1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-22 04:22:57 +02:00
llvm-mirror/lib/CodeGen/TargetRegisterInfo.cpp

421 lines
15 KiB
C++
Raw Normal View History

//===- TargetRegisterInfo.cpp - Target Register Information Implementation ===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
2002-12-17 05:03:08 +01:00
//
// This file implements the TargetRegisterInfo interface.
2002-12-17 05:03:08 +01:00
//
//===----------------------------------------------------------------------===//
#include "llvm/Target/TargetRegisterInfo.h"
#include "llvm/ADT/BitVector.h"
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/VirtRegMap.h"
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
#include "llvm/IR/Function.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Format.h"
#include "llvm/Support/raw_ostream.h"
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
#include "llvm/Target/TargetFrameLowering.h"
#define DEBUG_TYPE "target-reg-info"
using namespace llvm;
TargetRegisterInfo::TargetRegisterInfo(const TargetRegisterInfoDesc *ID,
regclass_iterator RCB, regclass_iterator RCE,
const char *const *SRINames,
const LaneBitmask *SRILaneMasks,
LaneBitmask SRICoveringLanes)
: InfoDesc(ID), SubRegIndexNames(SRINames),
SubRegIndexLaneMasks(SRILaneMasks),
RegClassBegin(RCB), RegClassEnd(RCE),
CoveringLanes(SRICoveringLanes) {
2002-12-17 05:03:08 +01:00
}
TargetRegisterInfo::~TargetRegisterInfo() {}
void TargetRegisterInfo::markSuperRegs(BitVector &RegisterSet, unsigned Reg)
const {
for (MCSuperRegIterator AI(Reg, this, true); AI.isValid(); ++AI)
RegisterSet.set(*AI);
}
bool TargetRegisterInfo::checkAllSuperRegsMarked(const BitVector &RegisterSet,
ArrayRef<MCPhysReg> Exceptions) const {
// Check that all super registers of reserved regs are reserved as well.
BitVector Checked(getNumRegs());
for (unsigned Reg : RegisterSet.set_bits()) {
if (Checked[Reg])
continue;
for (MCSuperRegIterator SR(Reg, this); SR.isValid(); ++SR) {
if (!RegisterSet[*SR] && !is_contained(Exceptions, Reg)) {
dbgs() << "Error: Super register " << PrintReg(*SR, this)
<< " of reserved register " << PrintReg(Reg, this)
<< " is not reserved.\n";
return false;
}
// We transitively check superregs. So we can remember this for later
// to avoid compiletime explosion in deep register hierarchies.
Checked.set(*SR);
}
}
return true;
}
namespace llvm {
Printable PrintReg(unsigned Reg, const TargetRegisterInfo *TRI,
unsigned SubIdx) {
return Printable([Reg, TRI, SubIdx](raw_ostream &OS) {
if (!Reg)
OS << "%noreg";
else if (TargetRegisterInfo::isStackSlot(Reg))
OS << "SS#" << TargetRegisterInfo::stackSlot2Index(Reg);
else if (TargetRegisterInfo::isVirtualRegister(Reg))
OS << "%vreg" << TargetRegisterInfo::virtReg2Index(Reg);
else if (TRI && Reg < TRI->getNumRegs())
OS << '%' << TRI->getName(Reg);
else
OS << "%physreg" << Reg;
if (SubIdx) {
if (TRI)
OS << ':' << TRI->getSubRegIndexName(SubIdx);
else
OS << ":sub(" << SubIdx << ')';
}
});
}
Printable PrintRegUnit(unsigned Unit, const TargetRegisterInfo *TRI) {
return Printable([Unit, TRI](raw_ostream &OS) {
// Generic printout when TRI is missing.
if (!TRI) {
OS << "Unit~" << Unit;
return;
}
// Check for invalid register units.
if (Unit >= TRI->getNumRegUnits()) {
OS << "BadUnit~" << Unit;
return;
}
// Normal units have at least one root.
