llvm-6502/lib/IR/Module.cpp
Duncan P. N. Exon Smith dad20b2ae2 IR: Split Metadata from Value
Split `Metadata` away from the `Value` class hierarchy, as part of
PR21532.  Assembly and bitcode changes are in the wings, but this is the
bulk of the change for the IR C++ API.

I have a follow-up patch prepared for `clang`.  If this breaks other
sub-projects, I apologize in advance :(.  Help me compile it on Darwin
I'll try to fix it.  FWIW, the errors should be easy to fix, so it may
be simpler to just fix it yourself.

This breaks the build for all metadata-related code that's out-of-tree.
Rest assured the transition is mechanical and the compiler should catch
almost all of the problems.

Here's a quick guide for updating your code:

  - `Metadata` is the root of a class hierarchy with three main classes:
    `MDNode`, `MDString`, and `ValueAsMetadata`.  It is distinct from
    the `Value` class hierarchy.  It is typeless -- i.e., instances do
    *not* have a `Type`.

  - `MDNode`'s operands are all `Metadata *` (instead of `Value *`).

  - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be
    replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively.

    If you're referring solely to resolved `MDNode`s -- post graph
    construction -- just use `MDNode*`.

  - `MDNode` (and the rest of `Metadata`) have only limited support for
    `replaceAllUsesWith()`.

    As long as an `MDNode` is pointing at a forward declaration -- the
    result of `MDNode::getTemporary()` -- it maintains a side map of its
    uses and can RAUW itself.  Once the forward declarations are fully
    resolved RAUW support is dropped on the ground.  This means that
    uniquing collisions on changing operands cause nodes to become
    "distinct".  (This already happened fairly commonly, whenever an
    operand went to null.)

    If you're constructing complex (non self-reference) `MDNode` cycles,
    you need to call `MDNode::resolveCycles()` on each node (or on a
    top-level node that somehow references all of the nodes).  Also,
    don't do that.  Metadata cycles (and the RAUW machinery needed to
    construct them) are expensive.

  - An `MDNode` can only refer to a `Constant` through a bridge called
    `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`).

    As a side effect, accessing an operand of an `MDNode` that is known
    to be, e.g., `ConstantInt`, takes three steps: first, cast from
    `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`;
    third, cast down to `ConstantInt`.

    The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have
    metadata schema owners transition away from using `Constant`s when
    the type isn't important (and they don't care about referring to
    `GlobalValue`s).

    In the meantime, I've added transitional API to the `mdconst`
    namespace that matches semantics with the old code, in order to
    avoid adding the error-prone three-step equivalent to every call
    site.  If your old code was:

        MDNode *N = foo();
        bar(isa             <ConstantInt>(N->getOperand(0)));
        baz(cast            <ConstantInt>(N->getOperand(1)));
        bak(cast_or_null    <ConstantInt>(N->getOperand(2)));
        bat(dyn_cast        <ConstantInt>(N->getOperand(3)));
        bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4)));

    you can trivially match its semantics with:

        MDNode *N = foo();
        bar(mdconst::hasa               <ConstantInt>(N->getOperand(0)));
        baz(mdconst::extract            <ConstantInt>(N->getOperand(1)));
        bak(mdconst::extract_or_null    <ConstantInt>(N->getOperand(2)));
        bat(mdconst::dyn_extract        <ConstantInt>(N->getOperand(3)));
        bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4)));

    and when you transition your metadata schema to `MDInt`:

        MDNode *N = foo();
        bar(isa             <MDInt>(N->getOperand(0)));
        baz(cast            <MDInt>(N->getOperand(1)));
        bak(cast_or_null    <MDInt>(N->getOperand(2)));
        bat(dyn_cast        <MDInt>(N->getOperand(3)));
        bay(dyn_cast_or_null<MDInt>(N->getOperand(4)));

  - A `CallInst` -- specifically, intrinsic instructions -- can refer to
    metadata through a bridge called `MetadataAsValue`.  This is a
    subclass of `Value` where `getType()->isMetadataTy()`.

    `MetadataAsValue` is the *only* class that can legally refer to a
    `LocalAsMetadata`, which is a bridged form of non-`Constant` values
    like `Argument` and `Instruction`.  It can also refer to any other
    `Metadata` subclass.

