llvm/lib/Target/ARM/ARMTargetMachine.cpp
Tim Northover badb137729 ARM: expand atomic ldrex/strex loops in IR
The previous situation where ATOMIC_LOAD_WHATEVER nodes were expanded
at MachineInstr emission time had grown to be extremely large and
involved, to account for the subtly different code needed for the
various flavours (8/16/32/64 bit, cmpxchg/add/minmax).

Moving this transformation into the IR clears up the code
substantially, and makes future optimisations much easier:

1. an atomicrmw followed by using the *new* value can be more
   efficient. As an IR pass, simple CSE could handle this
   efficiently.
2. Making use of cmpxchg success/failure orderings only has to be done
   in one (simpler) place.
3. The common "cmpxchg; did we store?" idiom can be exposed to
   optimisation.

I intend to gradually improve this situation within the ARM backend
and make sure there are no hidden issues before moving the code out
into CodeGen to be shared with (at least ARM64/AArch64, though I think
PPC & Mips could benefit too).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205525 91177308-0d34-0410-b5e6-96231b3b80d8
2014-04-03 11:44:58 +00:00

315 lines
11 KiB
C++

//===-- ARMTargetMachine.cpp - Define TargetMachine for ARM ---------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//
//===----------------------------------------------------------------------===//
#include "ARM.h"
#include "ARMTargetMachine.h"
#include "ARMFrameLowering.h"
#include "llvm/CodeGen/Passes.h"
#include "llvm/MC/MCAsmInfo.h"
#include "llvm/PassManager.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/FormattedStream.h"
#include "llvm/Support/TargetRegistry.h"
#include "llvm/Target/TargetOptions.h"
#include "llvm/Transforms/Scalar.h"
using namespace llvm;
static cl::opt<bool>
DisableA15SDOptimization("disable-a15-sd-optimization", cl::Hidden,
cl::desc("Inhibit optimization of S->D register accesses on A15"),
cl::init(false));
extern "C" void LLVMInitializeARMTarget() {
// Register the target.
RegisterTargetMachine<ARMLETargetMachine> X(TheARMLETarget);
RegisterTargetMachine<ARMBETargetMachine> Y(TheARMBETarget);
RegisterTargetMachine<ThumbLETargetMachine> A(TheThumbLETarget);
RegisterTargetMachine<ThumbBETargetMachine> B(TheThumbBETarget);
}
/// TargetMachine ctor - Create an ARM architecture model.
///
ARMBaseTargetMachine::ARMBaseTargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL,
bool isLittle)
: LLVMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
Subtarget(TT, CPU, FS, isLittle, Options),
JITInfo(),
InstrItins(Subtarget.getInstrItineraryData()) {
// Default to triple-appropriate float ABI
if (Options.FloatABIType == FloatABI::Default)
this->Options.FloatABIType =
Subtarget.isTargetHardFloat() ? FloatABI::Hard : FloatABI::Soft;
}
void ARMBaseTargetMachine::addAnalysisPasses(PassManagerBase &PM) {
// Add first the target-independent BasicTTI pass, then our ARM pass. This
// allows the ARM pass to delegate to the target independent layer when
// appropriate.
PM.add(createBasicTargetTransformInfoPass(this));
PM.add(createARMTargetTransformInfoPass(this));
}
void ARMTargetMachine::anchor() { }
static std::string computeDataLayout(ARMSubtarget &ST) {
std::string Ret = "";
if (ST.isLittle())
// Little endian.
Ret += "e";
else
// Big endian.
Ret += "E";
Ret += DataLayout::getManglingComponent(ST.getTargetTriple());
// Pointers are 32 bits and aligned to 32 bits.
Ret += "-p:32:32";
// On thumb, i16,i18 and i1 have natural aligment requirements, but we try to
// align to 32.
if (ST.isThumb())
Ret += "-i1:8:32-i8:8:32-i16:16:32";
// ABIs other than APCS have 64 bit integers with natural alignment.
if (!ST.isAPCS_ABI())
Ret += "-i64:64";
// We have 64 bits floats. The APCS ABI requires them to be aligned to 32
// bits, others to 64 bits. We always try to align to 64 bits.
if (ST.isAPCS_ABI())
Ret += "-f64:32:64";
// We have 128 and 64 bit vectors. The APCS ABI aligns them to 32 bits, others
// to 64. We always ty to give them natural alignment.
if (ST.isAPCS_ABI())
Ret += "-v64:32:64-v128:32:128";
else
Ret += "-v128:64:128";
// On thumb and APCS, only try to align aggregates to 32 bits (the default is
// 64 bits).
if (ST.isThumb() || ST.isAPCS_ABI())
Ret += "-a:0:32";
// Integer registers are 32 bits.
Ret += "-n32";
// The stack is 128 bit aligned on NaCl, 64 bit aligned on AAPCS and 32 bit
// aligned everywhere else.
if (ST.isTargetNaCl())
Ret += "-S128";
else if (ST.isAAPCS_ABI())
Ret += "-S64";
else
Ret += "-S32";
return Ret;
}
ARMTargetMachine::ARMTargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL,
bool isLittle)
: ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, isLittle),
InstrInfo(Subtarget),
DL(computeDataLayout(Subtarget)),
TLInfo(*this),
TSInfo(*this),
FrameLowering(Subtarget) {
initAsmInfo();
if (!Subtarget.hasARMOps())
report_fatal_error("CPU: '" + Subtarget.getCPUString() + "' does not "
"support ARM mode execution!");
}