MCRegUnitRootIterator Roots(Unit, TRI);
assert(Roots.isValid() && "Unit has no roots.");
OS << TRI->getName(*Roots);
for (++Roots; Roots.isValid(); ++Roots)
OS << '~' << TRI->getName(*Roots);
});
}
Printable PrintVRegOrUnit(unsigned Unit, const TargetRegisterInfo *TRI) {
return Printable([Unit, TRI](raw_ostream &OS) {
if (TRI && TRI->isVirtualRegister(Unit)) {
OS << "%vreg" << TargetRegisterInfo::virtReg2Index(Unit);
} else {
OS << PrintRegUnit(Unit, TRI);
}
});
2013-08-23 19:48:53 +02:00
}
} // End of llvm namespace
/// getAllocatableClass - Return the maximal subclass of the given register
/// class that is alloctable, or NULL.
const TargetRegisterClass *
TargetRegisterInfo::getAllocatableClass(const TargetRegisterClass *RC) const {
if (!RC || RC->isAllocatable())
return RC;
for (BitMaskClassIterator It(RC->getSubClassMask(), *this); It.isValid();
++It) {
const TargetRegisterClass *SubRC = getRegClass(It.getID());
if (SubRC->isAllocatable())
return SubRC;
}
return nullptr;
}
/// getMinimalPhysRegClass - Returns the Register Class of a physical
/// register of the given type, picking the most sub register class of
/// the right type that contains this physreg.
const TargetRegisterClass *
TargetRegisterInfo::getMinimalPhysRegClass(unsigned reg, MVT VT) const {
assert(isPhysicalRegister(reg) && "reg must be a physical register");
// Pick the most sub register class of the right type that contains
// this physreg.
const TargetRegisterClass* BestRC = nullptr;
for (const TargetRegisterClass* RC : regclasses()) {
if ((VT == MVT::Other || isTypeLegalForClass(*RC, VT)) &&
RC->contains(reg) && (!BestRC || BestRC->hasSubClass(RC)))
BestRC = RC;
}
assert(BestRC && "Couldn't find the register class");
return BestRC;
}
/// getAllocatableSetForRC - Toggle the bits that represent allocatable
/// registers for the specific register class.
2009-10-10 00:09:05 +02:00
static void getAllocatableSetForRC(const MachineFunction &MF,
const TargetRegisterClass *RC, BitVector &R){
assert(RC->isAllocatable() && "invalid for nonallocatable sets");
ArrayRef<MCPhysReg> Order = RC->getRawAllocationOrder(MF);
for (unsigned i = 0; i != Order.size(); ++i)
R.set(Order[i]);
}
2009-10-10 00:09:05 +02:00
BitVector TargetRegisterInfo::getAllocatableSet(const MachineFunction &MF,
const TargetRegisterClass *RC) const {
BitVector Allocatable(getNumRegs());
if (RC) {
// A register class with no allocatable subclass returns an empty set.
const TargetRegisterClass *SubClass = getAllocatableClass(RC);
if (SubClass)
getAllocatableSetForRC(MF, SubClass, Allocatable);
} else {
for (const TargetRegisterClass *C : regclasses())
if (C->isAllocatable())
getAllocatableSetForRC(MF, C, Allocatable);
}
// Mask out the reserved registers
BitVector Reserved = getReservedRegs(MF);
Allocatable &= Reserved.flip();
return Allocatable;
}
static inline
const TargetRegisterClass *firstCommonClass(const uint32_t *A,
const uint32_t *B,
[X86] Part 1 to fix x86-64 fp128 calling convention. Almost all these changes are conditioned and only apply to the new x86-64 f128 type configuration, which will be enabled in a follow up patch. They are required together to make new f128 work. If there is any error, we should fix or revert them as a whole. These changes should have no impact to current configurations. * Relax type legalization checks to accept new f128 type configuration, whose TypeAction is TypeSoftenFloat, not TypeLegal, but also has TLI.isTypeLegal true. * Relax GetSoftenedFloat to return in some cases f128 type SDValue, which is TLI.isTypeLegal but not "softened" to i128 node. * Allow customized FABS, FNEG, FCOPYSIGN on new f128 type configuration, to generate optimized bitwise operators for libm functions. * Enhance related Lower* functions to handle f128 type. * Enhance DAGTypeLegalizer::run, SoftenFloatResult, and related functions to keep new f128 type in register, and convert f128 operators to library calls. * Fix Combiner, Emitter, Legalizer routines that did not handle f128 type. * Add ExpandConstant to handle i128 constants, ExpandNode to handle ISD::Constant node. * Add one more parameter to getCommonSubClass and firstCommonClass, to guarantee that returned common sub class will contain the specified simple value type. This extra parameter is used by EmitCopyFromReg in InstrEmitter.cpp. * Fix infinite loop in getTypeLegalizationCost when f128 is the value type. * Fix printOperand to handle null operand. * Enhance ISD::BITCAST node to handle f128 constant. * Expand new f128 type for BR_CC, SELECT_CC, SELECT, SETCC nodes. * Enhance X86AsmPrinter to emit f128 values in comments. Differential Revision: http://reviews.llvm.org/D15134 llvm-svn: 254653
