├── .gitignore ├── LICENSE ├── README.md ├── doc ├── images │ ├── cfg_cons.dot │ ├── cfg_cons.png │ ├── cfg_if.dot │ ├── cfg_if.png │ ├── cfg_ifelse.dot │ ├── cfg_ifelse.png │ ├── cfg_while.dot │ ├── cfg_while.png │ ├── graph_stages.dot │ └── graph_stages.png └── ocd.md ├── ocd ├── src ├── control_flow.py ├── debug.py ├── decompile.py ├── disassemble.py ├── disassemblers │ ├── __init__.py │ ├── libdisassemble │ │ ├── LICENSE │ │ ├── README │ │ ├── __init__.py │ │ ├── disassemble.py │ │ └── opcode86.py │ └── x64.py ├── function_calls.py ├── objdump.py ├── ocd.py ├── output │ ├── __init__.py │ ├── c.py │ ├── conditions.py │ └── indent.py └── representation.py └── tests ├── Makefile ├── test_ack.c ├── test_ack_scanf.c ├── test_add.c ├── test_add_var.c ├── test_block.c ├── test_collapse.c ├── test_collapse2.c ├── test_div_zero.c ├── test_for.c ├── test_for_if.c ├── test_mod_zero.c ├── test_ptr.c ├── test_return_0.c ├── test_sum1.c ├── test_sum2.c ├── test_sum3.c ├── test_sum4.c ├── test_sum5.c ├── test_sum6.c ├── test_sum7.c ├── test_sznurki_alistra.c ├── test_sznurki_drx.c ├── test_variables.c ├── test_while.c └── test_zero.c /.gitignore: -------------------------------------------------------------------------------- 1 | # Ignore compiled python sources 2 | *.pyc 3 | 4 | # Ignore swap files 5 | *.swp 6 | 7 | # Ignore test executables, but not the sources 8 | tests/test_* 9 | !tests/test_*.c 10 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Copyright (c) 2010 Alek Balicki, Luke Zapart 2 | 3 | Permission is hereby granted, free of charge, to any person 4 | obtaining a copy of this software and associated documentation 5 | files (the "Software"), to deal in the Software without 6 | restriction, including without limitation the rights to use, 7 | copy, modify, merge, publish, distribute, sublicense, and/or sell 8 | copies of the Software, and to permit persons to whom the 9 | Software is furnished to do so, subject to the following 10 | conditions: 11 | 12 | The above copyright notice and this permission notice shall be 13 | included in all copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 16 | EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES 17 | OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND 18 | NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT 19 | HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, 20 | WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 21 | FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 22 | OTHER DEALINGS IN THE SOFTWARE. 23 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # ocd - ocd C decompiler 2 | 3 | **ocd** is a C decompiler written in Python, currently supporting decompilation of Linux programs compiled for the x64 architecture. 4 | 5 | The decompiler tries to infer the program structure, performing control and data flow analysis. 6 | 7 | If you'd like to learn how it works, check out my article on [how decompilers work](https://lukezapart.com/how-decompilers-work). 8 | 9 | More in-depth documentation is available in the [doc/ocd.md file](doc/ocd.md). 10 | 11 | ## Quick start 12 | 13 | Run `./ocd program` to decompile `program` (e.g. `./ocd /usr/bin/yes`). 14 | 15 | There are some test programs in the `tests/` directory. To check them out, do `cd tests; make`, then for example run `./ocd tests/test_ack`. 16 | 17 | Run `./ocd --help` to get more options. 18 | 19 | ## Requirements 20 | 21 | * Python 3 22 | * objdump 23 | 24 | ## Notes 25 | 26 | * ocd uses [Immunity libdisassembly v2.0](https://web.archive.org/web/20140209154423/http://www.immunitysec.com/resources-freesoftware.shtml) 27 | 28 | ## Example 29 | 30 | 96 | 97 |
31 | 32 | ```c 33 | int ack(int m, int n) 34 | { 35 | if (m == 0) 36 | return n + 1; 37 | else 38 | { 39 | if (n == 0) 40 | return ack(m-1, 1); 41 | else 42 | return ack(m-1, ack(m, n-1)); 43 | } 44 | } 45 | int main() 46 | { 47 | return printf("%d\n", ack(3, 4)); 48 | } 49 | 50 | 51 | 52 | 53 | 54 | 55 | 56 | 57 | 58 | 59 | 60 | 61 | ``` 62 | 63 | 64 | 65 | ```c 66 | int ack() 67 | { 68 | var_0 = temp_2; 69 | var_1 = temp_3; 70 | var_2 = var_0 - 0; 71 | if (var2) 72 | { 73 | var_2 = var_1 - 0; 74 | if (var2) 75 | { 76 | temp_4 = ack(temp_5 - 1, ack(var_0, temp_4 - 1)); 77 | } 78 | else 79 | { 80 | temp_4 = ack(temp_4 - 1, 1); 81 | } 82 | } 83 | else 84 | { 85 | temp_4 = temp_4 + 1; 86 | } 87 | return temp_4; 88 | } 89 | int main() 90 | { 91 | return unknown_function("%d\n", ack(3, 4)); 92 | } 93 | ``` 94 | 95 |
Original codeocd output
98 | -------------------------------------------------------------------------------- /doc/images/cfg_cons.dot: -------------------------------------------------------------------------------- 1 | digraph cfg_if { 2 | a [label=""]; 3 | b [label=""]; 4 | c [label=""]; 5 | a -> b; 6 | b -> c; 7 | } 8 | -------------------------------------------------------------------------------- /doc/images/cfg_cons.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/doc/images/cfg_cons.png -------------------------------------------------------------------------------- /doc/images/cfg_if.dot: -------------------------------------------------------------------------------- 1 | digraph cfg_if { 2 | a [label=""]; 3 | b [label=""]; 4 | c [label=""]; 5 | a -> b; 6 | b -> c; 7 | a -> c; 8 | } 9 | -------------------------------------------------------------------------------- /doc/images/cfg_if.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/doc/images/cfg_if.png -------------------------------------------------------------------------------- /doc/images/cfg_ifelse.dot: -------------------------------------------------------------------------------- 1 | digraph cfg_if { 2 | a [label=""]; 3 | b [label=""]; 4 | c [label=""]; 5 | d [label=""]; 6 | a -> b; 7 | a -> c; 8 | b -> d; 9 | c -> d; 10 | } 11 | -------------------------------------------------------------------------------- /doc/images/cfg_ifelse.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/doc/images/cfg_ifelse.png -------------------------------------------------------------------------------- /doc/images/cfg_while.dot: -------------------------------------------------------------------------------- 1 | digraph cfg_if { 2 | a [label=""]; 3 | b [label=""]; 4 | c [label=""]; 5 | d [label=""]; 6 | a -> b; 7 | b -> c; 8 | c -> b; 9 | b -> d; 10 | } 11 | -------------------------------------------------------------------------------- /doc/images/cfg_while.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/doc/images/cfg_while.png -------------------------------------------------------------------------------- /doc/images/graph_stages.dot: -------------------------------------------------------------------------------- 1 | digraph graph_stages { 2 | 1 [label="Object dump"]; 3 | 2 [label="Disassembly"]; 4 | subgraph cluster_cf { 5 | 3 [label="Control flow graph\ngeneration"]; 6 | 4 [label="Control flow graph\nreduction"]; 7 | label = "Control flow analysis"; 8 | style = dotted; 9 | } 10 | subgraph cluster_df { 11 | 5 [label="Variable inference"]; 12 | 6 [label="Computation collapse"]; 13 | label = "Data flow analysis"; 14 | style = dotted; 15 | } 16 | 7 [label="Program output"]; 17 | 1 -> 2 -> 3 -> 4 -> 5 -> 6 -> 7; 18 | } 19 | -------------------------------------------------------------------------------- /doc/images/graph_stages.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/doc/images/graph_stages.png -------------------------------------------------------------------------------- /doc/ocd.md: -------------------------------------------------------------------------------- 1 | # Introduction 2 | 3 | ocd is a C decompiler written in Python, currently supporting decompilation of programs compiled for the x64 architecture. The internal data structures used for instructions are quite universal, so it should be trivial to add additional architectures as they are needed, as well as new output languages. The decompiler performs some inferences as to the program structure (such as control and data flow analysis). 4 | 5 | # Using ocd 6 | 7 | ocd reads a specified object file, performs decompilation on it and prints the results on standard output. It can also output a set of control flow graphs in dot format. 8 | 9 | The usage of the program is as follows: 10 | 11 | Usage: ocd [options] file 12 | 13 | Options: 14 | -h, --help show this help message and exit 15 | -d OPTION, --debug=OPTION 16 | turn debug option on 17 | -g FILE, --graph=FILE 18 | output a control flow graph 19 | 20 | 21 | # Operation 22 | 23 | ## Overview 24 | 25 | The decompilation occurs in stages. The program is built to be modular and each stage is separate. 26 | 27 | ![Stages of decompilation](images/graph_stages.png?raw=true) 28 | 29 | ## Object dump 30 | 31 | Currently, ocd executes the GNU utility objdump and retrieves the list of object exports from it. 32 | 33 | ## Disassembly 34 | 35 | ocd supports decompilation of x86-64 and uses a slightly modified library called libdisassemble to disassemble code. 36 | 37 | ## Control flow analysis 38 | 39 | ### Control flow graph generation 40 | 41 | A block of code is a list of instructions. 42 | 43 | **B1** is a subblock of **B2** if **B1** is a sublist of **B2**. 44 | 45 | A block of code **B** is uninterruptible if both of these conditions hold: 46 | 47 | * no instruction jumps to an instruction within **B** other than the first (i.e. the only allowed entry point in **B** is its head) 48 | * no instruction in **B**, other than the last, jumps (i.e. the only exit point in **B** is its last instruction) 49 | 50 | A block of code **B** is maximally uninterruptible if it is interruptible and has both an entry point and an exit point. 51 | 52 | A control flow graph **cfgB** of a code block **B** is a digraph ****. The set of nodes **V** is the set of all maximally uninterruptible subblocks of **B**. An edge ** ∈ E** if and only if there is an instruction in **B1** that jumps to **B2**. 53 | 54 | Generating the cfg of a block of code consists of finding all jumps in the block and then chopping the block into maximally interruptible blocks based on the list of jumps. 55 | 56 | ### Control flow graph reduction 57 | 58 | A cfg of a block of code gives insight into how control flows through the block of code and ideally gives insight into what the program structure was originally. 59 | 60 | 61 | 62 | 63 | 64 | 65 | 66 | 67 | 68 | 69 | 70 | 71 | 72 | 73 | 74 | 75 | 76 | 77 | 78 | 79 | 80 | 81 | 82 |
CFG patterns
ifif/else
whilecons