void ARMLETargetMachine::anchor() { }
ARMLETargetMachine::
ARMLETargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS, const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: ARMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, true) {}
void ARMBETargetMachine::anchor() { }
ARMBETargetMachine::
ARMBETargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS, const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: ARMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, false) {}
void ThumbTargetMachine::anchor() { }
ThumbTargetMachine::ThumbTargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL,
bool isLittle)
: ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, isLittle),
InstrInfo(Subtarget.hasThumb2()
? ((ARMBaseInstrInfo*)new Thumb2InstrInfo(Subtarget))
: ((ARMBaseInstrInfo*)new Thumb1InstrInfo(Subtarget))),
DL(computeDataLayout(Subtarget)),
TLInfo(*this),
TSInfo(*this),
FrameLowering(Subtarget.hasThumb2()
? new ARMFrameLowering(Subtarget)
: (ARMFrameLowering*)new Thumb1FrameLowering(Subtarget)) {
initAsmInfo();
}
void ThumbLETargetMachine::anchor() { }
ThumbLETargetMachine::
ThumbLETargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS, const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: ThumbTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, true) {}
void ThumbBETargetMachine::anchor() { }
ThumbBETargetMachine::
ThumbBETargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS, const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: ThumbTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, false) {}
namespace {
/// ARM Code Generator Pass Configuration Options.
class ARMPassConfig : public TargetPassConfig {
public:
ARMPassConfig(ARMBaseTargetMachine *TM, PassManagerBase &PM)
: TargetPassConfig(TM, PM) {}
ARMBaseTargetMachine &getARMTargetMachine() const {
return getTM<ARMBaseTargetMachine>();
}
const ARMSubtarget &getARMSubtarget() const {
return *getARMTargetMachine().getSubtargetImpl();
}
bool addPreISel() override;
bool addInstSelector() override;
bool addPreRegAlloc() override;
bool addPreSched2() override;
bool addPreEmitPass() override;
};
} // namespace
TargetPassConfig *ARMBaseTargetMachine::createPassConfig(PassManagerBase &PM) {
return new ARMPassConfig(this, PM);
}
bool ARMPassConfig::addPreISel() {
const ARMSubtarget *Subtarget = &getARMSubtarget();
if (Subtarget->hasAnyDataBarrier() && !Subtarget->isThumb1Only())
addPass(createARMAtomicExpandPass(TM));
if (TM->getOptLevel() != CodeGenOpt::None)
addPass(createGlobalMergePass(TM));
return false;
}
bool ARMPassConfig::addInstSelector() {
addPass(createARMISelDag(getARMTargetMachine(), getOptLevel()));
const ARMSubtarget *Subtarget = &getARMSubtarget();
if (Subtarget->isTargetELF() && !Subtarget->isThumb1Only() &&
TM->Options.EnableFastISel)
addPass(createARMGlobalBaseRegPass());
return false;
}
bool ARMPassConfig::addPreRegAlloc() {
// FIXME: temporarily disabling load / store optimization pass for Thumb1.
if (getOptLevel() != CodeGenOpt::None && !getARMSubtarget().isThumb1Only())
addPass(createARMLoadStoreOptimizationPass(true));
if (getOptLevel() != CodeGenOpt::None && getARMSubtarget().isCortexA9())
addPass(createMLxExpansionPass());
// Since the A15SDOptimizer pass can insert VDUP instructions, it can only be
// enabled when NEON is available.
if (getOptLevel() != CodeGenOpt::None && getARMSubtarget().isCortexA15() &&
getARMSubtarget().hasNEON() && !DisableA15SDOptimization) {
addPass(createA15SDOptimizerPass());
}
return true;
}
bool ARMPassConfig::addPreSched2() {
// FIXME: temporarily disabling load / store optimization pass for Thumb1.
if (getOptLevel() != CodeGenOpt::None) {
if (!getARMSubtarget().isThumb1Only()) {
addPass(createARMLoadStoreOptimizationPass());
printAndVerify("After ARM load / store optimizer");
}
if (getARMSubtarget().hasNEON())
addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass));
}
// Expand some pseudo instructions into multiple instructions to allow
// proper scheduling.
addPass(createARMExpandPseudoPass());
if (getOptLevel() != CodeGenOpt::None) {
if (!getARMSubtarget().isThumb1Only()) {
// in v8, IfConversion depends on Thumb instruction widths
if (getARMSubtarget().restrictIT() &&
!getARMSubtarget().prefers32BitThumb())
addPass(createThumb2SizeReductionPass());
addPass(&IfConverterID);
}
}
if (getARMSubtarget().isThumb2())
addPass(createThumb2ITBlockPass());
return true;
}
bool ARMPassConfig::addPreEmitPass() {
if (getARMSubtarget().isThumb2()) {
if (!getARMSubtarget().prefers32BitThumb())
addPass(createThumb2SizeReductionPass());
// Constant island pass work on unbundled instructions.
addPass(&UnpackMachineBundlesID);
}
addPass(createARMOptimizeBarriersPass());
addPass(createARMConstantIslandPass());
return true;
}
bool ARMBaseTargetMachine::addCodeEmitter(PassManagerBase &PM,
JITCodeEmitter &JCE) {
// Machine code emitter pass for ARM.
PM.add(createARMJITCodeEmitterPass(*this, JCE));
return false;
}