(I'll break all your testcases in a follow-up commit, when I propagate
this change to assembly.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223802 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-09 18:38:53 +00:00

491 lines
17 KiB
C++

//===-- Module.cpp - Implement the Module class ---------------------------===//
//
// 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 Module class for the IR library.
//
//===----------------------------------------------------------------------===//
#include "llvm/IR/Module.h"
#include "SymbolTableListTraitsImpl.h"
#include "llvm/ADT/DenseSet.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/GVMaterializer.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/LeakDetector.h"
#include "llvm/IR/TypeFinder.h"
#include "llvm/Support/Dwarf.h"
#include "llvm/Support/Path.h"
#include "llvm/Support/RandomNumberGenerator.h"
#include <algorithm>
#include <cstdarg>
#include <cstdlib>
using namespace llvm;
//===----------------------------------------------------------------------===//
// Methods to implement the globals and functions lists.
//
// Explicit instantiations of SymbolTableListTraits since some of the methods
// are not in the public header file.
template class llvm::SymbolTableListTraits<Function, Module>;
template class llvm::SymbolTableListTraits<GlobalVariable, Module>;
template class llvm::SymbolTableListTraits<GlobalAlias, Module>;
//===----------------------------------------------------------------------===//
// Primitive Module methods.
//
Module::Module(StringRef MID, LLVMContext &C)
: Context(C), Materializer(), ModuleID(MID), RNG(nullptr), DL("") {
ValSymTab = new ValueSymbolTable();
NamedMDSymTab = new StringMap<NamedMDNode *>();
Context.addModule(this);
}
Module::~Module() {
Context.removeModule(this);
dropAllReferences();
GlobalList.clear();
FunctionList.clear();
AliasList.clear();
NamedMDList.clear();
delete ValSymTab;
delete static_cast<StringMap<NamedMDNode *> *>(NamedMDSymTab);
delete RNG;
}
/// getNamedValue - Return the first global value in the module with
/// the specified name, of arbitrary type. This method returns null
/// if a global with the specified name is not found.
GlobalValue *Module::getNamedValue(StringRef Name) const {
return cast_or_null<GlobalValue>(getValueSymbolTable().lookup(Name));
}
/// getMDKindID - Return a unique non-zero ID for the specified metadata kind.
/// This ID is uniqued across modules in the current LLVMContext.
unsigned Module::getMDKindID(StringRef Name) const {
return Context.getMDKindID(Name);
}
/// getMDKindNames - Populate client supplied SmallVector with the name for
/// custom metadata IDs registered in this LLVMContext. ID #0 is not used,
/// so it is filled in as an empty string.
void Module::getMDKindNames(SmallVectorImpl<StringRef> &Result) const {
return Context.getMDKindNames(Result);
}
//===----------------------------------------------------------------------===//
// Methods for easy access to the functions in the module.
//
// getOrInsertFunction - Look up the specified function in the module symbol
// table. If it does not exist, add a prototype for the function and return
// it. This is nice because it allows most passes to get away with not handling
// the symbol table directly for this common task.
//
Constant *Module::getOrInsertFunction(StringRef Name,
FunctionType *Ty,
AttributeSet AttributeList) {
// See if we have a definition for the specified function already.
GlobalValue *F = getNamedValue(Name);
if (!F) {
// Nope, add it
Function *New = Function::Create(Ty, GlobalVariable::ExternalLinkage, Name);
if (!New->isIntrinsic()) // Intrinsics get attrs set on construction
New->setAttributes(AttributeList);
FunctionList.push_back(New);
return New; // Return the new prototype.
}
// If the function exists but has the wrong type, return a bitcast to the
// right type.
if (F->getType() != PointerType::getUnqual(Ty))
return ConstantExpr::getBitCast(F, PointerType::getUnqual(Ty));
// Otherwise, we just found the existing function or a prototype.
