-//===-- SPUISelLowering.cpp - Cell SPU DAG Lowering Implementation --------===//
//
+//===-- SPUISelLowering.cpp - Cell SPU DAG Lowering Implementation --------===//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
#include "SPUISelLowering.h"
#include "SPUTargetMachine.h"
#include "SPUFrameInfo.h"
-#include "llvm/ADT/APInt.h"
-#include "llvm/ADT/VectorExtras.h"
+#include "llvm/Constants.h"
+#include "llvm/Function.h"
+#include "llvm/Intrinsics.h"
#include "llvm/CallingConv.h"
#include "llvm/CodeGen/CallingConvLower.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/SelectionDAG.h"
-#include "llvm/Constants.h"
-#include "llvm/Function.h"
-#include "llvm/Intrinsics.h"
+#include "llvm/Target/TargetLoweringObjectFile.h"
+#include "llvm/Target/TargetOptions.h"
+#include "llvm/ADT/VectorExtras.h"
#include "llvm/Support/Debug.h"
+#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
-#include "llvm/Target/TargetOptions.h"
-
+#include "llvm/Support/raw_ostream.h"
#include <map>
using namespace llvm;
#ifndef NDEBUG
if (retval == 0) {
- cerr << "getValueTypeMapEntry returns NULL for "
- << VT.getMVTString()
- << "\n";
- abort();
+ std::string msg;
+ raw_string_ostream Msg(msg);
+ Msg << "getValueTypeMapEntry returns NULL for "
+ << VT.getMVTString();
+ llvm_report_error(Msg.str());
}
#endif
TargetLowering::ArgListEntry Entry;
for (unsigned i = 0, e = Op.getNumOperands(); i != e; ++i) {
MVT ArgVT = Op.getOperand(i).getValueType();
- const Type *ArgTy = ArgVT.getTypeForMVT();
+ const Type *ArgTy = ArgVT.getTypeForMVT(*DAG.getContext());
Entry.Node = Op.getOperand(i);
Entry.Ty = ArgTy;
Entry.isSExt = isSigned;
TLI.getPointerTy());
// Splice the libcall in wherever FindInputOutputChains tells us to.
- const Type *RetTy = Op.getNode()->getValueType(0).getTypeForMVT();
+ const Type *RetTy =
+ Op.getNode()->getValueType(0).getTypeForMVT(*DAG.getContext());
std::pair<SDValue, SDValue> CallInfo =
TLI.LowerCallTo(InChain, RetTy, isSigned, !isSigned, false, false,
- CallingConv::C, false, Callee, Args, DAG,
- Op.getNode()->getDebugLoc());
+ 0, CallingConv::C, false, Callee, Args, DAG,
+ Op.getDebugLoc());
return CallInfo.first;
}
}
SPUTargetLowering::SPUTargetLowering(SPUTargetMachine &TM)
- : TargetLowering(TM),
- SPUTM(TM)
-{
+ : TargetLowering(TM, new TargetLoweringObjectFileELF()),
+ SPUTM(TM) {
// Fold away setcc operations if possible.
setPow2DivIsCheap();
setLoadExtAction(ISD::EXTLOAD, MVT::f32, Expand);
setLoadExtAction(ISD::EXTLOAD, MVT::f64, Expand);
+ setTruncStoreAction(MVT::i128, MVT::i64, Expand);
+ setTruncStoreAction(MVT::i128, MVT::i32, Expand);
+ setTruncStoreAction(MVT::i128, MVT::i16, Expand);
+ setTruncStoreAction(MVT::i128, MVT::i8, Expand);
+
+ setTruncStoreAction(MVT::f64, MVT::f32, Expand);
+
// SPU constant load actions are custom lowered:
setOperationAction(ISD::ConstantFP, MVT::f32, Legal);
setOperationAction(ISD::ConstantFP, MVT::f64, Custom);
// SPU has no intrinsics for these particular operations:
setOperationAction(ISD::MEMBARRIER, MVT::Other, Expand);
- // SPU has no SREM/UREM instructions
- setOperationAction(ISD::SREM, MVT::i32, Expand);
- setOperationAction(ISD::UREM, MVT::i32, Expand);
- setOperationAction(ISD::SREM, MVT::i64, Expand);
- setOperationAction(ISD::UREM, MVT::i64, Expand);
+ // SPU has no division/remainder instructions
+ setOperationAction(ISD::SREM, MVT::i8, Expand);
+ setOperationAction(ISD::UREM, MVT::i8, Expand);
+ setOperationAction(ISD::SDIV, MVT::i8, Expand);
+ setOperationAction(ISD::UDIV, MVT::i8, Expand);
+ setOperationAction(ISD::SDIVREM, MVT::i8, Expand);
+ setOperationAction(ISD::UDIVREM, MVT::i8, Expand);
+ setOperationAction(ISD::SREM, MVT::i16, Expand);
+ setOperationAction(ISD::UREM, MVT::i16, Expand);
+ setOperationAction(ISD::SDIV, MVT::i16, Expand);
+ setOperationAction(ISD::UDIV, MVT::i16, Expand);
+ setOperationAction(ISD::SDIVREM, MVT::i16, Expand);
+ setOperationAction(ISD::UDIVREM, MVT::i16, Expand);
+ setOperationAction(ISD::SREM, MVT::i32, Expand);
+ setOperationAction(ISD::UREM, MVT::i32, Expand);
+ setOperationAction(ISD::SDIV, MVT::i32, Expand);
+ setOperationAction(ISD::UDIV, MVT::i32, Expand);
+ setOperationAction(ISD::SDIVREM, MVT::i32, Expand);
+ setOperationAction(ISD::UDIVREM, MVT::i32, Expand);
+ setOperationAction(ISD::SREM, MVT::i64, Expand);
+ setOperationAction(ISD::UREM, MVT::i64, Expand);
+ setOperationAction(ISD::SDIV, MVT::i64, Expand);
+ setOperationAction(ISD::UDIV, MVT::i64, Expand);
+ setOperationAction(ISD::SDIVREM, MVT::i64, Expand);
+ setOperationAction(ISD::UDIVREM, MVT::i64, Expand);
+ setOperationAction(ISD::SREM, MVT::i128, Expand);
+ setOperationAction(ISD::UREM, MVT::i128, Expand);
+ setOperationAction(ISD::SDIV, MVT::i128, Expand);
+ setOperationAction(ISD::UDIV, MVT::i128, Expand);
+ setOperationAction(ISD::SDIVREM, MVT::i128, Expand);
+ setOperationAction(ISD::UDIVREM, MVT::i128, Expand);
// We don't support sin/cos/sqrt/fmod
setOperationAction(ISD::FSIN , MVT::f64, Expand);
setOperationAction(ISD::MUL, MVT::i32, Legal);
setOperationAction(ISD::MUL, MVT::i64, Legal);
+ // Expand double-width multiplication
+ // FIXME: It would probably be reasonable to support some of these operations
+ setOperationAction(ISD::UMUL_LOHI, MVT::i8, Expand);
+ setOperationAction(ISD::SMUL_LOHI, MVT::i8, Expand);
+ setOperationAction(ISD::MULHU, MVT::i8, Expand);
+ setOperationAction(ISD::MULHS, MVT::i8, Expand);
+ setOperationAction(ISD::UMUL_LOHI, MVT::i16, Expand);
+ setOperationAction(ISD::SMUL_LOHI, MVT::i16, Expand);
+ setOperationAction(ISD::MULHU, MVT::i16, Expand);
+ setOperationAction(ISD::MULHS, MVT::i16, Expand);
+ setOperationAction(ISD::UMUL_LOHI, MVT::i32, Expand);
+ setOperationAction(ISD::SMUL_LOHI, MVT::i32, Expand);
+ setOperationAction(ISD::MULHU, MVT::i32, Expand);
+ setOperationAction(ISD::MULHS, MVT::i32, Expand);
+ setOperationAction(ISD::UMUL_LOHI, MVT::i64, Expand);
+ setOperationAction(ISD::SMUL_LOHI, MVT::i64, Expand);
+ setOperationAction(ISD::MULHU, MVT::i64, Expand);
+ setOperationAction(ISD::MULHS, MVT::i64, Expand);
+
// Need to custom handle (some) common i8, i64 math ops
setOperationAction(ISD::ADD, MVT::i8, Custom);
