llvm/lib/Target/Mips/MipsSubtarget.cpp

221 lines
7.4 KiB
C++
Raw Normal View History

//===-- MipsSubtarget.cpp - Mips Subtarget Information --------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements the Mips specific subclass of TargetSubtargetInfo.
//
//===----------------------------------------------------------------------===//
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
#include "MipsMachineFunction.h"
#include "Mips.h"
#include "MipsRegisterInfo.h"
#include "MipsSubtarget.h"
#include "MipsTargetMachine.h"
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
#include "llvm/IR/Attributes.h"
#include "llvm/IR/Function.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/TargetRegistry.h"
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
#define DEBUG_TYPE "mips-subtarget"
#define GET_SUBTARGETINFO_TARGET_DESC
#define GET_SUBTARGETINFO_CTOR
#include "MipsGenSubtargetInfo.inc"
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
// FIXME: Maybe this should be on by default when Mips16 is specified
//
static cl::opt<bool> Mixed16_32(
"mips-mixed-16-32",
cl::init(false),
cl::desc("Allow for a mixture of Mips16 "
"and Mips32 code in a single source file"),
cl::Hidden);
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179185 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-10 16:58:04 +00:00
static cl::opt<bool> Mips_Os16(
"mips-os16",
cl::init(false),
cl::desc("Compile all functions that don' use "
"floating point as Mips 16"),
cl::Hidden);
static cl::opt<bool>
Mips16HardFloat("mips16-hard-float", cl::NotHidden,
cl::desc("MIPS: mips16 hard float enable."),
cl::init(false));
static cl::opt<bool>
Mips16ConstantIslands(
"mips16-constant-islands", cl::NotHidden,
cl::desc("MIPS: mips16 constant islands enable."),
cl::init(true));
/// Select the Mips CPU for the given triple and cpu name.
/// FIXME: Merge with the copy in MipsMCTargetDesc.cpp
static StringRef selectMipsCPU(Triple TT, StringRef CPU) {
if (CPU.empty() || CPU == "generic") {
if (TT.getArch() == Triple::mips || TT.getArch() == Triple::mipsel)
CPU = "mips32";
else
CPU = "mips64";
}
return CPU;
}
void MipsSubtarget::anchor() { }
static std::string computeDataLayout(const MipsSubtarget &ST) {
std::string Ret = "";
// There are both little and big endian mips.
if (ST.isLittle())
Ret += "e";
else
Ret += "E";
Ret += "-m:m";
// Pointers are 32 bit on some ABIs.
if (!ST.isABI_N64())
Ret += "-p:32:32";
// 8 and 16 bit integers only need no have natural alignment, but try to
// align them to 32 bits. 64 bit integers have natural alignment.
Ret += "-i8:8:32-i16:16:32-i64:64";
// 32 bit registers are always available and the stack is at least 64 bit
// aligned. On N64 64 bit registers are also available and the stack is
// 128 bit aligned.
if (ST.isABI_N64() || ST.isABI_N32())
Ret += "-n32:64-S128";
else
Ret += "-n32-S64";
return Ret;
}
MipsSubtarget::MipsSubtarget(const std::string &TT, const std::string &CPU,
const std::string &FS, bool little,
const MipsTargetMachine *_TM)
: MipsGenSubtargetInfo(TT, CPU, FS), MipsArchVersion(Mips32),
MipsABI(UnknownABI), IsLittle(little), IsSingleFloat(false),
IsFPXX(false), NoABICalls(false), IsFP64bit(false), UseOddSPReg(true),
IsNaN2008bit(false), IsGP64bit(false), HasVFPU(false), HasCnMips(false),
IsLinux(true), HasMips3_32(false), HasMips3_32r2(false),
HasMips4_32(false), HasMips4_32r2(false), HasMips5_32r2(false),
InMips16Mode(false), InMips16HardFloat(Mips16HardFloat),
InMicroMipsMode(false), HasDSP(false), HasDSPR2(false),
AllowMixed16_32(Mixed16_32 | Mips_Os16), Os16(Mips_Os16),
HasMSA(false), TM(_TM), TargetTriple(TT),
DL(computeDataLayout(initializeSubtargetDependencies(CPU, FS, TM))),
TSInfo(DL), InstrInfo(MipsInstrInfo::create(*this)),
FrameLowering(MipsFrameLowering::create(*this)),
TLInfo(MipsTargetLowering::create(*TM, *this)) {
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
PreviousInMips16Mode = InMips16Mode;
// Don't even attempt to generate code for MIPS-I, MIPS-II, MIPS-III, and
// MIPS-V. They have not been tested and currently exist for the integrated
// assembler only.
