hanchenye-llvm-project/llvm/lib/Target/X86/X86TargetMachine.cpp

220 lines
7.7 KiB
C++
Raw Normal View History

//===-- X86TargetMachine.cpp - Define TargetMachine for the X86 -----------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file defines the X86 specific subclass of TargetMachine.
//
//===----------------------------------------------------------------------===//
#include "X86TargetMachine.h"
#include "X86.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/Passes.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"
using namespace llvm;
2011-02-17 20:23:50 +08:00
extern "C" void LLVMInitializeX86Target() {
// Register the target.
RegisterTargetMachine<X86_32TargetMachine> X(TheX86_32Target);
RegisterTargetMachine<X86_64TargetMachine> Y(TheX86_64Target);
}
void X86_32TargetMachine::anchor() { }
X86_32TargetMachine::X86_32TargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: X86TargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, false),
DL(getSubtargetImpl()->isTargetDarwin() ?
"e-p:32:32-f64:32:64-i64:32:64-f80:128:128-f128:128:128-"
"n8:16:32-S128" :
(getSubtargetImpl()->isTargetCygMing() ||
getSubtargetImpl()->isTargetWindows()) ?
"e-p:32:32-f64:64:64-i64:64:64-f80:32:32-f128:128:128-"
"n8:16:32-S32" :
"e-p:32:32-f64:32:64-i64:32:64-f80:32:32-f128:128:128-"
"n8:16:32-S128"),
InstrInfo(*this),
TLInfo(*this),
TSInfo(*this),
Switch TargetTransformInfo from an immutable analysis pass that requires a TargetMachine to construct (and thus isn't always available), to an analysis group that supports layered implementations much like AliasAnalysis does. This is a pretty massive change, with a few parts that I was unable to easily separate (sorry), so I'll walk through it. The first step of this conversion was to make TargetTransformInfo an analysis group, and to sink the nonce implementations in ScalarTargetTransformInfo and VectorTargetTranformInfo into a NoTargetTransformInfo pass. This allows other passes to add a hard requirement on TTI, and assume they will always get at least on implementation. The TargetTransformInfo analysis group leverages the delegation chaining trick that AliasAnalysis uses, where the base class for the analysis group delegates to the previous analysis *pass*, allowing all but tho NoFoo analysis passes to only implement the parts of the interfaces they support. It also introduces a new trick where each pass in the group retains a pointer to the top-most pass that has been initialized. This allows passes to implement one API in terms of another API and benefit when some other pass above them in the stack has more precise results for the second API. The second step of this conversion is to create a pass that implements the TargetTransformInfo analysis using the target-independent abstractions in the code generator. This replaces the ScalarTargetTransformImpl and VectorTargetTransformImpl classes in lib/Target with a single pass in lib/CodeGen called BasicTargetTransformInfo. This class actually provides most of the TTI functionality, basing it upon the TargetLowering abstraction and other information in the target independent code generator. The third step of the conversion adds support to all TargetMachines to register custom analysis passes. This allows building those passes with access to TargetLowering or other target-specific classes, and it also allows each target to customize the set of analysis passes desired in the pass manager. The baseline LLVMTargetMachine implements this interface to add the BasicTTI pass to the pass manager, and all of the tools that want to support target-aware TTI passes call this routine on whatever target machine they end up with to add the appropriate passes. The fourth step of the conversion created target-specific TTI analysis passes for the X86 and ARM backends. These passes contain the custom logic that was previously in their extensions of the ScalarTargetTransformInfo and VectorTargetTransformInfo interfaces. I separated them into their own file, as now all of the interface bits are private and they just expose a function to create the pass itself. Then I extended these target machines to set up a custom set of analysis passes, first adding BasicTTI as a fallback, and then adding their customized TTI implementations. The fourth step required logic that was shared between the target independent layer and the specific targets to move to a different interface, as they no longer derive from each other. As a consequence, a helper functions were added to TargetLowering representing the common logic needed both in the target implementation and the codegen implementation of the TTI pass. While technically this is the only change that could have been committed separately, it would have been a nightmare to extract. The final step of the conversion was just to delete all the old boilerplate. This got rid of the ScalarTargetTransformInfo and VectorTargetTransformInfo classes, all of the support in all of the targets for producing instances of them, and all of the support in the tools for manually constructing a pass based around them. Now that TTI is a relatively normal analysis group, two things become straightforward. First, we can sink it into lib/Analysis which is a more natural layer for it to live. Second, clients of this interface can depend on it *always* being available which will simplify their code and behavior. These (and other) simplifications will follow in subsequent commits, this one is clearly big enough. Finally, I'm very aware that much of the comments and documentation needs to be updated. As soon as I had this working, and plausibly well commented, I wanted to get it committed and in front of the build bots. I'll be doing a few passes over documentation later if it sticks. Commits to update DragonEgg and Clang will be made presently. llvm-svn: 171681
