#include "ARMBaseInstrInfo.h"
#include "ARMBaseRegisterInfo.h"
#include "ARMMachineFunctionInfo.h"
-#include "ARMRegisterInfo.h"
#include "MCTargetDesc/ARMAddressingModes.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
bool IsLoad;
bool isUpdating;
bool hasWritebackOperand;
- NEONRegSpacing RegSpacing;
- unsigned char NumRegs; // D registers loaded or stored
- unsigned char RegElts; // elements per D register; used for lane ops
+ uint8_t RegSpacing; // One of type NEONRegSpacing
+ uint8_t NumRegs; // D registers loaded or stored
+ uint8_t RegElts; // elements per D register; used for lane ops
// FIXME: Temporary flag to denote whether the real instruction takes
// a single register (like the encoding) or all of the registers in
// the list (like the asm syntax and the isel DAG). When all definitions
const NEONLdStTableEntry *TableEntry = LookupNEONLdSt(MI.getOpcode());
assert(TableEntry && TableEntry->IsLoad && "NEONLdStTable lookup failed");
- NEONRegSpacing RegSpc = TableEntry->RegSpacing;
+ NEONRegSpacing RegSpc = (NEONRegSpacing)TableEntry->RegSpacing;
unsigned NumRegs = TableEntry->NumRegs;
MachineInstrBuilder MIB = BuildMI(MBB, MBBI, MI.getDebugLoc(),
const NEONLdStTableEntry *TableEntry = LookupNEONLdSt(MI.getOpcode());
assert(TableEntry && !TableEntry->IsLoad && "NEONLdStTable lookup failed");
- NEONRegSpacing RegSpc = TableEntry->RegSpacing;
+ NEONRegSpacing RegSpc = (NEONRegSpacing)TableEntry->RegSpacing;
unsigned NumRegs = TableEntry->NumRegs;
MachineInstrBuilder MIB = BuildMI(MBB, MBBI, MI.getDebugLoc(),
MIB.addOperand(MI.getOperand(OpIdx++));
bool SrcIsKill = MI.getOperand(OpIdx).isKill();
+ bool SrcIsUndef = MI.getOperand(OpIdx).isUndef();
unsigned SrcReg = MI.getOperand(OpIdx++).getReg();
unsigned D0, D1, D2, D3;
GetDSubRegs(SrcReg, RegSpc, TRI, D0, D1, D2, D3);
- MIB.addReg(D0);
+ MIB.addReg(D0, getUndefRegState(SrcIsUndef));
if (NumRegs > 1 && TableEntry->copyAllListRegs)
- MIB.addReg(D1);
+ MIB.addReg(D1, getUndefRegState(SrcIsUndef));
if (NumRegs > 2 && TableEntry->copyAllListRegs)
- MIB.addReg(D2);
+ MIB.addReg(D2, getUndefRegState(SrcIsUndef));
if (NumRegs > 3 && TableEntry->copyAllListRegs)
- MIB.addReg(D3);
+ MIB.addReg(D3, getUndefRegState(SrcIsUndef));
// Copy the predicate operands.
MIB.addOperand(MI.getOperand(OpIdx++));
MIB.addOperand(MI.getOperand(OpIdx++));
- if (SrcIsKill) // Add an implicit kill for the super-reg.
+ if (SrcIsKill && !SrcIsUndef) // Add an implicit kill for the super-reg.