2015-12-03 23:02:40 +01:00
const TargetRegisterInfo *TRI,
const MVT::SimpleValueType SVT =
MVT::SimpleValueType::Any) {
const MVT VT(SVT);
for (unsigned I = 0, E = TRI->getNumRegClasses(); I < E; I += 32)
[X86] Part 1 to fix x86-64 fp128 calling convention. Almost all these changes are conditioned and only apply to the new x86-64 f128 type configuration, which will be enabled in a follow up patch. They are required together to make new f128 work. If there is any error, we should fix or revert them as a whole. These changes should have no impact to current configurations. * Relax type legalization checks to accept new f128 type configuration, whose TypeAction is TypeSoftenFloat, not TypeLegal, but also has TLI.isTypeLegal true. * Relax GetSoftenedFloat to return in some cases f128 type SDValue, which is TLI.isTypeLegal but not "softened" to i128 node. * Allow customized FABS, FNEG, FCOPYSIGN on new f128 type configuration, to generate optimized bitwise operators for libm functions. * Enhance related Lower* functions to handle f128 type. * Enhance DAGTypeLegalizer::run, SoftenFloatResult, and related functions to keep new f128 type in register, and convert f128 operators to library calls. * Fix Combiner, Emitter, Legalizer routines that did not handle f128 type. * Add ExpandConstant to handle i128 constants, ExpandNode to handle ISD::Constant node. * Add one more parameter to getCommonSubClass and firstCommonClass, to guarantee that returned common sub class will contain the specified simple value type. This extra parameter is used by EmitCopyFromReg in InstrEmitter.cpp. * Fix infinite loop in getTypeLegalizationCost when f128 is the value type. * Fix printOperand to handle null operand. * Enhance ISD::BITCAST node to handle f128 constant. * Expand new f128 type for BR_CC, SELECT_CC, SELECT, SETCC nodes. * Enhance X86AsmPrinter to emit f128 values in comments. Differential Revision: http://reviews.llvm.org/D15134 llvm-svn: 254653
2015-12-03 23:02:40 +01:00
if (unsigned Common = *A++ & *B++) {
const TargetRegisterClass *RC =
TRI->getRegClass(I + countTrailingZeros(Common));
if (SVT == MVT::SimpleValueType::Any || TRI->isTypeLegalForClass(*RC, VT))
[X86] Part 1 to fix x86-64 fp128 calling convention. Almost all these changes are conditioned and only apply to the new x86-64 f128 type configuration, which will be enabled in a follow up patch. They are required together to make new f128 work. If there is any error, we should fix or revert them as a whole. These changes should have no impact to current configurations. * Relax type legalization checks to accept new f128 type configuration, whose TypeAction is TypeSoftenFloat, not TypeLegal, but also has TLI.isTypeLegal true. * Relax GetSoftenedFloat to return in some cases f128 type SDValue, which is TLI.isTypeLegal but not "softened" to i128 node. * Allow customized FABS, FNEG, FCOPYSIGN on new f128 type configuration, to generate optimized bitwise operators for libm functions. * Enhance related Lower* functions to handle f128 type. * Enhance DAGTypeLegalizer::run, SoftenFloatResult, and related functions to keep new f128 type in register, and convert f128 operators to library calls. * Fix Combiner, Emitter, Legalizer routines that did not handle f128 type. * Add ExpandConstant to handle i128 constants, ExpandNode to handle ISD::Constant node. * Add one more parameter to getCommonSubClass and firstCommonClass, to guarantee that returned common sub class will contain the specified simple value type. This extra parameter is used by EmitCopyFromReg in InstrEmitter.cpp. * Fix infinite loop in getTypeLegalizationCost when f128 is the value type. * Fix printOperand to handle null operand. * Enhance ISD::BITCAST node to handle f128 constant. * Expand new f128 type for BR_CC, SELECT_CC, SELECT, SETCC nodes. * Enhance X86AsmPrinter to emit f128 values in comments. Differential Revision: http://reviews.llvm.org/D15134 llvm-svn: 254653
2015-12-03 23:02:40 +01:00
return RC;
}
return nullptr;
}
const TargetRegisterClass *
TargetRegisterInfo::getCommonSubClass(const TargetRegisterClass *A,
