//
// The LLVM Compiler Infrastructure
//
-// This file was developed by Nate Begeman and is distributed under the
-// University of Illinois Open Source License. See LICENSE.TXT for details.
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This pass combines dag nodes to form fewer, simpler DAG nodes. It can be run
// both before and after the DAG is legalized.
-//
-// FIXME: Missing folds
-// sdiv, udiv, srem, urem (X, const) where X is an integer can be expanded into
-// a sequence of multiplies, shifts, and adds. This should be controlled by
-// some kind of hint from the target that int div is expensive.
-// various folds of mulh[s,u] by constants such as -1, powers of 2, etc.
-//
-// FIXME: select C, pow2, pow2 -> something smart
-// FIXME: trunc(select X, Y, Z) -> select X, trunc(Y), trunc(Z)
-// FIXME: Dead stores -> nuke
-// FIXME: shr X, (and Y,31) -> shr X, Y (TRICKY!)
-// FIXME: mul (x, const) -> shifts + adds
-// FIXME: undef values
-// FIXME: divide by zero is currently left unfolded. do we want to turn this
-// into an undef?
-// FIXME: select ne (select cc, 1, 0), 0, true, false -> select cc, true, false
//
//===----------------------------------------------------------------------===//
#define DEBUG_TYPE "dagcombine"
#include "llvm/CodeGen/SelectionDAG.h"
+#include "llvm/CodeGen/MachineFunction.h"
+#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Target/TargetData.h"
+#include "llvm/Target/TargetFrameInfo.h"
#include "llvm/Target/TargetLowering.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Target/TargetOptions.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/Statistic.h"
-#include "llvm/Support/Alignment.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
SDOperand To[] = { Res0, Res1 };
return CombineTo(N, To, 2, AddTo);
}
+
private:
/// SimplifyDemandedBits - Check the specified integer node value to see if
/// it can be simplified or if things it uses can be simplified by bit
/// propagation. If so, return true.
bool SimplifyDemandedBits(SDOperand Op, uint64_t Demanded = ~0ULL) {
- TargetLowering::TargetLoweringOpt TLO(DAG);
+ TargetLowering::TargetLoweringOpt TLO(DAG, AfterLegalize);
uint64_t KnownZero, KnownOne;
Demanded &= MVT::getIntVTBitMask(Op.getValueType());
if (!TLI.SimplifyDemandedBits(Op, Demanded, KnownZero, KnownOne, TLO))
SDOperand XformToShuffleWithZero(SDNode *N);
SDOperand ReassociateOps(unsigned Opc, SDOperand LHS, SDOperand RHS);
+ SDOperand visitShiftByConstant(SDNode *N, unsigned Amt);
+
bool SimplifySelectOps(SDNode *SELECT, SDOperand LHS, SDOperand RHS);
SDOperand SimplifyBinOpWithSameOpcodeHands(SDNode *N);
SDOperand SimplifySelect(SDOperand N0, SDOperand N1, SDOperand N2);
bool NotExtCompare = false);
SDOperand SimplifySetCC(MVT::ValueType VT, SDOperand N0, SDOperand N1,
ISD::CondCode Cond, bool foldBooleans = true);
- bool SimplifyNodeWithTwoResults(SDNode *N, unsigned LoOp, unsigned HiOp);
+ SDOperand SimplifyNodeWithTwoResults(SDNode *N, unsigned LoOp,
+ unsigned HiOp);
SDOperand ConstantFoldBIT_CONVERTofBUILD_VECTOR(SDNode *, MVT::ValueType);
SDOperand BuildSDIV(SDNode *N);
SDOperand BuildUDIV(SDNode *N);
GetNegatedExpression(Op.getOperand(1), DAG, Depth+1));
case ISD::FP_EXTEND:
- case ISD::FP_ROUND:
case ISD::FSIN:
return DAG.getNode(Op.getOpcode(), Op.getValueType(),
GetNegatedExpression(Op.getOperand(0), DAG, Depth+1));
+ case ISD::FP_ROUND:
+ return DAG.getNode(ISD::FP_ROUND, Op.getValueType(),
+ GetNegatedExpression(Op.getOperand(0), DAG, Depth+1),
+ Op.getOperand(1));
}
}
SDOperand RV = combine(N);
- if (RV.Val) {
- ++NodesCombined;
- // If we get back the same node we passed in, rather than a new node or
- // zero, we know that the node must have defined multiple values and
- // CombineTo was used. Since CombineTo takes care of the worklist
- // mechanics for us, we have no work to do in this case.
- if (RV.Val != N) {
- assert(N->getOpcode() != ISD::DELETED_NODE &&
- RV.Val->getOpcode() != ISD::DELETED_NODE &&
- "Node was deleted but visit returned new node!");
-
- DOUT << "\nReplacing.3 "; DEBUG(N->dump(&DAG));
- DOUT << "\nWith: "; DEBUG(RV.Val->dump(&DAG));
- DOUT << '\n';
- std::vector<SDNode*> NowDead;
- if (N->getNumValues() == RV.Val->getNumValues())
- DAG.ReplaceAllUsesWith(N, RV.Val, &NowDead);
- else {
- assert(N->getValueType(0) == RV.getValueType() && "Type mismatch");
- SDOperand OpV = RV;
- DAG.ReplaceAllUsesWith(N, &OpV, &NowDead);
- }
-
- // Push the new node and any users onto the worklist
- AddToWorkList(RV.Val);
- AddUsersToWorkList(RV.Val);
-
- // Nodes can be reintroduced into the worklist. Make sure we do not
- // process a node that has been replaced.
