mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-07-27 13:29:50 +00:00
6a0f060f64
The VSX instruction set has two types of FMA instructions: A-type (where the
addend is taken from the output register) and M-type (where one of the product
operands is taken from the output register). This adds a small pass that runs
just after MI scheduling (and, thus, just before register allocation) that
mutates A-type instructions (that are created during isel) into M-type
instructions when:
1. This will eliminate an otherwise-necessary copy of the addend
2. One of the product operands is killed by the instruction
The "right" moment to make this decision is in between scheduling and register
allocation, because only there do we know whether or not one of the product
operands is killed by any particular instruction. Unfortunately, this also
makes the implementation somewhat complicated, because the MIs are not in SSA
form and we need to preserve the LiveIntervals analysis.
As a simple example, if we have:
%vreg5<def> = COPY %vreg9; VSLRC:%vreg5,%vreg9
%vreg5<def,tied1> = XSMADDADP %vreg5<tied0>, %vreg17, %vreg16,
%RM<imp-use>; VSLRC:%vreg5,%vreg17,%vreg16
...
%vreg9<def,tied1> = XSMADDADP %vreg9<tied0>, %vreg17, %vreg19,
%RM<imp-use>; VSLRC:%vreg9,%vreg17,%vreg19
...
We can eliminate the copy by changing from the A-type to the
M-type instruction. This means:
%vreg5<def,tied1> = XSMADDADP %vreg5<tied0>, %vreg17, %vreg16,
%RM<imp-use>; VSLRC:%vreg5,%vreg17,%vreg16
is replaced by:
%vreg16<def,tied1> = XSMADDMDP %vreg16<tied0>, %vreg18, %vreg9,
%RM<imp-use>; VSLRC:%vreg16,%vreg18,%vreg9
and we remove: %vreg5<def> = COPY %vreg9; VSLRC:%vreg5,%vreg9
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204768 91177308-0d34-0410-b5e6-96231b3b80d8
104 lines
3.3 KiB
C++
104 lines
3.3 KiB
C++
//===-- PPC.h - Top-level interface for PowerPC Target ----------*- C++ -*-===//
|
|
//
|
|
// The LLVM Compiler Infrastructure
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This file contains the entry points for global functions defined in the LLVM
|
|
// PowerPC back-end.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#ifndef LLVM_TARGET_POWERPC_H
|
|
#define LLVM_TARGET_POWERPC_H
|
|
|
|
#include "MCTargetDesc/PPCMCTargetDesc.h"
|
|
#include <string>
|
|
|
|
// GCC #defines PPC on Linux but we use it as our namespace name
|
|
#undef PPC
|
|
|
|
namespace llvm {
|
|
class PPCTargetMachine;
|
|
class PassRegistry;
|
|
class FunctionPass;
|
|
class ImmutablePass;
|
|
class JITCodeEmitter;
|
|
class MachineInstr;
|
|
class AsmPrinter;
|
|
class MCInst;
|
|
|
|
FunctionPass *createPPCCTRLoops(PPCTargetMachine &TM);
|
|
#ifndef NDEBUG
|
|
FunctionPass *createPPCCTRLoopsVerify();
|
|
#endif
|
|
FunctionPass *createPPCEarlyReturnPass();
|
|
FunctionPass *createPPCVSXCopyPass();
|
|
FunctionPass *createPPCVSXFMAMutatePass();
|
|
FunctionPass *createPPCBranchSelectionPass();
|
|
FunctionPass *createPPCISelDag(PPCTargetMachine &TM);
|
|
FunctionPass *createPPCJITCodeEmitterPass(PPCTargetMachine &TM,
|
|
JITCodeEmitter &MCE);
|
|
void LowerPPCMachineInstrToMCInst(const MachineInstr *MI, MCInst &OutMI,
|
|
AsmPrinter &AP, bool isDarwin);
|
|
|
|
/// \brief Creates an PPC-specific Target Transformation Info pass.
|
|
ImmutablePass *createPPCTargetTransformInfoPass(const PPCTargetMachine *TM);
|
|
|
|
void initializePPCVSXFMAMutatePass(PassRegistry&);
|
|
extern char &PPCVSXFMAMutateID;
|
|
|
|
namespace PPCII {
|
|
|
|
/// Target Operand Flag enum.
|
|
enum TOF {
|
|
//===------------------------------------------------------------------===//
|
|
// PPC Specific MachineOperand flags.
|
|
MO_NO_FLAG,
|
|
|
|
/// MO_DARWIN_STUB - On a symbol operand "FOO", this indicates that the
|
|
/// reference is actually to the "FOO$stub" symbol. This is used for calls
|
|
/// and jumps to external functions on Tiger and earlier.
|
|
MO_DARWIN_STUB = 1,
|
|
|
|
/// MO_PIC_FLAG - If this bit is set, the symbol reference is relative to
|
|
/// the function's picbase, e.g. lo16(symbol-picbase).
|
|
MO_PIC_FLAG = 2,
|
|
|
|
/// MO_NLP_FLAG - If this bit is set, the symbol reference is actually to
|
|
/// the non_lazy_ptr for the global, e.g. lo16(symbol$non_lazy_ptr-picbase).
|
|
MO_NLP_FLAG = 4,
|
|
|
|
/// MO_NLP_HIDDEN_FLAG - If this bit is set, the symbol reference is to a
|
|
/// symbol with hidden visibility. This causes a different kind of
|
|
/// non-lazy-pointer to be generated.
|
|
MO_NLP_HIDDEN_FLAG = 8,
|
|
|
|
/// The next are not flags but distinct values.
|
|
MO_ACCESS_MASK = 0xf0,
|
|
|
|
/// MO_LO, MO_HA - lo16(symbol) and ha16(symbol)
|
|
MO_LO = 1 << 4,
|
|
MO_HA = 2 << 4,
|
|
|
|
MO_TPREL_LO = 4 << 4,
|
|
MO_TPREL_HA = 3 << 4,
|
|
|
|
/// These values identify relocations on immediates folded
|
|
/// into memory operations.
|
|
MO_DTPREL_LO = 5 << 4,
|
|
MO_TLSLD_LO = 6 << 4,
|
|
MO_TOC_LO = 7 << 4,
|
|
|
|
// Symbol for VK_PPC_TLS fixup attached to an ADD instruction
|
|
MO_TLS = 8 << 4
|
|
};
|
|
} // end namespace PPCII
|
|
|
|
} // end namespace llvm;
|
|
|
|
#endif
|