2013-01-07 09:37:14 +08:00
JITInfo(*this) {
}
void X86_64TargetMachine::anchor() { }
X86_64TargetMachine::X86_64TargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL)
: X86TargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, true),
DL("e-p:64:64-s:64-f64:64:64-i64:64:64-f80:128:128-f128:128:128-"
"n8:16:32:64-S128"),
InstrInfo(*this),
TLInfo(*this),
TSInfo(*this),
Switch TargetTransformInfo from an immutable analysis pass that requires a TargetMachine to construct (and thus isn't always available), to an analysis group that supports layered implementations much like AliasAnalysis does. This is a pretty massive change, with a few parts that I was unable to easily separate (sorry), so I'll walk through it. The first step of this conversion was to make TargetTransformInfo an analysis group, and to sink the nonce implementations in ScalarTargetTransformInfo and VectorTargetTranformInfo into a NoTargetTransformInfo pass. This allows other passes to add a hard requirement on TTI, and assume they will always get at least on implementation. The TargetTransformInfo analysis group leverages the delegation chaining trick that AliasAnalysis uses, where the base class for the analysis group delegates to the previous analysis *pass*, allowing all but tho NoFoo analysis passes to only implement the parts of the interfaces they support. It also introduces a new trick where each pass in the group retains a pointer to the top-most pass that has been initialized. This allows passes to implement one API in terms of another API and benefit when some other pass above them in the stack has more precise results for the second API. The second step of this conversion is to create a pass that implements the TargetTransformInfo analysis using the target-independent abstractions in the code generator. This replaces the ScalarTargetTransformImpl and VectorTargetTransformImpl classes in lib/Target with a single pass in lib/CodeGen called BasicTargetTransformInfo. This class actually provides most of the TTI functionality, basing it upon the TargetLowering abstraction and other information in the target independent code generator. The third step of the conversion adds support to all TargetMachines to register custom analysis passes. This allows building those passes with access to TargetLowering or other target-specific classes, and it also allows each target to customize the set of analysis passes desired in the pass manager. The baseline LLVMTargetMachine implements this interface to add the BasicTTI pass to the pass manager, and all of the tools that want to support target-aware TTI passes call this routine on whatever target machine they end up with to add the appropriate passes. The fourth step of the conversion created target-specific TTI analysis passes for the X86 and ARM backends. These passes contain the custom logic that was previously in their extensions of the ScalarTargetTransformInfo and VectorTargetTransformInfo interfaces. I separated them into their own file, as now all of the interface bits are private and they just expose a function to create the pass itself. Then I extended these target machines to set up a custom set of analysis passes, first adding BasicTTI as a fallback, and then adding their customized TTI implementations. The fourth step required logic that was shared between the target independent layer and the specific targets to move to a different interface, as they no longer derive from each other. As a consequence, a helper functions were added to TargetLowering representing the common logic needed both in the target implementation and the codegen implementation of the TTI pass. While technically this is the only change that could have been committed separately, it would have been a nightmare to extract. The final step of the conversion was just to delete all the old boilerplate. This got rid of the ScalarTargetTransformInfo and VectorTargetTransformInfo classes, all of the support in all of the targets for producing instances of them, and all of the support in the tools for manually constructing a pass based around them. Now that TTI is a relatively normal analysis group, two things become straightforward. First, we can sink it into lib/Analysis which is a more natural layer for it to live. Second, clients of this interface can depend on it *always* being available which will simplify their code and behavior. These (and other) simplifications will follow in subsequent commits, this one is clearly big enough. Finally, I'm very aware that much of the comments and documentation needs to be updated. As soon as I had this working, and plausibly well commented, I wanted to get it committed and in front of the build bots. I'll be doing a few passes over documentation later if it sticks. Commits to update DragonEgg and Clang will be made presently. llvm-svn: 171681
2013-01-07 09:37:14 +08:00
JITInfo(*this) {
}
/// X86TargetMachine ctor - Create an X86 target.