- removeFromWorkList(N);
- for (unsigned i = 0, e = NowDead.size(); i != e; ++i)
- removeFromWorkList(NowDead[i]);
-
- // Finally, since the node is now dead, remove it from the graph.
- DAG.DeleteNode(N);
- }
+ if (RV.Val == 0)
+ continue;
+
+ ++NodesCombined;
+
+ // If we get back the same node we passed in, rather than a new node or
+ // zero, we know that the node must have defined multiple values and
+ // CombineTo was used. Since CombineTo takes care of the worklist
+ // mechanics for us, we have no work to do in this case.
+ if (RV.Val == N)
+ continue;
+
+ assert(N->getOpcode() != ISD::DELETED_NODE &&
+ RV.Val->getOpcode() != ISD::DELETED_NODE &&
+ "Node was deleted but visit returned new node!");
+
+ DOUT << "\nReplacing.3 "; DEBUG(N->dump(&DAG));
+ DOUT << "\nWith: "; DEBUG(RV.Val->dump(&DAG));
+ DOUT << '\n';
+ std::vector<SDNode*> NowDead;
+ if (N->getNumValues() == RV.Val->getNumValues())
+ DAG.ReplaceAllUsesWith(N, RV.Val, &NowDead);
+ else {
+ assert(N->getValueType(0) == RV.getValueType() &&
+ N->getNumValues() == 1 && "Type mismatch");
+ SDOperand OpV = RV;
+ DAG.ReplaceAllUsesWith(N, &OpV, &NowDead);
}
+
+ // Push the new node and any users onto the worklist
+ AddToWorkList(RV.Val);
+ AddUsersToWorkList(RV.Val);
+
+ // Add any uses of the old node to the worklist in case this node is the
+ // last one that uses them. They may become dead after this node is
+ // deleted.
+ for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i)
+ AddToWorkList(N->getOperand(i).Val);
+
+ // Nodes can be reintroduced into the worklist. Make sure we do not
+ // process a node that has been replaced.
+ removeFromWorkList(N);
+ for (unsigned i = 0, e = NowDead.size(); i != e; ++i)
+ removeFromWorkList(NowDead[i]);
+
+ // Finally, since the node is now dead, remove it from the graph.
+ DAG.DeleteNode(N);
}
// If the root changed (e.g. it was a dead load, update the root).
RHS.getOpcode() == ISD::Constant &&
cast<ConstantSDNode>(RHS)->isNullValue()) {
std::swap(LHS, RHS);
- bool isInt = MVT::isInteger(isSlctCC ? Slct.getOperand(0).getValueType()
- : Slct.getOperand(0).getOperand(0).getValueType());
+ SDOperand Op0 = Slct.getOperand(0);
+ bool isInt = MVT::isInteger(isSlctCC ? Op0.getValueType()
+ : Op0.getOperand(0).getValueType());
CC = ISD::getSetCCInverse(CC, isInt);
DoXform = true;
InvCC = true;
DAG.MaskedValueIsZero(N0, SignBit))
return DAG.getNode(ISD::UREM, VT, N0, N1);
- // Unconditionally lower X%C -> X-X/C*C. This allows the X/C logic to hack on
- // the remainder operation.
+ // If X/C can be simplified by the division-by-constant logic, lower
+ // X%C to the equivalent of X-X/C*C.
if (N1C && !N1C->isNullValue()) {
SDOperand Div = DAG.getNode(ISD::SDIV, VT, N0, N1);
- SDOperand Mul = DAG.getNode(ISD::MUL, VT, Div, N1);
- SDOperand Sub = DAG.getNode(ISD::SUB, VT, N0, Mul);
AddToWorkList(Div.Val);
- AddToWorkList(Mul.Val);
- return Sub;
+ SDOperand OptimizedDiv = combine(Div.Val);
+ if (OptimizedDiv.Val && OptimizedDiv.Val != Div.Val) {
+ SDOperand Mul = DAG.getNode(ISD::MUL, VT, OptimizedDiv, N1);
+ SDOperand Sub = DAG.getNode(ISD::SUB, VT, N0, Mul);
+ AddToWorkList(Mul.Val);
+ return Sub;
+ }
}
// undef % X -> 0
}
}
- // Unconditionally lower X%C -> X-X/C*C. This allows the X/C logic to hack on
- // the remainder operation.