[X86] Part 1 to fix x86-64 fp128 calling convention. Almost all these changes are conditioned and only apply to the new x86-64 f128 type configuration, which will be enabled in a follow up patch. They are required together to make new f128 work. If there is any error, we should fix or revert them as a whole. These changes should have no impact to current configurations. * Relax type legalization checks to accept new f128 type configuration, whose TypeAction is TypeSoftenFloat, not TypeLegal, but also has TLI.isTypeLegal true. * Relax GetSoftenedFloat to return in some cases f128 type SDValue, which is TLI.isTypeLegal but not "softened" to i128 node. * Allow customized FABS, FNEG, FCOPYSIGN on new f128 type configuration, to generate optimized bitwise operators for libm functions. * Enhance related Lower* functions to handle f128 type. * Enhance DAGTypeLegalizer::run, SoftenFloatResult, and related functions to keep new f128 type in register, and convert f128 operators to library calls. * Fix Combiner, Emitter, Legalizer routines that did not handle f128 type. * Add ExpandConstant to handle i128 constants, ExpandNode to handle ISD::Constant node. * Add one more parameter to getCommonSubClass and firstCommonClass, to guarantee that returned common sub class will contain the specified simple value type. This extra parameter is used by EmitCopyFromReg in InstrEmitter.cpp. * Fix infinite loop in getTypeLegalizationCost when f128 is the value type. * Fix printOperand to handle null operand. * Enhance ISD::BITCAST node to handle f128 constant. * Expand new f128 type for BR_CC, SELECT_CC, SELECT, SETCC nodes. * Enhance X86AsmPrinter to emit f128 values in comments. Differential Revision: http://reviews.llvm.org/D15134 llvm-svn: 254653
2015-12-03 23:02:40 +01:00
const TargetRegisterClass *B,
const MVT::SimpleValueType SVT) const {
// First take care of the trivial cases.
if (A == B)
return A;
if (!A || !B)
return nullptr;
// Register classes are ordered topologically, so the largest common
// sub-class it the common sub-class with the smallest ID.
[X86] Part 1 to fix x86-64 fp128 calling convention. Almost all these changes are conditioned and only apply to the new x86-64 f128 type configuration, which will be enabled in a follow up patch. They are required together to make new f128 work. If there is any error, we should fix or revert them as a whole. These changes should have no impact to current configurations. * Relax type legalization checks to accept new f128 type configuration, whose TypeAction is TypeSoftenFloat, not TypeLegal, but also has TLI.isTypeLegal true. * Relax GetSoftenedFloat to return in some cases f128 type SDValue, which is TLI.isTypeLegal but not "softened" to i128 node. * Allow customized FABS, FNEG, FCOPYSIGN on new f128 type configuration, to generate optimized bitwise operators for libm functions. * Enhance related Lower* functions to handle f128 type. * Enhance DAGTypeLegalizer::run, SoftenFloatResult, and related functions to keep new f128 type in register, and convert f128 operators to library calls. * Fix Combiner, Emitter, Legalizer routines that did not handle f128 type. * Add ExpandConstant to handle i128 constants, ExpandNode to handle ISD::Constant node. * Add one more parameter to getCommonSubClass and firstCommonClass, to guarantee that returned common sub class will contain the specified simple value type. This extra parameter is used by EmitCopyFromReg in InstrEmitter.cpp. * Fix infinite loop in getTypeLegalizationCost when f128 is the value type. * Fix printOperand to handle null operand. * Enhance ISD::BITCAST node to handle f128 constant. * Expand new f128 type for BR_CC, SELECT_CC, SELECT, SETCC nodes. * Enhance X86AsmPrinter to emit f128 values in comments. Differential Revision: http://reviews.llvm.org/D15134 llvm-svn: 254653
2015-12-03 23:02:40 +01:00
return firstCommonClass(A->getSubClassMask(), B->getSubClassMask(), this, SVT);
}
const TargetRegisterClass *
TargetRegisterInfo::getMatchingSuperRegClass(const TargetRegisterClass *A,
const TargetRegisterClass *B,
unsigned Idx) const {
assert(A && B && "Missing register class");
assert(Idx && "Bad sub-register index");
// Find Idx in the list of super-register indices.
for (SuperRegClassIterator RCI(B, this); RCI.isValid(); ++RCI)
if (RCI.getSubReg() == Idx)
// The bit mask contains all register classes that are projected into B
// by Idx. Find a class that is also a sub-class of A.