83 | 84 | Logical structures can be recognized by finding patterns in the cfg. The patterns are listed above. The patterns are used to define a decreasing graph rewriting system (not unlike a context-sensitive grammar) in which a single step finds a pattern and contracts it to a single node containing information about the pattern's respective logical structure. Ideally, the fixed point of such a transformation will result in a single node containing the abstract syntax tree of the original program. 85 | 86 | ## Data flow analysis 87 | 88 | ### Variable inference 89 | 90 | In this phase, instruction operands are substituted with newly assigned names. 91 | 92 | ## Computation collapse 93 | 94 | In this phase, the program's computation is analyzed. Dead instructions are removed and others are folded into more coherent expressions. For example, 95 | 96 | a = a + 1 97 | b = a * 2 98 | n = 2 99 | n = b - a 100 | 101 | might be converted into: 102 | 103 | n = (a+1)*2-a+1 104 | 105 | ## Program output 106 | 107 | The program is output recursively on the cfgs of its symbols. 108 | -------------------------------------------------------------------------------- /ocd: -------------------------------------------------------------------------------- 1 | ./src/ocd.py -------------------------------------------------------------------------------- /src/control_flow.py: -------------------------------------------------------------------------------- 1 | from collections import defaultdict 2 | from itertools import dropwhile, permutations 3 | 4 | graphfile = None 5 | 6 | class Graph: 7 | ''' 8 | Basic graph structure based on successor and predecessor lists. 9 | ''' 10 | def __init__(self): 11 | ''' 12 | Initialize the graph with empty dicts of edges and vertices. 13 | ''' 14 | self._pred = defaultdict(list) 15 | self._succ = defaultdict(list) 16 | self._edge_values = {} 17 | self._vertices = {} 18 | 19 | def __str__(self): 20 | ''' 21 | Return the string representation of the graph. 22 | ''' 23 | return 'Graph(vertices=' + str(self._vertices) + ', pred=' + str(self._pred) + ', succ=' + str(self._succ) + ')' 24 | 25 | def __contains__(self, vertex): 26 | ''' 27 | Check if the graph contains a vertex. 28 | ''' 29 | v, t = vertex 30 | if t == 'in': 31 | return (v in self._pred) 32 | if t == 'out': 33 | return (v in self._succ) 34 | if t == None: 35 | return (v in self._pred or v in self._succ) 36 | 37 | def successors(self, key): 38 | ''' 39 | Return a list of successors for a given vertex. 40 | ''' 41 | return self._succ[key] 42 | 43 | def predecessors(self, key): 44 | ''' 45 | Return a list of predecessors for a given vertex. 46 | ''' 47 | return self._pred[key] 48 | 49 | def vertex(self, k): 50 | ''' 51 | Return a vertex. 52 | ''' 53 | return self._vertices[k] 54 | 55 | def set_vertex(self, k, v=None): 56 | ''' 57 | Set a vertex. 58 | ''' 59 | self._vertices[k] = v 60 | 61 | def remove_vertices(self, ks): 62 | ''' 63 | Remove vertices, along with edges adjacent to them. 64 | ''' 65 | for k in ks: 66 | for p in self._pred[k]: 67 | self._succ[p].remove(k) 68 | for s in self._succ[k]: 69 | self._pred[s].remove(k) 70 | del self._pred[k] 71 | del self._succ[k] 72 | del self._vertices[k] 73 | 74 | def vertices(self): 75 | ''' 76 | Return the list of vertices. 77 | ''' 78 | return self._vertices 79 | 80 | def deg_in(self, k): 81 | ''' 82 | Return the in-degree of a vertex. 83 | ''' 84 | return len(self._pred[k]) 85 | 86 | def deg_out(self, k): 87 | ''' 88 | Return the out-degree of a vertex. 89 | ''' 90 | return len(self._succ[k]) 91 | 92 | def add_edge(self, v_in, v_out, value=None): 93 | ''' 94 | Add an edge to the graph, optionally specifying a value. 95 | ''' 96 | self._succ[v_in].append(v_out) 97 | self._pred[v_out].append(v_in) 98 | self._edge_values[(v_in, v_out)] = value 99 | 100 | def move_predecessors(self, v, v_new): 101 | ''' 102 | Move edges pointing to a vertex so they point to another. 103 | ''' 104 | for pred in self.predecessors(v)[:]: 105 | self.add_edge(pred, v_new, self.edge(pred, v)) 106 | self.remove_edge(pred, v) 107 | 108 | def move_successors(self, v, v_new): 109 | ''' 110 | Move edges coming out of a vertex so they come out of another. 111 | ''' 112 | for succ in self.successors(v)[:]: 113 | self.add_edge(v_new, succ, self.edge(v, succ)) 114 | self.remove_edge(v, succ) 115 | 116 | def edge(self, v_in, v_out): 117 | ''' 118 | Return an edge. 119 | ''' 120 | return self._edge_values[(v_in, v_out)] 121 | 122 | def remove_edge(self, v_in, v_out): 123 | ''' 124 | Remove an edge. 125 | ''' 126 | self._succ[v_in].remove(v_out) 127 | self._pred[v_out].remove(v_in) 128 | del self._edge_values[(v_in, v_out)] 129 | 130 | def sortedvertices(self): 131 | ''' 132 | Return an iterator of sorted vertices (according to their original address). 133 | ''' 134 | def key_f(arg): 135 | ''' 136 | Use the vertex start address as key. 137 | ''' 138 | arg_v, arg_t = arg 139 | a, b = arg_v 140 | return b 141 | return sorted(self.vertices().items(), key=key_f) 142 | 143 | def iterblocks(self): 144 | """ 145 | Inorder traversal of graph blocks. 146 | 147 | Yields block and its depth. 148 | """ 149 | def traverse(b, depth): 150 | if type(b) is not tuple: 151 | return 152 | 153 | t, v = b 154 | v_type, v_start = t 155 | 156 | if v_type == 'block': 157 | yield v, depth 158 | 159 | else: 160 | for block in v: 161 | for y in traverse(block, depth+1): 162 | yield y 163 | 164 | for v in self.sortedvertices(): 165 | for y in traverse(v, 1): 166 | yield y 167 | 168 | def export(self, f, name, random=False): 169 | ''' 170 | Export a graph to graphviz format. 171 | ''' 172 | def block_label(block_type, block_loc, block): 173 | if block_type == 'block': 174 | return "block\\n{0:x}:{1:x}".format(block_loc, block[-1]['loc']) 175 | else: 176 | return "{0}\\n{1:x}".format(block_type, block_loc) 177 | 178 | if random: 179 | import random 180 | name += '_'+''.join(random.sample('0123456789abcdef', 8)) 181 | 182 | f.write("\tsubgraph {0} {{\n".format(name)) 183 | 184 | entries = filter(lambda k: self.deg_in(k) == 0, self.vertices()) 185 | 186 | f.write("\t\t{0}_entry [label=\"{0}\"];\n".format(name)) 187 | for e_type, e_loc in entries: 188 | f.write("\t\t{0}_entry -> {0}_{1}_{2:x};\n".format(name, e_type, e_loc)) 189 | 190 | for (block_type, block_loc), block in self.vertices().items(): 191 | f.write("\t\t{0}_{1}_{2:x} [label=\"{3}\"];\n".format(name, block_type, block_loc, block_label(block_type, block_loc, block))) 192 | for v_type, v_out in self.successors((block_type, block_loc)): 193 | label = '' 194 | value = self.edge((block_type, block_loc), (v_type, v_out)) 195 | if value is not None: 196 | label = '[label="{0}"]'.format(value) 197 | f.write("\t\t{0}_{1}_{2:x} -> {0}_{3}_{4:x} {label};\n".format(name, block_type, block_loc, v_type, v_out, label=label)) 198 | f.write("\t}\n") 199 | 200 | def control_flow_graph(function, labels, name): 201 | ''' 202 | Analyze the control flow graph (cfg) of the function. 203 | ''' 204 | 205 | graph = Graph() 206 | block_acc = [] 207 | block_cur = None 208 | block_last = None 209 | block_change = True 210 | pass_through = False 211 | for ins in function: 212 | if ins['loc'] in labels or block_change: 213 | if block_last is not None: 214 | graph.set_vertex(block_last, block_acc) 215 | block_cur = ('block', ins['loc']) 216 | block_acc = [] 217 | graph.set_vertex(block_cur) 218 | 219 | if ins['loc'] in labels and pass_through: 220 | # pass-through edge 221 | graph.add_edge(block_last, ('block', ins['loc'])) 222 | 223 | block_change = False 224 | pass_through = True 225 | 226 | if ins['ins']['op'] == 'jump': # jumps (and only jumps) 227 | loc_next = ins['loc']+ins['length'] 228 | loc_j = loc_next+ins['ins']['dest']['repr'] 229 | graph.add_edge(block_cur, ('block', loc_j), ins['ins']['cond']) 230 | block_change = True 231 | if ins['ins']['cond'] != 'true': 232 | graph.add_edge(block_cur, ('block', loc_next)) 233 | else: 234 | pass_through = False 235 | block_last = block_cur 236 | block_acc.append(ins) 237 | 238 | graph.set_vertex(block_last, block_acc) 239 | 240 | if graphfile: 241 | graph.export(graphfile, name) 242 | 243 | return graph_transform(graph) 244 | 245 | def flip(x): 246 | ''' 247 | flip(x)(f) = f(x) 248 | ''' 249 | def f(g): 250 | return g(x) 251 | return f 252 | 253 | def graph_transform(graph): 254 | ''' 255 | Perform one step transformations of the graph according to 256 | preset rules until no more transformations can be performed. 257 | 258 | The rewriting system used here is decreasing (i.e. all transformations 259 | decrease the size of the graph), because every transformation rule is 260 | guaranteed to be decreasing, therefore it is total (it will always stop 261 | at some point. 262 | 263 | Transformation rules are functions of the type 264 | 265 | Graph -> (Boolean, Graph) 266 | 267 | The boolean value signifies whether the transformation condition was 268 | satisfied and the transformation was applied successfully. 269 | 270 | If no further transformation can be performed, the process is stopped 271 | and the resulting graph is returned. 272 | ''' 273 | def t_trivial(graph): 274 | return False, graph 275 | 276 | def t_while(graph): 277 | for v in graph.vertices(): 278 | if graph.deg_out(v) == 2: 279 | succs = graph.successors(v) 280 | for s1, s2 in permutations(succs): 281 | if graph.deg_out(s1) == 1 and graph.deg_in(s1) == 1 and v in graph.successors(s1): 282 | v_type, v_start = v 283 | v_new = ('while', v_start) 284 | condition = graph.edge(v, s1) 285 | v_new_value = (condition, (v, graph.vertex(v)), (s1, graph.vertex(s1))) 286 | graph.set_vertex(v_new, v_new_value) 287 | for pred in graph.predecessors(v): 288 | graph.add_edge(pred, v_new) 289 | graph.add_edge(v_new, s2) 290 | graph.remove_vertices([v,s1]) 291 | 292 | return (True, graph) 293 | 294 | return (False, graph) 295 | 296 | def t_if(graph): 297 | for v in graph.vertices(): 298 | if graph.deg_out(v) == 2: 299 | succs = graph.successors(v) 300 | for s1, s2 in permutations(succs): 301 | if graph.deg_out(s1) == 1 and graph.deg_in(s1) == 1: 302 | s1_succ = graph.successors(s1)[0] 303 | if s1_succ == s2: 304 | s1_type, s1_start = s1 305 | v_new = ('if', s1_start) 306 | snd = lambda x,y: y 307 | condition = '!'