+ // If X/C can be simplified by the division-by-constant logic, lower
+ // X%C to the equivalent of X-X/C*C.
if (N1C && !N1C->isNullValue()) {
SDOperand Div = DAG.getNode(ISD::UDIV, VT, N0, N1);
- SDOperand Mul = DAG.getNode(ISD::MUL, VT, Div, N1);
- SDOperand Sub = DAG.getNode(ISD::SUB, VT, N0, Mul);
- AddToWorkList(Div.Val);
- AddToWorkList(Mul.Val);
- return Sub;
+ SDOperand OptimizedDiv = combine(Div.Val);
+ if (OptimizedDiv.Val && OptimizedDiv.Val != Div.Val) {
+ SDOperand Mul = DAG.getNode(ISD::MUL, VT, OptimizedDiv, N1);
+ SDOperand Sub = DAG.getNode(ISD::SUB, VT, N0, Mul);
+ AddToWorkList(Mul.Val);
+ return Sub;
+ }
}
// undef % X -> 0
/// compute two values. LoOp and HiOp give the opcodes for the two computations
/// that are being performed. Return true if a simplification was made.
///
-bool DAGCombiner::SimplifyNodeWithTwoResults(SDNode *N,
- unsigned LoOp, unsigned HiOp) {
+SDOperand DAGCombiner::SimplifyNodeWithTwoResults(SDNode *N, unsigned LoOp,
+ unsigned HiOp) {
// If the high half is not needed, just compute the low half.
- if (!N->hasAnyUseOfValue(1) &&
+ bool HiExists = N->hasAnyUseOfValue(1);
+ if (!HiExists &&
(!AfterLegalize ||
TLI.isOperationLegal(LoOp, N->getValueType(0)))) {
- DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 0),
- DAG.getNode(LoOp, N->getValueType(0),
- N->op_begin(),
- N->getNumOperands()));
- return true;
+ SDOperand Res = DAG.getNode(LoOp, N->getValueType(0), N->op_begin(),
+ N->getNumOperands());
+ return CombineTo(N, Res, Res);
}
// If the low half is not needed, just compute the high half.
- if (!N->hasAnyUseOfValue(0) &&
+ bool LoExists = N->hasAnyUseOfValue(0);
+ if (!LoExists &&
(!AfterLegalize ||
TLI.isOperationLegal(HiOp, N->getValueType(1)))) {
- DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 1),
- DAG.getNode(HiOp, N->getValueType(1),
- N->op_begin(),
- N->getNumOperands()));
- return true;
+ SDOperand Res = DAG.getNode(HiOp, N->getValueType(1), N->op_begin(),
+ N->getNumOperands());
+ return CombineTo(N, Res, Res);
}
- // If the two computed results can be siplified separately, separate them.
- SDOperand Lo = DAG.getNode(LoOp, N->getValueType(0),
- N->op_begin(), N->getNumOperands());
- SDOperand Hi = DAG.getNode(HiOp, N->getValueType(1),
- N->op_begin(), N->getNumOperands());
- unsigned LoExists = !Lo.use_empty();
- unsigned HiExists = !Hi.use_empty();
- SDOperand LoOpt = Lo;
- SDOperand HiOpt = Hi;
- if (!LoExists || !HiExists) {
- SDOperand Pair = DAG.getNode(ISD::BUILD_PAIR, MVT::Other, Lo, Hi);
- assert(Pair.use_empty() && "Pair with type MVT::Other already exists!");
- LoOpt = combine(Lo.Val);
- HiOpt = combine(Hi.Val);
- if (!LoOpt.Val)
- LoOpt = Pair.getOperand(0);
- if (!HiOpt.Val)
- HiOpt = Pair.getOperand(1);
- DAG.DeleteNode(Pair.Val);
- }
- if ((LoExists || LoOpt != Lo) &&
- (HiExists || HiOpt != Hi) &&
- TLI.isOperationLegal(LoOpt.getOpcode(), LoOpt.getValueType()) &&
- TLI.isOperationLegal(HiOpt.getOpcode(), HiOpt.getValueType())) {
- DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 0), LoOpt);
- DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 1), HiOpt);
- return true;
- }
+ // If both halves are used, return as it is.
+ if (LoExists && HiExists)
+ return SDOperand();
- return false;
+ // If the two computed results can be simplified separately, separate them.