return F;
}
Constant *Module::getOrInsertFunction(StringRef Name,
FunctionType *Ty) {
return getOrInsertFunction(Name, Ty, AttributeSet());
}
// getOrInsertFunction - Look up the specified function in the module symbol
// table. If it does not exist, add a prototype for the function and return it.
// This version of the method takes a null terminated list of function
// arguments, which makes it easier for clients to use.
//
Constant *Module::getOrInsertFunction(StringRef Name,
AttributeSet AttributeList,
Type *RetTy, ...) {
va_list Args;
va_start(Args, RetTy);
// Build the list of argument types...
std::vector<Type*> ArgTys;
while (Type *ArgTy = va_arg(Args, Type*))
ArgTys.push_back(ArgTy);
va_end(Args);
// Build the function type and chain to the other getOrInsertFunction...
return getOrInsertFunction(Name,
FunctionType::get(RetTy, ArgTys, false),
AttributeList);
}
Constant *Module::getOrInsertFunction(StringRef Name,
Type *RetTy, ...) {
va_list Args;
va_start(Args, RetTy);
// Build the list of argument types...
std::vector<Type*> ArgTys;
while (Type *ArgTy = va_arg(Args, Type*))
ArgTys.push_back(ArgTy);
va_end(Args);
// Build the function type and chain to the other getOrInsertFunction...
return getOrInsertFunction(Name,
FunctionType::get(RetTy, ArgTys, false),
AttributeSet());
}
// getFunction - Look up the specified function in the module symbol table.
// If it does not exist, return null.
//
Function *Module::getFunction(StringRef Name) const {
return dyn_cast_or_null<Function>(getNamedValue(Name));
}
//===----------------------------------------------------------------------===//
// Methods for easy access to the global variables in the module.
//
/// getGlobalVariable - Look up the specified global variable in the module
/// symbol table. If it does not exist, return null. The type argument
/// should be the underlying type of the global, i.e., it should not have
/// the top-level PointerType, which represents the address of the global.
/// If AllowLocal is set to true, this function will return types that
/// have an local. By default, these types are not returned.
///
GlobalVariable *Module::getGlobalVariable(StringRef Name, bool AllowLocal) {
if (GlobalVariable *Result =
dyn_cast_or_null<GlobalVariable>(getNamedValue(Name)))
if (AllowLocal || !Result->hasLocalLinkage())
return Result;
return nullptr;
}
/// getOrInsertGlobal - Look up the specified global in the module symbol table.
/// 1. If it does not exist, add a declaration of the global and return it.
/// 2. Else, the global exists but has the wrong type: return the function
/// with a constantexpr cast to the right type.
/// 3. Finally, if the existing global is the correct declaration, return the
/// existing global.
Constant *Module::getOrInsertGlobal(StringRef Name, Type *Ty) {
// See if we have a definition for the specified global already.
GlobalVariable *GV = dyn_cast_or_null<GlobalVariable>(getNamedValue(Name));
if (!GV) {
// Nope, add it
GlobalVariable *New =
new GlobalVariable(*this, Ty, false, GlobalVariable::ExternalLinkage,
nullptr, Name);
return New; // Return the new declaration.
}
// If the variable exists but has the wrong type, return a bitcast to the
// right type.
Type *GVTy = GV->getType();
PointerType *PTy = PointerType::get(Ty, GVTy->getPointerAddressSpace());
if (GVTy != PTy)
return ConstantExpr::getBitCast(GV, PTy);
// Otherwise, we just found the existing function or a prototype.
return GV;
}
//===----------------------------------------------------------------------===//
// Methods for easy access to the global variables in the module.
//
// getNamedAlias - Look up the specified global in the module symbol table.
// If it does not exist, return null.
//
GlobalAlias *Module::getNamedAlias(StringRef Name) const {
return dyn_cast_or_null<GlobalAlias>(getNamedValue(Name));
}
/// getNamedMetadata - Return the first NamedMDNode in the module with the
/// specified name. This method returns null if a NamedMDNode with the