setOperationAction(ISD::ADD, MVT::i64, Legal);
setOperationAction(ISD::CTPOP, MVT::i16, Custom);
setOperationAction(ISD::CTPOP, MVT::i32, Custom);
setOperationAction(ISD::CTPOP, MVT::i64, Custom);
+ setOperationAction(ISD::CTPOP, MVT::i128, Expand);
+ setOperationAction(ISD::CTTZ , MVT::i8, Expand);
+ setOperationAction(ISD::CTTZ , MVT::i16, Expand);
setOperationAction(ISD::CTTZ , MVT::i32, Expand);
setOperationAction(ISD::CTTZ , MVT::i64, Expand);
+ setOperationAction(ISD::CTTZ , MVT::i128, Expand);
+ setOperationAction(ISD::CTLZ , MVT::i8, Promote);
+ setOperationAction(ISD::CTLZ , MVT::i16, Promote);
setOperationAction(ISD::CTLZ , MVT::i32, Legal);
+ setOperationAction(ISD::CTLZ , MVT::i64, Expand);
+ setOperationAction(ISD::CTLZ , MVT::i128, Expand);
// SPU has a version of select that implements (a&~c)|(b&c), just like
// select ought to work:
// Custom lower i128 -> i64 truncates
setOperationAction(ISD::TRUNCATE, MVT::i64, Custom);
+ setOperationAction(ISD::FP_TO_SINT, MVT::i8, Promote);
+ setOperationAction(ISD::FP_TO_UINT, MVT::i8, Promote);
+ setOperationAction(ISD::FP_TO_SINT, MVT::i16, Promote);
+ setOperationAction(ISD::FP_TO_UINT, MVT::i16, Promote);
// SPU has a legal FP -> signed INT instruction for f32, but for f64, need
// to expand to a libcall, hence the custom lowering:
setOperationAction(ISD::FP_TO_SINT, MVT::i32, Custom);
setOperationAction(ISD::FP_TO_UINT, MVT::i32, Custom);
+ setOperationAction(ISD::FP_TO_SINT, MVT::i64, Expand);
+ setOperationAction(ISD::FP_TO_UINT, MVT::i64, Expand);
+ setOperationAction(ISD::FP_TO_SINT, MVT::i128, Expand);
+ setOperationAction(ISD::FP_TO_UINT, MVT::i128, Expand);
// FDIV on SPU requires custom lowering
setOperationAction(ISD::FDIV, MVT::f64, Expand); // to libcall
return ((i != node_names.end()) ? i->second : 0);
}
+/// getFunctionAlignment - Return the Log2 alignment of this function.
+unsigned SPUTargetLowering::getFunctionAlignment(const Function *) const {
+ return 3;
+}
+
//===----------------------------------------------------------------------===//
// Return the Cell SPU's SETCC result type
//===----------------------------------------------------------------------===//
// Simplify the base pointer for this case:
basePtr = basePtr.getOperand(0);
if ((offset & ~0xf) > 0) {
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant((offset & ~0xf), PtrVT));
}
// Convert the (add <ptr>, <const>) to an indirect address contained
// in a register. Note that this is done because we need to avoid
// creating a 0(reg) d-form address due to the SPU's block loads.
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT, Op0, Op1);
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Op0, Op1);
the_chain = DAG.getCopyToReg(the_chain, dl, VReg, basePtr, Flag);
basePtr = DAG.getCopyFromReg(the_chain, dl, VReg, PtrVT);
} else {
// Convert the (add <arg1>, <arg2>) to an indirect address, which
// will likely be lowered as a reg(reg) x-form address.
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT, Op0, Op1);
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Op0, Op1);
}
} else {
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant(0, PtrVT));
}
case ISD::POST_INC:
case ISD::POST_DEC:
case ISD::LAST_INDEXED_MODE:
- cerr << "LowerLOAD: Got a LoadSDNode with an addr mode other than "
+ {
+ std::string msg;
+ raw_string_ostream Msg(msg);
+ Msg << "LowerLOAD: Got a LoadSDNode with an addr mode other than "
"UNINDEXED\n";
- cerr << (unsigned) LN->getAddressingMode() << "\n";
- abort();
- /*NOTREACHED*/
+ Msg << (unsigned) LN->getAddressingMode();
+ llvm_report_error(Msg.str());
+ /*NOTREACHED*/
+ }
}
return SDValue();
// Simplify the base pointer for this case:
basePtr = basePtr.getOperand(0);
- insertEltOffs = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ insertEltOffs = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant((offset & 0xf), PtrVT));
if ((offset & ~0xf) > 0) {
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant((offset & ~0xf), PtrVT));
}
} else {
// Otherwise, assume it's at byte 0 of basePtr
- insertEltOffs = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ insertEltOffs = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant(0, PtrVT));
}
// Convert the (add <ptr>, <const>) to an indirect address contained
// in a register. Note that this is done because we need to avoid
// creating a 0(reg) d-form address due to the SPU's block loads.
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT, Op0, Op1);
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Op0, Op1);
the_chain = DAG.getCopyToReg(the_chain, dl, VReg, basePtr, Flag);
basePtr = DAG.getCopyFromReg(the_chain, dl, VReg, PtrVT);
} else {
// Convert the (add <arg1>, <arg2>) to an indirect address, which
// will likely be lowered as a reg(reg) x-form address.
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT, Op0, Op1);
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Op0, Op1);
}
} else {
- basePtr = DAG.getNode(SPUISD::IndirectAddr, PtrVT,
+ basePtr = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT,
basePtr,
DAG.getConstant(0, PtrVT));
}
result = DAG.getNode(SPUISD::SHUFB, dl, vecVT,
vectorizeOp, alignLoadVec,
- DAG.getNode(ISD::BIT_CONVERT, dl,
+ DAG.getNode(ISD::BIT_CONVERT, dl,
MVT::v4i32, insertEltOp));
result = DAG.getStore(the_chain, dl, result, basePtr,
case ISD::POST_INC:
case ISD::POST_DEC:
case ISD::LAST_INDEXED_MODE:
- cerr << "LowerLOAD: Got a LoadSDNode with an addr mode other than "
+ {
+ std::string msg;
+ raw_string_ostream Msg(msg);
+ Msg << "LowerLOAD: Got a LoadSDNode with an addr mode other than "
"UNINDEXED\n";
- cerr << (unsigned) SN->getAddressingMode() << "\n";
- abort();
- /*NOTREACHED*/
+ Msg << (unsigned) SN->getAddressingMode();
+ llvm_report_error(Msg.str());
+ /*NOTREACHED*/
+ }
}
return SDValue();
SDValue CPI = DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment());
SDValue Zero = DAG.getConstant(0, PtrVT);
const TargetMachine &TM = DAG.getTarget();
+ // FIXME there is no actual debug info here
+ DebugLoc dl = Op.getDebugLoc();
if (TM.getRelocationModel() == Reloc::Static) {
if (!ST->usingLargeMem()) {
// Just return the SDValue with the constant pool address in it.