+ if (LoExists) {
+ SDOperand Lo = DAG.getNode(LoOp, N->getValueType(0),
+ N->op_begin(), N->getNumOperands());
+ AddToWorkList(Lo.Val);
+ SDOperand LoOpt = combine(Lo.Val);
+ if (LoOpt.Val && LoOpt.Val != Lo.Val &&
+ TLI.isOperationLegal(LoOpt.getOpcode(), LoOpt.getValueType()))
+ return CombineTo(N, LoOpt, LoOpt);
+ }
+
+ if (HiExists) {
+ SDOperand Hi = DAG.getNode(HiOp, N->getValueType(1),
+ N->op_begin(), N->getNumOperands());
+ AddToWorkList(Hi.Val);
+ SDOperand HiOpt = combine(Hi.Val);
+ if (HiOpt.Val && HiOpt != Hi &&
+ TLI.isOperationLegal(HiOpt.getOpcode(), HiOpt.getValueType()))
+ return CombineTo(N, HiOpt, HiOpt);
+ }
+ return SDOperand();
}
SDOperand DAGCombiner::visitSMUL_LOHI(SDNode *N) {
-
- if (SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHS))
- return SDOperand();
+ SDOperand Res = SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHS);
+ if (Res.Val) return Res;
return SDOperand();
}
SDOperand DAGCombiner::visitUMUL_LOHI(SDNode *N) {
-
- if (SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHU))
- return SDOperand();
+ SDOperand Res = SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHU);
+ if (Res.Val) return Res;
return SDOperand();
}
SDOperand DAGCombiner::visitSDIVREM(SDNode *N) {
-
- if (SimplifyNodeWithTwoResults(N, ISD::SDIV, ISD::SREM))
- return SDOperand();
+ SDOperand Res = SimplifyNodeWithTwoResults(N, ISD::SDIV, ISD::SREM);
+ if (Res.Val) return Res;
return SDOperand();
}
SDOperand DAGCombiner::visitUDIVREM(SDNode *N) {
-
- if (SimplifyNodeWithTwoResults(N, ISD::UDIV, ISD::UREM))
- return SDOperand();
+ SDOperand Res = SimplifyNodeWithTwoResults(N, ISD::UDIV, ISD::UREM);
+ if (Res.Val) return Res;
return SDOperand();
}
if (N1C && N0.getOpcode() == ISD::LOAD) {
LoadSDNode *LN0 = cast<LoadSDNode>(N0);
if (LN0->getExtensionType() != ISD::SEXTLOAD &&
- LN0->getAddressingMode() == ISD::UNINDEXED &&
- N0.hasOneUse()) {
+ LN0->isUnindexed() && N0.hasOneUse()) {
MVT::ValueType EVT, LoadedVT;
if (N1C->getValue() == 255)
EVT = MVT::i8;
// For big endian targets, we need to add an offset to the pointer to
// load the correct bytes. For little endian systems, we merely need to
// read fewer bytes from the same pointer.
- unsigned PtrOff =
- (MVT::getSizeInBits(LoadedVT) - MVT::getSizeInBits(EVT)) / 8;
+ unsigned LVTStoreBytes = MVT::getStoreSizeInBits(LoadedVT)/8;
+ unsigned EVTStoreBytes = MVT::getStoreSizeInBits(EVT)/8;
+ unsigned PtrOff = LVTStoreBytes - EVTStoreBytes;
unsigned Alignment = LN0->getAlignment();
SDOperand NewPtr = LN0->getBasePtr();
if (!TLI.isLittleEndian()) {
return SDOperand();
}
+/// visitShiftByConstant - Handle transforms common to the three shifts, when
+/// the shift amount is a constant.
+SDOperand DAGCombiner::visitShiftByConstant(SDNode *N, unsigned Amt) {
+ SDNode *LHS = N->getOperand(0).Val;
+ if (!LHS->hasOneUse()) return SDOperand();
+
+ // We want to pull some binops through shifts, so that we have (and (shift))
+ // instead of (shift (and)), likewise for add, or, xor, etc. This sort of
+ // thing happens with address calculations, so it's important to canonicalize
+ // it.
+ bool HighBitSet = false; // Can we transform this if the high bit is set?
+
+ switch (LHS->getOpcode()) {
+ default: return SDOperand();
+ case ISD::OR:
+ case ISD::XOR:
+ HighBitSet = false; // We can only transform sra if the high bit is clear.
+ break;
+ case ISD::AND:
+ HighBitSet = true; // We can only transform sra if the high bit is set.
+ break;
+ case ISD::ADD:
+ if (N->getOpcode() != ISD::SHL)
+ return SDOperand(); // only shl(add) not sr[al](add).
+ HighBitSet = false; // We can only transform sra if the high bit is clear.
+ break;
+ }
+
+ // We require the RHS of the binop to be a constant as well.
+ ConstantSDNode *BinOpCst = dyn_cast<ConstantSDNode>(LHS->getOperand(1));
+ if (!BinOpCst) return SDOperand();
+
+
+ // FIXME: disable this for unless the input to the binop is a shift by a
+ // constant. If it is not a shift, it pessimizes some common cases like:
+ //
+ //void foo(int *X, int i) { X[i & 1235] = 1; }
+ //int bar(int *X, int i) { return X[i & 255]; }
+ SDNode *BinOpLHSVal = LHS->getOperand(0).Val;
+ if ((BinOpLHSVal->getOpcode() != ISD::SHL &&
+ BinOpLHSVal->getOpcode() != ISD::SRA &&
+ BinOpLHSVal->getOpcode() != ISD::SRL) ||
+ !isa<ConstantSDNode>(BinOpLHSVal->getOperand(1)))
+ return SDOperand();
+
+ MVT::ValueType VT = N->getValueType(0);
+
+ // If this is a signed shift right, and the high bit is modified
+ // by the logical operation, do not perform the transformation.
+ // The highBitSet boolean indicates the value of the high bit of
+ // the constant which would cause it to be modified for this
+ // operation.
+ if (N->getOpcode() == ISD::SRA) {
+ uint64_t BinOpRHSSign = BinOpCst->getValue() >> MVT::getSizeInBits(VT)-1;
+ if ((bool)BinOpRHSSign != HighBitSet)
+ return SDOperand();
+ }
+
+ // Fold the constants, shifting the binop RHS by the shift amount.