+graph.edge(v, s2) 308 | v_new_value = (condition, (s1, graph.vertex(s1))) 309 | graph.set_vertex(v_new, v_new_value) 310 | graph.add_edge(v, v_new) 311 | graph.add_edge(v_new, s2) 312 | graph.remove_edge(v, s2) 313 | graph.remove_vertices([s1]) 314 | return (True, graph) 315 | 316 | return (False, graph) 317 | 318 | def t_ifelse(graph): 319 | for v in graph.vertices(): 320 | succs = graph.successors(v) 321 | if len(succs) == 2: 322 | for s, t in permutations(succs): 323 | if graph.edge(v, s) is None: 324 | continue 325 | s_s, t_s = map(graph.successors, succs) 326 | s_p, t_p = map(graph.predecessors, succs) 327 | if [len(x) for x in (s_s, t_s, s_p, t_p)] == [1]*4 and s_s == t_s: 328 | s_type, s_start = s 329 | v_new = ('ifelse', s_start) 330 | condition = graph.edge(v,s) 331 | # modify v 332 | v_new_value = (condition, 333 | (s, graph.vertex(s)), (t, graph.vertex(t))) 334 | graph.set_vertex(v_new, v_new_value) 335 | graph.add_edge(v_new, s_s[0]) 336 | graph.add_edge(v, v_new) 337 | graph.remove_vertices([s, t]) 338 | return (True, graph) 339 | 340 | return (False, graph) 341 | 342 | def t_cons(graph): 343 | for v in graph.vertices(): 344 | if graph.deg_out(v) == 1: 345 | s = graph.successors(v)[0] 346 | if graph.deg_in(s) == 1 and v != s: 347 | v_type, v_start = v 348 | s_type, s_start = s 349 | if v_type == 'cons': 350 | graph.vertex(v).append((s, graph.vertex(s))) 351 | #graph.set_vertex(v, v_value) 352 | for succ in graph.successors(s): 353 | graph.add_edge(v, succ) 354 | graph.remove_vertices([s]) 355 | else: 356 | v_new = ('cons', v_start) 357 | v_new_value = [(v, graph.vertex(v)), (s, graph.vertex(s))] 358 | graph.set_vertex(v_new, v_new_value) 359 | graph.move_predecessors(v, v_new) 360 | graph.move_successors(s, v_new) 361 | graph.remove_vertices([v,s]) 362 | return (True, graph) 363 | 364 | return (False, graph) 365 | 366 | rules = [t_trivial, t_ifelse, t_if, t_while, t_cons] 367 | 368 | i = dropwhile(lambda x: not x[0], map(flip(graph), rules)) 369 | try: 370 | true, graph = i.__next__() 371 | if graphfile: 372 | graph.export(graphfile, 'step', random=True) 373 | return graph_transform(graph) 374 | except StopIteration: 375 | if graphfile: 376 | graph.export(graphfile, 'transformed') 377 | return graph 378 | -------------------------------------------------------------------------------- /src/debug.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Debug module 3 | 4 | Acceptable debug levels: all, graph, misc 5 | ''' 6 | 7 | options = ['all', 'graph', 'misc', 'asm_rw'] 8 | 9 | __debug = [] 10 | 11 | def check(cond): 12 | global __debug 13 | return (cond in __debug or 'all' in __debug) 14 | 15 | def set(cond=None): 16 | global __debug 17 | if cond is None: 18 | cond = 'all' 19 | 20 | __debug.append(cond) 21 | 22 | def sprint(str, cond): 23 | if debug_check(cond): 24 | return str 25 | else: 26 | return '' 27 | -------------------------------------------------------------------------------- /src/decompile.py: -------------------------------------------------------------------------------- 1 | from control_flow import control_flow_graph 2 | from copy import copy, deepcopy 3 | from itertools import count 4 | import debug 5 | import function_calls 6 | import disassemblers.libdisassemble.opcode86 as opcode86 7 | import re 8 | import representation 9 | import zlib 10 | 11 | def is_constant(x): 12 | ''' 13 | Test of the argument is a constant. 14 | ''' 15 | return re.match("-?0x.*", x) 16 | 17 | def is_register(x): 18 | ''' 19 | Test if the argument is a register. 20 | ''' 21 | registers = map(lambda x: x[0], opcode86.regs) 22 | return x in registers 23 | 24 | def get_labels(functions): 25 | ''' 26 | Find addresses that need to be labeled. 27 | ''' 28 | labels = {} 29 | for function in functions: 30 | for ins in functions[function]: 31 | if ins['ins']['op'] == 'jump': 32 | addr = ins['loc']+ins['length']+ins['ins']['dest']['repr'] 33 | labels[addr] = 'loc_{0:x}'.format(addr) 34 | 35 | return labels 36 | 37 | def infer_signature(asm): 38 | ''' 39 | Infer the signature of a function. 40 | ''' 41 | return ('int', []) 42 | 43 | def variable_inference(cfg): 44 | ''' 45 | Infer the variables in code. 46 | 47 | Change all temporary variables (registers) into temp_0, temp_1, etc. 48 | and other variables into var_0, var_1, etc. 49 | ''' 50 | def variable_inference_arg(arg): 51 | if 'op' in arg: 52 | variable_inference_ins(arg) 53 | elif arg['value'] in vars: 54 | arg.update(vars[arg['value']]) 55 | else: 56 | if arg['w'] or arg['r']: 57 | if arg['r'] and is_constant(arg['value']): 58 | type = 'const' 59 | if arg['value'] == arg['repr']: 60 | repr = int(arg['value'], 16) 61 | else: 62 | repr = arg['repr'] 63 | elif is_register(arg['value']): 64 | type = 'temp' 65 | repr = temp_names.__next__() 66 | else: 67 | type = 'var' 68 | repr = var_names.__next__() 69 | info = {'type':type, 'repr':repr, 'value': arg['value']} 70 | vars[arg['value']] = info 71 | arg.update(info) 72 | 73 | def variable_inference_ins(ins): 74 | for k, arg in ins.items(): 75 | if k == 'args': 76 | for a in arg: 77 | variable_inference_arg(a) 78 | if k not in ('src', 'dest'): 79 | continue 80 | 81 | variable_inference_arg(ins[k]) 82 | 83 | 84 | var_names = new_var_name() 85 | temp_names = new_temp_name() 86 | vars = {} 87 | 88 | for block, depth in cfg.iterblocks(): 89 | for i, line in enumerate(block): 90 | variable_inference_ins(line['ins']) 91 | 92 | var_number_indicator = var_names.__next__() 93 | m = re.match( "var_(.*)", var_number_indicator ) 94 | return int(m.group(1)) 95 | 96 | def computation_collapse(cfg): 97 | ''' 98 | Collapse the computation tree in the graph. 99 | 100 | For example, this code: 101 | 102 | a = 5 103 | b = a + 2 104 | c = b * 2 105 | 106 | would be converted to: 107 | 108 | c = (5+2)*2 109 | ''' 110 | def is_writable(ins, k): 111 | ''' 112 | Check if the argument is writable. 113 | ''' 114 | return k in ins and 'w' in ins[k] and ins[k]['w'] 115 | 116 | def is_temp_comp(ins, k): 117 | ''' 118 | Check if the argument is a temporary. 119 | ''' 120 | return k in ins and 'type' in ins[k] and ins[k]['type'] == 'temp' 121 | 122 | def lookup_vars(ins, mem): 123 | ''' 124 | Lookup variables in memory. 125 | ''' 126 | for k in ('src','dest'): 127 | if k in ins and 'op' in ins[k]: 128 | ins[k] = lookup_vars(ins[k], mem) 129 | elif k != 'dest' and k in ins and 'repr' in ins[k] and ins[k]['repr'] in mem: 130 | ins[k] = mem[ins[k]['repr']] 131 | if 'op' in ins and ins['op'] == 'apply': 132 | for i, arg in enumerate(ins['args']): 133 | ins['args'][i] = lookup_vars({'src': arg}, mem)['src'] 134 | return ins 135 | 136 | def collapse_line(line, mem): 137 | ''' 138 | Try to collapse a line. 139 | ''' 140 | ins = line['ins'] 141 | for k in ('dest',): 142 | if k in ins and is_writable(ins, k) and is_temp_comp(ins, k): 143 | key = ins[k]['repr'] 144 | mem[key] = lookup_vars(deepcopy(ins), mem) 145 | line['ins'] = lookup_vars(ins, mem) 146 | 147 | def collapse_vertex(vertex, mem): 148 | ''' 149 | Collapse a vertex of the cfg. 150 | ''' 151 | t, v = vertex 152 | v_type, v_start = t 153 | 154 | if v_type == 'block': 155 | for line in v: 156 | collapse_line(line, mem) 157 | 158 | elif v_type == 'cons': 159 | for block in v: 160 | collapse_vertex(block, mem) 161 | 162 | elif v_type == 'if': 163 | cond, block = v 164 | collapse_vertex(block, deepcopy(mem)) 165 | 166 | elif v_type == 'ifelse': 167 | cond, true, false = v 168 | for block in (true, false): 169 | collapse_vertex(block, deepcopy(mem)) 170 | 171 | elif v_type == 'while': 172 | cond, pre, loop = v 173 | for block in (pre, loop): 174 | collapse_vertex(block, deepcopy(mem)) 175 | 176 | for v in cfg.sortedvertices(): 177 | collapse_vertex(v, {}) 178 | 179 | def cremate(cfg): 180 | ''' 181 | Remove dead instructions. 182 | 183 | Specifically, traverse the code reverse inorder and: 184 | * record every read of a temporary variable 185 | * if you encounter a write to a temporary variable, 186 | erase it from the set 187 | 188 | This way record every necessary write to a temporary variable 189 | and remove the unnecessary ones. 190 | ''' 191 | 192 | def get_read_arg(arg): 193 | if 'op' in arg: 194 | return get_read(arg) 195 | if arg['r'] and arg['type'] == 'temp': 196 | return {arg['repr']} 197 | 198 | return set() 199 | 200 | def get_read(ins): 201 | ''' 202 | Get the set of variables read from in the instruction. 203 | ''' 204 | r = set() 205 | if 'src' in ins: 206 | r |= get_read_arg(ins['src']) 207 | return r 208 | 209 | def get_written(ins): 210 | ''' 211 | Get the set of variables written to in the instruction. 212 | ''' 213 | w = set() 214 | if 'dest' in ins and ins['dest']['w'] and ins['dest']['type'] == 'temp': 215 | w |= {ins['dest']['repr']} 216 | return w 217 | 218 | def variable_declarations(cfg): 219 | ''' 220 | Finds information about used variables and adds the declarations for them. 221 | ''' 222 | 223 | def consume_block(block, reads_in): 224 | ''' 225 | Try to consume lines in a block of code. 226 | ''' 227 | reads = deepcopy(reads_in) 228 | t, v = block 229 | v_type, v_start = t 230 | 231 | if v_type == 'if': 232 | cond, true = v 233 | return consume_block(true, reads) 234 | 235 | if v_type == 'while': 236 | cond, pre, loop = v 237 | reads = consume_block(loop, reads) 238 | return consume_block(pre, reads) 239 | 240 | if v_type == 'cons': 241 | for b in reversed(v): 242 | reads = consume_block(b, reads) 243 | 244 | return reads 245 | 246 | if v_type == 'ifelse': 247 | cond, true, false = v 248 | reads_true = consume_block(true, reads) 249 | reads_false = consume_block(false, reads) 250 | return reads_true | reads_false 251 | 252 | for line in reversed(block[1]): 253 | w = get_written(line['ins']) 254 | r = get_read(line['ins']) 255 | if w <= reads: 256 | pass 257 | else: 258 | line['display'] = False 259 | reads -= w 260 | reads |= r 261 | 262 | return reads 263 | 264 | vars = set() 265 | for vertex in list(cfg.sortedvertices()): 266 | vars |= consume_block(vertex, set()) 267 | 268 | return vars 269 | 270 | def new_var_name(): 271 | ''' 272 | Generator of new variable names. 273 | ''' 274 | for n in count(0): 275 | yield "var_{0}".format(n) 276 | 277 | def new_temp_name(): 278 | ''' 279 | Generator of new temporary variable names. 280 | ''' 281 | for n in count(0): 282 | yield "temp_{0}".format(n) 283 | 284 | def add_declarations(cfg, var_number, temps_used): 285 | to_declare = temps_used 286 | for i in range(0, var_number): 287 | to_declare |= set(["var_{0}".format(i)]) 288 | 289 | #print(to_declare) 290 | 291 | def decompile_function(asm, labels, name, symbols): 292 | ''' 293 | Decompile a function. 294 | ''' 295 | signature = infer_signature(asm) 296 | 297 | cfg = control_flow_graph(asm, labels, name) 298 | symbols_rev = {symbols[s]['start']: s for s in symbols} 299 | function_calls.fold(cfg, symbols_rev) 300 | 301 | var_number = variable_inference(cfg) 302 | computation_collapse(cfg) 303 | temps_used = cremate(cfg) 304 | add_declarations(cfg, var_number, temps_used) 305 | 306 | return cfg, signature 307 | 308 | def decompile_functions(functions, symbols): 309 | ''' 310 | Decompile all functions. 311 | ''' 312 | labels = get_labels(functions) 313 | 314 | decompiled_functions = {} 315 | 316 | for name, symbol in functions.items(): 317 | decompiled_functions[name] = decompile_function(functions[name], labels, name, symbols) 318 | 319 | return decompiled_functions 320 | -------------------------------------------------------------------------------- /src/disassemble.py: -------------------------------------------------------------------------------- 1 | import disassemblers.x64 2 | 3 | def disassemble(buf, virt, sections, binary, arch='x64'): 4 | archs = { 5 | 'x64': disassemblers.x64 6 | 7 | } 8 | return archs[arch].disassemble(buf, virt, sections, binary) 9 | -------------------------------------------------------------------------------- /src/disassemblers/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/src/disassemblers/__init__.py -------------------------------------------------------------------------------- /src/disassemblers/libdisassemble/LICENSE: -------------------------------------------------------------------------------- 1 | GNU LESSER GENERAL PUBLIC LICENSE 2 | Version 2.1, February 1999 3 | 4 | Copyright (C) 1991, 1999 Free Software Foundation, Inc. 5 | 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 6 | Everyone is permitted to copy and distribute verbatim copies 7 | of this license document, but changing it is not allowed. 