return firstCommonClass(RCI.getMask(), A->getSubClassMask(), this);
return nullptr;
}
const TargetRegisterClass *TargetRegisterInfo::
getCommonSuperRegClass(const TargetRegisterClass *RCA, unsigned SubA,
const TargetRegisterClass *RCB, unsigned SubB,
unsigned &PreA, unsigned &PreB) const {
assert(RCA && SubA && RCB && SubB && "Invalid arguments");
// Search all pairs of sub-register indices that project into RCA and RCB
// respectively. This is quadratic, but usually the sets are very small. On
// most targets like X86, there will only be a single sub-register index
// (e.g., sub_16bit projecting into GR16).
//
// The worst case is a register class like DPR on ARM.
// We have indices dsub_0..dsub_7 projecting into that class.
//
// It is very common that one register class is a sub-register of the other.
// Arrange for RCA to be the larger register so the answer will be found in
// the first iteration. This makes the search linear for the most common
// case.
const TargetRegisterClass *BestRC = nullptr;
unsigned *BestPreA = &PreA;
unsigned *BestPreB = &PreB;
if (getRegSizeInBits(*RCA) < getRegSizeInBits(*RCB)) {
std::swap(RCA, RCB);
std::swap(SubA, SubB);
std::swap(BestPreA, BestPreB);
}
// Also terminate the search one we have found a register class as small as
// RCA.
unsigned MinSize = getRegSizeInBits(*RCA);
for (SuperRegClassIterator IA(RCA, this, true); IA.isValid(); ++IA) {
unsigned FinalA = composeSubRegIndices(IA.getSubReg(), SubA);
for (SuperRegClassIterator IB(RCB, this, true); IB.isValid(); ++IB) {
// Check if a common super-register class exists for this index pair.
const TargetRegisterClass *RC =
firstCommonClass(IA.getMask(), IB.getMask(), this);
if (!RC || getRegSizeInBits(*RC) < MinSize)
continue;
// The indexes must compose identically: PreA+SubA == PreB+SubB.
unsigned FinalB = composeSubRegIndices(IB.getSubReg(), SubB);
if (FinalA != FinalB)
continue;
// Is RC a better candidate than BestRC?
if (BestRC && getRegSizeInBits(*RC) >= getRegSizeInBits(*BestRC))
continue;
// Yes, RC is the smallest super-register seen so far.
BestRC = RC;
*BestPreA = IA.getSubReg();
*BestPreB = IB.getSubReg();
// Bail early if we reached MinSize. We won't find a better candidate.
if (getRegSizeInBits(*BestRC) == MinSize)
return BestRC;
}
}
return BestRC;
}
/// \brief Check if the registers defined by the pair (RegisterClass, SubReg)
/// share the same register file.
static bool shareSameRegisterFile(const TargetRegisterInfo &TRI,
const TargetRegisterClass *DefRC,
unsigned DefSubReg,
const TargetRegisterClass *SrcRC,
unsigned SrcSubReg) {
// Same register class.
if (DefRC == SrcRC)
return true;
// Both operands are sub registers. Check if they share a register class.
unsigned SrcIdx, DefIdx;
if (SrcSubReg && DefSubReg) {
return TRI.getCommonSuperRegClass(SrcRC, SrcSubReg, DefRC, DefSubReg,
SrcIdx, DefIdx) != nullptr;
}
// At most one of the register is a sub register, make it Src to avoid
// duplicating the test.
if (!SrcSubReg) {
std::swap(DefSubReg, SrcSubReg);
std::swap(DefRC, SrcRC);
}
// One of the register is a sub register, check if we can get a superclass.
if (SrcSubReg)
return TRI.getMatchingSuperRegClass(SrcRC, DefRC, SrcSubReg) != nullptr;
// Plain copy.
return TRI.getCommonSubClass(DefRC, SrcRC) != nullptr;
}
bool TargetRegisterInfo::shouldRewriteCopySrc(const TargetRegisterClass *DefRC,
unsigned DefSubReg,
const TargetRegisterClass *SrcRC,
unsigned SrcSubReg) const {
// If this source does not incur a cross register bank copy, use it.
return shareSameRegisterFile(*this, DefRC, DefSubReg, SrcRC, SrcSubReg);
}
// Compute target-independent register allocator hints to help eliminate copies.
void
TargetRegisterInfo::getRegAllocationHints(unsigned VirtReg,
ArrayRef<MCPhysReg> Order,
SmallVectorImpl<MCPhysReg> &Hints,
const MachineFunction &MF,
const VirtRegMap *VRM,
const LiveRegMatrix *Matrix) const {
const MachineRegisterInfo &MRI = MF.getRegInfo();
std::pair<unsigned, unsigned> Hint = MRI.getRegAllocationHint(VirtReg);
// Hints with HintType != 0 were set by target-dependent code.