+ SDOperand NewRHS = DAG.getNode(N->getOpcode(), N->getValueType(0),
+ LHS->getOperand(1), N->getOperand(1));
+
+ // Create the new shift.
+ SDOperand NewShift = DAG.getNode(N->getOpcode(), VT, LHS->getOperand(0),
+ N->getOperand(1));
+
+ // Create the new binop.
+ return DAG.getNode(LHS->getOpcode(), VT, NewShift, NewRHS);
+}
+
+
SDOperand DAGCombiner::visitSHL(SDNode *N) {
SDOperand N0 = N->getOperand(0);
SDOperand N1 = N->getOperand(1);
if (N1C && N0.getOpcode() == ISD::SRA && N1 == N0.getOperand(1))
return DAG.getNode(ISD::AND, VT, N0.getOperand(0),
DAG.getConstant(~0ULL << N1C->getValue(), VT));
- return SDOperand();
+
+ return N1C ? visitShiftByConstant(N, N1C->getValue()) : SDOperand();
}
SDOperand DAGCombiner::visitSRA(SDNode *N) {
// If the sign bit is known to be zero, switch this to a SRL.
if (DAG.MaskedValueIsZero(N0, MVT::getIntVTSignBit(VT)))
return DAG.getNode(ISD::SRL, VT, N0, N1);
- return SDOperand();
+
+ return N1C ? visitShiftByConstant(N, N1C->getValue()) : SDOperand();
}
SDOperand DAGCombiner::visitSRL(SDNode *N) {
if (N1C && SimplifyDemandedBits(SDOperand(N, 0)))
return SDOperand(N, 0);
- return SDOperand();
+ return N1C ? visitShiftByConstant(N, N1C->getValue()) : SDOperand();
}
SDOperand DAGCombiner::visitCTLZ(SDNode *N) {
return DAG.getNode(ISD::TRUNCATE, VT, XORNode);
}
// fold select C, 0, X -> ~C & X
- if (VT == VT0 && N1C && N1C->isNullValue()) {
+ if (VT == VT0 && VT == MVT::i1 && N1C && N1C->isNullValue()) {
SDOperand XORNode = DAG.getNode(ISD::XOR, VT, N0, DAG.getConstant(1, VT));
AddToWorkList(XORNode.Val);
return DAG.getNode(ISD::AND, VT, XORNode, N2);
}
// fold select C, X, 1 -> ~C | X
- if (VT == VT0 && N2C && N2C->getValue() == 1) {
+ if (VT == VT0 && VT == MVT::i1 && N2C && N2C->getValue() == 1) {
SDOperand XORNode = DAG.getNode(ISD::XOR, VT, N0, DAG.getConstant(1, VT));
AddToWorkList(XORNode.Val);
return DAG.getNode(ISD::OR, VT, XORNode, N1);
if (SOp == Trunc)
Ops.push_back(ExtLoad);
else
- Ops.push_back(DAG.getNode(ISD::SIGN_EXTEND, VT, SOp));
+ Ops.push_back(DAG.getNode(ISD::ZERO_EXTEND, VT, SOp));
}
Ops.push_back(SetCC->getOperand(2));
CombineTo(SetCC, DAG.getNode(ISD::SETCC, SetCC->getValueType(0),
MVT::ValueType PtrType = N0.getOperand(1).getValueType();
// For big endian targets, we need to adjust the offset to the pointer to
// load the correct bytes.
- if (!TLI.isLittleEndian())
- ShAmt = MVT::getSizeInBits(N0.getValueType()) - ShAmt - EVTBits;
+ if (!TLI.isLittleEndian()) {
+ unsigned LVTStoreBits = MVT::getStoreSizeInBits(N0.getValueType());
+ unsigned EVTStoreBits = MVT::getStoreSizeInBits(EVT);
+ ShAmt = LVTStoreBits - EVTStoreBits - ShAmt;
+ }
uint64_t PtrOff = ShAmt / 8;
unsigned NewAlign = MinAlign(LN0->getAlignment(), PtrOff);
SDOperand NewPtr = DAG.getNode(ISD::ADD, PtrType, LN0->getBasePtr(),
SDOperand DAGCombiner::visitFP_ROUND(SDNode *N) {
SDOperand N0 = N->getOperand(0);
+ SDOperand N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
MVT::ValueType VT = N->getValueType(0);
// fold (fp_round c1fp) -> c1fp
if (N0CFP && N0.getValueType() != MVT::ppcf128)
- return DAG.getNode(ISD::FP_ROUND, VT, N0);
+ return DAG.getNode(ISD::FP_ROUND, VT, N0, N1);
// fold (fp_round (fp_extend x)) -> x
if (N0.getOpcode() == ISD::FP_EXTEND && VT == N0.getOperand(0).getValueType())
return N0.getOperand(0);
+ // fold (fp_round (fp_round x)) -> (fp_round x)
+ if (N0.getOpcode() == ISD::FP_ROUND) {
+ // This is a value preserving truncation if both round's are.