/// specified name is not found.
NamedMDNode *Module::getNamedMetadata(const Twine &Name) const {
SmallString<256> NameData;
StringRef NameRef = Name.toStringRef(NameData);
return static_cast<StringMap<NamedMDNode*> *>(NamedMDSymTab)->lookup(NameRef);
}
/// getOrInsertNamedMetadata - Return the first named MDNode in the module
/// with the specified name. This method returns a new NamedMDNode if a
/// NamedMDNode with the specified name is not found.
NamedMDNode *Module::getOrInsertNamedMetadata(StringRef Name) {
NamedMDNode *&NMD =
(*static_cast<StringMap<NamedMDNode *> *>(NamedMDSymTab))[Name];
if (!NMD) {
NMD = new NamedMDNode(Name);
NMD->setParent(this);
NamedMDList.push_back(NMD);
}
return NMD;
}
/// eraseNamedMetadata - Remove the given NamedMDNode from this module and
/// delete it.
void Module::eraseNamedMetadata(NamedMDNode *NMD) {
static_cast<StringMap<NamedMDNode *> *>(NamedMDSymTab)->erase(NMD->getName());
NamedMDList.erase(NMD);
}
bool Module::isValidModFlagBehavior(Metadata *MD, ModFlagBehavior &MFB) {
if (ConstantInt *Behavior = mdconst::dyn_extract<ConstantInt>(MD)) {
uint64_t Val = Behavior->getLimitedValue();
if (Val >= ModFlagBehaviorFirstVal && Val <= ModFlagBehaviorLastVal) {
MFB = static_cast<ModFlagBehavior>(Val);
return true;
}
}
return false;
}
/// getModuleFlagsMetadata - Returns the module flags in the provided vector.
void Module::
getModuleFlagsMetadata(SmallVectorImpl<ModuleFlagEntry> &Flags) const {
const NamedMDNode *ModFlags = getModuleFlagsMetadata();
if (!ModFlags) return;
for (const MDNode *Flag : ModFlags->operands()) {
ModFlagBehavior MFB;
if (Flag->getNumOperands() >= 3 &&
isValidModFlagBehavior(Flag->getOperand(0), MFB) &&
isa<MDString>(Flag->getOperand(1))) {
// Check the operands of the MDNode before accessing the operands.
// The verifier will actually catch these failures.
MDString *Key = cast<MDString>(Flag->getOperand(1));
Metadata *Val = Flag->getOperand(2);
Flags.push_back(ModuleFlagEntry(MFB, Key, Val));
}
}
}
/// Return the corresponding value if Key appears in module flags, otherwise
/// return null.
Metadata *Module::getModuleFlag(StringRef Key) const {
SmallVector<Module::ModuleFlagEntry, 8> ModuleFlags;
getModuleFlagsMetadata(ModuleFlags);
for (const ModuleFlagEntry &MFE : ModuleFlags) {
if (Key == MFE.Key->getString())
return MFE.Val;
}
return nullptr;
}
/// getModuleFlagsMetadata - Returns the NamedMDNode in the module that
/// represents module-level flags. This method returns null if there are no
/// module-level flags.
NamedMDNode *Module::getModuleFlagsMetadata() const {
return getNamedMetadata("llvm.module.flags");
}
/// getOrInsertModuleFlagsMetadata - Returns the NamedMDNode in the module that
/// represents module-level flags. If module-level flags aren't found, it
/// creates the named metadata that contains them.
NamedMDNode *Module::getOrInsertModuleFlagsMetadata() {
return getOrInsertNamedMetadata("llvm.module.flags");
}
/// addModuleFlag - Add a module-level flag to the module-level flags
/// metadata. It will create the module-level flags named metadata if it doesn't
/// already exist.
void Module::addModuleFlag(ModFlagBehavior Behavior, StringRef Key,
Metadata *Val) {
Type *Int32Ty = Type::getInt32Ty(Context);
Metadata *Ops[3] = {
ConstantAsMetadata::get(ConstantInt::get(Int32Ty, Behavior)),
MDString::get(Context, Key), Val};
getOrInsertModuleFlagsMetadata()->addOperand(MDNode::get(Context, Ops));
}
void Module::addModuleFlag(ModFlagBehavior Behavior, StringRef Key,
Constant *Val) {
addModuleFlag(Behavior, Key, ConstantAsMetadata::get(Val));
}
void Module::addModuleFlag(ModFlagBehavior Behavior, StringRef Key,
uint32_t Val) {
Type *Int32Ty = Type::getInt32Ty(Context);
addModuleFlag(Behavior, Key, ConstantInt::get(Int32Ty, Val));
}
void Module::addModuleFlag(MDNode *Node) {
assert(Node->getNumOperands() == 3 &&
"Invalid number of operands for module flag!");
assert(mdconst::hasa<ConstantInt>(Node->getOperand(0)) &&
isa<MDString>(Node->getOperand(1)) &&
"Invalid operand types for module flag!");