8 | 9 | [This is the first released version of the Lesser GPL. It also counts 10 | as the successor of the GNU Library Public License, version 2, hence 11 | the version number 2.1.] 12 | 13 | Preamble 14 | 15 | The licenses for most software are designed to take away your 16 | freedom to share and change it. By contrast, the GNU General Public 17 | Licenses are intended to guarantee your freedom to share and change 18 | free software--to make sure the software is free for all its users. 19 | 20 | This license, the Lesser General Public License, applies to some 21 | specially designated software packages--typically libraries--of the 22 | Free Software Foundation and other authors who decide to use it. You 23 | can use it too, but we suggest you first think carefully about whether 24 | this license or the ordinary General Public License is the better 25 | strategy to use in any particular case, based on the explanations below. 26 | 27 | When we speak of free software, we are referring to freedom of use, 28 | not price. Our General Public Licenses are designed to make sure that 29 | you have the freedom to distribute copies of free software (and charge 30 | for this service if you wish); that you receive source code or can get 31 | it if you want it; that you can change the software and use pieces of 32 | it in new free programs; and that you are informed that you can do 33 | these things. 34 | 35 | To protect your rights, we need to make restrictions that forbid 36 | distributors to deny you these rights or to ask you to surrender these 37 | rights. These restrictions translate to certain responsibilities for 38 | you if you distribute copies of the library or if you modify it. 39 | 40 | For example, if you distribute copies of the library, whether gratis 41 | or for a fee, you must give the recipients all the rights that we gave 42 | you. You must make sure that they, too, receive or can get the source 43 | code. If you link other code with the library, you must provide 44 | complete object files to the recipients, so that they can relink them 45 | with the library after making changes to the library and recompiling 46 | it. And you must show them these terms so they know their rights. 47 | 48 | We protect your rights with a two-step method: (1) we copyright the 49 | library, and (2) we offer you this license, which gives you legal 50 | permission to copy, distribute and/or modify the library. 51 | 52 | To protect each distributor, we want to make it very clear that 53 | there is no warranty for the free library. Also, if the library is 54 | modified by someone else and passed on, the recipients should know 55 | that what they have is not the original version, so that the original 56 | author's reputation will not be affected by problems that might be 57 | introduced by others. 58 | 59 | Finally, software patents pose a constant threat to the existence of 60 | any free program. We wish to make sure that a company cannot 61 | effectively restrict the users of a free program by obtaining a 62 | restrictive license from a patent holder. Therefore, we insist that 63 | any patent license obtained for a version of the library must be 64 | consistent with the full freedom of use specified in this license. 65 | 66 | Most GNU software, including some libraries, is covered by the 67 | ordinary GNU General Public License. This license, the GNU Lesser 68 | General Public License, applies to certain designated libraries, and 69 | is quite different from the ordinary General Public License. We use 70 | this license for certain libraries in order to permit linking those 71 | libraries into non-free programs. 72 | 73 | When a program is linked with a library, whether statically or using 74 | a shared library, the combination of the two is legally speaking a 75 | combined work, a derivative of the original library. The ordinary 76 | General Public License therefore permits such linking only if the 77 | entire combination fits its criteria of freedom. The Lesser General 78 | Public License permits more lax criteria for linking other code with 79 | the library. 80 | 81 | We call this license the "Lesser" General Public License because it 82 | does Less to protect the user's freedom than the ordinary General 83 | Public License. It also provides other free software developers Less 84 | of an advantage over competing non-free programs. These disadvantages 85 | are the reason we use the ordinary General Public License for many 86 | libraries. However, the Lesser license provides advantages in certain 87 | special circumstances. 88 | 89 | For example, on rare occasions, there may be a special need to 90 | encourage the widest possible use of a certain library, so that it becomes 91 | a de-facto standard. To achieve this, non-free programs must be 92 | allowed to use the library. A more frequent case is that a free 93 | library does the same job as widely used non-free libraries. In this 94 | case, there is little to gain by limiting the free library to free 95 | software only, so we use the Lesser General Public License. 96 | 97 | In other cases, permission to use a particular library in non-free 98 | programs enables a greater number of people to use a large body of 99 | free software. For example, permission to use the GNU C Library in 100 | non-free programs enables many more people to use the whole GNU 101 | operating system, as well as its variant, the GNU/Linux operating 102 | system. 103 | 104 | Although the Lesser General Public License is Less protective of the 105 | users' freedom, it does ensure that the user of a program that is 106 | linked with the Library has the freedom and the wherewithal to run 107 | that program using a modified version of the Library. 108 | 109 | The precise terms and conditions for copying, distribution and 110 | modification follow. Pay close attention to the difference between a 111 | "work based on the library" and a "work that uses the library". The 112 | former contains code derived from the library, whereas the latter must 113 | be combined with the library in order to run. 114 | 115 | GNU LESSER GENERAL PUBLIC LICENSE 116 | TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 117 | 118 | 0. This License Agreement applies to any software library or other 119 | program which contains a notice placed by the copyright holder or 120 | other authorized party saying it may be distributed under the terms of 121 | this Lesser General Public License (also called "this License"). 122 | Each licensee is addressed as "you". 123 | 124 | A "library" means a collection of software functions and/or data 125 | prepared so as to be conveniently linked with application programs 126 | (which use some of those functions and data) to form executables. 127 | 128 | The "Library", below, refers to any such software library or work 129 | which has been distributed under these terms. A "work based on the 130 | Library" means either the Library or any derivative work under 131 | copyright law: that is to say, a work containing the Library or a 132 | portion of it, either verbatim or with modifications and/or translated 133 | straightforwardly into another language. (Hereinafter, translation is 134 | included without limitation in the term "modification".) 135 | 136 | "Source code" for a work means the preferred form of the work for 137 | making modifications to it. For a library, complete source code means 138 | all the source code for all modules it contains, plus any associated 139 | interface definition files, plus the scripts used to control compilation 140 | and installation of the library. 141 | 142 | Activities other than copying, distribution and modification are not 143 | covered by this License; they are outside its scope. The act of 144 | running a program using the Library is not restricted, and output from 145 | such a program is covered only if its contents constitute a work based 146 | on the Library (independent of the use of the Library in a tool for 147 | writing it). Whether that is true depends on what the Library does 148 | and what the program that uses the Library does. 149 | 150 | 1. You may copy and distribute verbatim copies of the Library's 151 | complete source code as you receive it, in any medium, provided that 152 | you conspicuously and appropriately publish on each copy an 153 | appropriate copyright notice and disclaimer of warranty; keep intact 154 | all the notices that refer to this License and to the absence of any 155 | warranty; and distribute a copy of this License along with the 156 | Library. 157 | 158 | You may charge a fee for the physical act of transferring a copy, 159 | and you may at your option offer warranty protection in exchange for a 160 | fee. 161 | 162 | 2. You may modify your copy or copies of the Library or any portion 163 | of it, thus forming a work based on the Library, and copy and 164 | distribute such modifications or work under the terms of Section 1 165 | above, provided that you also meet all of these conditions: 166 | 167 | a) The modified work must itself be a software library. 168 | 169 | b) You must cause the files modified to carry prominent notices 170 | stating that you changed the files and the date of any change. 171 | 172 | c) You must cause the whole of the work to be licensed at no 173 | charge to all third parties under the terms of this License. 174 | 175 | d) If a facility in the modified Library refers to a function or a 176 | table of data to be supplied by an application program that uses 177 | the facility, other than as an argument passed when the facility 178 | is invoked, then you must make a good faith effort to ensure that, 179 | in the event an application does not supply such function or 180 | table, the facility still operates, and performs whatever part of 181 | its purpose remains meaningful. 182 | 183 | (For example, a function in a library to compute square roots has 184 | a purpose that is entirely well-defined independent of the 185 | application. Therefore, Subsection 2d requires that any 186 | application-supplied function or table used by this function must 187 | be optional: if the application does not supply it, the square 188 | root function must still compute square roots.) 