MIB->addRegisterKilled(SrcReg, TRI, true);
TransferImpOps(MI, MIB, MIB);
const NEONLdStTableEntry *TableEntry = LookupNEONLdSt(MI.getOpcode());
assert(TableEntry && "NEONLdStTable lookup failed");
- NEONRegSpacing RegSpc = TableEntry->RegSpacing;
+ NEONRegSpacing RegSpc = (NEONRegSpacing)TableEntry->RegSpacing;
unsigned NumRegs = TableEntry->NumRegs;
unsigned RegElts = TableEntry->RegElts;
MachineInstr &MI = *MBBI;
unsigned Opcode = MI.getOpcode();
unsigned PredReg = 0;
- ARMCC::CondCodes Pred = llvm::getInstrPredicate(&MI, PredReg);
+ ARMCC::CondCodes Pred = getInstrPredicate(&MI, PredReg);
unsigned DstReg = MI.getOperand(0).getReg();
bool DstIsDead = MI.getOperand(0).isDead();
bool isCC = Opcode == ARM::MOVCCi32imm || Opcode == ARM::t2MOVCCi32imm;
"base pointer without frame pointer?");
if (AFI->isThumb2Function()) {
- llvm::emitT2RegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
- FramePtr, -NumBytes, ARMCC::AL, 0, *TII);
+ emitT2RegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
+ FramePtr, -NumBytes, ARMCC::AL, 0, *TII);
} else if (AFI->isThumbFunction()) {
- llvm::emitThumbRegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
- FramePtr, -NumBytes, *TII, RI);
+ emitThumbRegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
+ FramePtr, -NumBytes, *TII, RI);
} else {
- llvm::emitARMRegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
- FramePtr, -NumBytes, ARMCC::AL, 0,
- *TII);
+ emitARMRegPlusImmediate(MBB, MBBI, MI.getDebugLoc(), ARM::R6,
+ FramePtr, -NumBytes, ARMCC::AL, 0,
+ *TII);
}
// If there's dynamic realignment, adjust for it.
if (RI.needsStackRealignment(MF)) {
if (isARM) {
AddDefaultPred(MIB3);
if (Opcode == ARM::MOV_ga_pcrel_ldr)
- MIB2->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
+ MIB3->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
}
TransferImpOps(MI, MIB1, MIB3);
MI.eraseFromParent();
BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(NewOpc));
unsigned OpIdx = 0;
unsigned SrcReg = MI.getOperand(1).getReg();
- unsigned Lane = getARMRegisterNumbering(SrcReg) & 1;
+ unsigned Lane = TRI->getEncodingValue(SrcReg) & 1;
unsigned DReg = TRI->getMatchingSuperReg(SrcReg,
Lane & 1 ? ARM::ssub_1 : ARM::ssub_0,
&ARM::DPR_VFP2RegClass);
ExpandLaneOp(MBBI);
return true;
+ case ARM::VSETLNi8Q:
+ case ARM::VSETLNi16Q: {
+ // Expand VSETLNs acting on a Q register to equivalent VSETLNs acting
+ // on the respective D register.
+
+ unsigned QReg = MI.getOperand(1).getReg();
+ unsigned QLane = MI.getOperand(3).getImm();
+
+ unsigned NewOpcode, DLane, DSubReg;
+ switch (Opcode) {
+ default: llvm_unreachable("Invalid opcode!");
+ case ARM::VSETLNi8Q:
+ // 4 possible 8-bit lanes per DPR:
+ NewOpcode = ARM::VSETLNi8;
+ DLane = QLane % 8;
+ DSubReg = (QLane / 8) ? ARM::dsub_1 : ARM::dsub_0;
+ break;
+ case ARM::VSETLNi16Q:
+ // 4 possible 16-bit lanes per DPR.
+ NewOpcode = ARM::VSETLNi16;
+ DLane = QLane % 4;
+ DSubReg = (QLane / 4) ? ARM::dsub_1 : ARM::dsub_0;
+ break;
+ }
+
+ MachineInstrBuilder MIB =
+ BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(NewOpcode));
+
+ unsigned DReg = TRI->getSubReg(QReg, DSubReg);
+
+ MIB.addReg(DReg, RegState::Define); // Output DPR
+ MIB.addReg(DReg); // Input DPR
+ MIB.addOperand(MI.getOperand(2)); // Input GPR
+ MIB.addImm(DLane); // Lane
+
+ // Add the predicate operands.
+ MIB.addOperand(MI.getOperand(4));
+ MIB.addOperand(MI.getOperand(5));
+
+ if (MI.getOperand(1).isKill()) // Add an implicit kill for the Q register.
+ MIB->addRegisterKilled(QReg, TRI, true);
+ // And an implicit def of the output register (which should always be the
+ // same as the input register).
+ MIB->addRegisterDefined(QReg, TRI);
+
+ TransferImpOps(MI, MIB, MIB);
+
+ MI.eraseFromParent();
+ return true;
+ }
+
case ARM::VTBL3Pseudo: ExpandVTBL(MBBI, ARM::VTBL3, false); return true;
case ARM::VTBL4Pseudo: ExpandVTBL(MBBI, ARM::VTBL4, false); return true;
case ARM::VTBX3Pseudo: ExpandVTBL(MBBI, ARM::VTBX3, true); return true;
projects / oota-llvm.git / blobdiff