// Such targets must provide their own implementation of
// TRI::getRegAllocationHints to interpret those hint types.
assert(Hint.first == 0 && "Target must implement TRI::getRegAllocationHints");
// Target-independent hints are either a physical or a virtual register.
unsigned Phys = Hint.second;
if (VRM && isVirtualRegister(Phys))
Phys = VRM->getPhys(Phys);
// Check that Phys is a valid hint in VirtReg's register class.
if (!isPhysicalRegister(Phys))
return;
if (MRI.isReserved(Phys))
return;
// Check that Phys is in the allocation order. We shouldn't heed hints
// from VirtReg's register class if they aren't in the allocation order. The
// target probably has a reason for removing the register.
if (!is_contained(Order, Phys))
return;
// All clear, tell the register allocator to prefer this register.
Hints.push_back(Phys);
}
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
bool TargetRegisterInfo::canRealignStack(const MachineFunction &MF) const {
return !MF.getFunction()->hasFnAttribute("no-realign-stack");
}
bool TargetRegisterInfo::needsStackRealignment(
const MachineFunction &MF) const {
const MachineFrameInfo &MFI = MF.getFrameInfo();
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
const TargetFrameLowering *TFI = MF.getSubtarget().getFrameLowering();
const Function *F = MF.getFunction();
unsigned StackAlign = TFI->getStackAlignment();
bool requiresRealignment = ((MFI.getMaxAlignment() > StackAlign) ||
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
F->hasFnAttribute(Attribute::StackAlignment));
if (MF.getFunction()->hasFnAttribute("stackrealign") || requiresRealignment) {
Targets: commonize some stack realignment code This patch does the following: * Fix FIXME on `needsStackRealignment`: it is now shared between multiple targets, implemented in `TargetRegisterInfo`, and isn't `virtual` anymore. This will break out-of-tree targets, silently if they used `virtual` and with a build error if they used `override`. * Factor out `canRealignStack` as a `virtual` function on `TargetRegisterInfo`, by default only looks for the `no-realign-stack` function attribute. Multiple targets duplicated the same `needsStackRealignment` code: - Aarch64. - ARM. - Mips almost: had extra `DEBUG` diagnostic, which the default implementation now has. - PowerPC. - WebAssembly. - x86 almost: has an extra `-force-align-stack` option, which the default implementation now has. The default implementation of `needsStackRealignment` used to just return `false`. My current patch changes the behavior by simply using the above shared behavior. This affects: - AMDGPU - BPF - CppBackend - MSP430 - NVPTX - Sparc - SystemZ - XCore - Out-of-tree targets This is a breaking change! `make check` passes. The only implementation of the `virtual` function (besides the slight different in x86) was Hexagon (which did `MF.getFrameInfo()->getMaxAlignment() > 8`), and potentially some out-of-tree targets. Hexagon now uses the default implementation. `needsStackRealignment` was being overwritten in `<Target>GenRegisterInfo.inc`, to return `false` as the default also did. That was odd and is now gone. Reviewers: sunfish Subscribers: aemerson, llvm-commits, jfb Differential Revision: http://reviews.llvm.org/D11160 llvm-svn: 242727
2015-07-21 00:51:32 +02:00
if (canRealignStack(MF))
return true;
DEBUG(dbgs() << "Can't realign function's stack: " << F->getName() << "\n");
}
return false;
}
bool TargetRegisterInfo::regmaskSubsetEqual(const uint32_t *mask0,
const uint32_t *mask1) const {
unsigned N = (getNumRegs()+31) / 32;
for (unsigned I = 0; I < N; ++I)
if ((mask0[I] & mask1[I]) != mask0[I])
return false;
return true;
}
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
LLVM_DUMP_METHOD
void TargetRegisterInfo::dumpReg(unsigned Reg, unsigned SubRegIndex,
const TargetRegisterInfo *TRI) {
dbgs() << PrintReg(Reg, TRI, SubRegIndex) << "\n";
}
#endif