/* ### * IP: GHIDRA * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ /// \file sleigh.hh /// \brief Classes and utilities for the main SLEIGH engine #ifndef __SLEIGH__ #define __SLEIGH__ #include "sleighbase.hh" class LoadImage; /// \brief Class for describing a relative p-code branch destination /// /// An intra-instruction p-code branch takes a \e relative operand. /// The actual value produced during p-code generation is calculated at /// the last second using \b this. It stores the index of the BRANCH /// instruction and a reference to its destination operand. This initially /// holds a reference to a destination \e label symbol, but is later updated /// with the final relative value. struct RelativeRecord { VarnodeData *dataptr; ///< Varnode indicating relative offset uintb calling_index; ///< Index of instruction containing relative offset }; /// \brief Data for building one p-code instruction /// /// Raw data used by the emitter to produce a single PcodeOp struct PcodeData { OpCode opc; ///< The op code VarnodeData *outvar; ///< Output Varnode data (or null) VarnodeData *invar; ///< Array of input Varnode data int4 isize; ///< Number of input Varnodes }; /// \brief Class for caching a chunk of p-code, prior to emitting /// /// The engine accumulates PcodeData and VarnodeData objects for /// a single instruction. Once the full instruction is constructed, /// the objects are passed to the emitter (PcodeEmit) via the emit() method. /// The class acts as a pool of memory for PcodeData and VarnodeData objects /// that can be reused repeatedly to emit multiple instructions. class PcodeCacher { VarnodeData *poolstart; ///< Start of the pool of VarnodeData objects VarnodeData *curpool; ///< First unused VarnodeData VarnodeData *endpool; ///< End of the pool of VarnodeData objects vector issued; ///< P-code ops issued for the current instruction list label_refs; ///< References to labels vector labels; ///< Locations of labels VarnodeData *expandPool(uint4 size); ///< Expand the memory pool public: PcodeCacher(void); ///< Constructor ~PcodeCacher(void); ///< Destructor /// \brief Allocate data objects for a new set of Varnodes /// /// \param size is the number of objects to allocate /// \return a pointer to the array of available VarnodeData objects VarnodeData *allocateVarnodes(uint4 size) { VarnodeData *newptr = curpool + size; if (newptr <= endpool) { VarnodeData *res = curpool; curpool = newptr; return res; } return expandPool(size); } /// \brief Allocate a data object for a new p-code operation /// /// \return the new PcodeData object PcodeData *allocateInstruction(void) { issued.emplace_back(); PcodeData *res = &issued.back(); res->outvar = (VarnodeData *)0; res->invar = (VarnodeData *)0; return res; } void addLabelRef(VarnodeData *ptr); ///< Denote a Varnode holding a \e relative \e branch offset void addLabel(uint4 id); ///< Attach a label to the \e next p-code instruction void clear(void); ///< Reset the cache so that all objects are unallocated void resolveRelatives(void); ///< Rewrite branch target Varnodes as \e relative offsets void emit(const Address &addr,PcodeEmit *emt) const; ///< Pass the cached p-code data to the emitter }; /// \brief A container for disassembly context used by the SLEIGH engine /// /// This acts as a factor for the ParserContext objects which are used to disassemble /// a single instruction. These all share a ContextCache which is a front end for /// accessing the ContextDatabase and resolving context variables from the SLEIGH spec. /// ParserContext objects are stored in a hash-table keyed by the address of the instruction. class DisassemblyCache { ContextCache *contextcache; ///< Cached values from the ContextDatabase AddrSpace *constspace; ///< The constant address space int4 minimumreuse; ///< Can call getParserContext this many times, before a ParserContext is reused uint4 mask; ///< Size of the hashtable in form 2^n-1 ParserContext **list; ///< (circular) array of currently cached ParserContext objects int4 nextfree; ///< Current end/beginning of circular list ParserContext **hashtable; ///< Hashtable for looking up ParserContext via Address void initialize(int4 min,int4 hashsize); ///< Initialize the hash-table of ParserContexts void free(void); ///< Free the hash-table of ParserContexts public: DisassemblyCache(ContextCache *ccache,AddrSpace *cspace,int4 cachesize,int4 windowsize); ///< Constructor ~DisassemblyCache(void) { free(); } ///< Destructor ParserContext *getParserContext(const Address &addr); ///< Get the parser for a particular Address }; /// \brief Build p-code from a