getOrInsertModuleFlagsMetadata()->addOperand(Node);
}
void Module::setDataLayout(StringRef Desc) {
DL.reset(Desc);
if (Desc.empty()) {
DataLayoutStr = "";
} else {
DataLayoutStr = DL.getStringRepresentation();
// DataLayoutStr is now equivalent to Desc, but since the representation
// is not unique, they may not be identical.
}
}
void Module::setDataLayout(const DataLayout *Other) {
if (!Other) {
DataLayoutStr = "";
DL.reset("");
} else {
DL = *Other;
DataLayoutStr = DL.getStringRepresentation();
}
}
const DataLayout *Module::getDataLayout() const {
if (DataLayoutStr.empty())
return nullptr;
return &DL;
}
// We want reproducible builds, but ModuleID may be a full path so we just use
// the filename to salt the RNG (although it is not guaranteed to be unique).
RandomNumberGenerator &Module::getRNG() const {
if (RNG == nullptr) {
StringRef Salt = sys::path::filename(ModuleID);
RNG = new RandomNumberGenerator(Salt);
}
return *RNG;
}
//===----------------------------------------------------------------------===//
// Methods to control the materialization of GlobalValues in the Module.
//
void Module::setMaterializer(GVMaterializer *GVM) {
assert(!Materializer &&
"Module already has a GVMaterializer. Call MaterializeAllPermanently"
" to clear it out before setting another one.");
Materializer.reset(GVM);
}
bool Module::isDematerializable(const GlobalValue *GV) const {
if (Materializer)
return Materializer->isDematerializable(GV);
return false;
}
std::error_code Module::materialize(GlobalValue *GV) {
if (!Materializer)
return std::error_code();
return Materializer->materialize(GV);
}
void Module::Dematerialize(GlobalValue *GV) {
if (Materializer)
return Materializer->Dematerialize(GV);
}
std::error_code Module::materializeAll() {
if (!Materializer)
return std::error_code();
return Materializer->MaterializeModule(this);
}
std::error_code Module::materializeAllPermanently() {
if (std::error_code EC = materializeAll())
return EC;
Materializer.reset();
return std::error_code();
}
//===----------------------------------------------------------------------===//
// Other module related stuff.
//
std::vector<StructType *> Module::getIdentifiedStructTypes() const {
// If we have a materializer, it is possible that some unread function
// uses a type that is currently not visible to a TypeFinder, so ask
// the materializer which types it created.
if (Materializer)
return Materializer->getIdentifiedStructTypes();
std::vector<StructType *> Ret;
TypeFinder SrcStructTypes;
SrcStructTypes.run(*this, true);
Ret.assign(SrcStructTypes.begin(), SrcStructTypes.end());
return Ret;
}
// dropAllReferences() - This function causes all the subelements to "let go"
// of all references that they are maintaining. This allows one to 'delete' a
// whole module at a time, even though there may be circular references... first
// all references are dropped, and all use counts go to zero. Then everything
// is deleted for real. Note that no operations are valid on an object that
// has "dropped all references", except operator delete.
//
void Module::dropAllReferences() {
for (Function &F : *this)
F.dropAllReferences();
for (GlobalVariable &GV : globals())
GV.dropAllReferences();
for (GlobalAlias &GA : aliases())
GA.dropAllReferences();
}
unsigned Module::getDwarfVersion() const {
auto *Val = cast_or_null<ConstantAsMetadata>(getModuleFlag("Dwarf Version"));
if (!Val)
return dwarf::DWARF_VERSION;
return cast<ConstantInt>(Val->getValue())->getZExtValue();
}
Comdat *Module::getOrInsertComdat(StringRef Name) {
auto &Entry = *ComdatSymTab.insert(std::make_pair(Name, Comdat())).first;
Entry.second.Name = &Entry;
return &Entry.second;
}
PICLevel::Level Module::getPICLevel() const {
auto *Val = cast_or_null<ConstantAsMetadata>(getModuleFlag("PIC Level"));
if (Val == NULL)
return PICLevel::Default;
return static_cast<PICLevel::Level>(
cast<ConstantInt>(Val->getValue())->getZExtValue());
}
void Module::setPICLevel(PICLevel::Level PL) {
addModuleFlag(ModFlagBehavior::Error, "PIC Level", PL);
}