parser.h

当前支持 16-bit, 32-bit and 64-bit 的二进制文件

// parser.cpp
// Copyright (C) 2008 Willow Schlanger

#ifndef l_crudcom__parser_h__included
#define l_crudcom__parser_h__included

#include <list>
#include <map>
#include <stdexcept>
#include <cstddef>
#include <fstream>
#include <iostream>

#include "../x86s/types.h"
#include "../x86s/x86s_common.h"
#include "../x86s/x86s_decode.h"
#include "../crudasm/asmwriter.h"

#include <set>
#include <cstdio>

#define ICODE_PEEP_SIZE 4	/* must be > 1, and must be a power of 2 */

namespace x86s {

struct metadata_t
{
	U1 size : 4;		// been here? 1 for invalid opcodes, bytes_left for insn's that overhang image size.
	U1 local : 1;		// part of current procedure?
	U1 entry : 1;		// is this the start of a procedure?
	U1 branch : 1;
	U1 decompiled : 1;
	
	void clear(bool clear_entry)
	{
		size = 0;
		local = 0;
		if(clear_entry)
			entry = 0;
		decompiled = 0;
		branch = 0;
	}
};

struct memory_t
{
	U1 *image;
	metadata_t *meta;
	U8 image_size;
	U8 image_base;	// origin -- image[0] is really here
	U8 cs_base;	// offset in image that cs:0 corresponds to [?]
	U8 rip_mask;
};

struct parser_call_t
{
	bool decompile;
	std::set<U8> calls;
	parser_call_t()
	{
		decompile = true;
	}
};

typedef std::map<U8, parser_call_t> call_graph_t;

class parser_t
{
	memory_t &memory;
	public:
	int dsz;
	
	call_graph_t xgraph;

	parser_t(memory_t &mem, int dsz_t);
	
	// Call this between reparsing.
	void reset();
	
	// After calling reset, call this at least once.
	void add_entrypoint(U8 target);
	
	// Then call this to do the parsing.
	// Returns true if reparsing is necessary.
	bool parse();

	private:
	bool reparse;
	void do_parse(U8 target);
	std::set<U8> wanted_procs;
};

}	// namespace x86s

#endif	// l_crudcom__parser_h__included