pre-parsed instruction /// /// Through the build() method, \b this walks the parse tree and prepares data /// for final emission as p-code. (The final emitting is done separately through the /// PcodeCacher.emit() method). Generally, only p-code for one instruction is prepared. /// But, through the \b delay-slot mechanism, build() may recursively visit /// additional instructions. class SleighBuilder : public PcodeBuilder { virtual void dump( OpTpl *op ); AddrSpace *const_space; ///< The constant address space AddrSpace *uniq_space; ///< The unique address space uintb uniquemask; ///< Mask of address bits to use to uniquify temporary registers uintb uniqueoffset; ///< Uniquifier bits for \b this instruction DisassemblyCache *discache; ///< Cache of disassembled instructions PcodeCacher *cache; ///< Cache accumulating p-code data for the instruction void buildEmpty(Constructor *ct,int4 secnum); void generateLocation(const VarnodeTpl *vntpl,VarnodeData &vn); AddrSpace *generatePointer(const VarnodeTpl *vntpl,VarnodeData &vn); void generatePointerAdd(PcodeData *op,const VarnodeTpl *vntpl); void setUniqueOffset(const Address &addr); ///< Set uniquifying bits for the current instruction public: SleighBuilder(ParserWalker *w,DisassemblyCache *dcache,PcodeCacher *pc,AddrSpace *cspc,AddrSpace *uspc,uint4 umask); virtual void appendBuild(OpTpl *bld,int4 secnum); virtual void delaySlot(OpTpl *op); virtual void setLabel(OpTpl *op); virtual void appendCrossBuild(OpTpl *bld,int4 secnum); }; /// \brief A full SLEIGH engine /// /// Its provided with a LoadImage of the bytes to be disassembled and /// a ContextDatabase. /// /// Assembly is produced via the printAssembly() method, provided with an /// AssemblyEmit object and an Address. /// /// P-code is produced via the oneInstruction() method, provided with a PcodeEmit /// object and an Address. class Sleigh : public SleighBase { LoadImage *loader; ///< The mapped bytes in the program ContextDatabase *context_db; ///< Database of context values steering disassembly ContextCache *cache; ///< Cache of recently used context values mutable DisassemblyCache *discache; ///< Cache of recently parsed instructions mutable PcodeCacher pcode_cache; ///< Cache of p-code data just prior to emitting void clearForDelete(void); ///< Delete the context and disassembly caches protected: ParserContext *obtainContext(const Address &addr,int4 state) const; void resolve(ParserContext &pos) const; ///< Generate a parse tree suitable for disassembly void resolveHandles(ParserContext &pos) const; ///< Prepare the parse tree for p-code generation public: Sleigh(LoadImage *ld,ContextDatabase *c_db); ///< Constructor virtual ~Sleigh(void); ///< Destructor void reset(LoadImage *ld,ContextDatabase *c_db); ///< Reset the engine for a new program virtual void initialize(DocumentStorage &store); virtual void registerContext(const string &name,int4 sbit,int4 ebit); virtual void setContextDefault(const string &nm,uintm val); virtual void allowContextSet(bool val) const; virtual int4 instructionLength(const Address &baseaddr) const; virtual int4 oneInstruction(PcodeEmit &emit,const Address &baseaddr) const; virtual int4 printAssembly(AssemblyEmit &emit,const Address &baseaddr) const; }; /** \page sleigh SLEIGH \section sleightoc Table of Contents - \ref sleighoverview - \ref sleighbuild - \ref sleighuse - \subpage sleighAPIbasic - \subpage sleighAPIemulate \b Key \b Classes - \ref Translate - \ref AssemblyEmit - \ref PcodeEmit - \ref LoadImage - \ref ContextDatabase \section sleighoverview Overview Welcome to \b SLEIGH, a machine language translation and dissassembly engine. SLEIGH is both a processor specification language and the associated library and tools for using such a specification to generate assembly and to generate \b pcode, a reverse engineering Register Transfer Language (RTL), from binary machine instructions. SLEIGH was originally based on \b SLED, a \e Specification \e Language \e for \e Encoding \e and \e Decoding, designed by Norman Ramsey and Mary F. Fernandez, which performed disassembly (and assembly). SLEIGH extends SLED by providing semantic descriptions (via the RTL) of machine instructions and other practical enhancements for doing real world reverse engineering. SLEIGH is part of Project \b GHIDRA. It provides the core of the GHIDRA disassembler and the data-flow and decompilation analysis. However, SLEIGH can serve as a standalone library for use in other applications for providing a generic disassembly and RTL translation interface. \section sleighbuild Building SLEIGH There are a couple of \e make