if (MipsArchVersion == Mips1)
report_fatal_error("Code generation for MIPS-I is not implemented", false);
if (MipsArchVersion == Mips2)
report_fatal_error("Code generation for MIPS-II is not implemented", false);
if (MipsArchVersion == Mips3)
report_fatal_error("Code generation for MIPS-III is not implemented",
false);
if (MipsArchVersion == Mips5)
report_fatal_error("Code generation for MIPS-V is not implemented", false);
// Assert exactly one ABI was chosen.
assert(MipsABI != UnknownABI);
assert((((getFeatureBits() & Mips::FeatureO32) != 0) +
((getFeatureBits() & Mips::FeatureEABI) != 0) +
((getFeatureBits() & Mips::FeatureN32) != 0) +
((getFeatureBits() & Mips::FeatureN64) != 0)) == 1);
// Check if Architecture and ABI are compatible.
assert(((!isGP64bit() && (isABI_O32() || isABI_EABI())) ||
(isGP64bit() && (isABI_N32() || isABI_N64()))) &&
"Invalid Arch & ABI pair.");
if (hasMSA() && !isFP64bit())
report_fatal_error("MSA requires a 64-bit FPU register file (FR=1 mode). "
"See -mattr=+fp64.",
false);
[mips] Add support for -modd-spreg/-mno-odd-spreg Summary: When -mno-odd-spreg is in effect, 32-bit floating point values are not permitted in odd FPU registers. The option also prohibits 32-bit and 64-bit floating point comparison results from being written to odd registers. This option has three purposes: * It allows support for certain MIPS implementations such as loongson-3a that do not allow the use of odd registers for single precision arithmetic. * When using -mfpxx, -mno-odd-spreg is the default and this allows us to statically check that code is compliant with the O32 FPXX ABI since mtc1/mfc1 instructions to/from odd registers are guaranteed not to appear for any reason. Once this has been established, the user can then re-enable -modd-spreg to regain the use of all 32 single-precision registers. * When using -mfp64 and -mno-odd-spreg together, an O32 extension named O32 FP64A is used as the ABI. This is intended to provide almost all functionality of an FR=1 processor but can also be executed on a FR=0 core with the assistance of a hardware compatibility mode which emulates FR=0 behaviour on an FR=1 processor. * Added '.module oddspreg' and '.module nooddspreg' each of which update the .MIPS.abiflags section appropriately * Moved setFpABI() call inside emitDirectiveModuleFP() so that the caller doesn't have to remember to do it. * MipsABIFlags now calculates the flags1 and flags2 member on demand rather than trying to maintain them in the same format they will be emitted in. There is one portion of the -mfp64 and -mno-odd-spreg combination that is not implemented yet. Moves to/from odd-numbered double-precision registers must not use mtc1. I will fix this in a follow-up. Differential Revision: http://reviews.llvm.org/D4383 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212717 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 13:38:23 +00:00
if (!isABI_O32() && !useOddSPReg())
report_fatal_error("-mattr=+nooddspreg requires the O32 ABI.", false);
[mips] Add support for -modd-spreg/-mno-odd-spreg Summary: When -mno-odd-spreg is in effect, 32-bit floating point values are not permitted in odd FPU registers. The option also prohibits 32-bit and 64-bit floating point comparison results from being written to odd registers. This option has three purposes: * It allows support for certain MIPS implementations such as loongson-3a that do not allow the use of odd registers for single precision arithmetic. * When using -mfpxx, -mno-odd-spreg is the default and this allows us to statically check that code is compliant with the O32 FPXX ABI since mtc1/mfc1 instructions to/from odd registers are guaranteed not to appear for any reason. Once this has been established, the user can then re-enable -modd-spreg to regain the use of all 32 single-precision registers. * When using -mfp64 and -mno-odd-spreg together, an O32 extension named O32 FP64A is used as the ABI. This is intended to provide almost all functionality of an FR=1 processor but can also be executed on a FR=0 core with the assistance of a hardware compatibility mode which emulates FR=0 behaviour on an FR=1 processor. * Added '.module oddspreg' and '.module nooddspreg' each of which update the .MIPS.abiflags section appropriately * Moved setFpABI() call inside emitDirectiveModuleFP() so that the caller doesn't have to remember to do it. * MipsABIFlags now calculates the flags1 and flags2 member on demand rather than trying to maintain them in the same format they will be emitted in. There is one portion of the -mfp64 and -mno-odd-spreg combination that is not implemented yet. Moves to/from odd-numbered double-precision registers must not use mtc1. I will fix this in a follow-up. Differential Revision: http://reviews.llvm.org/D4383 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212717 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 13:38:23 +00:00
if (IsFPXX && (isABI_N32() || isABI_N64()))
report_fatal_error("FPXX is not permitted for the N32/N64 ABI's.", false);
if (hasMips32r6()) {
StringRef ISA = hasMips64r6() ? "MIPS64r6" : "MIPS32r6";
assert(isFP64bit());
assert(isNaN2008());
if (hasDSP())
report_fatal_error(ISA + " is not compatible with the DSP ASE", false);
}
// Is the target system Linux ?