- return DAG.getNode(SPUISD::AFormAddr, PtrVT, CPI, Zero);
+ return DAG.getNode(SPUISD::AFormAddr, dl, PtrVT, CPI, Zero);
} else {
- SDValue Hi = DAG.getNode(SPUISD::Hi, PtrVT, CPI, Zero);
- SDValue Lo = DAG.getNode(SPUISD::Lo, PtrVT, CPI, Zero);
- return DAG.getNode(SPUISD::IndirectAddr, PtrVT, Hi, Lo);
+ SDValue Hi = DAG.getNode(SPUISD::Hi, dl, PtrVT, CPI, Zero);
+ SDValue Lo = DAG.getNode(SPUISD::Lo, dl, PtrVT, CPI, Zero);
+ return DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Hi, Lo);
}
}
- assert(0 &&
- "LowerConstantPool: Relocation model other than static"
- " not supported.");
+ llvm_unreachable("LowerConstantPool: Relocation model other than static"
+ " not supported.");
return SDValue();
}
SDValue JTI = DAG.getTargetJumpTable(JT->getIndex(), PtrVT);
SDValue Zero = DAG.getConstant(0, PtrVT);
const TargetMachine &TM = DAG.getTarget();
+ // FIXME there is no actual debug info here
+ DebugLoc dl = Op.getDebugLoc();
if (TM.getRelocationModel() == Reloc::Static) {
if (!ST->usingLargeMem()) {
- return DAG.getNode(SPUISD::AFormAddr, PtrVT, JTI, Zero);
+ return DAG.getNode(SPUISD::AFormAddr, dl, PtrVT, JTI, Zero);
} else {
- SDValue Hi = DAG.getNode(SPUISD::Hi, PtrVT, JTI, Zero);
- SDValue Lo = DAG.getNode(SPUISD::Lo, PtrVT, JTI, Zero);
- return DAG.getNode(SPUISD::IndirectAddr, PtrVT, Hi, Lo);
+ SDValue Hi = DAG.getNode(SPUISD::Hi, dl, PtrVT, JTI, Zero);
+ SDValue Lo = DAG.getNode(SPUISD::Lo, dl, PtrVT, JTI, Zero);
+ return DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Hi, Lo);
}
}
- assert(0 &&
- "LowerJumpTable: Relocation model other than static not supported.");
+ llvm_unreachable("LowerJumpTable: Relocation model other than static"
+ " not supported.");
return SDValue();
}
SDValue GA = DAG.getTargetGlobalAddress(GV, PtrVT, GSDN->getOffset());
const TargetMachine &TM = DAG.getTarget();
SDValue Zero = DAG.getConstant(0, PtrVT);
+ // FIXME there is no actual debug info here
+ DebugLoc dl = Op.getDebugLoc();
if (TM.getRelocationModel() == Reloc::Static) {
if (!ST->usingLargeMem()) {
- return DAG.getNode(SPUISD::AFormAddr, PtrVT, GA, Zero);
+ return DAG.getNode(SPUISD::AFormAddr, dl, PtrVT, GA, Zero);
} else {
- SDValue Hi = DAG.getNode(SPUISD::Hi, PtrVT, GA, Zero);
- SDValue Lo = DAG.getNode(SPUISD::Lo, PtrVT, GA, Zero);
- return DAG.getNode(SPUISD::IndirectAddr, PtrVT, Hi, Lo);
+ SDValue Hi = DAG.getNode(SPUISD::Hi, dl, PtrVT, GA, Zero);
+ SDValue Lo = DAG.getNode(SPUISD::Lo, dl, PtrVT, GA, Zero);
+ return DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, Hi, Lo);
}
} else {
- cerr << "LowerGlobalAddress: Relocation model other than static not "
- << "supported.\n";
- abort();
+ llvm_report_error("LowerGlobalAddress: Relocation model other than static"
+ "not supported.");
/*NOTREACHED*/
}
static SDValue
LowerConstantFP(SDValue Op, SelectionDAG &DAG) {
MVT VT = Op.getValueType();
+ // FIXME there is no actual debug info here
+ DebugLoc dl = Op.getDebugLoc();
if (VT == MVT::f64) {
ConstantFPSDNode *FP = cast<ConstantFPSDNode>(Op.getNode());
uint64_t dbits = DoubleToBits(FP->getValueAPF().convertToDouble());
SDValue T = DAG.getConstant(dbits, MVT::i64);
- SDValue Tvec = DAG.getNode(ISD::BUILD_VECTOR, MVT::v2i64, T, T);
- return DAG.getNode(SPUISD::VEC2PREFSLOT, VT,
- DAG.getNode(ISD::BIT_CONVERT, MVT::v2f64, Tvec));
+ SDValue Tvec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v2i64, T, T);
+ return DAG.getNode(SPUISD::VEC2PREFSLOT, dl, VT,
+ DAG.getNode(ISD::BIT_CONVERT, dl, MVT::v2f64, Tvec));
}
return SDValue();
switch (ObjectVT.getSimpleVT()) {
default: {
- cerr << "LowerFORMAL_ARGUMENTS Unhandled argument type: "
- << ObjectVT.getMVTString()
- << "\n";
- abort();
+ std::string msg;
+ raw_string_ostream Msg(msg);
+ Msg << "LowerFORMAL_ARGUMENTS Unhandled argument type: "
+ << ObjectVT.getMVTString();
+ llvm_report_error(Msg.str());
}
case MVT::i8:
ArgRegClass = &SPU::R8CRegClass;
ArgOffset += StackSlotSize;
}
if (!MemOps.empty())
- Root = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
+ Root = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
&MemOps[0], MemOps.size());
}
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr, PtrOff);
switch (Arg.getValueType().getSimpleVT()) {
- default: assert(0 && "Unexpected ValueType for argument!");
+ default: llvm_unreachable("Unexpected ValueType for argument!");
case MVT::i8:
case MVT::i16:
case MVT::i32:
// and flag operands which copy the outgoing args into the appropriate regs.
SDValue InFlag;
for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) {
- Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
+ Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
RegsToPass[i].second, InFlag);
InFlag = Chain.getValue(1);
}
// This may be an unsafe assumption for JIT and really large compilation
// units.
if (GV->isDeclaration()) {
- Callee = DAG.getNode(SPUISD::AFormAddr, CalleeVT, GA, Zero);
+ Callee = DAG.getNode(SPUISD::AFormAddr, dl, CalleeVT, GA, Zero);
} else {
- Callee = DAG.getNode(SPUISD::PCRelAddr, CalleeVT, GA, Zero);
+ Callee = DAG.getNode(SPUISD::PCRelAddr, dl, CalleeVT, GA, Zero);
}
} else {
// "Large memory" mode: Turn all calls into indirect calls with a X-form
// address pairs:
- Callee = DAG.getNode(SPUISD::IndirectAddr, PtrVT, GA, Zero);
+ Callee = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, GA, Zero);
}
} else if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) {
MVT CalleeVT = Callee.getValueType();
Callee.getValueType());
if (!ST->usingLargeMem()) {
- Callee = DAG.getNode(SPUISD::AFormAddr, CalleeVT, ExtSym, Zero);
+ Callee = DAG.getNode(SPUISD::AFormAddr, dl, CalleeVT, ExtSym, Zero);
} else {
- Callee = DAG.getNode(SPUISD::IndirectAddr, PtrVT, ExtSym, Zero);
+ Callee = DAG.getNode(SPUISD::IndirectAddr, dl, PtrVT, ExtSym, Zero);
}
} else if (SDNode *Dest = isLSAAddress(Callee, DAG)) {
// If this is an absolute destination address that appears to be a legal
// If the call has results, copy the values out of the ret val registers.