189 | 190 | These requirements apply to the modified work as a whole. If 191 | identifiable sections of that work are not derived from the Library, 192 | and can be reasonably considered independent and separate works in 193 | themselves, then this License, and its terms, do not apply to those 194 | sections when you distribute them as separate works. But when you 195 | distribute the same sections as part of a whole which is a work based 196 | on the Library, the distribution of the whole must be on the terms of 197 | this License, whose permissions for other licensees extend to the 198 | entire whole, and thus to each and every part regardless of who wrote 199 | it. 200 | 201 | Thus, it is not the intent of this section to claim rights or contest 202 | your rights to work written entirely by you; rather, the intent is to 203 | exercise the right to control the distribution of derivative or 204 | collective works based on the Library. 205 | 206 | In addition, mere aggregation of another work not based on the Library 207 | with the Library (or with a work based on the Library) on a volume of 208 | a storage or distribution medium does not bring the other work under 209 | the scope of this License. 210 | 211 | 3. You may opt to apply the terms of the ordinary GNU General Public 212 | License instead of this License to a given copy of the Library. To do 213 | this, you must alter all the notices that refer to this License, so 214 | that they refer to the ordinary GNU General Public License, version 2, 215 | instead of to this License. (If a newer version than version 2 of the 216 | ordinary GNU General Public License has appeared, then you can specify 217 | that version instead if you wish.) Do not make any other change in 218 | these notices. 219 | 220 | Once this change is made in a given copy, it is irreversible for 221 | that copy, so the ordinary GNU General Public License applies to all 222 | subsequent copies and derivative works made from that copy. 223 | 224 | This option is useful when you wish to copy part of the code of 225 | the Library into a program that is not a library. 226 | 227 | 4. You may copy and distribute the Library (or a portion or 228 | derivative of it, under Section 2) in object code or executable form 229 | under the terms of Sections 1 and 2 above provided that you accompany 230 | it with the complete corresponding machine-readable source code, which 231 | must be distributed under the terms of Sections 1 and 2 above on a 232 | medium customarily used for software interchange. 233 | 234 | If distribution of object code is made by offering access to copy 235 | from a designated place, then offering equivalent access to copy the 236 | source code from the same place satisfies the requirement to 237 | distribute the source code, even though third parties are not 238 | compelled to copy the source along with the object code. 239 | 240 | 5. A program that contains no derivative of any portion of the 241 | Library, but is designed to work with the Library by being compiled or 242 | linked with it, is called a "work that uses the Library". Such a 243 | work, in isolation, is not a derivative work of the Library, and 244 | therefore falls outside the scope of this License. 245 | 246 | However, linking a "work that uses the Library" with the Library 247 | creates an executable that is a derivative of the Library (because it 248 | contains portions of the Library), rather than a "work that uses the 249 | library". The executable is therefore covered by this License. 250 | Section 6 states terms for distribution of such executables. 251 | 252 | When a "work that uses the Library" uses material from a header file 253 | that is part of the Library, the object code for the work may be a 254 | derivative work of the Library even though the source code is not. 255 | Whether this is true is especially significant if the work can be 256 | linked without the Library, or if the work is itself a library. The 257 | threshold for this to be true is not precisely defined by law. 258 | 259 | If such an object file uses only numerical parameters, data 260 | structure layouts and accessors, and small macros and small inline 261 | functions (ten lines or less in length), then the use of the object 262 | file is unrestricted, regardless of whether it is legally a derivative 263 | work. (Executables containing this object code plus portions of the 264 | Library will still fall under Section 6.) 265 | 266 | Otherwise, if the work is a derivative of the Library, you may 267 | distribute the object code for the work under the terms of Section 6. 268 | Any executables containing that work also fall under Section 6, 269 | whether or not they are linked directly with the Library itself. 270 | 271 | 6. As an exception to the Sections above, you may also combine or 272 | link a "work that uses the Library" with the Library to produce a 273 | work containing portions of the Library, and distribute that work 274 | under terms of your choice, provided that the terms permit 275 | modification of the work for the customer's own use and reverse 276 | engineering for debugging such modifications. 277 | 278 | You must give prominent notice with each copy of the work that the 279 | Library is used in it and that the Library and its use are covered by 280 | this License. You must supply a copy of this License. If the work 281 | during execution displays copyright notices, you must include the 282 | copyright notice for the Library among them, as well as a reference 283 | directing the user to the copy of this License. Also, you must do one 284 | of these things: 285 | 286 | a) Accompany the work with the complete corresponding 287 | machine-readable source code for the Library including whatever 288 | changes were used in the work (which must be distributed under 289 | Sections 1 and 2 above); and, if the work is an executable linked 290 | with the Library, with the complete machine-readable "work that 291 | uses the Library", as object code and/or source code, so that the 292 | user can modify the Library and then relink to produce a modified 293 | executable containing the modified Library. (It is understood 294 | that the user who changes the contents of definitions files in the 295 | Library will not necessarily be able to recompile the application 296 | to use the modified definitions.) 297 | 298 | b) Use a suitable shared library mechanism for linking with the 299 | Library. A suitable mechanism is one that (1) uses at run time a 300 | copy of the library already present on the user's computer system, 301 | rather than copying library functions into the executable, and (2) 302 | will operate properly with a modified version of the library, if 303 | the user installs one, as long as the modified version is 304 | interface-compatible with the version that the work was made with. 305 | 306 | c) Accompany the work with a written offer, valid for at 307 | least three years, to give the same user the materials 308 | specified in Subsection 6a, above, for a charge no more 309 | than the cost of performing this distribution. 310 | 311 | d) If distribution of the work is made by offering access to copy 312 | from a designated place, offer equivalent access to copy the above 313 | specified materials from the same place. 314 | 315 | e) Verify that the user has already received a copy of these 316 | materials or that you have already sent this user a copy. 317 | 318 | For an executable, the required form of the "work that uses the 319 | Library" must include any data and utility programs needed for 320 | reproducing the executable from it. However, as a special exception, 321 | the materials to be distributed need not include anything that is 322 | normally distributed (in either source or binary form) with the major 323 | components (compiler, kernel, and so on) of the operating system on 324 | which the executable runs, unless that component itself accompanies 325 | the executable. 326 | 327 | It may happen that this requirement contradicts the license 328 | restrictions of other proprietary libraries that do not normally 329 | accompany the operating system. Such a contradiction means you cannot 330 | use both them and the Library together in an executable that you 331 | distribute. 332 | 333 | 7. You may place library facilities that are a work based on the 334 | Library side-by-side in a single library together with other library 335 | facilities not covered by this License, and distribute such a combined 336 | library, provided that the separate distribution of the work based on 337 | the Library and of the other library facilities is otherwise 338 | permitted, and provided that you do these two things: 339 | 340 | a) Accompany the combined library with a copy of the same work 341 | based on the Library, uncombined with any other library 342 | facilities. This must be distributed under the terms of the 343 | Sections above. 344 | 345 | b) Give prominent notice with the combined library of the fact 346 | that part of it is a work based on the Library, and explaining 347 | where to find the accompanying uncombined form of the same work. 348 | 349 | 8. You may not copy, modify, sublicense, link with, or distribute 350 | the Library except as expressly provided under this License. Any 351 | attempt otherwise to copy, modify, sublicense, link with, or 352 | distribute the Library is void, and will automatically terminate your 353 | rights under this License. However, parties who have received copies, 354 | or rights, from you under this License will not have their licenses 355 | terminated so long as such parties remain in full compliance. 356 | 357 | 9. You are not required to accept this License, since you have not 358 | signed it. However, nothing else grants you permission to modify or 359 | distribute the Library or its derivative works. These actions are 360 | prohibited by law if you do not accept this License. Therefore, by 361 | modifying or distributing the Library (or any work based on the 362 | Library), you indicate your acceptance of this License to do so, and 363 | all its terms and conditions for copying, distributing or modifying 364 | the Library or works based on it. 365 | 366 | 10. Each time you redistribute the Library (or any work based on the 367 | Library), the recipient automatically receives a license from the 368 | original licensor to copy, distribute, link with or modify the Library 369 | subject to these terms and conditions. You may not impose any further 370 | restrictions on the recipients' exercise of the rights granted herein. 371 | You are not responsible for enforcing compliance by third parties with 372 | this License. 