+ bool IsTrunc = N->getConstantOperandVal(1) == 1 &&
+ N0.Val->getConstantOperandVal(1) == 1;
+ return DAG.getNode(ISD::FP_ROUND, VT, N0.getOperand(0),
+ DAG.getIntPtrConstant(IsTrunc));
+ }
+
// fold (fp_round (copysign X, Y)) -> (copysign (fp_round X), Y)
if (N0.getOpcode() == ISD::FCOPYSIGN && N0.Val->hasOneUse()) {
- SDOperand Tmp = DAG.getNode(ISD::FP_ROUND, VT, N0.getOperand(0));
+ SDOperand Tmp = DAG.getNode(ISD::FP_ROUND, VT, N0.getOperand(0), N1);
AddToWorkList(Tmp.Val);
return DAG.getNode(ISD::FCOPYSIGN, VT, Tmp, N0.getOperand(1));
}
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
MVT::ValueType VT = N->getValueType(0);
+ // If this is fp_round(fpextend), don't fold it, allow ourselves to be folded.
+ if (N->hasOneUse() && (*N->use_begin())->getOpcode() == ISD::FP_ROUND)
+ return SDOperand();
+
// fold (fp_extend c1fp) -> c1fp
if (N0CFP && VT != MVT::ppcf128)
return DAG.getNode(ISD::FP_EXTEND, VT, N0);
-
- // fold (fpext (load x)) -> (fpext (fpround (extload x)))
+
+ // Turn fp_extend(fp_round(X, 1)) -> x since the fp_round doesn't affect the
+ // value of X.
+ if (N0.getOpcode() == ISD::FP_ROUND && N0.Val->getConstantOperandVal(1) == 1){
+ SDOperand In = N0.getOperand(0);
+ if (In.getValueType() == VT) return In;
+ if (VT < In.getValueType())
+ return DAG.getNode(ISD::FP_ROUND, VT, In, N0.getOperand(1));
+ return DAG.getNode(ISD::FP_EXTEND, VT, In);
+ }
+
+ // fold (fpext (load x)) -> (fpext (fptrunc (extload x)))
if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() &&
(!AfterLegalize||TLI.isLoadXLegal(ISD::EXTLOAD, N0.getValueType()))) {
LoadSDNode *LN0 = cast<LoadSDNode>(N0);
LN0->isVolatile(),
LN0->getAlignment());
CombineTo(N, ExtLoad);
- CombineTo(N0.Val, DAG.getNode(ISD::FP_ROUND, N0.getValueType(), ExtLoad),
+ CombineTo(N0.Val, DAG.getNode(ISD::FP_ROUND, N0.getValueType(), ExtLoad,
+ DAG.getIntPtrConstant(1)),
ExtLoad.getValue(1));
return SDOperand(N, 0); // Return N so it doesn't get rechecked!
}
SDOperand Ptr;
MVT::ValueType VT;
if (LoadSDNode *LD = dyn_cast<LoadSDNode>(N)) {
- if (LD->getAddressingMode() != ISD::UNINDEXED)
+ if (LD->isIndexed())
return false;
VT = LD->getLoadedVT();
if (!TLI.isIndexedLoadLegal(ISD::PRE_INC, VT) &&
return false;
Ptr = LD->getBasePtr();
} else if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) {
- if (ST->getAddressingMode() != ISD::UNINDEXED)
+ if (ST->isIndexed())
return false;
VT = ST->getStoredVT();
if (!TLI.isIndexedStoreLegal(ISD::PRE_INC, VT) &&
SDOperand Ptr;
MVT::ValueType VT;
if (LoadSDNode *LD = dyn_cast<LoadSDNode>(N)) {
- if (LD->getAddressingMode() != ISD::UNINDEXED)
+ if (LD->isIndexed())
return false;
VT = LD->getLoadedVT();
if (!TLI.isIndexedLoadLegal(ISD::POST_INC, VT) &&
return false;
Ptr = LD->getBasePtr();
} else if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) {
- if (ST->getAddressingMode() != ISD::UNINDEXED)
+ if (ST->isIndexed())
return false;
VT = ST->getStoredVT();
if (!TLI.isIndexedStoreLegal(ISD::POST_INC, VT) &&
return false;
}
+/// InferAlignment - If we can infer some alignment information from this
+/// pointer, return it.
+static unsigned InferAlignment(SDOperand Ptr, SelectionDAG &DAG) {
+ // If this is a direct reference to a stack slot, use information about the
+ // stack slot's alignment.
+ int FrameIdx = 1 << 31;
+ int64_t FrameOffset = 0;
+ if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(Ptr)) {
+ FrameIdx = FI->getIndex();
+ } else if (Ptr.getOpcode() == ISD::ADD &&
+ isa<ConstantSDNode>(Ptr.getOperand(1)) &&
+ isa<FrameIndexSDNode>(Ptr.getOperand(0))) {
+ FrameIdx = cast<FrameIndexSDNode>(Ptr.getOperand(0))->getIndex();
+ FrameOffset = Ptr.getConstantOperandVal(1);
+ }
+
+ if (FrameIdx != (1 << 31)) {
+ // FIXME: Handle FI+CST.