switch (TheCall->getValueType(0).getSimpleVT()) {
- default: assert(0 && "Unexpected ret value!");
+ default: llvm_unreachable("Unexpected ret value!");
case MVT::Other: break;
case MVT::i32:
if (TheCall->getValueType(1) == MVT::i32) {
- Chain = DAG.getCopyFromReg(Chain, dl, SPU::R4,
+ Chain = DAG.getCopyFromReg(Chain, dl, SPU::R4,
MVT::i32, InFlag).getValue(1);
ResultVals[0] = Chain.getValue(0);
Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i32,
ResultVals[1] = Chain.getValue(0);
NumResults = 2;
} else {
- Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i32,
+ Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i32,
InFlag).getValue(1);
ResultVals[0] = Chain.getValue(0);
NumResults = 1;
}
break;
case MVT::i64:
- Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i64,
+ Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i64,
InFlag).getValue(1);
ResultVals[0] = Chain.getValue(0);
NumResults = 1;
break;
case MVT::i128:
- Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i128,
+ Chain = DAG.getCopyFromReg(Chain, dl, SPU::R3, MVT::i128,
InFlag).getValue(1);
ResultVals[0] = Chain.getValue(0);
NumResults = 1;
unsigned CC = DAG.getMachineFunction().getFunction()->getCallingConv();
bool isVarArg = DAG.getMachineFunction().getFunction()->isVarArg();
DebugLoc dl = Op.getDebugLoc();
- CCState CCInfo(CC, isVarArg, TM, RVLocs);
+ CCState CCInfo(CC, isVarArg, TM, RVLocs, *DAG.getContext());
CCInfo.AnalyzeReturn(Op.getNode(), RetCC_SPU);
// If this is the first return lowered for this function, add the regs to the
}
}
- return 0; // All UNDEF: use implicit def.; not Constant node
+ return 0;
}
/// get_vec_i18imm - Test if this vector is a vector filled with the same value
return SDValue();
}
-// If this is a vector of constants or undefs, get the bits. A bit in
-// UndefBits is set if the corresponding element of the vector is an
-// ISD::UNDEF value. For undefs, the corresponding VectorBits values are
-// zero. Return true if this is not an array of constants, false if it is.
-//
-static bool GetConstantBuildVectorBits(SDNode *BV, uint64_t VectorBits[2],
- uint64_t UndefBits[2]) {
- // Start with zero'd results.
- VectorBits[0] = VectorBits[1] = UndefBits[0] = UndefBits[1] = 0;
-
- unsigned EltBitSize = BV->getOperand(0).getValueType().getSizeInBits();
- for (unsigned i = 0, e = BV->getNumOperands(); i != e; ++i) {
- SDValue OpVal = BV->getOperand(i);
-
- unsigned PartNo = i >= e/2; // In the upper 128 bits?
- unsigned SlotNo = e/2 - (i & (e/2-1))-1; // Which subpiece of the uint64_t.
-
- uint64_t EltBits = 0;
- if (OpVal.getOpcode() == ISD::UNDEF) {
- uint64_t EltUndefBits = ~0ULL >> (64-EltBitSize);
- UndefBits[PartNo] |= EltUndefBits << (SlotNo*EltBitSize);
- continue;
- } else if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(OpVal)) {
- EltBits = CN->getZExtValue() & (~0ULL >> (64-EltBitSize));
- } else if (ConstantFPSDNode *CN = dyn_cast<ConstantFPSDNode>(OpVal)) {
- const APFloat &apf = CN->getValueAPF();
- EltBits = (CN->getValueType(0) == MVT::f32
- ? FloatToBits(apf.convertToFloat())
- : DoubleToBits(apf.convertToDouble()));
- } else {
- // Nonconstant element.
- return true;
- }
-
- VectorBits[PartNo] |= EltBits << (SlotNo*EltBitSize);
- }
-
- //printf("%llx %llx %llx %llx\n",
- // VectorBits[0], VectorBits[1], UndefBits[0], UndefBits[1]);
- return false;
-}
-
-/// If this is a splat (repetition) of a value across the whole vector, return
-/// the smallest size that splats it. For example, "0x01010101010101..." is a
-/// splat of 0x01, 0x0101, and 0x01010101. We return SplatBits = 0x01 and
-/// SplatSize = 1 byte.
-static bool isConstantSplat(const uint64_t Bits128[2],
- const uint64_t Undef128[2],
- int MinSplatBits,
- uint64_t &SplatBits, uint64_t &SplatUndef,
- int &SplatSize) {
- // Don't let undefs prevent splats from matching. See if the top 64-bits are
- // the same as the lower 64-bits, ignoring undefs.
- uint64_t Bits64 = Bits128[0] | Bits128[1];
- uint64_t Undef64 = Undef128[0] & Undef128[1];
- uint32_t Bits32 = uint32_t(Bits64) | uint32_t(Bits64 >> 32);
- uint32_t Undef32 = uint32_t(Undef64) & uint32_t(Undef64 >> 32);
- uint16_t Bits16 = uint16_t(Bits32) | uint16_t(Bits32 >> 16);
- uint16_t Undef16 = uint16_t(Undef32) & uint16_t(Undef32 >> 16);
-
- if ((Bits128[0] & ~Undef128[1]) == (Bits128[1] & ~Undef128[0])) {
- if (MinSplatBits < 64) {
-
- // Check that the top 32-bits are the same as the lower 32-bits, ignoring
- // undefs.
- if ((Bits64 & (~Undef64 >> 32)) == ((Bits64 >> 32) & ~Undef64)) {
- if (MinSplatBits < 32) {
-
- // If the top 16-bits are different than the lower 16-bits, ignoring
- // undefs, we have an i32 splat.
- if ((Bits32 & (~Undef32 >> 16)) == ((Bits32 >> 16) & ~Undef32)) {
- if (MinSplatBits < 16) {
- // If the top 8-bits are different than the lower 8-bits, ignoring
- // undefs, we have an i16 splat.
- if ((Bits16 & (uint16_t(~Undef16) >> 8))
- == ((Bits16 >> 8) & ~Undef16)) {
- // Otherwise, we have an 8-bit splat.
- SplatBits = uint8_t(Bits16) | uint8_t(Bits16 >> 8);
- SplatUndef = uint8_t(Undef16) & uint8_t(Undef16 >> 8);
- SplatSize = 1;
- return true;
- }
- } else {
- SplatBits = Bits16;
- SplatUndef = Undef16;
- SplatSize = 2;
- return true;
- }
- }
- } else {
- SplatBits = Bits32;
- SplatUndef = Undef32;
- SplatSize = 4;
- return true;
- }
- }
- } else {
- SplatBits = Bits128[0];
- SplatUndef = Undef128[0];
- SplatSize = 8;
- return true;
- }
- }
-
- return false; // Can't be a splat if two pieces don't match.
-}
-
//! Lower a BUILD_VECTOR instruction creatively:
SDValue
LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) {
MVT VT = Op.getValueType();
+ MVT EltVT = VT.getVectorElementType();
DebugLoc dl = Op.getDebugLoc();
- // If this is a vector of constants or undefs, get the bits. A bit in
- // UndefBits is set if the corresponding element of the vector is an
- // ISD::UNDEF value. For undefs, the corresponding VectorBits values are
- // zero.
- uint64_t VectorBits[2];
- uint64_t UndefBits[2];
- uint64_t SplatBits, SplatUndef;
- int SplatSize;
- if (GetConstantBuildVectorBits(Op.getNode(), VectorBits, UndefBits)
- || !isConstantSplat(VectorBits, UndefBits,
- VT.getVectorElementType().getSizeInBits(),
- SplatBits, SplatUndef, SplatSize))
- return SDValue(); // Not a constant vector, not a splat.