///
X86TargetMachine::X86TargetMachine(const Target &T, StringRef TT,
StringRef CPU, StringRef FS,
const TargetOptions &Options,
Reloc::Model RM, CodeModel::Model CM,
CodeGenOpt::Level OL,
bool is64Bit)
: LLVMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
Subtarget(TT, CPU, FS, Options.StackAlignmentOverride, is64Bit),
FrameLowering(*this, Subtarget),
InstrItins(Subtarget.getInstrItineraryData()){
// Determine the PICStyle based on the target selected.
if (getRelocationModel() == Reloc::Static) {
// Unless we're in PIC or DynamicNoPIC mode, set the PIC style to None.
Subtarget.setPICStyle(PICStyles::None);
} else if (Subtarget.is64Bit()) {
// PIC in 64 bit mode is always rip-rel.
Subtarget.setPICStyle(PICStyles::RIPRel);
} else if (Subtarget.isTargetCygMing()) {
Subtarget.setPICStyle(PICStyles::None);
} else if (Subtarget.isTargetDarwin()) {
if (getRelocationModel() == Reloc::PIC_)
Subtarget.setPICStyle(PICStyles::StubPIC);
else {
assert(getRelocationModel() == Reloc::DynamicNoPIC);
Subtarget.setPICStyle(PICStyles::StubDynamicNoPIC);
}
} else if (Subtarget.isTargetELF()) {
Subtarget.setPICStyle(PICStyles::GOT);
}
// default to hard float ABI
if (Options.FloatABIType == FloatABI::Default)
2012-02-03 13:12:30 +08:00
this->Options.FloatABIType = FloatABI::Hard;
2006-02-04 02:59:39 +08:00
}
//===----------------------------------------------------------------------===//
// Command line options for x86
//===----------------------------------------------------------------------===//
static cl::opt<bool>
UseVZeroUpper("x86-use-vzeroupper",
cl::desc("Minimize AVX to SSE transition penalty"),
cl::init(true));
// Temporary option to control early if-conversion for x86 while adding machine
// models.
static cl::opt<bool>
X86EarlyIfConv("x86-early-ifcvt",
cl::desc("Enable early if-conversion on X86"));
Switch TargetTransformInfo from an immutable analysis pass that requires a TargetMachine to construct (and thus isn't always available), to an analysis group that supports layered implementations much like AliasAnalysis does. This is a pretty massive change, with a few parts that I was unable to easily separate (sorry), so I'll walk through it. The first step of this conversion was to make TargetTransformInfo an analysis group, and to sink the nonce implementations in ScalarTargetTransformInfo and VectorTargetTranformInfo into a NoTargetTransformInfo pass. This allows other passes to add a hard requirement on TTI, and assume they will always get at least on implementation. The TargetTransformInfo analysis group leverages the delegation chaining trick that AliasAnalysis uses, where the base class for the analysis group delegates to the previous analysis *pass*, allowing all but tho NoFoo analysis passes to only implement the parts of the interfaces they support. It also introduces a new trick where each pass in the group retains a pointer to the top-most pass that has been initialized. This allows passes to implement one API in terms of another API and benefit when some other pass above them in the stack has more precise results for the second API. The second step of this conversion is to create a pass that implements the TargetTransformInfo analysis using the target-independent abstractions in the code generator. This replaces the ScalarTargetTransformImpl and VectorTargetTransformImpl classes in lib/Target with a single pass in lib/CodeGen called BasicTargetTransformInfo. This class actually provides most of the TTI functionality, basing it upon the TargetLowering abstraction and other information in the target independent code generator. The third step of the conversion adds support to all TargetMachines to register custom analysis passes. This allows building those passes with access to TargetLowering or other target-specific classes, and it also allows each target to customize the set of analysis passes desired in the pass manager. The baseline LLVMTargetMachine implements this interface to add the BasicTTI pass to the pass manager, and all of the tools that want to support target-aware TTI passes call this routine on whatever target machine they end up with to add the appropriate passes. The fourth step of the conversion created target-specific TTI analysis passes for the X86 and ARM backends. These passes contain the custom logic that was previously in their extensions of the ScalarTargetTransformInfo and VectorTargetTransformInfo interfaces. I separated them into their own file, as now all of the interface bits are private and they just expose a function to create the pass itself. Then I extended these target machines to