373 | 374 | 11. If, as a consequence of a court judgment or allegation of patent 375 | infringement or for any other reason (not limited to patent issues), 376 | conditions are imposed on you (whether by court order, agreement or 377 | otherwise) that contradict the conditions of this License, they do not 378 | excuse you from the conditions of this License. If you cannot 379 | distribute so as to satisfy simultaneously your obligations under this 380 | License and any other pertinent obligations, then as a consequence you 381 | may not distribute the Library at all. For example, if a patent 382 | license would not permit royalty-free redistribution of the Library by 383 | all those who receive copies directly or indirectly through you, then 384 | the only way you could satisfy both it and this License would be to 385 | refrain entirely from distribution of the Library. 386 | 387 | If any portion of this section is held invalid or unenforceable under any 388 | particular circumstance, the balance of the section is intended to apply, 389 | and the section as a whole is intended to apply in other circumstances. 390 | 391 | It is not the purpose of this section to induce you to infringe any 392 | patents or other property right claims or to contest validity of any 393 | such claims; this section has the sole purpose of protecting the 394 | integrity of the free software distribution system which is 395 | implemented by public license practices. Many people have made 396 | generous contributions to the wide range of software distributed 397 | through that system in reliance on consistent application of that 398 | system; it is up to the author/donor to decide if he or she is willing 399 | to distribute software through any other system and a licensee cannot 400 | impose that choice. 401 | 402 | This section is intended to make thoroughly clear what is believed to 403 | be a consequence of the rest of this License. 404 | 405 | 12. If the distribution and/or use of the Library is restricted in 406 | certain countries either by patents or by copyrighted interfaces, the 407 | original copyright holder who places the Library under this License may add 408 | an explicit geographical distribution limitation excluding those countries, 409 | so that distribution is permitted only in or among countries not thus 410 | excluded. In such case, this License incorporates the limitation as if 411 | written in the body of this License. 412 | 413 | 13. The Free Software Foundation may publish revised and/or new 414 | versions of the Lesser General Public License from time to time. 415 | Such new versions will be similar in spirit to the present version, 416 | but may differ in detail to address new problems or concerns. 417 | 418 | Each version is given a distinguishing version number. If the Library 419 | specifies a version number of this License which applies to it and 420 | "any later version", you have the option of following the terms and 421 | conditions either of that version or of any later version published by 422 | the Free Software Foundation. If the Library does not specify a 423 | license version number, you may choose any version ever published by 424 | the Free Software Foundation. 425 | 426 | 14. If you wish to incorporate parts of the Library into other free 427 | programs whose distribution conditions are incompatible with these, 428 | write to the author to ask for permission. For software which is 429 | copyrighted by the Free Software Foundation, write to the Free 430 | Software Foundation; we sometimes make exceptions for this. Our 431 | decision will be guided by the two goals of preserving the free status 432 | of all derivatives of our free software and of promoting the sharing 433 | and reuse of software generally. 434 | 435 | NO WARRANTY 436 | 437 | 15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO 438 | WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW. 439 | EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR 440 | OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY 441 | KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE 442 | IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 443 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE 444 | LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME 445 | THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 446 | 447 | 16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN 448 | WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY 449 | AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU 450 | FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR 451 | CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE 452 | LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING 453 | RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A 454 | FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF 455 | SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH 456 | DAMAGES. 457 | 458 | END OF TERMS AND CONDITIONS 459 | 460 | 461 | -------------------------------------------------------------------------------- /src/disassemblers/libdisassemble/README: -------------------------------------------------------------------------------- 1 |  Immunity libdisassembly v2.0 2 | ~~~~~~~~ ~~~~~~~~~~~~~~ ~~~~ 3 | http://www.immunitysec.com 4 | June 5, 2007 5 | 6 | 7 | Credits to atlas and Matthew Carpenter 8 | for the 2.0 release! 9 | 10 | 11 | Libdisassembly is simply a python library for disassembling x86 opcodes. It has been made for Immunity's PDB Project (a vulnerability development focused debugger), and is partially based on mammon libdisasm opcode list (http://www.eccentrix.com/members/mammon/). There is still a lot of work to do with the Metadata, but the library tries to return as much information it can get off of an opcode. 12 | 13 | USAGE: 14 | 15 | ie: 16 | $ objdump -x /bin/cat | grep .text 17 | 11 .text 00001e48 08048b00 08048b00 00000b00 2**4 18 | $ ./disassemble.py /bin/cat 0xb00 0x1e48 19 | 20 | Disassembling file /bin/cat at offset: 0x2816 21 | 00002816: mov 0x8(%ebp),%edx 22 | 00002819: mov 0xc(%ebp),%eax 23 | 0000281C: mov %edx,(%esp,1) 24 | [...] 25 | 26 | LIBRARY USAGE: 27 | Getting the string information of an opcode is really simple: 28 | 29 | from disassemble import * 30 | data="\xeb\xfe" 31 | p=Opcode(data) 32 | print p.printOpcode("AT&T") 33 | 34 | The output would be: 35 | jmp 0xfffffffe 36 | 37 | If for example, we want to put that information in a fancy GUI, we could request for a string separated in mnemonic, source, dest (or dest, source) with getOpcode(FORMAT). 38 | 39 | p.getOpcode("AT&T") 40 | 41 | and it will return a table: 42 | ['jmp', '0xfffffffe'] 43 | 44 | Metadata info: 45 | ~~~~~~~~ ~~~~~ 46 | Every opcode has 3 variables: opcode, source, dest. 47 | 48 | - opcode, is the string of the mnenonic 49 | - source, is an object or None 50 | - dest, is an object or None 51 | 52 | There are 3 types of objects, Register, Address, SIB and Expression. There are a couple of common methods you can use: 53 | 54 | - object.printOpcode(FORMAT), the string representation of the object. 55 | - object.getSize(), size 56 | - object.getType, information of the opcode table. 57 | - object.getSFlag, object permission ("R"ead, "W"rite, e"X"ecute) 58 | - etc (You can browse the source to get more information) 59 | 60 | -------------------------------------------------------------------------------- /src/disassemblers/libdisassemble/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/src/disassemblers/libdisassemble/__init__.py -------------------------------------------------------------------------------- /src/disassemblers/libdisassemble/disassemble.py: -------------------------------------------------------------------------------- 1 | #! /usr/bin/env python 2 | 3 | 4 | # Immunity libdisassemble 5 | # 6 | # Most of the functions are ported from the great libdisassm since we 7 | # we are using their opcode map. 8 | 9 | # TODO: 10 | # - Fix the getSize(), doesn't work well with all the opcodes 11 | # - Enhance the metadata info with more information on opcode. 12 | # i.e. we need a way to know if an address is an immediate, a relative offset, etc 13 | # - Fix the jmp (SIB) opcode in at&t that it has different output that the others. 14 | # - support all the PREFIX* 15 | 16 | # NOTE: This is less than a week work, so it might be full of bugs (we love bugs!) 17 | # 18 | # Any question, comments, hate mail: dave@immunitysec.com 19 | 20 | 21 | # 22 | # This library is free software; you can redistribute it and/or 23 | # modify it under the terms of the GNU Lesser General Public 24 | # License as published by the Free Software Foundation; either 25 | # version 2.1 of the License, or (at your option) any later version. 26 | # 27 | # This library is distributed in the hope that it will be useful, 28 | # but WITHOUT ANY WARRANTY; without even the implied warranty of 29 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 30 | # Lesser General Public License for more details. 31 | # 32 | # You should have received a copy of the GNU Lesser General Public 33 | # License along with this library; if not, write to the Free Software 34 | # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 35 | 36 | # This code largely copyright Immunity, Inc (2004), parts 37 | # copyright mammon and used under the LGPL by permission 38 | 39 | import disassemblers.libdisassemble.opcode86 as opcode86 40 | import struct 41 | from sys import * 42 | 43 | table86=opcode86.tables86 44 | OP_PERM_MASK = 0x0000007 45 | OP_TYPE_MASK = 0x0000F00 46 | OP_MOD_MASK = 0x00FF000 47 | OP_SEG_MASK = 0x00F0000 48 | OP_SIZE_MASK = 0x0F000000 49 | 50 | 51 | class Mode: 52 | def __init__(self, type, mode=32): 53 | self.type = type #type & 0x700 54 | #self.flag = type & 0x7 55 | self.flag = type & OP_PERM_MASK 56 | self.length = 0 57 | self.mode = mode 58 | 59 | # format "AT&T" or "INTEL" 60 | def printOpcode(self, format, eip=0): 61 | return "Not available" 62 | 63 | def getType(self): 64 | return self.type 65 | 66 | def getSize(self): 67 | return self.length 68 | 69 | def getFlag(self): 70 | return self.flag 71 | 72 | def getSFlag(self): 73 | return ("R", "W", "X")[self.flag/2] 74 | 75 | def getOpSize(self): 76 | return (self.type & OP_SIZE_MASK) 77 | 78 | def getAddrMeth(self): 79 | return (self.type & opcode86.ADDRMETH_MASK) 80 | 81 | class Register(Mode): 82 | def __init__(self, regndx, type=opcode86.OP_REG): 83 | Mode.__init__(self, type) 84 | #print regndx 85 | (self.name, self.detail, self.length)=opcode86.regs[regndx] 86 | 87 | def printOpcode(self, format="INTEL", eip=0): 88 | if format == "INTEL": 89 | return self.name 90 | else: 91 | return "%%%s" % self.name 92 | 93 | def getName(self): 94 | return self.name 95 | 96 | class Address(Mode): 97 | def __init__(self, data, length, type=opcode86.ADDEXP_DISP_OFFSET, signed=1, relative = None): 98 | Mode.