+ const MachineFrameInfo &MFI = *DAG.getMachineFunction().getFrameInfo();
+ if (MFI.isFixedObjectIndex(FrameIdx)) {
+ int64_t ObjectOffset = MFI.getObjectOffset(FrameIdx);
+
+ // The alignment of the frame index can be determined from its offset from
+ // the incoming frame position. If the frame object is at offset 32 and
+ // the stack is guaranteed to be 16-byte aligned, then we know that the
+ // object is 16-byte aligned.
+ unsigned StackAlign = DAG.getTarget().getFrameInfo()->getStackAlignment();
+ unsigned Align = MinAlign(ObjectOffset, StackAlign);
+
+ // Finally, the frame object itself may have a known alignment. Factor
+ // the alignment + offset into a new alignment. For example, if we know
+ // the FI is 8 byte aligned, but the pointer is 4 off, we really have a
+ // 4-byte alignment of the resultant pointer. Likewise align 4 + 4-byte
+ // offset = 4-byte alignment, align 4 + 1-byte offset = align 1, etc.
+ unsigned FIInfoAlign = MinAlign(MFI.getObjectAlignment(FrameIdx),
+ FrameOffset);
+ return std::max(Align, FIInfoAlign);
+ }
+ }
+
+ return 0;
+}
SDOperand DAGCombiner::visitLOAD(SDNode *N) {
LoadSDNode *LD = cast<LoadSDNode>(N);
SDOperand Chain = LD->getChain();
SDOperand Ptr = LD->getBasePtr();
+
+ // Try to infer better alignment information than the load already has.
+ if (LD->isUnindexed()) {
+ if (unsigned Align = InferAlignment(Ptr, DAG)) {
+ if (Align > LD->getAlignment())
+ return DAG.getExtLoad(LD->getExtensionType(), LD->getValueType(0),
+ Chain, Ptr, LD->getSrcValue(),
+ LD->getSrcValueOffset(), LD->getLoadedVT(),
+ LD->isVolatile(), Align);
+ }
+ }
+
// If load is not volatile and there are no uses of the loaded value (and
// the updated indexed value in case of indexed loads), change uses of the
if (!LD->isVolatile()) {
if (N->getValueType(1) == MVT::Other) {
// Unindexed loads.
- if (N->hasNUsesOfValue(0, 0))
- return CombineTo(N, DAG.getNode(ISD::UNDEF, N->getValueType(0)), Chain);
+ if (N->hasNUsesOfValue(0, 0)) {
+ // It's not safe to use the two value CombineTo variant here. e.g.
+ // v1, chain2 = load chain1, loc
+ // v2, chain3 = load chain2, loc
+ // v3 = add v2, c
+ // Now we replace use of chain2 with chain1. This makes the second load
+ // isomorphic to the one we are deleting, and thus makes this load live.
+ std::vector<SDNode*> NowDead;
+ DOUT << "\nReplacing.6 "; DEBUG(N->dump(&DAG));
+ DOUT << "\nWith chain: "; DEBUG(Chain.Val->dump(&DAG));
+ DOUT << "\n";
+ DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 1), Chain, &NowDead);
+ for (unsigned i = 0, e = NowDead.size(); i != e; ++i)
+ removeFromWorkList(NowDead[i]);
+ if (N->use_empty()) {
+ removeFromWorkList(N);
+ DAG.DeleteNode(N);
+ }
+ return SDOperand(N, 0); // Return N so it doesn't get rechecked!
+ }
} else {
// Indexed loads.
assert(N->getValueType(2) == MVT::Other && "Malformed indexed loads?");
if (N->hasNUsesOfValue(0, 0) && N->hasNUsesOfValue(0, 1)) {
- SDOperand Undef0 = DAG.getNode(ISD::UNDEF, N->getValueType(0));
- SDOperand Undef1 = DAG.getNode(ISD::UNDEF, N->getValueType(1));
- SDOperand To[] = { Undef0, Undef1, Chain };
- return CombineTo(N, To, 3);
+ std::vector<SDNode*> NowDead;
+ SDOperand Undef = DAG.getNode(ISD::UNDEF, N->getValueType(0));
+ DOUT << "\nReplacing.6 "; DEBUG(N->dump(&DAG));
+ DOUT << "\nWith: "; DEBUG(Undef.Val->dump(&DAG));
+ DOUT << " and 2 other values\n";
+ DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 0), Undef, &NowDead);
+ DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 1),
+ DAG.getNode(ISD::UNDEF, N->getValueType(1)),
+ &NowDead);
+ DAG.ReplaceAllUsesOfValueWith(SDOperand(N, 2), Chain, &NowDead);
+ removeFromWorkList(N);
+ for (unsigned i = 0, e = NowDead.size(); i != e; ++i)
+ removeFromWorkList(NowDead[i]);
+ DAG.DeleteNode(N);
+ return SDOperand(N, 0); // Return N so it doesn't get rechecked!
}
}
}
return SDOperand();
}
+
SDOperand DAGCombiner::visitSTORE(SDNode *N) {
StoreSDNode *ST = cast<StoreSDNode>(N);
SDOperand Chain = ST->getChain();
SDOperand Value = ST->getValue();
SDOperand Ptr = ST->getBasePtr();
+ // Try to infer better alignment information than the store already has.