set up a custom set of analysis passes, first adding BasicTTI as a fallback, and then adding their customized TTI implementations. The fourth step required logic that was shared between the target independent layer and the specific targets to move to a different interface, as they no longer derive from each other. As a consequence, a helper functions were added to TargetLowering representing the common logic needed both in the target implementation and the codegen implementation of the TTI pass. While technically this is the only change that could have been committed separately, it would have been a nightmare to extract. The final step of the conversion was just to delete all the old boilerplate. This got rid of the ScalarTargetTransformInfo and VectorTargetTransformInfo classes, all of the support in all of the targets for producing instances of them, and all of the support in the tools for manually constructing a pass based around them. Now that TTI is a relatively normal analysis group, two things become straightforward. First, we can sink it into lib/Analysis which is a more natural layer for it to live. Second, clients of this interface can depend on it *always* being available which will simplify their code and behavior. These (and other) simplifications will follow in subsequent commits, this one is clearly big enough. Finally, I'm very aware that much of the comments and documentation needs to be updated. As soon as I had this working, and plausibly well commented, I wanted to get it committed and in front of the build bots. I'll be doing a few passes over documentation later if it sticks. Commits to update DragonEgg and Clang will be made presently. llvm-svn: 171681
2013-01-07 09:37:14 +08:00
//===----------------------------------------------------------------------===//
// X86 Analysis Pass Setup
//===----------------------------------------------------------------------===//
void X86TargetMachine::addAnalysisPasses(PassManagerBase &PM) {
// Add first the target-independent BasicTTI pass, then our X86 pass. This
// allows the X86 pass to delegate to the target independent layer when
// appropriate.
PM.add(createBasicTargetTransformInfoPass(getTargetLowering()));
PM.add(createX86TargetTransformInfoPass(this));
}
//===----------------------------------------------------------------------===//
// Pass Pipeline Configuration
//===----------------------------------------------------------------------===//
2003-08-06 00:34:44 +08:00
namespace {
/// X86 Code Generator Pass Configuration Options.
class X86PassConfig : public TargetPassConfig {
public:
X86PassConfig(X86TargetMachine *TM, PassManagerBase &PM)
: TargetPassConfig(TM, PM) {}
X86TargetMachine &getX86TargetMachine() const {
return getTM<X86TargetMachine>();
}
const X86Subtarget &getX86Subtarget() const {
return *getX86TargetMachine().getSubtargetImpl();
}
virtual bool addInstSelector();
virtual bool addPreRegAlloc();
virtual bool addPostRegAlloc();
virtual bool addPreEmitPass();
};
} // namespace
TargetPassConfig *X86TargetMachine::createPassConfig(PassManagerBase &PM) {
X86PassConfig *PC = new X86PassConfig(this, PM);
if (X86EarlyIfConv && Subtarget.hasCMov())
PC->enablePass(&EarlyIfConverterID);
return PC;
}
bool X86PassConfig::addInstSelector() {
// Install an instruction selector.
addPass(createX86ISelDag(getX86TargetMachine(), getOptLevel()));
// For ELF, cleanup any local-dynamic TLS accesses.
if (getX86Subtarget().isTargetELF() && getOptLevel() != CodeGenOpt::None)
addPass(createCleanupLocalDynamicTLSPass());
// For 32-bit, prepend instructions to set the "global base reg" for PIC.
if (!getX86Subtarget().is64Bit())
addPass(createGlobalBaseRegPass());
return false;
}
bool X86PassConfig::addPreRegAlloc() {
return false; // -print-machineinstr shouldn't print after this.
}
bool X86PassConfig::addPostRegAlloc() {
addPass(createX86FloatingPointStackifierPass());
return true; // -print-machineinstr should print after this.
}
bool X86PassConfig::addPreEmitPass() {
bool ShouldPrint = false;
if (getOptLevel() != CodeGenOpt::None && getX86Subtarget().hasSSE2()) {
addPass(createExecutionDependencyFixPass(&X86::VR128RegClass));
ShouldPrint = true;
}
if (getX86Subtarget().hasAVX() && UseVZeroUpper) {
addPass(createX86IssueVZeroUpperPass());
ShouldPrint = true;
}
if (getOptLevel() != CodeGenOpt::None &&
getX86Subtarget().padShortFunctions()) {
addPass(createX86PadShortFunctions());
ShouldPrint = true;
}
return ShouldPrint;
}
bool X86TargetMachine::addCodeEmitter(PassManagerBase &PM,
JITCodeEmitter &JCE) {
PM.add(createX86JITCodeEmitterPass(*this, JCE));
return false;
}