if (TT.find("linux") == std::string::npos)
IsLinux = false;
// Set UseSmallSection.
// TODO: Investigate the IsLinux check. I suspect it's really checking for
// bare-metal.
UseSmallSection = !IsLinux && (TM->getRelocationModel() == Reloc::Static);
}
/// This overrides the PostRAScheduler bit in the SchedModel for any CPU.
bool MipsSubtarget::enablePostMachineScheduler() const { return true; }
void MipsSubtarget::getCriticalPathRCs(RegClassVector &CriticalPathRCs) const {
CriticalPathRCs.clear();
CriticalPathRCs.push_back(isGP64bit() ?
&Mips::GPR64RegClass : &Mips::GPR32RegClass);
}
CodeGenOpt::Level MipsSubtarget::getOptLevelToEnablePostRAScheduler() const {
return CodeGenOpt::Aggressive;
}
This patch enables llvm to switch between compiling for mips32/mips64 and mips16 on a per function basis. Because this patch is somewhat involved I have provide an overview of the key pieces of it. The patch is written so as to not change the behavior of the non mixed mode. We have tested this a lot but it is something new to switch subtargets so we don't want any chance of regression in the mainline compiler until we have more confidence in this. Mips32/64 are very different from Mip16 as is the case of ARM vs Thumb1. For that reason there are derived versions of the register info, frame info, instruction info and instruction selection classes. Now we register three separate passes for instruction selection. One which is used to switch subtargets (MipsModuleISelDAGToDAG.cpp) and then one for each of the current subtargets (Mips16ISelDAGToDAG.cpp and MipsSEISelDAGToDAG.cpp). When the ModuleISel pass runs, it determines if there is a need to switch subtargets and if so, the owning pointers in MipsTargetMachine are appropriately changed. When 16Isel or SEIsel is run, they will return immediately without doing any work if the current subtarget mode does not apply to them. In addition, MipsAsmPrinter needs to be reset on a function basis. The pass BasicTargetTransformInfo is substituted with a null pass since the pass is immutable and really needs to be a function pass for it to be used with changing subtargets. This will be fixed in a follow on patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-04-09 19:46:01 +00:00
MipsSubtarget &
MipsSubtarget::initializeSubtargetDependencies(StringRef CPU, StringRef FS,
const TargetMachine *TM) {
std::string CPUName = selectMipsCPU(TargetTriple, CPU);
// Parse features string.
ParseSubtargetFeatures(CPUName, FS);
// Initialize scheduling itinerary for the specified CPU.
InstrItins = getInstrItineraryForCPU(CPUName);
if (InMips16Mode && !TM->Options.UseSoftFloat)
InMips16HardFloat = true;
return *this;
}
bool MipsSubtarget::abiUsesSoftFloat() const {
return TM->Options.UseSoftFloat && !InMips16HardFloat;
}
bool MipsSubtarget::useConstantIslands() {
DEBUG(dbgs() << "use constant islands " << Mips16ConstantIslands << "\n");
return Mips16ConstantIslands;
}
Reloc::Model MipsSubtarget::getRelocationModel() const {
return TM->getRelocationModel();
}