__init__(self, type) 99 | 100 | self.signed=signed 101 | self.length = length 102 | #print "'%s' %d %x, %s"%(data, length, type, relative) 103 | fmt = "<" 104 | if self.signed: 105 | fmt += ("b", "h", "l")[length//2] 106 | else: 107 | fmt += ("B", "H", "L")[length//2] 108 | 109 | if (self.getAddrMeth() == opcode86.ADDRMETH_A): 110 | fmt += "H" 111 | length += 2 112 | self.value, self.segment = struct.unpack(fmt, data[:length]) 113 | else: 114 | self.value, = struct.unpack(fmt, data[:length]) 115 | self.segment = None 116 | 117 | self.relative = relative 118 | 119 | 120 | def printOpcode(self, format="INTEL", eip=0, exp=0): 121 | value = self.value 122 | segment = self.segment 123 | 124 | if (self.relative): 125 | value += eip 126 | 127 | if format == "INTEL": 128 | tmp="" 129 | if (segment): 130 | tmp += "0x%04x:"%(segment) 131 | if self.signed: 132 | if value < 0: 133 | return "%s-0x%x" % (tmp, value * -1) 134 | return "%s0x%x" % (tmp,self.value) 135 | else: 136 | #if self.length == 4 or not self.signed: 137 | return "%s0x%x" % (tmp,self.value) 138 | #else: 139 | 140 | else: 141 | pre="" 142 | #if self.getAddrMeth() == opcode86.ADDRMETH_E and not exp: 143 | if (self.getAddrMeth() == opcode86.ADDRMETH_I or self.getAddrMeth() == opcode86.ADDRMETH_A or self.type & opcode86.OP_IMM) and not exp: 144 | pre+="$" 145 | if segment: 146 | pre = "$0x%04x:%s"%(segment,pre) 147 | if (value < 0): 148 | if (self.signed): 149 | return "%s0x%0x" % (pre, ((1< 0 and tmp: 179 | tmp+="+" 180 | tmp += self.disp.printOpcode(format, eip, 1) 181 | pre="" 182 | optype=self.getOpSize() 183 | 184 | addr_meth=self.getAddrMeth() 185 | if addr_meth == opcode86.ADDRMETH_E: 186 | if optype == opcode86.OPTYPE_b: 187 | pre="BYTE PTR" 188 | elif optype== opcode86.OPTYPE_w: 189 | pre="WORD PTR" 190 | else : 191 | pre="DWORD PTR" 192 | tmp="%s [%s]" % (pre, tmp) 193 | else: 194 | if self.base: 195 | tmp+="(%s)" % self.base.printOpcode(format, eip) 196 | if self.disp: 197 | tmp= "%s%s" % (self.disp.printOpcode(format, eip, 1), tmp) 198 | #tmp="Not available" 199 | return tmp 200 | class SIB(Mode): 201 | def __init__(self, scale, base, index): 202 | self.scale = scale 203 | self.base = base 204 | self.index = index 205 | def printOpcode(self, format="INTEL", eip=0): 206 | tmp="" 207 | if format == "INTEL": 208 | if self.base: 209 | tmp+="%s" % self.base.printOpcode(format, eip) 210 | if self.scale > 1: 211 | tmp+= "*%d" % self.scale 212 | if self.index: 213 | if tmp: 214 | tmp+="+" 215 | tmp+="%s" % self.index.printOpcode(format, eip) 216 | else: 217 | if self.base: 218 | tmp+="%s" % self.base.printOpcode(format, eip) 219 | if self.index: 220 | #if tmp: 221 | #tmp+="," 222 | tmp += ", %s" % self.index.printOpcode(format, eip) 223 | if self.scale: 224 | if (self.scale > 1 or self.index): 225 | tmp+=", %d" % self.scale 226 | 227 | return tmp 228 | return tmp 229 | 230 | class Prefix: 231 | def __init__(self, ndx, ptr): 232 | self.ptr = ptr 233 | self.type = opcode86.prefix_table[ndx] 234 | 235 | def getType(self): 236 | return self.type 237 | 238 | def getName(self): 239 | if self.ptr[6]: 240 | return self.ptr[6] 241 | else: 242 | return "" 243 | 244 | class Opcode: 245 | def __init__(self, data, mode=32): 246 | self.length = 0 247 | self.mode = mode 248 | if mode == 64: 249 | self.addr_size = 4 250 | else: 251 | self.addr_size = mode/8 # 32-bit mode = 4 bytes. 16-bit mode = 2 bytes 252 | self.data = data 253 | self.off = 0 254 | self.source = "" 255 | self.dest = "" 256 | self.aux = "" 257 | self.prefix = [] 258 | self.parse(table86[0], self.off) 259 | 260 | def getOpcodetype(self): 261 | return self.opcodetype 262 | 263 | def parse(self, table, off): 264 | """ 265 | Opcode.parse() is the core logic of libdisassemble. It recurses through the supplied bytes digesting prefixes and opcodes, and then handles operands. 266 | """ 267 | try: ## Crash Gracefully with a "invalid" opcode 268 | self.addr_size = 4 269 | ndx = self.data[off] 270 | 271 | ### This next line slices and dices the opcode to make it fit correctly into the current lookup table. 272 | # 273 | # byte min shift mask 274 | # (tbl_0F, 0, 0xff, 0, 0xff), 275 | # (tbl_80, 3, 0x07, 0, 0xff), 276 | # 277 | # simple subtraction 278 | # shift bits right (eg. >> 4 makes each line in the table valid for 16 numbers... ie 0xc0-0xcf are all one entry in the table) 279 | # mask part of the byte (eg. & 0x7 only makes use of the 00000111 bits...) 280 | if (ndx > table[4]): 281 | table = table86[table[5]] # if the opcode byte falls outside the bounds of accepted values, use the table pointed to as table[5] 282 | ndx = ( (ndx - table[3]) >> table[1]) & table[2] 283 | ptr = table[0][ndx] # index from table 284 | 285 | if ptr[1] == opcode86.INSTR_PREFIX or (ptr[1] & opcode86.INSTR_PREFIX64 and self.mode == 64): 286 | # You can have more than one prefix (!?) 287 | if ptr[0] != 0 and len(self.data) > off and self.data[off+1] == 0x0F: 288 | self.parse(table86[ptr[0]], off+2) # eg. 660Fxx, F20Fxx, F30Fxx, etc... 289 | else: 290 | self.prefix.append( Prefix(ndx, ptr) ) 291 | self.parse(table, off+1) # parse next instruction 292 | 293 | return 294 | if ptr[0] != 0: 295 | # > 1 byte length opcode 296 | self.parse(table86[ptr[0]], off+1) 297 | return 298 | 299 | ### End Recursion, we hit a leaf node. 300 | 301 | self.opcode = ptr[6] 302 | self.opcodetype = ptr[1] 303 | self.cpu = ptr[5] 304 | self.off = off + 1 # This instruction 305 | 306 | if table[2] != 0xff: # Is this byte split between opcode and operand? If so, let's not skip over this byte quite yet... 307 | self.off-=1 308 | #print >>stderr,(" opcode = %s\n opcodetype = %x\n cpu = %x\n off = %d"%(ptr[6], ptr[1], ptr[5], off+1)) 309 | 310 | n_bytes=0 311 | # src dst aux 312 | values=['', '', '' ] 313 | r = [False]*3 314 | w = [False]*3 315 | #print self.off 316 | 317 | for a in range(2, 5): 318 | ret = (0, None) 319 | 320 | tmp =ptr[a] 321 | addr_meth = tmp & opcode86.ADDRMETH_MASK; 322 | if addr_meth == opcode86.OP_REG: 323 | # what do i supposed to do? 324 | pass 325 | 326 | operand_size = self.get_operand_size(tmp) 327 | #print operand_size 328 | if operand_size == 1: 329 | genreg = opcode86.REG_BYTE_OFFSET 330 | elif operand_size == 2: 331 | genreg = opcode86.REG_WORD_OFFSET 332 | else: 333 | genreg= opcode86.REG_DWORD_OFFSET 334 | 335 | # Begin hashing on the ADDRMETH for this operand. This should determine the number of bytes to advance in the data. 336 | if addr_meth == opcode86.ADDRMETH_E: 337 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, genreg, self.addr_size, tmp) 338 | 339 | elif addr_meth == opcode86.ADDRMETH_M: 340 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, genreg, self.addr_size, tmp) 341 | 342 | elif addr_meth == opcode86.ADDRMETH_N: 343 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, opcode86.REG_MMX_OFFSET, self.addr_size, tmp) 344 | 345 | elif addr_meth == opcode86.ADDRMETH_Q: 346 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, opcode86.REG_MMX_OFFSET, self.addr_size, tmp) 347 | 348 | elif addr_meth == opcode86.ADDRMETH_R: 349 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, genreg, self.addr_size, tmp) 350 | 351 | elif addr_meth == opcode86.ADDRMETH_W: 352 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, opcode86.REG_SIMD_OFFSET, self.addr_size, tmp) 353 | 354 | elif addr_meth == opcode86.ADDRMETH_C: 355 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, opcode86.REG_CTRL_OFFSET, self.addr_size, tmp) 356 | 357 | elif addr_meth == opcode86.ADDRMETH_D: 358 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, opcode86.REG_DEBUG_OFFSET, self.addr_size, tmp) 359 | 360 | elif addr_meth == opcode86.ADDRMETH_G: 361 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, genreg, self.addr_size, tmp) 362 | 363 | elif addr_meth == opcode86.ADDRMETH_P: 364 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, opcode86.REG_MMX_OFFSET, self.addr_size, tmp) 365 | 366 | elif addr_meth == opcode86.ADDRMETH_S: 367 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, opcode86.REG_SEG_OFFSET, self.addr_size, tmp) 368 | 369 | #elif addr_meth == opcode86.ADDRMETH_T: #TEST REGISTERS?: 370 | #ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, opcode86.REG_TEST_OFFSET, self.addr_size, tmp) 371 | 372 | elif addr_meth == opcode86.ADDRMETH_U: 373 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_EA, opcode86.REG_SIMD_OFFSET, self.addr_size, tmp) 374 | 375 | elif addr_meth == opcode86.ADDRMETH_V: 376 | ret=self.get_modrm(self.data[self.off:], opcode86.MODRM_reg, opcode86.REG_SIMD_OFFSET, self.addr_size, tmp) 377 | 378 | elif addr_meth == opcode86.ADDRMETH_A: 379 | ret= (self.addr_size, Address(self.data[self.off:], self.addr_size, tmp, signed=0)) 380 | 381 | elif addr_meth == opcode86.ADDRMETH_F: 382 | # eflags, so what? 383 | pass 384 | 385 | elif addr_meth == opcode86.ADDRMETH_I: 386 | if tmp & opcode86.OP_SIGNED: 387 | ret = (operand_size, Address( self.data[self.off+n_bytes:], operand_size, tmp)) 388 | #ret = (self.addr_size, Address( self.data[self.off+bytes:], operand_size, tmp)) 389 | else: 390 | ret = (operand_size, Address( self.data[self.off+n_bytes:], operand_size,tmp, signed=0)) 391 | #ret = (self.addr_size, Address( self.data[self.off+bytes:], operand_size,tmp, signed=0)) 392 | 393 | elif addr_meth == opcode86.ADDRMETH_J: 394 | ret = (operand_size, Address( self.data[self.off+n_bytes:], operand_size, tmp, signed=1, relative=True)) 395 | #ret = (self.addr_size, Address( self.data[self.off+bytes:], operand_size, tmp, signed=1, relative=True)) 396 | 397 | elif addr_meth == opcode86.ADDRMETH_O: 398 | ret = (self.addr_size, Address( self.data[self.off:], self.addr_size, tmp, signed=0)) 399 | 400 | elif addr_meth == opcode86.ADDRMETH_X: 401 | ret = (0, Register(6+opcode86.REG_DWORD_OFFSET, tmp)) 402 | 403 | elif addr_meth == opcode86.ADDRMETH_Y: 404 | ret = (0, Register(7+opcode86.REG_DWORD_OFFSET, tmp)) 405 | 406 | else: 407 | if tmp & opcode86.OP_REG: 408 | regoff = 0 409 | if self.mode == 64 and self.opcodetype in [opcode86.INS_PUSH, opcode86.INS_POP]: 410 | regoff = opcode86.REG_QWORD_OFFSET - opcode86.REG_DWORD_OFFSET 411 | if self.rex('w'): 412 | regoff -= 16 413 | if self.rex('b'): 414 | regoff += 8 415 | ret = (0, Register(ptr[5+a]+regoff, tmp)) 416 | elif tmp & opcode86.OP_IMM: 417 | ret = (0, Address(bytes([ptr[5+a]]), 1, signed=0)) 418 | else: 419 | ret= (0, None) 420 | if ret[1]: 421 | if isinstance(ret[1], Expression): 422 | ret[1].setPsize(operand_size) 423 | values[a-2]=ret[1] 424 | r[a-2] = (tmp & opcode86.OP_R) != 0 425 | w[a-2] = (tmp & opcode86.OP_W) != 0 426 | n_bytes += ret[0] 427 | 428 | self.source = values[0] 429 | self.dest = values[1] 430 | self.aux = values[2] 431 | self.r = r 432 | self.w = w 433 | 434 | self.off += n_bytes 435 | #self.data = self.data[:self.off] 436 | except IndexError: 437 | output = "" 438 | for i in range(len(self.data)): 439 | output += "%02x"%self.data[i] 440 | 441 | print (("Error Parsing Opcode - Data: %s\t Offset: 0x%x"%(output,self.off)), file=stderr) 442 | 443 | x,y,z = exc_info() 444 | excepthook(x,y,z) 445 | 446 | def getSize(self): 447 | return self.off 448 | 449 | def get_operand_size(self, opflag): 450 | """ 451 | get_operand_size() gets the operand size, not the address-size or the size of the opcode itself. 