targets for building the SLEIGH library from source. These are: \code make libsla.a # Build the main library make libsla_dbg.a # Build the library with debug symbols \endcode The source code file \e sleighexample.cc has a complete example of initializing the Translate engine and using it to generate assembly and pcode. The source has a hard-coded file name, \e x86testcode, as the example binary executable it attempts to decode, but this can easily be changed. It also needs a SLEIGH specification file (\e .sla) to be present. Building the example application can be done with something similar to the following makefile fragment. \code # The C compiler CXX=g++ # Debug flags DBG_CXXFLAGS=-g -Wall -Wno-sign-compare OPT_CXXFLAGS=-O2 -Wall -Wno-sign-compare # libraries INCLUDES=-I./src LNK=src/libsla_dbg.a sleighexample.o: sleighexample.cc $(CXX) -c $(DBG_CXXFLAGS) -o sleighexample sleighexample.o $(LNK) clean: rm -rf *.o sleighexample \endcode \section sleighuse Using SLEIGH SLEIGH is a generic reverse engineering tool in the sense that the API is designed to be completely processor independent. In order to process binary executables for a specific processor, The library reads in a \e specification \e file, which describes how instructions are encoded and how they are interpreted by the processor. An application which needs to do disassembly or generate \b pcode can design to the SLEIGH API once, and then the application will automatically support any processor for which there is a specification. For working with a single processor, the SLEIGH library needs to load a single \e compiled form of the processor specification, which is traditionally given a ".sla" suffix. Most common processors already have a ".sla" file available. So to use SLEIGH with these processors, the library merely needs to be made aware of the desired file. This documentation covers the use of the SLEIGH API, assuming that this specification file is available. The ".sla" files themselves are created by running the \e compiler on a file written in the formal SLEIGH language. These files traditionally have the suffix ".slaspec" For those who want to design such a specification for a new processor, please refer to the document, "SLEIGH: A Language for Rapid Processor Specification." */ /** \page sleighAPIbasic The Basic SLEIGH Interface To use SLEIGH as a library within an application, there are basically five classes that you need to be aware of. - \ref sleightranslate - \ref sleighassememit - \ref sleighpcodeemit - \ref sleighloadimage - \ref sleighcontext \section sleightranslate Translate (or Sleigh) The core SLEIGH class is Sleigh, which is derived from the interface, Translate. In order to instantiate it in your code, you need a LoadImage object, and a ContextDatabase object. The load image is responsible for retrieving instruction bytes, based on address, from a binary executable. The context database provides the library extra mode information that may be necessary to do the disassembly or translation. This can be used, for instance, to specify that an x86 binary is running in 32-bit mode, or to specify that an ARM processor is running in THUMB mode. Once these objects are built, the Sleigh object can be immediately instantiated. \code LoadImageBfd *loader; ContextDatabase *context; Translate *trans; // Set up the loadimage // Providing an executable name and architecture string loadimagename = "x86testcode"; string bfdtarget= "default"; loader = new LoadImageBfd(loadimagename,bfdtarget); loader->open(); // Load the executable from file context = new ContextInternal(); // Create a processor context trans = new Sleigh(loader,context); // Instantiate the translator \endcode Once the Sleigh object is in hand, the only required initialization step left is to inform it of the ".sla" file. The file is in XML format and needs to be read in using SLEIGH's built-in XML parser. The following code accomplishes this. \code string sleighfilename = "specfiles/x86.sla"; DocumentStorage docstorage; Element *sleighroot = docstorage.openDocument(sleighfilename)->getRoot(); docstorage.registerTag(sleighroot); trans->initialize(docstorage); // Initialize the translator \endcode \section sleighassememit AssemblyEmit In order to do disassembly, you need to derive a class from AssemblyEmit, and implement the method \e dump. The library will call this method exactly once, for each instruction disassembled. This routine simply needs to decide how (and where) to print the corresponding portion of the disassembly. For instance, \code class AssemblyRaw : public AssemblyEmit { public: virtual