+ BuildVectorSDNode *BCN = dyn_cast<BuildVectorSDNode>(Op.getNode());
+ assert(BCN != 0 && "Expected BuildVectorSDNode in SPU LowerBUILD_VECTOR");
+ unsigned minSplatBits = EltVT.getSizeInBits();
+
+ if (minSplatBits < 16)
+ minSplatBits = 16;
+
+ APInt APSplatBits, APSplatUndef;
+ unsigned SplatBitSize;
+ bool HasAnyUndefs;
+
+ if (!BCN->isConstantSplat(APSplatBits, APSplatUndef, SplatBitSize,
+ HasAnyUndefs, minSplatBits)
+ || minSplatBits < SplatBitSize)
+ return SDValue(); // Wasn't a constant vector or splat exceeded min
+
+ uint64_t SplatBits = APSplatBits.getZExtValue();
switch (VT.getSimpleVT()) {
- default:
- cerr << "CellSPU: Unhandled VT in LowerBUILD_VECTOR, VT = "
- << VT.getMVTString()
- << "\n";
- abort();
+ default: {
+ std::string msg;
+ raw_string_ostream Msg(msg);
+ Msg << "CellSPU: Unhandled VT in LowerBUILD_VECTOR, VT = "
+ << VT.getMVTString();
+ llvm_report_error(Msg.str());
/*NOTREACHED*/
+ }
case MVT::v4f32: {
uint32_t Value32 = uint32_t(SplatBits);
- assert(SplatSize == 4
+ assert(SplatBitSize == 32
&& "LowerBUILD_VECTOR: Unexpected floating point vector element.");
// NOTE: pretend the constant is an integer. LLVM won't load FP constants
SDValue T = DAG.getConstant(Value32, MVT::i32);
return DAG.getNode(ISD::BIT_CONVERT, dl, MVT::v4f32,
- DAG.getNode(ISD::BUILD_VECTOR, dl,
- MVT::v4i32, T, T, T, T));
+ DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, T,T,T,T));
break;
}
case MVT::v2f64: {
uint64_t f64val = uint64_t(SplatBits);
- assert(SplatSize == 8
+ assert(SplatBitSize == 64
&& "LowerBUILD_VECTOR: 64-bit float vector size > 8 bytes.");
// NOTE: pretend the constant is an integer. LLVM won't load FP constants
SDValue T = DAG.getConstant(f64val, MVT::i64);
}
case MVT::v16i8: {
// 8-bit constants have to be expanded to 16-bits
- unsigned short Value16 = SplatBits | (SplatBits << 8);
- SDValue Ops[8];
- for (int i = 0; i < 8; ++i)
- Ops[i] = DAG.getConstant(Value16, MVT::i16);
+ unsigned short Value16 = SplatBits /* | (SplatBits << 8) */;
+ SmallVector<SDValue, 8> Ops;
+
+ Ops.assign(8, DAG.getConstant(Value16, MVT::i16));
return DAG.getNode(ISD::BIT_CONVERT, dl, VT,
- DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i16, Ops, 8));
+ DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i16, &Ops[0], Ops.size()));
}
case MVT::v8i16: {
- unsigned short Value16;
- if (SplatSize == 2)
- Value16 = (unsigned short) (SplatBits & 0xffff);
- else
- Value16 = (unsigned short) (SplatBits | (SplatBits << 8));
- SDValue T = DAG.getConstant(Value16, VT.getVectorElementType());
- SDValue Ops[8];
- for (int i = 0; i < 8; ++i) Ops[i] = T;
- return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ops, 8);
+ unsigned short Value16 = SplatBits;
+ SDValue T = DAG.getConstant(Value16, EltVT);
+ SmallVector<SDValue, 8> Ops;
+
+ Ops.assign(8, T);
+ return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Ops[0], Ops.size());
}
case MVT::v4i32: {
- unsigned int Value = SplatBits;
- SDValue T = DAG.getConstant(Value, VT.getVectorElementType());
+ SDValue T = DAG.getConstant(unsigned(SplatBits), VT.getVectorElementType());
return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, T, T, T, T);
}
case MVT::v2i32: {
- unsigned int Value = SplatBits;
- SDValue T = DAG.getConstant(Value, VT.getVectorElementType());
+ SDValue T = DAG.getConstant(unsigned(SplatBits), VT.getVectorElementType());
return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, T, T);
}
case MVT::v2i64: {
- return SPU::LowerSplat_v2i64(VT, DAG, SplatBits, dl);
+ return SPU::LowerV2I64Splat(VT, DAG, SplatBits, dl);
}
}
return SDValue();
}
+/*!
+ */
SDValue
-SPU::LowerSplat_v2i64(MVT OpVT, SelectionDAG& DAG, uint64_t SplatVal,
- DebugLoc dl) {
+SPU::LowerV2I64Splat(MVT OpVT, SelectionDAG& DAG, uint64_t SplatVal,
+ DebugLoc dl) {
uint32_t upper = uint32_t(SplatVal >> 32);
uint32_t lower = uint32_t(SplatVal);
DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
Val, Val, Val, Val));
} else {
- SDValue LO32;
- SDValue HI32;
- SmallVector<SDValue, 16> ShufBytes;
- SDValue Result;
bool upper_special, lower_special;
// NOTE: This code creates common-case shuffle masks that can be easily
upper_special = (upper == 0 || upper == 0xffffffff || upper == 0x80000000);
lower_special = (lower == 0 || lower == 0xffffffff || lower == 0x80000000);
+ // Both upper and lower are special, lower to a constant pool load:
+ if (lower_special && upper_special) {
+ SDValue SplatValCN = DAG.getConstant(SplatVal, MVT::i64);
+ return DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v2i64,
+ SplatValCN, SplatValCN);
+ }
+
+ SDValue LO32;
+ SDValue HI32;
+ SmallVector<SDValue, 16> ShufBytes;
+ SDValue Result;
+
// Create lower vector if not a special pattern
if (!lower_special) {
SDValue LO32C = DAG.getConstant(lower, MVT::i32);
LO32 = HI32;
if (upper_special)
HI32 = LO32;
- if (lower_special && upper_special) {
- // Unhappy situation... both upper and lower are special, so punt with
- // a target constant:
- SDValue Zero = DAG.getConstant(0, MVT::i32);
- HI32 = LO32 = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Zero, Zero,
- Zero, Zero);
- }
for (int i = 0; i < 4; ++i) {
uint64_t val = 0;
/// \note
/// SPUISD::SHUFB is eventually selected as Cell's <i>shufb</i> instructions.
static SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) {
+ const ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(Op);
SDValue V1 = Op.getOperand(0);
SDValue V2 = Op.getOperand(1);
- SDValue PermMask = Op.getOperand(2);
DebugLoc dl = Op.getDebugLoc();
if (V2.getOpcode() == ISD::UNDEF) V2 = V1;
} else if (EltVT == MVT::i64 || EltVT == MVT::f64) {
V2EltIdx0 = 2;
} else
- assert(0 && "Unhandled vector type in LowerVECTOR_SHUFFLE");
+ llvm_unreachable("Unhandled vector type in LowerVECTOR_SHUFFLE");
- for (unsigned i = 0; i != PermMask.getNumOperands(); ++i) {
- if (PermMask.getOperand(i).getOpcode() != ISD::UNDEF) {
- unsigned SrcElt = cast<ConstantSDNode > (PermMask.getOperand(i))->getZExtValue();
+ for (unsigned i = 0; i != MaxElts; ++i) {
+ if (SVN->getMaskElt(i) < 0)
+ continue;
+
+ unsigned SrcElt = SVN->getMaskElt(i);
- if (monotonic) {
- if (SrcElt >= V2EltIdx0) {
- if (1 >= (++EltsFromV2)) {
- V2Elt = (V2EltIdx0 - SrcElt) << 2;
- }
- } else if (CurrElt != SrcElt) {
- monotonic = false;
+ if (monotonic) {
+ if (SrcElt >= V2EltIdx0) {
+ if (1 >= (++EltsFromV2)) {
+ V2Elt = (V2EltIdx0 - SrcElt) << 2;
}
-
- ++CurrElt;
+ } else if (CurrElt != SrcElt) {
+ monotonic = false;
}
- if (rotate) {
- if (PrevElt > 0 && SrcElt < MaxElts) {
- if ((PrevElt == SrcElt - 1)
- || (PrevElt == MaxElts - 1 && SrcElt == 0)) {
- PrevElt = SrcElt;
- if (SrcElt == 0)
- V0Elt = i;
- } else {
- rotate = false;
- }
- } else if (PrevElt == 0) {
- // First time through, need to keep track of previous element
+ ++CurrElt;
+ }
+
+ if (rotate) {
+ if (PrevElt > 0 && SrcElt < MaxElts) {
+ if ((PrevElt == SrcElt - 1)
+ || (PrevElt == MaxElts - 1 && SrcElt == 0)) {
PrevElt = SrcElt;
+ if (SrcElt == 0)
+ V0Elt = i;
} else {
- // This isn't a rotation, takes elements from vector 2
rotate = false;
}
+ } else if (PrevElt == 0) {
+ // First time through, need to keep track of previous element
+ PrevElt = SrcElt;
+ } else {
+ // This isn't a rotation, takes elements from vector 2
+ rotate = false;
}
}
}
DAG.getTargetConstant(V2Elt, MVT::i32),
DAG.getCopyFromReg(InitTempReg, dl, VReg, PtrVT));
// Use shuffle mask in SHUFB synthetic instruction:
- return DAG.getNode(SPUISD::SHUFB, dl, V1.getValueType(), V2, V1,
+ return DAG.getNode(SPUISD::SHUFB, dl, V1.getValueType(), V2, V1,
ShufMaskOp);
} else if (rotate) {
int rotamt = (MaxElts - V0Elt) * EltVT.getSizeInBits()/8;
unsigned BytesPerElement = EltVT.getSizeInBits()/8;
SmallVector<SDValue, 16> ResultMask;
- for (unsigned i = 0, e = PermMask.getNumOperands(); i != e; ++i) {
- unsigned SrcElt;
- if (PermMask.getOperand(i).getOpcode() == ISD::UNDEF)
- SrcElt = 0;
- else
- SrcElt = cast<ConstantSDNode>(PermMask.getOperand(i))->getZExtValue();
-
- for (unsigned j = 0; j < BytesPerElement; ++j) {
- ResultMask.push_back(DAG.getConstant(SrcElt*BytesPerElement+j,
- MVT::i8));
- }
+ for (unsigned i = 0, e = MaxElts; i != e; ++i) {
+ unsigned SrcElt = SVN->getMaskElt(i) < 0 ? 0 : SVN->getMaskElt(i);
+
+ for (unsigned j = 0; j < BytesPerElement; ++j)
+ ResultMask.push_back(DAG.getConstant(SrcElt*BytesPerElement+j,MVT::i8));
}
SDValue VPermMask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v16i8,
// Create a constant vector:
switch (Op.getValueType().getSimpleVT()) {
- default: assert(0 && "Unexpected constant value type in "
- "LowerSCALAR_TO_VECTOR");
+ default: llvm_unreachable("Unexpected constant value type in "
+ "LowerSCALAR_TO_VECTOR");
case MVT::v16i8: n_copies = 16; VT = MVT::i8; break;
case MVT::v8i16: n_copies = 8; VT = MVT::i16; break;
case MVT::v4i32: n_copies = 4; VT = MVT::i32; break;
} else {
// Otherwise, copy the value from one register to another:
switch (Op0.getValueType().getSimpleVT()) {
- default: assert(0 && "Unexpected value type in LowerSCALAR_TO_VECTOR");
+ default: llvm_unreachable("Unexpected value type in LowerSCALAR_TO_VECTOR");
case MVT::i8:
case MVT::i16:
case MVT::i32:
// sanity checks:
if (VT == MVT::i8 && EltNo >= 16)
- assert(0 && "SPU LowerEXTRACT_VECTOR_ELT: i8 extraction slot > 15");
+ llvm_unreachable("SPU LowerEXTRACT_VECTOR_ELT: i8 extraction slot > 15");
else if (VT == MVT::i16 && EltNo >= 8)
- assert(0 && "SPU LowerEXTRACT_VECTOR_ELT: i16 extraction slot > 7");
+ llvm_unreachable("SPU LowerEXTRACT_VECTOR_ELT: i16 extraction slot > 7");
else if (VT == MVT::i32 && EltNo >= 4)
- assert(0 && "SPU LowerEXTRACT_VECTOR_ELT: i32 extraction slot > 4");
+ llvm_unreachable("SPU LowerEXTRACT_VECTOR_ELT: i32 extraction slot > 4");
else if (VT == MVT::i64 && EltNo >= 2)
- assert(0 && "SPU LowerEXTRACT_VECTOR_ELT: i64 extraction slot > 2");
+ llvm_unreachable("SPU LowerEXTRACT_VECTOR_ELT: i64 extraction slot > 2");
if (EltNo == 0 && (VT == MVT::i32 || VT == MVT::i64)) {
// i32 and i64: Element 0 is the preferred slot
ShufMask[i] = DAG.getConstant(bits, MVT::i32);
}
- SDValue ShufMaskVec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
- &ShufMask[0],
- sizeof(ShufMask) / sizeof(ShufMask[0]));
+ SDValue ShufMaskVec =
+ DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
+ &ShufMask[0], sizeof(ShufMask)/sizeof(ShufMask[0]));
retval = DAG.getNode(SPUISD::VEC2PREFSLOT, dl, VT,
DAG.getNode(SPUISD::SHUFB, dl, N.getValueType(),
// slot 0 across the vector
MVT VecVT = N.getValueType();
if (!VecVT.isSimple() || !VecVT.isVector() || !VecVT.is128BitVector()) {
- cerr << "LowerEXTRACT_VECTOR_ELT: Must have a simple, 128-bit vector type!\n";
- abort();
+ llvm_report_error("LowerEXTRACT_VECTOR_ELT: Must have a simple, 128-bit"
+ "vector type!");
}
// Make life easier by making sure the index is zero-extended to i32
switch (VT.getSimpleVT()) {
default:
- cerr << "LowerEXTRACT_VECTOR_ELT(varable): Unhandled vector type\n";
- abort();
+ llvm_report_error("LowerEXTRACT_VECTOR_ELT(varable): Unhandled vector"
+ "type");
/*NOTREACHED*/
case MVT::i8: {
SDValue factor = DAG.getConstant(0x00000000, MVT::i32);
- replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, factor, factor,
- factor, factor);
+ replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
+ factor, factor, factor, factor);
break;
}
case MVT::i16: {
SDValue factor = DAG.getConstant(0x00010001, MVT::i32);
- replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, factor, factor,
- factor, factor);
+ replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
+ factor, factor, factor, factor);
break;
}
case MVT::i32:
case MVT::f32: {
SDValue factor = DAG.getConstant(0x00010203, MVT::i32);
- replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, factor, factor,
- factor, factor);
+ replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
+ factor, factor, factor, factor);
break;
}
case MVT::i64:
case MVT::f64: {
SDValue loFactor = DAG.getConstant(0x00010203, MVT::i32);
SDValue hiFactor = DAG.getConstant(0x04050607, MVT::i32);
- replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
+ replicate = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
loFactor, hiFactor, loFactor, hiFactor);
break;
}
assert(Op.getValueType() == MVT::i8);
switch (Opc) {
default:
- assert(0 && "Unhandled i8 math operator");
+ llvm_unreachable("Unhandled i8 math operator");
/*NOTREACHED*/
break;
case ISD::ADD: {
case ISD::ROTR:
case ISD::ROTL: {
SDValue N1 = Op.getOperand(1);
- unsigned N1Opc;
- N0 = (N0.getOpcode() != ISD::Constant
- ? DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i16, N0)
- : DAG.getConstant(cast<ConstantSDNode>(N0)->getZExtValue(),
- MVT::i16));
- N1Opc = N1.getValueType().bitsLT(ShiftVT)
- ? ISD::ZERO_EXTEND
- : ISD::TRUNCATE;
- N1 = (N1.getOpcode() != ISD::Constant
- ? DAG.getNode(N1Opc, dl, ShiftVT, N1)
- : DAG.getConstant(cast<ConstantSDNode>(N1)->getZExtValue(),
- TLI.getShiftAmountTy()));
+ MVT N1VT = N1.getValueType();
+
+ N0 = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i16, N0);
+ if (!N1VT.bitsEq(ShiftVT)) {
+ unsigned N1Opc = N1.getValueType().bitsLT(ShiftVT)
+ ? ISD::ZERO_EXTEND
+ : ISD::TRUNCATE;
+ N1 = DAG.getNode(N1Opc, dl, ShiftVT, N1);
+ }
+
+ // Replicate lower 8-bits into upper 8:
SDValue ExpandArg =
DAG.getNode(ISD::OR, dl, MVT::i16, N0,
DAG.getNode(ISD::SHL, dl, MVT::i16,
N0, DAG.getConstant(8, MVT::i32)));
+
+ // Truncate back down to i8
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i8,
DAG.getNode(Opc, dl, MVT::i16, ExpandArg, N1));
}
case ISD::SRL:
case ISD::SHL: {
SDValue N1 = Op.getOperand(1);
- unsigned N1Opc;
- N0 = (N0.getOpcode() != ISD::Constant
- ? DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i16, N0)
- : DAG.getConstant(cast<ConstantSDNode>(N0)->getZExtValue(),
- MVT::i32));
- N1Opc = N1.getValueType().bitsLT(ShiftVT)
- ? ISD::ZERO_EXTEND
- : ISD::TRUNCATE;
- N1 = (N1.getOpcode() != ISD::Constant
- ? DAG.getNode(N1Opc, dl, ShiftVT, N1)
- : DAG.getConstant(cast<ConstantSDNode>(N1)->getZExtValue(), ShiftVT));
+ MVT N1VT = N1.getValueType();
+
+ N0 = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i16, N0);
+ if (!N1VT.bitsEq(ShiftVT)) {
+ unsigned N1Opc = ISD::ZERO_EXTEND;
+
+ if (N1.getValueType().bitsGT(ShiftVT))
+ N1Opc = ISD::TRUNCATE;
+
+ N1 = DAG.getNode(N1Opc, dl, ShiftVT, N1);
+ }
+
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i8,
DAG.getNode(Opc, dl, MVT::i16, N0, N1));
}
case ISD::SRA: {
SDValue N1 = Op.getOperand(1);
- unsigned N1Opc;
- N0 = (N0.getOpcode() != ISD::Constant
- ? DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::i16, N0)
- : DAG.getConstant(cast<ConstantSDNode>(N0)->getSExtValue(),
- MVT::i16));
- N1Opc = N1.getValueType().bitsLT(ShiftVT)
- ? ISD::SIGN_EXTEND
- : ISD::TRUNCATE;
- N1 = (N1.getOpcode() != ISD::Constant
- ? DAG.getNode(N1Opc, dl, ShiftVT, N1)
- : DAG.getConstant(cast<ConstantSDNode>(N1)->getZExtValue(),
- ShiftVT));
+ MVT N1VT = N1.getValueType();
+
+ N0 = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::i16, N0);
+ if (!N1VT.bitsEq(ShiftVT)) {
+ unsigned N1Opc = ISD::SIGN_EXTEND;
+
+ if (N1VT.bitsGT(ShiftVT))
+ N1Opc = ISD::TRUNCATE;
+ N1 = DAG.getNode(N1Opc, dl, ShiftVT, N1);
+ }
+
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i8,
DAG.getNode(Opc, dl, MVT::i16, N0, N1));
}
case ISD::MUL: {
SDValue N1 = Op.getOperand(1);
- unsigned N1Opc;
- N0 = (N0.getOpcode() != ISD::Constant
- ? DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::i16, N0)
- : DAG.getConstant(cast<ConstantSDNode>(N0)->getZExtValue(),
- MVT::i16));
- N1Opc = N1.getValueType().bitsLT(MVT::i16) ? ISD::SIGN_EXTEND : ISD::TRUNCATE;
- N1 = (N1.getOpcode() != ISD::Constant
- ? DAG.getNode(N1Opc, dl, MVT::i16, N1)
- : DAG.getConstant(cast<ConstantSDNode>(N1)->getSExtValue(),
- MVT::i16));
+
+ N0 = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::i16, N0);
+ N1 = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::i16, N1);
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i8,
DAG.getNode(Opc, dl, MVT::i16, N0, N1));
break;
return SDValue();
}
-//! Generate the carry-generate shuffle mask.
-SDValue SPU::getCarryGenerateShufMask(SelectionDAG &DAG, DebugLoc dl) {
- SmallVector<SDValue, 16 > ShufBytes;
-
- // Create the shuffle mask for "rotating" the borrow up one register slot
- // once the borrow is generated.
- ShufBytes.push_back(DAG.getConstant(0x04050607, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0x80808080, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0x0c0d0e0f, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0x80808080, MVT::i32));
-
- return DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
- &ShufBytes[0], ShufBytes.size());
-}
-
-//! Generate the borrow-generate shuffle mask
-SDValue SPU::getBorrowGenerateShufMask(SelectionDAG &DAG, DebugLoc dl) {
- SmallVector<SDValue, 16 > ShufBytes;
-
- // Create the shuffle mask for "rotating" the borrow up one register slot
- // once the borrow is generated.
