#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/Support/DataTypes.h"
+#include "llvm/Support/MemoryObject.h"
+#include <vector>
namespace llvm {
namespace object {
class ObjectFile;
+ class MachOObjectFile;
}
class MCBasicBlock;
class MCFunction;
class MCInstrAnalysis;
class MCModule;
+class MCObjectSymbolizer;
/// \brief Disassemble an ObjectFile to an MCModule and MCFunctions.
/// This class builds on MCDisassembler to disassemble whole sections, creating
MCModule *buildEmptyModule();
+ typedef std::vector<uint64_t> AddressSetTy;
+ /// \name Create a new MCFunction.
+ MCFunction *createFunction(MCModule *Module, uint64_t BeginAddr,
+ AddressSetTy &CallTargets,
+ AddressSetTy &TailCallTargets);
+
+ /// \brief Set the region on which to fallback if disassembly was requested
+ /// somewhere not accessible in the object file.
+ /// This is used for dynamic disassembly (see RawMemoryObject).
+ void setFallbackRegion(std::unique_ptr<MemoryObject> Region) {
+ FallbackRegion = std::move(Region);
+ }
+
+ /// \brief Set the symbolizer to use to get information on external functions.
+ /// Note that this isn't used to do instruction-level symbolization (that is,
+ /// plugged into MCDisassembler), but to symbolize function call targets.
+ void setSymbolizer(MCObjectSymbolizer *ObjectSymbolizer) {
+ MOS = ObjectSymbolizer;
+ }
+
/// \brief Get the effective address of the entrypoint, or 0 if there is none.
virtual uint64_t getEntrypoint();
const object::ObjectFile &Obj;
const MCDisassembler &Dis;
const MCInstrAnalysis &MIA;
+ MCObjectSymbolizer *MOS;
+
+ /// \brief The fallback memory region, outside the object file.
+ std::unique_ptr<MemoryObject> FallbackRegion;
+
+ /// \brief Return a memory region suitable for reading starting at \p Addr.
+ /// In most cases, this returns a StringRefMemoryObject backed by the
+ /// containing section. When no section was found, this returns the
+ /// FallbackRegion, if it is suitable.
+ /// If it is not, or if there is no fallback region, this returns 0.
+ MemoryObject *getRegionFor(uint64_t Addr);
private:
/// \brief Fill \p Module by creating an atom for each section.
/// When the CFG is built, contiguous instructions that were previously in a
/// single MCTextAtom will be split in multiple basic block atoms.
void buildCFG(MCModule *Module);
+
+ MCBasicBlock *getBBAt(MCModule *Module, MCFunction *MCFN, uint64_t BeginAddr,
+ AddressSetTy &CallTargets,
+ AddressSetTy &TailCallTargets);
+};
+
+class MCMachOObjectDisassembler : public MCObjectDisassembler {
+ const object::MachOObjectFile &MOOF;
+
+ uint64_t VMAddrSlide;
+ uint64_t HeaderLoadAddress;
+
+ // __DATA;__mod_init_func support.
+ llvm::StringRef ModInitContents;
+ // __DATA;__mod_exit_func support.
+ llvm::StringRef ModExitContents;
+
+public:
+ /// \brief Construct a Mach-O specific object disassembler.
+ /// \param VMAddrSlide The virtual address slide applied by dyld.
+ /// \param HeaderLoadAddress The load address of the mach_header for this
+ /// object.
+ MCMachOObjectDisassembler(const object::MachOObjectFile &MOOF,
+ const MCDisassembler &Dis,
+ const MCInstrAnalysis &MIA, uint64_t VMAddrSlide,
+ uint64_t HeaderLoadAddress);
+
+protected:
+ uint64_t getEffectiveLoadAddr(uint64_t Addr) override;
+ uint64_t getOriginalLoadAddr(uint64_t EffectiveAddr) override;
+ uint64_t getEntrypoint() override;
+
+ ArrayRef<uint64_t> getStaticInitFunctions() override;
+ ArrayRef<uint64_t> getStaticExitFunctions() override;
};
}