void dump(const Address &addr,const string &mnem,const string &body) { addr.printRaw(cout); cout << ": " << mnem << ' ' << body << endl; } }; \endcode This is a minimal implementation that simply dumps the disassembly straight to standard out. Once this object is instantiated, the Sleigh object can use it to write out assembly via the Translate::printAssembly() method. \code AssemblyEmit *assememit = new AssemblyRaw(); Address addr(trans->getDefaultCodeSpace(),0x80484c0); int4 length; // Length of instruction in bytes length = trans->printAssembly(*assememit,addr); addr = addr + length; // Advance to next instruction length = trans->printAssembly(*assememit,addr); addr = addr + length; length = trans->printAssembly(*assememit,addr); \endcode \section sleighpcodeemit PcodeEmit In order to generate a \b pcode translation of a machine instruction, you need to derive a class from PcodeEmit and implement the virtual method \e dump. This method will be invoked once for each \b pcode operation in the translation of a machine instruction. There will likely be multiple calls per instruction. Each call passes in a single \b pcode operation, complete with its possible varnode output, and all of its varnode inputs. Here is an example of a PcodeEmit object that simply prints out the \b pcode. \code class PcodeRawOut : public PcodeEmit { public: virtual void dump(const Address &addr,OpCode opc,VarnodeData *outvar,VarnodeData *vars,int4 isize); }; static void print_vardata(ostream &s,VarnodeData &data) { s << '(' << data.space->getName() << ','; data.space->printOffset(s,data.offset); s << ',' << dec << data.size << ')'; } void PcodeRawOut::dump(const Address &addr,OpCode opc,VarnodeData *outvar,VarnodeData *vars,int4 isize) { if (outvar != (VarnodeData *)0) { // The output is optional print_vardata(cout,*outvar); cout << " = "; } cout << get_opname(opc); // Possibly check for a code reference or a space reference for(int4 i=0;igetDefaultCodeSpace(),0x80484c0); int4 length; // Length of instruction in bytes length = trans->oneInstruction(*pcodeemit,addr); addr = addr + length; // Advance to next instruction length = trans->oneInstruction(*pcodeemit,addr); addr = addr + length; length = trans->oneInstruction(*pcodeemit,addr); \endcode For an application to properly \e follow \e flow, while translating machine instructions into pcode, the emitted pcode must be inspected for the various branch operations. \section sleighloadimage LoadImage A LoadImage holds all the binary data from an executable file in the format similar to how it would exist when being executed by a real processor. The interface to this from SLEIGH is actually very simple, although it can hide a complicated structure. One method does most of the work, LoadImage::loadFill(). It takes a byte pointer, a size, and an Address. The method is expected to fill in the \e ptr array with \e size bytes taken from the load image, corresponding to the address \e addr. There are two more virtual methods that are required for a complete implementation of LoadImage, \e getArchType and \e adjustVma, but these do not need to be implemented fully. \code class MyLoadImage : public LoadImage { public: MyLoadImage(const string &nm) : Loadimage(nm) {} virtual void loadFill(uint1 *ptr,int4 size,const Address &addr); virtual string getArchType(void) const { return "mytype"; } virtual void adjustVma(long adjust) {} }; \endcode \section sleighcontext ContextDatabase The ContextDatabase needs to keep track of any possible context variable and its value, over different address ranges. In most cases, you probably don't need to override the class yourself, but can use the built-in class, ContextInternal. This provides the basic functionality required and will work for different architectures. What you may need to do is set values for certain variables, depending on the processor and the environment it is running in. For instance, for the x86 platform, you need to set the \e addrsize and \e opsize bits, to indicate the processor would be running in 32-bit mode. The context variables specific to a particular processor are established by the SLEIGH spec. So the variables can only be set \e after the spec has been loaded. \code ... context = new ContextInternal(); trans = new Sleigh(loader,context); DocumentStorage docstorage; Element *root = docstorage.openDocument("specfiles/x86.sla")->getRoot(); docstorage.registerTag(root); trans->initialize(docstorage); context->setVariableDefault("addrsize",1); // Address size is 32-bits context->setVariableDefault("opsize",1); // Operand size is 32-bits \endcode */ #endif