- ShufBytes.push_back(DAG.getConstant(0x04050607, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0xc0c0c0c0, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0x0c0d0e0f, MVT::i32));
- ShufBytes.push_back(DAG.getConstant(0xc0c0c0c0, MVT::i32));
-
- return DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
- &ShufBytes[0], ShufBytes.size());
-}
-
//! Lower byte immediate operations for v16i8 vectors:
static SDValue
LowerByteImmed(SDValue Op, SelectionDAG &DAG) {
}
if (ConstVec.getNode()->getOpcode() == ISD::BUILD_VECTOR) {
- uint64_t VectorBits[2];
- uint64_t UndefBits[2];
- uint64_t SplatBits, SplatUndef;
- int SplatSize;
-
- if (!GetConstantBuildVectorBits(ConstVec.getNode(), VectorBits, UndefBits)
- && isConstantSplat(VectorBits, UndefBits,
- VT.getVectorElementType().getSizeInBits(),
- SplatBits, SplatUndef, SplatSize)) {
- SDValue tcVec[16];
+ BuildVectorSDNode *BCN = dyn_cast<BuildVectorSDNode>(ConstVec.getNode());
+ assert(BCN != 0 && "Expected BuildVectorSDNode in SPU LowerByteImmed");
+
+ APInt APSplatBits, APSplatUndef;
+ unsigned SplatBitSize;
+ bool HasAnyUndefs;
+ unsigned minSplatBits = VT.getVectorElementType().getSizeInBits();
+
+ if (BCN->isConstantSplat(APSplatBits, APSplatUndef, SplatBitSize,
+ HasAnyUndefs, minSplatBits)
+ && minSplatBits <= SplatBitSize) {
+ uint64_t SplatBits = APSplatBits.getZExtValue();
SDValue tc = DAG.getTargetConstant(SplatBits & 0xff, MVT::i8);
- const size_t tcVecSize = sizeof(tcVec) / sizeof(tcVec[0]);
-
- // Turn the BUILD_VECTOR into a set of target constants:
- for (size_t i = 0; i < tcVecSize; ++i)
- tcVec[i] = tc;
+ SmallVector<SDValue, 16> tcVec;
+ tcVec.assign(16, tc);
return DAG.getNode(Op.getNode()->getOpcode(), dl, VT, Arg,
- DAG.getNode(ISD::BUILD_VECTOR, dl, VT,
- tcVec, tcVecSize));
+ DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &tcVec[0], tcVec.size()));
}
}
SDValue Comp1 =
DAG.getNode(ISD::SRL, dl, MVT::i32,
- DAG.getCopyFromReg(CNTB_rescopy, dl, CNTB_reg, MVT::i32),
+ DAG.getCopyFromReg(CNTB_rescopy, dl, CNTB_reg, MVT::i32),
Shift1);
SDValue Sum1 =
return ExpandLibCall(LC, Op, DAG, false, Dummy, TLI);
}
- return Op; // return unmolested, legalized op
+ return Op;
}
//! Lower ISD::SINT_TO_FP, ISD::UINT_TO_FP for i32
return ExpandLibCall(LC, Op, DAG, false, Dummy, TLI);
}
- return Op; // return unmolested, legalized
+ return Op;
}
//! Lower ISD::SETCC
static SDValue LowerSETCC(SDValue Op, SelectionDAG &DAG,
const TargetLowering &TLI) {
CondCodeSDNode *CC = dyn_cast<CondCodeSDNode>(Op.getOperand(2));
- DebugLoc dl = Op.getNode()->getDebugLoc();
+ DebugLoc dl = Op.getDebugLoc();
assert(CC != 0 && "LowerSETCC: CondCodeSDNode should not be null here!\n");
SDValue lhs = Op.getOperand(0);
ISD::SETGT));
}
- SDValue i64rhs = DAG.getNode(ISD::BIT_CONVERT, IntVT, rhs);
+ SDValue i64rhs = DAG.getNode(ISD::BIT_CONVERT, dl, IntVT, rhs);
SDValue rhsHi32 =
DAG.getNode(ISD::TRUNCATE, dl, MVT::i32,
DAG.getNode(ISD::SRL, dl, IntVT,
case ISD::SETONE:
compareOp = ISD::SETNE; break;
default:
- cerr << "CellSPU ISel Select: unimplemented f64 condition\n";
- abort();
- break;
+ llvm_report_error("CellSPU ISel Select: unimplemented f64 condition");
}
SDValue result =
- DAG.getSetCC(dl, ccResultVT, lhsSelect, rhsSelect,
+ DAG.getSetCC(dl, ccResultVT, lhsSelect, rhsSelect,
(ISD::CondCode) compareOp);
if ((CC->get() & 0x8) == 0) {
SDValue trueval = Op.getOperand(2);
SDValue falseval = Op.getOperand(3);
SDValue condition = Op.getOperand(4);
+ DebugLoc dl = Op.getDebugLoc();
// NOTE: SELB's arguments: $rA, $rB, $mask
//
// legalizer insists on combining SETCC/SELECT into SELECT_CC, so we end up
// with another "cannot select select_cc" assert:
- SDValue compare = DAG.getNode(ISD::SETCC,
+ SDValue compare = DAG.getNode(ISD::SETCC, dl,
TLI.getSetCCResultType(Op.getValueType()),
lhs, rhs, condition);
- return DAG.getNode(SPUISD::SELB, VT, falseval, trueval, compare);
+ return DAG.getNode(SPUISD::SELB, dl, VT, falseval, trueval, compare);
}
//! Custom lower ISD::TRUNCATE
static SDValue LowerTRUNCATE(SDValue Op, SelectionDAG &DAG)
{
+ // Type to truncate to
MVT VT = Op.getValueType();
MVT::SimpleValueType simpleVT = VT.getSimpleVT();
MVT VecVT = MVT::getVectorVT(VT, (128 / VT.getSizeInBits()));
+ DebugLoc dl = Op.getDebugLoc();
+ // Type to truncate from
SDValue Op0 = Op.getOperand(0);
MVT Op0VT = Op0.getValueType();
- MVT Op0VecVT = MVT::getVectorVT(Op0VT, (128 / Op0VT.getSizeInBits()));
if (Op0VT.getSimpleVT() == MVT::i128 && simpleVT == MVT::i64) {
// Create shuffle mask, least significant doubleword of quadword
unsigned maskHigh = 0x08090a0b;
unsigned maskLow = 0x0c0d0e0f;
// Use a shuffle to perform the truncation
- SDValue shufMask = DAG.getNode(ISD::BUILD_VECTOR, MVT::v4i32,
+ SDValue shufMask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32,
DAG.getConstant(maskHigh, MVT::i32),
DAG.getConstant(maskLow, MVT::i32),
DAG.getConstant(maskHigh, MVT::i32),
DAG.getConstant(maskLow, MVT::i32));
+ SDValue truncShuffle = DAG.getNode(SPUISD::SHUFB, dl, VecVT,
+ Op0, Op0, shufMask);
- SDValue PromoteScalar = DAG.getNode(SPUISD::PREFSLOT2VEC, Op0VecVT, Op0);
-
- SDValue truncShuffle = DAG.getNode(SPUISD::SHUFB, Op0VecVT,
- PromoteScalar, PromoteScalar, shufMask);
-
- return DAG.getNode(SPUISD::VEC2PREFSLOT, VT,
- DAG.getNode(ISD::BIT_CONVERT, VecVT, truncShuffle));
+ return DAG.getNode(SPUISD::VEC2PREFSLOT, dl, VT, truncShuffle);
}
return SDValue(); // Leave the truncate unmolested
switch (Opc) {
default: {
+#ifndef NDEBUG
cerr << "SPUTargetLowering::LowerOperation(): need to lower this!\n";
cerr << "Op.getOpcode() = " << Opc << "\n";
cerr << "*Op.getNode():\n";
Op.getNode()->dump();
- abort();
+#endif
+ llvm_unreachable(0);
}
case ISD::LOAD:
case ISD::EXTLOAD:
MVT NodeVT = N->getValueType(0); // The node's value type
MVT Op0VT = Op0.getValueType(); // The first operand's result
SDValue Result; // Initially, empty result
+ DebugLoc dl = N->getDebugLoc();
switch (N->getOpcode()) {
default: break;
}
#endif
- return DAG.getNode(SPUISD::IndirectAddr, Op0VT,
+ return DAG.getNode(SPUISD::IndirectAddr, dl, Op0VT,
IndirectArg, combinedValue);
}
}
}
#endif
- return DAG.getNode(SPUISD::IndirectAddr, Op0VT,
+ return DAG.getNode(SPUISD::IndirectAddr, dl, Op0VT,
Op0.getOperand(0), Op0.getOperand(1));
}
}
const SelectionDAG &DAG,
unsigned Depth ) const {
#if 0
- const uint64_t uint64_sizebits = sizeof(uint64_t) * 8;
+ const uint64_t uint64_sizebits = sizeof(uint64_t) * CHAR_BIT;
switch (Op.getOpcode()) {
default: