//===----------------------------------------------------------------------===//
#include "llvm/ADT/StringMap.h"
-#include "llvm/ADT/OwningPtr.h"
-#include "llvm/Object/MachOObject.h"
+#include "llvm/DebugInfo/DIContext.h"
+#include "llvm/ExecutionEngine/ObjectBuffer.h"
+#include "llvm/ExecutionEngine/ObjectImage.h"
+#include "llvm/ExecutionEngine/RuntimeDyld.h"
+#include "llvm/Object/MachO.h"
#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/DynamicLibrary.h"
#include "llvm/Support/ManagedStatic.h"
#include "llvm/Support/Memory.h"
#include "llvm/Support/MemoryBuffer.h"
+#include "llvm/Support/PrettyStackTrace.h"
+#include "llvm/Support/Signals.h"
#include "llvm/Support/raw_ostream.h"
-#include "llvm/Support/system_error.h"
+#include <system_error>
using namespace llvm;
using namespace llvm::object;
+using std::error_code;
-static cl::opt<std::string>
-InputFile(cl::Positional, cl::desc("<input file>"), cl::init("-"));
+static cl::list<std::string>
+InputFileList(cl::Positional, cl::ZeroOrMore,
+ cl::desc("<input file>"));
enum ActionType {
- AC_Execute
+ AC_Execute,
+ AC_PrintLineInfo
};
static cl::opt<ActionType>
cl::init(AC_Execute),
cl::values(clEnumValN(AC_Execute, "execute",
"Load, link, and execute the inputs."),
+ clEnumValN(AC_PrintLineInfo, "printline",
+ "Load, link, and print line information for each function."),
clEnumValEnd));
-/* *** */
-
-static const char *ProgramName;
-
-static void Message(const char *Type, const Twine &Msg) {
- errs() << ProgramName << ": " << Type << ": " << Msg << "\n";
-}
+static cl::opt<std::string>
+EntryPoint("entry",
+ cl::desc("Function to call as entry point."),
+ cl::init("_main"));
-static int Error(const Twine &Msg) {
- Message("error", Msg);
- return 1;
-}
+static cl::list<std::string>
+Dylibs("dylib",
+ cl::desc("Add library."),
+ cl::ZeroOrMore);
/* *** */
-static bool
-loadSegment32(const MachOObject *Obj,
- sys::MemoryBlock &Data,
- const MachOObject::LoadCommandInfo *SegmentLCI,
- const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC,
- StringMap<void*> &SymbolTable) {
- InMemoryStruct<macho::SegmentLoadCommand> Segment32LC;
- Obj->ReadSegmentLoadCommand(*SegmentLCI, Segment32LC);
- if (!Segment32LC)
- return Error("unable to load segment load command");
-
- // Map the segment into memory.
- std::string ErrorStr;
- Data = sys::Memory::AllocateRWX(Segment32LC->VMSize, 0, &ErrorStr);
- if (!Data.base())
- return Error("unable to allocate memory block: '" + ErrorStr + "'");
- memcpy(Data.base(), Obj->getData(Segment32LC->FileOffset,
- Segment32LC->FileSize).data(),
- Segment32LC->FileSize);
- memset((char*)Data.base() + Segment32LC->FileSize, 0,
- Segment32LC->VMSize - Segment32LC->FileSize);
-
- // Bind the section indices to address.
- void **SectionBases = new void*[Segment32LC->NumSections];
- for (unsigned i = 0; i != Segment32LC->NumSections; ++i) {
- InMemoryStruct<macho::Section> Sect;
- Obj->ReadSection(*SegmentLCI, i, Sect);
- if (!Sect)
- return Error("unable to load section: '" + Twine(i) + "'");
-
- // FIXME: We don't support relocations yet.
- if (Sect->NumRelocationTableEntries != 0)
- return Error("not yet implemented: relocations!");
-
- // FIXME: Improve check.
- if (Sect->Flags != 0x80000400)
- return Error("unsupported section type!");
-
- SectionBases[i] = (char*) Data.base() + Sect->Address;
- }
- // Bind all the symbols to address.
- for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
- InMemoryStruct<macho::SymbolTableEntry> STE;
- Obj->ReadSymbolTableEntry(SymtabLC->SymbolTableOffset, i, STE);
- if (!STE)
- return Error("unable to read symbol: '" + Twine(i) + "'");
- if (STE->SectionIndex == 0)
- return Error("unexpected undefined symbol!");
-
- unsigned Index = STE->SectionIndex - 1;
- if (Index >= Segment32LC->NumSections)
- return Error("invalid section index for symbol: '" + Twine() + "'");
+// A trivial memory manager that doesn't do anything fancy, just uses the
+// support library allocation routines directly.
+class TrivialMemoryManager : public RTDyldMemoryManager {
+public:
+ SmallVector<sys::MemoryBlock, 16> FunctionMemory;
+ SmallVector<sys::MemoryBlock, 16> DataMemory;
+
+ uint8_t *allocateCodeSection(uintptr_t Size, unsigned Alignment,
+ unsigned SectionID,
+ StringRef SectionName) override;
+ uint8_t *allocateDataSection(uintptr_t Size, unsigned Alignment,
+ unsigned SectionID, StringRef SectionName,
+ bool IsReadOnly) override;
+
+ void *getPointerToNamedFunction(const std::string &Name,
+ bool AbortOnFailure = true) override {
+ return nullptr;
+ }
- // Get the symbol name.
- StringRef Name = Obj->getStringAtIndex(STE->StringIndex);
+ bool finalizeMemory(std::string *ErrMsg) override { return false; }
- // Get the section base address.
- void *SectionBase = SectionBases[Index];
+ // Invalidate instruction cache for sections with execute permissions.
+ // Some platforms with separate data cache and instruction cache require
+ // explicit cache flush, otherwise JIT code manipulations (like resolved
+ // relocations) will get to the data cache but not to the instruction cache.
+ virtual void invalidateInstructionCache();
+};
- // Get the symbol address.
- void *Address = (char*) SectionBase + STE->Value;
+uint8_t *TrivialMemoryManager::allocateCodeSection(uintptr_t Size,
+ unsigned Alignment,
+ unsigned SectionID,
+ StringRef SectionName) {
+ sys::MemoryBlock MB = sys::Memory::AllocateRWX(Size, nullptr, nullptr);
+ FunctionMemory.push_back(MB);
+ return (uint8_t*)MB.base();
+}
- // FIXME: Check the symbol type and flags.
- if (STE->Type != 0xF)
- return Error("unexpected symbol type!");
- if (STE->Flags != 0x0)
- return Error("unexpected symbol type!");
+uint8_t *TrivialMemoryManager::allocateDataSection(uintptr_t Size,
+ unsigned Alignment,
+ unsigned SectionID,
+ StringRef SectionName,
+ bool IsReadOnly) {
+ sys::MemoryBlock MB = sys::Memory::AllocateRWX(Size, nullptr, nullptr);
+ DataMemory.push_back(MB);
+ return (uint8_t*)MB.base();
+}
- SymbolTable[Name] = Address;
- }
+void TrivialMemoryManager::invalidateInstructionCache() {
+ for (int i = 0, e = FunctionMemory.size(); i != e; ++i)
+ sys::Memory::InvalidateInstructionCache(FunctionMemory[i].base(),
+ FunctionMemory[i].size());
- delete SectionBases;
- return false;
+ for (int i = 0, e = DataMemory.size(); i != e; ++i)
+ sys::Memory::InvalidateInstructionCache(DataMemory[i].base(),
+ DataMemory[i].size());
}
-static bool
-loadSegment64(const MachOObject *Obj,
- sys::MemoryBlock &Data,
- const MachOObject::LoadCommandInfo *SegmentLCI,
- const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC,
- StringMap<void*> &SymbolTable) {
- InMemoryStruct<macho::Segment64LoadCommand> Segment64LC;
- Obj->ReadSegment64LoadCommand(*SegmentLCI, Segment64LC);
- if (!Segment64LC)
- return Error("unable to load segment load command");
-
- // Map the segment into memory.
- std::string ErrorStr;
- Data = sys::Memory::AllocateRWX(Segment64LC->VMSize, 0, &ErrorStr);
- if (!Data.base())
- return Error("unable to allocate memory block: '" + ErrorStr + "'");
- memcpy(Data.base(), Obj->getData(Segment64LC->FileOffset,
- Segment64LC->FileSize).data(),
- Segment64LC->FileSize);
- memset((char*)Data.base() + Segment64LC->FileSize, 0,
- Segment64LC->VMSize - Segment64LC->FileSize);
-
- // Bind the section indices to address.
- void **SectionBases = new void*[Segment64LC->NumSections];
- for (unsigned i = 0; i != Segment64LC->NumSections; ++i) {
- InMemoryStruct<macho::Section64> Sect;
- Obj->ReadSection64(*SegmentLCI, i, Sect);
- if (!Sect)
- return Error("unable to load section: '" + Twine(i) + "'");
-
- // FIXME: We don't support relocations yet.
- if (Sect->NumRelocationTableEntries != 0)
- return Error("not yet implemented: relocations!");
-
- // FIXME: Improve check.
- if (Sect->Flags != 0x80000400)
- return Error("unsupported section type!");
-
- SectionBases[i] = (char*) Data.base() + Sect->Address;
- }
+static const char *ProgramName;
- // Bind all the symbols to address.
- for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
- InMemoryStruct<macho::Symbol64TableEntry> STE;
- Obj->ReadSymbol64TableEntry(SymtabLC->SymbolTableOffset, i, STE);
- if (!STE)
- return Error("unable to read symbol: '" + Twine(i) + "'");
- if (STE->SectionIndex == 0)
- return Error("unexpected undefined symbol!");
+static void Message(const char *Type, const Twine &Msg) {
+ errs() << ProgramName << ": " << Type << ": " << Msg << "\n";
+}
- unsigned Index = STE->SectionIndex - 1;
- if (Index >= Segment64LC->NumSections)
- return Error("invalid section index for symbol: '" + Twine() + "'");
+static int Error(const Twine &Msg) {
+ Message("error", Msg);
+ return 1;
+}
- // Get the symbol name.
- StringRef Name = Obj->getStringAtIndex(STE->StringIndex);
+static void loadDylibs() {
+ for (const std::string &Dylib : Dylibs) {
+ if (sys::fs::is_regular_file(Dylib)) {
+ std::string ErrMsg;
+ if (sys::DynamicLibrary::LoadLibraryPermanently(Dylib.c_str(), &ErrMsg))
+ llvm::errs() << "Error loading '" << Dylib << "': "
+ << ErrMsg << "\n";
+ } else
+ llvm::errs() << "Dylib not found: '" << Dylib << "'.\n";
+ }
+}
- // Get the section base address.
- void *SectionBase = SectionBases[Index];
- // Get the symbol address.
- void *Address = (char*) SectionBase + STE->Value;
+/* *** */
- // FIXME: Check the symbol type and flags.
- if (STE->Type != 0xF)
- return Error("unexpected symbol type!");
- if (STE->Flags != 0x0)
- return Error("unexpected symbol type!");
+static int printLineInfoForInput() {
+ // Load any dylibs requested on the command line.
+ loadDylibs();
+
+ // If we don't have any input files, read from stdin.
+ if (!InputFileList.size())
+ InputFileList.push_back("-");
+ for(unsigned i = 0, e = InputFileList.size(); i != e; ++i) {
+ // Instantiate a dynamic linker.
+ TrivialMemoryManager MemMgr;
+ RuntimeDyld Dyld(&MemMgr);
+
+ // Load the input memory buffer.
+ std::unique_ptr<MemoryBuffer> InputBuffer;
+ std::unique_ptr<ObjectImage> LoadedObject;
+ if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFileList[i],
+ InputBuffer))
+ return Error("unable to read input: '" + ec.message() + "'");
+
+ // Load the object file
+ LoadedObject.reset(Dyld.loadObject(new ObjectBuffer(InputBuffer.release())));
+ if (!LoadedObject) {
+ return Error(Dyld.getErrorString());
+ }
- SymbolTable[Name] = Address;
+ // Resolve all the relocations we can.
