mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-25 04:02:41 +01:00
8b6358ead9
VSX is an ISA extension supported on the POWER7 and later cores that enhances
floating-point vector and scalar capabilities. Among other things, this adds
<2 x double> support and generally helps to reduce register pressure.
The interesting part of this ISA feature is the register configuration: there
are 64 new 128-bit vector registers, the 32 of which are super-registers of the
existing 32 scalar floating-point registers, and the second 32 of which overlap
with the 32 Altivec vector registers. This makes things like vector insertion
and extraction tricky: this can be free but only if we force a restriction to
the right register subclass when needed. A new "minipass" PPCVSXCopy takes care
of this (although it could do a more-optimal job of it; see the comment about
unnecessary copies below).
Please note that, currently, VSX is not enabled by default when targeting
anything because it is not yet ready for that. The assembler and disassembler
are fully implemented and tested. However:
- CodeGen support causes miscompiles; test-suite runtime failures:
MultiSource/Benchmarks/FreeBench/distray/distray
MultiSource/Benchmarks/McCat/08-main/main
MultiSource/Benchmarks/Olden/voronoi/voronoi
MultiSource/Benchmarks/mafft/pairlocalalign
MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4
SingleSource/Benchmarks/CoyoteBench/almabench
SingleSource/Benchmarks/Misc/matmul_f64_4x4
- The lowering currently falls back to using Altivec instructions far more
than it should. Worse, there are some things that are scalarized through the
stack that shouldn't be.
- A lot of unnecessary copies make it past the optimizers, and this needs to
be fixed.
- Many more regression tests are needed.
Normally, I'd fix these things prior to committing, but there are some
students and other contributors who would like to work this, and so it makes
sense to move this development process upstream where it can be subject to the
regular code-review procedures.
llvm-svn: 203768
99 lines
3.2 KiB
C++
99 lines
3.2 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 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 *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);
|
|
|
|
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
|