmallVectorImpl<MachineInstr *> &Insts = Block.getInsts();
return false;
}
- if (!VPTState::isValid())
+ if (!VPTState::isValid(RDA))
return false;
if (!ValidateLiveOuts()) {
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vpush {d8, d9, d10, d11, d12, d13, d14, d15}
; CHECK-NEXT: ldrd r5, r12, [sp, #80]
-; CHECK-NEXT: cmp.w r12, #4
-; CHECK-NEXT: mov r4, r12
; CHECK-NEXT: vmvn.i32 q0, #0x80000000
-; CHECK-NEXT: it ge
-; CHECK-NEXT: movge r4, #4
; CHECK-NEXT: vmov.i32 q1, #0x3f
-; CHECK-NEXT: sub.w r4, r12, r4
; CHECK-NEXT: vmov.i32 q2, #0x1
-; CHECK-NEXT: add.w lr, r4, #3
-; CHECK-NEXT: movs r4, #1
-; CHECK-NEXT: add.w lr, r4, lr, lsr #2
-; CHECK-NEXT: dls lr, lr
+; CHECK-NEXT: dlstp.32 lr, r12
; CHECK-NEXT: .LBB0_1: @ %bb6
; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: vctp.32 r12
-; CHECK-NEXT: vpst
-; CHECK-NEXT: vldrwt.u32 q4, [r1], #16
+; CHECK-NEXT: vldrw.u32 q4, [r1], #16
; CHECK-NEXT: vabs.s32 q5, q4
; CHECK-NEXT: vcls.s32 q3, q5
; CHECK-NEXT: vshl.u32 q5, q5, q3
; CHECK-NEXT: vqshl.s32 q5, q5, #1
; CHECK-NEXT: vpt.s32 lt, q4, zr
; CHECK-NEXT: vnegt.s32 q5, q5
-; CHECK-NEXT: vctp.32 r12
-; CHECK-NEXT: sub.w r12, r12, #4
-; CHECK-NEXT: vpst
-; CHECK-NEXT: vldrwt.u32 q4, [r0], #16
+; CHECK-NEXT: vldrw.u32 q4, [r0], #16
; CHECK-NEXT: vqrdmulh.s32 q4, q4, q5
-; CHECK-NEXT: vpstt
-; CHECK-NEXT: vstrwt.32 q4, [r2], #16
-; CHECK-NEXT: vstrwt.32 q3, [r3], #16
-; CHECK-NEXT: le lr, .LBB0_1
+; CHECK-NEXT: vstrw.32 q4, [r2], #16
+; CHECK-NEXT: vstrw.32 q3, [r3], #16
+; CHECK-NEXT: letp lr, .LBB0_1
; CHECK-NEXT: @ %bb.2: @ %bb44
; CHECK-NEXT: vpop {d8, d9, d10, d11, d12, d13, d14, d15}
; CHECK-NEXT: pop {r4, r5, r7, pc}
; CHECK: bb.1.vector.ph:
; CHECK: successors: %bb.2(0x80000000)
; CHECK: liveins: $r0, $r1, $r2
- ; CHECK: renamable $r3, dead $cpsr = tADDi3 renamable $r1, 3, 14 /* CC::al */, $noreg
; CHECK: renamable $q0 = MVE_VMOVimmi32 0, 0, $noreg, undef renamable $q0
- ; CHECK: renamable $r3 = t2BICri killed renamable $r3, 3, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: renamable $r12 = t2SUBri killed renamable $r3, 4, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: renamable $r3, dead $cpsr = tMOVi8 1, 14 /* CC::al */, $noreg
- ; CHECK: renamable $lr = nuw nsw t2ADDrs killed renamable $r3, killed renamable $r12, 19, 14 /* CC::al */, $noreg, $noreg
; CHECK: renamable $r3, dead $cpsr = nsw tRSB renamable $r2, 14 /* CC::al */, $noreg
- ; CHECK: $lr = t2DLS killed renamable $lr
+ ; CHECK: $lr = MVE_DLSTP_32 killed renamable $r1
; CHECK: bb.2.vector.body:
; CHECK: successors: %bb.2(0x7c000000), %bb.3(0x04000000)
- ; CHECK: liveins: $lr, $q0, $r0, $r1, $r2, $r3
- ; CHECK: renamable $vpr = MVE_VCTP32 renamable $r1, 0, $noreg
- ; CHECK: MVE_VPST 8, implicit $vpr
- ; CHECK: renamable $q1 = MVE_VLDRWU32 renamable $r0, 0, 1, killed renamable $vpr
- ; CHECK: MVE_VPTv4s32r 14, renamable $q1, renamable $r2, 10, implicit-def $vpr
+ ; CHECK: liveins: $lr, $q0, $r0, $r2, $r3
+ ; CHECK: renamable $q1 = MVE_VLDRWU32 renamable $r0, 0, 0, killed $noreg
+ ; CHECK: MVE_VPTv4s32r 12, renamable $q1, renamable $r2, 10, implicit-def $vpr
; CHECK: renamable $vpr = MVE_VCMPs32r killed renamable $q1, renamable $r3, 13, 1, killed renamable $vpr
- ; CHECK: renamable $vpr = MVE_VCTP32 renamable $r1, 2, killed renamable $vpr
; CHECK: renamable $r0 = MVE_VSTRWU32_post renamable $q0, killed renamable $r0, 16, 2, killed renamable $vpr
- ; CHECK: renamable $r1, dead $cpsr = tSUBi8 killed renamable $r1, 4, 14 /* CC::al */, $noreg
- ; CHECK: $lr = t2LEUpdate killed renamable $lr, %bb.2
+ ; CHECK: $lr = MVE_LETP killed renamable $lr, %bb.2
; CHECK: bb.3.for.cond.cleanup:
; CHECK: frame-destroy tPOP_RET 14 /* CC::al */, $noreg, def $r7, def $pc
;