+ Dyld.resolveRelocations();
+
+ std::unique_ptr<DIContext> Context(
+ DIContext::getDWARFContext(LoadedObject->getObjectFile()));
+
+ // Use symbol info to iterate functions in the object.
+ for (object::symbol_iterator I = LoadedObject->begin_symbols(),
+ E = LoadedObject->end_symbols();
+ I != E; ++I) {
+ object::SymbolRef::Type SymType;
+ if (I->getType(SymType)) continue;
+ if (SymType == object::SymbolRef::ST_Function) {
+ StringRef Name;
+ uint64_t Addr;
+ uint64_t Size;
+ if (I->getName(Name)) continue;
+ if (I->getAddress(Addr)) continue;
+ if (I->getSize(Size)) continue;
+
+ outs() << "Function: " << Name << ", Size = " << Size << "\n";
+
+ DILineInfoTable Lines = Context->getLineInfoForAddressRange(Addr, Size);
+ DILineInfoTable::iterator Begin = Lines.begin();
+ DILineInfoTable::iterator End = Lines.end();
+ for (DILineInfoTable::iterator It = Begin; It != End; ++It) {
+ outs() << " Line info @ " << It->first - Addr << ": "
+ << It->second.FileName << ", line:" << It->second.Line << "\n";
+ }
+ }
+ }
}
- delete SectionBases;
- return false;
+ return 0;
}
static int executeInput() {
- // Load the input memory buffer.
- OwningPtr<MemoryBuffer> InputBuffer;
- if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFile, InputBuffer))
- return Error("unable to read input: '" + ec.message() + "'");
-
- // Load the Mach-O wrapper object.
- std::string ErrorStr;
- OwningPtr<MachOObject> Obj(
- MachOObject::LoadFromBuffer(InputBuffer.take(), &ErrorStr));
- if (!Obj)
- return Error("unable to load object: '" + ErrorStr + "'");
-
- // Validate that the load commands match what we expect.
- const MachOObject::LoadCommandInfo *SegmentLCI = 0, *SymtabLCI = 0,
- *DysymtabLCI = 0;
- for (unsigned i = 0; i != Obj->getHeader().NumLoadCommands; ++i) {
- const MachOObject::LoadCommandInfo &LCI = Obj->getLoadCommandInfo(i);
- switch (LCI.Command.Type) {
- case macho::LCT_Segment:
- case macho::LCT_Segment64:
- if (SegmentLCI)
- return Error("unexpected input object (multiple segments)");
- SegmentLCI = &LCI;
- break;
- case macho::LCT_Symtab:
- if (SymtabLCI)
- return Error("unexpected input object (multiple symbol tables)");
- SymtabLCI = &LCI;
- break;
- case macho::LCT_Dysymtab:
- if (DysymtabLCI)
- return Error("unexpected input object (multiple symbol tables)");
- DysymtabLCI = &LCI;
- break;
- default:
- return Error("unexpected input object (unexpected load command");
+ // Load any dylibs requested on the command line.
+ loadDylibs();
+
+ // Instantiate a dynamic linker.
+ TrivialMemoryManager MemMgr;
+ RuntimeDyld Dyld(&MemMgr);
+
+ // If we don't have any input files, read from stdin.
+ if (!InputFileList.size())
+ InputFileList.push_back("-");
+ for(unsigned i = 0, e = InputFileList.size(); i != e; ++i) {
+ // Load the input memory buffer.
+ std::unique_ptr<MemoryBuffer> InputBuffer;
+ std::unique_ptr<ObjectImage> LoadedObject;
+ if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFileList[i],
+ InputBuffer))
+ return Error("unable to read input: '" + ec.message() + "'");
+
+ // Load the object file
+ LoadedObject.reset(Dyld.loadObject(new ObjectBuffer(InputBuffer.release())));
+ if (!LoadedObject) {
+ return Error(Dyld.getErrorString());
}
}
- if (!SymtabLCI)
- return Error("no symbol table found in object");
- if (!SegmentLCI)
- return Error("no symbol table found in object");
-
- // Read and register the symbol table data.
- InMemoryStruct<macho::SymtabLoadCommand> SymtabLC;
- Obj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC);
- if (!SymtabLC)
- return Error("unable to load symbol table load command");
- Obj->RegisterStringTable(*SymtabLC);
-
- // Read the dynamic link-edit information, if present (not present in static
- // objects).
- if (DysymtabLCI) {
- InMemoryStruct<macho::DysymtabLoadCommand> DysymtabLC;
- Obj->ReadDysymtabLoadCommand(*DysymtabLCI, DysymtabLC);
- if (!DysymtabLC)
- return Error("unable to load dynamic link-exit load command");
-
- // FIXME: We don't support anything interesting yet.
- if (DysymtabLC->LocalSymbolsIndex != 0)
- return Error("NOT YET IMPLEMENTED: local symbol entries");
- if (DysymtabLC->ExternalSymbolsIndex != 0)
- return Error("NOT YET IMPLEMENTED: non-external symbol entries");
- if (DysymtabLC->UndefinedSymbolsIndex != SymtabLC->NumSymbolTableEntries)
- return Error("NOT YET IMPLEMENTED: undefined symbol entries");
+ // Resolve all the relocations we can.
+ Dyld.resolveRelocations();
+ // Clear instruction cache before code will be executed.
+ MemMgr.invalidateInstructionCache();
+
+ // FIXME: Error out if there are unresolved relocations.
+
+ // Get the address of the entry point (_main by default).
+ void *MainAddress = Dyld.getSymbolAddress(EntryPoint);
+ if (!MainAddress)
+ return Error("no definition for '" + EntryPoint + "'");
+
+ // Invalidate the instruction cache for each loaded function.
+ for (unsigned i = 0, e = MemMgr.FunctionMemory.size(); i != e; ++i) {
+ sys::MemoryBlock &Data = MemMgr.FunctionMemory[i];
+ // Make sure the memory is executable.
+ std::string ErrorStr;
+ sys::Memory::InvalidateInstructionCache(Data.base(), Data.size());
+ if (!sys::Memory::setExecutable(Data, &ErrorStr))
+ return Error("unable to mark function executable: '" + ErrorStr + "'");
}
- // Load the segment load command.
- sys::MemoryBlock Data;
- StringMap<void*> SymbolTable;
- if (SegmentLCI->Command.Type == macho::LCT_Segment) {
- if (loadSegment32(Obj.get(), Data, SegmentLCI, SymtabLC, SymbolTable))
- return true;
- } else {
- if (loadSegment64(Obj.get(), Data, SegmentLCI, SymtabLC, SymbolTable))
- return true;
- }
-
- // Get the address of "_main".
- StringMap<void*>::iterator it = SymbolTable.find("_main");
- if (it == SymbolTable.end())
- return Error("no definition for '_main'");
-
- // Invalidate the instruction cache.
- sys::Memory::InvalidateInstructionCache(Data.base(), Data.size());
-
- // Make sure the memory is executable.
- if (!sys::Memory::setExecutable(Data, &ErrorStr))
- return Error("unable to mark function executable: '" + ErrorStr + "'");
-
// Dispatch to _main().
- void *MainAddress = it->second;
- errs() << "loaded '_main' at: " << MainAddress << "\n";
+ errs() << "loaded '" << EntryPoint << "' at: " << (void*)MainAddress << "\n";
int (*Main)(int, const char**) =
(int(*)(int,const char**)) uintptr_t(MainAddress);
const char **Argv = new const char*[2];
- Argv[0] = InputFile.c_str();
- Argv[1] = 0;
+ // Use the name of the first input object module as argv[0] for the target.
+ Argv[0] = InputFileList[0].c_str();
+ Argv[1] = nullptr;
return Main(1, Argv);
}
int main(int argc, char **argv) {
+ sys::PrintStackTraceOnErrorSignal();
+ PrettyStackTraceProgram X(argc, argv);
+
ProgramName = argv[0];
llvm_shutdown_obj Y; // Call llvm_shutdown() on exit.
cl::ParseCommandLineOptions(argc, argv, "llvm MC-JIT tool\n");
switch (Action) {
- default:
case AC_Execute:
return executeInput();
+ case AC_PrintLineInfo:
+ return printLineInfoForInput();
}
-
- return 0;
}