+ if (ST->isUnindexed()) {
+ if (unsigned Align = InferAlignment(Ptr, DAG)) {
+ if (Align > ST->getAlignment())
+ return DAG.getTruncStore(Chain, Value, Ptr, ST->getSrcValue(),
+ ST->getSrcValueOffset(), ST->getStoredVT(),
+ ST->isVolatile(), Align);
+ }
+ }
+
// If this is a store of a bit convert, store the input value if the
// resultant store does not need a higher alignment than the original.
if (Value.getOpcode() == ISD::BIT_CONVERT && !ST->isTruncatingStore() &&
- ST->getAddressingMode() == ISD::UNINDEXED) {
+ ST->isUnindexed()) {
unsigned Align = ST->getAlignment();
MVT::ValueType SVT = Value.getOperand(0).getValueType();
unsigned OrigAlign = TLI.getTargetMachine().getTargetData()->
SDOperand ReplStore;
if (ST->isTruncatingStore()) {
ReplStore = DAG.getTruncStore(BetterChain, Value, Ptr,
- ST->getSrcValue(), ST->getSrcValueOffset(), ST->getStoredVT(),
- ST->isVolatile(), ST->getAlignment());
+ ST->getSrcValue(),ST->getSrcValueOffset(),
+ ST->getStoredVT(),
+ ST->isVolatile(), ST->getAlignment());
} else {
ReplStore = DAG.getStore(BetterChain, Value, Ptr,
- ST->getSrcValue(), ST->getSrcValueOffset(),
- ST->isVolatile(), ST->getAlignment());
+ ST->getSrcValue(), ST->getSrcValueOffset(),
+ ST->isVolatile(), ST->getAlignment());
}
// Create token to keep both nodes around.
if (CombineToPreIndexedLoadStore(N) || CombineToPostIndexedLoadStore(N))
return SDOperand(N, 0);
- // FIXME: is there such a think as a truncating indexed store?
- if (ST->isTruncatingStore() && ST->getAddressingMode() == ISD::UNINDEXED &&
+ // FIXME: is there such a thing as a truncating indexed store?
+ if (ST->isTruncatingStore() && ST->isUnindexed() &&
MVT::isInteger(Value.getValueType())) {
// See if we can simplify the input to this truncstore with knowledge that
// only the low bits are being used. For example:
return SDOperand(N, 0);
}
+ // If this is a load followed by a store to the same location, then the store
+ // is dead/noop.
+ if (LoadSDNode *Ld = dyn_cast<LoadSDNode>(Value)) {
+ if (Ld->getBasePtr() == Ptr && ST->getStoredVT() == Ld->getLoadedVT() &&
+ ST->isUnindexed() && !ST->isVolatile() &&
+ // There can't be any side effects between the load and store, such as
+ // a call or store.
+ Chain.reachesChainWithoutSideEffects(SDOperand(Ld, 1))) {
+ // The store is dead, remove it.
+ return Chain;
+ }
+ }
+
+ // If this is an FP_ROUND or TRUNC followed by a store, fold this into a
+ // truncating store. We can do this even if this is already a truncstore.
+ if ((Value.getOpcode() == ISD::FP_ROUND || Value.getOpcode() == ISD::TRUNCATE)
+ && TLI.isTypeLegal(Value.getOperand(0).getValueType()) &&
+ Value.Val->hasOneUse() && ST->isUnindexed() &&
+ TLI.isTruncStoreLegal(Value.getOperand(0).getValueType(),
+ ST->getStoredVT())) {
+ return DAG.getTruncStore(Chain, Value.getOperand(0), Ptr, ST->getSrcValue(),
+ ST->getSrcValueOffset(), ST->getStoredVT(),
+ ST->isVolatile(), ST->getAlignment());
+ }
+
return SDOperand();
}
unsigned NumElts = MVT::getVectorNumElements(VT);
if (InVec.getOpcode() == ISD::BIT_CONVERT) {
MVT::ValueType BCVT = InVec.getOperand(0).getValueType();
- if (NumElts != MVT::getVectorNumElements(BCVT))
+ if (!MVT::isVector(BCVT) ||
+ NumElts != MVT::getVectorNumElements(BCVT))
return SDOperand();
InVec = InVec.getOperand(0);
EVT = MVT::getVectorElementType(BCVT);
// Otherwise, use InIdx + VecSize
unsigned Idx = cast<ConstantSDNode>(Extract.getOperand(1))->getValue();
- BuildVecIndices.push_back(DAG.getConstant(Idx+NumInScalars,
- TLI.getPointerTy()));
+ BuildVecIndices.push_back(DAG.getIntPtrConstant(Idx+NumInScalars));
}
// Add count and size info.
- MVT::ValueType BuildVecVT =
- MVT::getVectorType(TLI.getPointerTy(), NumElts);
+ MVT::ValueType BuildVecVT = MVT::getVectorType(TLI.getPointerTy(), NumElts);
// Return the new VECTOR_SHUFFLE node.
SDOperand Ops[5];