452 | But it's also bastardized, because it manipulates self.addr_size at the top 453 | """ 454 | size=self.mode / 8 #initial setting (4 for 32-bit mode) 455 | if self.mode == 64: 456 | size = 4 457 | 458 | flag = opflag & opcode86.OPTYPE_MASK 459 | 460 | #print "flag=%x mode=%d"%(flag,self.mode) 461 | if (flag in opcode86.OPERSIZE.keys()): # lookup the value in the table 462 | size = opcode86.OPERSIZE[flag][size >> 2] 463 | 464 | 465 | for a in self.prefix: 466 | if a.getType() & opcode86.PREFIX_OP_SIZE and size > 2: 467 | size = 2 468 | if a.getType() & opcode86.PREFIX_ADDR_SIZE: 469 | # this simply swaps between 16- to 32-bit (default is determined on a "system-wide" level. This will require changing for 64-bit mode 470 | if (self.addr_size == 2): 471 | self.addr_size = 4 472 | else: 473 | self.addr_size = 2 474 | 475 | return size 476 | 477 | """ 478 | ### THIS IS THE OLD LIBDISASSEMBLE CODE... 479 | #print flag 480 | if flag == opcode86.OPTYPE_c: 481 | size = (1,2)[size==4] 482 | elif (flag == opcode86.OPTYPE_a) or (flag == opcode86.OPTYPE_v) or (flag == opcode86.OPTYPE_z): 483 | size = (2,4)[size==4] 484 | elif flag == opcode86.OPTYPE_p: 485 | size = (4,6)[size==4] 486 | elif flag == opcode86.OPTYPE_b: 487 | size = 1 488 | elif flag == opcode86.OPTYPE_w: 489 | size = 2 490 | elif flag == opcode86.OPTYPE_d: 491 | size = 4 492 | elif flag & opcode86.OPTYPE_s: 493 | size = 6 494 | elif flag & opcode86.OPTYPE_q: 495 | size = 8 496 | # - a lot more to add 497 | """ 498 | 499 | def get_reg(self, regtable, num): 500 | return regtable[num] 501 | 502 | def get_sib(self, data, mod): 503 | count = 1 504 | sib = data[0] 505 | s=None 506 | #print "SIB: %s" % hex(ord(data[0])) 507 | 508 | scale = (sib >> 6) & 0x3 # XX 509 | index = (sib & 56) >>3 # XXX 510 | base = sib & 0x7 # XXX 511 | 512 | base2 = None 513 | index2= None 514 | #print base, index, scale 515 | # Especial case 516 | if base == 5 and not mod: 517 | base2 = Address(data[1:], 4) 518 | count += 4 519 | else: 520 | if self.rex('b'): 521 | base += 8 522 | base2 = Register(base + 16) 523 | 524 | index2=None 525 | # Yeah, i know, this is really ugly 526 | if index != 4: # ESP 527 | if self.rex('x'): 528 | index += 8 529 | index2=Register( index + 16) 530 | else: 531 | scale = 0 532 | s= SIB( 1<> 6) & 0x3 # XX 543 | reg = (modrm >> 3) & 0x7 # XXX 544 | rm = modrm & 0x7 # XXX 545 | 546 | result = None 547 | disp = None 548 | base = None 549 | 550 | rmoff = 0 551 | regoff = 0 552 | if self.rex('w'): 553 | rmoff -= 16 554 | regoff -= 16 555 | if self.rex('b'): 556 | rmoff += 8 557 | if self.rex('r'): 558 | regoff += 8 559 | 560 | if flags == opcode86.MODRM_EA: 561 | if mod == 3: # 11 562 | result=Register(rm+reg_type+rmoff, type_flag) 563 | elif mod == 0: # 0 564 | if rm == 5: 565 | disp= Address(data[count:], self.addr_size, type_flag) 566 | count+= self.addr_size 567 | elif rm == 4: 568 | (tmpcount, base) =self.get_sib(data[count:], mod) 569 | count+=tmpcount 570 | else: 571 | #print ("mod:%d\t reg:%d\t rm:%d"%(mod,reg,rm)) 572 | base=Register(rm+reg_type+rmoff, type_flag) 573 | else: 574 | 575 | if rm ==4: 576 | disp_base = 2 577 | (tmpcount, base) =self.get_sib(data[count:], mod) 578 | count+=tmpcount 579 | else: 580 | disp_base = 1 581 | base=Register(rm+reg_type+rmoff, type_flag) 582 | #print ">BASE: %s" % base.printOpcode() 583 | if mod == 1: 584 | disp= Address(data[disp_base:], 1, type_flag) 585 | count+=1 586 | else: 587 | disp= Address(data[disp_base:], self.addr_size, type_flag) 588 | count+= self.addr_size 589 | if disp or base: 590 | result=Expression(disp, base, type_flag) 591 | else: 592 | result=Register(reg+reg_type+regoff, type_flag) 593 | count=0 594 | 595 | return (count, result) 596 | # FIX: 597 | # 598 | def getOpcode(self, FORMAT, eip = 0): 599 | opcode=[] 600 | if not self.opcode: 601 | return [0] 602 | if FORMAT == "INTEL": 603 | opcode.append("%s" % self.opcode) 604 | #tmp="%-06s %s" % (self.opcode, " " * space) 605 | if self.source: 606 | opcode.append(self.source.printOpcode(FORMAT, eip)) 607 | #tmp+=" %s" % self.source.printOpcode(FORMAT, eip) 608 | if self.dest: 609 | opcode.append(self.dest.printOpcode(FORMAT, eip)) 610 | #tmp+=", %s" % self.dest.printOpcode(FORMAT, eip) 611 | if self.aux: 612 | opcode.append(self.aux.printOpcode(FORMAT, eip)) 613 | 614 | else: 615 | mnemonic = self.opcode 616 | post=[] 617 | if self.source and self.dest: 618 | addr_meth = self.source.getAddrMeth() 619 | optype = self.source.getOpSize() 620 | mnemonic = self.opcode 621 | 622 | if addr_meth == opcode86.ADDRMETH_E and\ 623 | not (isinstance(self.source, Register) or\ 624 | isinstance(self.dest, Register)): 625 | if optype == opcode86.OPTYPE_b: 626 | mnemonic+="b" 627 | elif optype== opcode86.OPTYPE_w: 628 | mnemonic+="" 629 | else : 630 | mnemonic+="l" 631 | 632 | ##first="%-06s %s" % (mnemonic, " " * space) 633 | post= [self.dest.printOpcode(FORMAT, eip), self.source.printOpcode(FORMAT, eip)] 634 | if self.aux: 635 | post.append(self.aux.printOpcode(FORMAT, eip)) 636 | #post = "%s, %s" % (self.dest.printOpcode(FORMAT,eip), self.source.printOpcode(FORMAT, eip)) 637 | elif self.source: 638 | #second="%-06s %s" % (mnemonic, " " * space) 639 | opt = self.getOpcodetype() 640 | tmp="" 641 | if (opt== opcode86.INS_CALL or\ 642 | opt== opcode86.INS_BRANCH)\ 643 | and self.source.getAddrMeth() == opcode86.ADDRMETH_E: 644 | 645 | tmp = "*" 646 | post=[tmp + self.source.printOpcode(FORMAT, eip)] 647 | #post += "%s" % self.source.printOpcode(FORMAT, eip) 648 | opcode = [mnemonic] + post 649 | 650 | return (opcode, self.r, self.w) 651 | 652 | def printOpcode(self, FORMAT, eip = 0, space=6): 653 | opcode=self.getOpcode(FORMAT, eip + self.getSize()) 654 | prefix=self.getPrefix(); 655 | if opcode[0]==0: 656 | return "invalid" 657 | if len(opcode) ==2: 658 | return "%-08s%s%s" % (prefix+opcode[0], " " * space, opcode[1]) 659 | #return "%-08s%s%s" % (prefix+opcode[0], " " * 6, opcode[1]) 660 | elif len(opcode) ==3: 661 | return "%-08s%s%s, %s" % (prefix+opcode[0], " " * space, opcode[1], opcode[2]) 662 | #return "%-08s%s%s, %s" % (prefix+ opcode[0], " " * 6, opcode[1], opcode[2]) 663 | elif len(opcode) ==4: 664 | return "%-08s%s%s, %s, %s" % (prefix+opcode[0], " " * space, opcode[3], opcode[1], opcode[2]) 665 | else: 666 | return "%-08s" % (prefix+opcode[0]) 667 | return tmp 668 | def rex(self, f): 669 | if self.mode != 64: 670 | return False 671 | b, w, x, r = False, False, False, False 672 | for a in self.prefix: 673 | type = a.getType() 674 | if type & opcode86.PREFIX_REX: 675 | if type & opcode86.PREFIX_REXB: 676 | b = True 677 | if type & opcode86.PREFIX_REXW: 678 | w = True 679 | if type & opcode86.PREFIX_REXX: 680 | x = True 681 | if type & opcode86.PREFIX_REXR: 682 | r = True 683 | 684 | if f == 'w': 685 | return w 686 | if f == 'x': 687 | return x 688 | if f == 'b': 689 | return b 690 | if f == 'r': 691 | return r 692 | 693 | return False 694 | 695 | def getPrefix(self): 696 | prefix="" 697 | for a in self.prefix: 698 | type = a.getType() 699 | if type in [opcode86.PREFIX_LOCK, opcode86.PREFIX_REPNZ, opcode86.PREFIX_REP]: 700 | prefix+= a.getName() + " " 701 | if self.mode == 64: 702 | if (type & opcode86.PREFIX_REX): 703 | rex = '' 704 | if type & opcode86.PREFIX_REXB: 705 | rex += 'b' 706 | if type & opcode86.PREFIX_REXW: 707 | rex += 'w' 708 | if type & opcode86.PREFIX_REXX: 709 | rex += 'x' 710 | if type & opcode86.PREFIX_REXR: 711 | rex += 'r' 712 | if rex: 713 | prefix += 'REX.'+rex+' ' 714 | else: 715 | prefix += 'REX ' 716 | return prefix 717 | if __name__=="__main__": 718 | # To get this information, just 719 | import sys 720 | if len(sys.argv) != 4: 721 | print ("usage: {} ".format(sys.argv[0])) 722 | print ("\t file:\t file to disassemble") 723 | print ("\t offset:\t offset to beggining of code(hexa)") 724 | print ("\t size:\t amount of bytes to dissasemble (hexa)\n") 725 | 726 | 727 | sys.exit(0) 728 | f=open(sys.argv[1]) 729 | offset= int(sys.argv[2], 16) 730 | f.seek( offset ) 731 | buf=f.read(int(sys.argv[3], 16) ) 732 | off=0 733 | FORMAT="AT&T" 734 | print("Disassembling file %s at offset: 0x%x" % (sys.argv[1], offset)) 735 | while 1: 736 | try: 737 | p=Opcode(buf[off:]) 738 | 739 | print(" %08X: %s" % (off+offset, p.printOpcode(FORMAT))) 740 | off+=p.getSize() 741 | except ValueError: 742 | break 743 | -------------------------------------------------------------------------------- /src/disassemblers/x64.py: -------------------------------------------------------------------------------- 1 | from disassemblers.libdisassemble.disassemble import * 2 | import debug 3 | 4 | def repr_ins(ins, r, w, objbounds, sections, binary): 5 | ''' 6 | Represent an x64 instruction in the immediate instruction format. 7 | ''' 8 | def arg(n): 9 | ''' 10 | Helper function for converting arguments. 11 | ''' 12 | value = ins[n+1] 13 | arg_repr = value 14 | 15 | if value.startswith('0x'): 16 | val = int(value, 16) 17 | if val >= objbounds[0] and val < objbounds[1]: 18 | '''Likely candidate for a constant''' 19 | 20 | ro = sections['.rodata'] 21 | if val >= ro['virt'] and val < ro['virt']+ro['length']: 22 | addr = ro['start']+val-ro['virt'] 23 | for i, b in enumerate(binary[addr:]): 24 | if b == 0: 25 | s = binary[addr:addr+i].decode() 26 | s = '"' + repr(s)[1:-1] + '"' 27 | arg_repr = s 28 | break 29 | if i > 100: 30 | break 31 | 32 | return {'value': value, 'repr': arg_repr, 'r': r[n], 'w': w[n]} 33 | 34 | def translate(op): 35 | ''' 36 | Some instructions are easy to translate, but have weird names, 37 | for example imul -> mul. 38 | ''' 39 | ins[0] = op 40 | return repr_ins(ins, r, w, objbounds, sections, binary) 41 | 42 | def jump_dest(): 43 | ''' 44 | Calculate a jump operand. 45 | ''' 46 | dest = {'type':'const', 'value': ins[1], 'r': True, 'w': False} 47 | try: 48 | dest['repr'] = int(ins[1],16) 49 | except ValueError: 50 | dest['repr'] = ins[1] 51 | return dest 52 | 53 | def parse_addr(addr, r=True, w=False): 54 | ''' 55 | Parse the addressing mode (e.g. in a LEA). 56 | ''' 57 | addr = addr.strip().strip('[]') 58 | 59 | for op, repr in [('sub', '-'), ('add', '+'), ('mul', '*')]: 60 | parts = addr.partition(repr) 61 | if parts[1]: 62 | return {'op': op, 'dest': parse_addr(parts[0], r, w), 'src': parse_addr(parts[2], r, w)} 63 | 64 | return {'value': addr, 'repr': addr, 'r': r, 'w': w} 65 | 66 | 67 | nop = {'op': 'nop'} 68 | 69 | if ins[0][0] == 'j': 70 | cond = ins[0][1:] 71 | if cond == 'mp': 72 | cond = 'true' 73 | return {'op': 'jump', 'cond': cond, 'dest': jump_dest()} 74 | elif ins[0] == 'call': 75 | return {'op': 'call', 'dest': jump_dest()} 76 | elif ins[0] == 'arpl': 77 | return nop 78 | elif ins[0] == 'cwde': 79 | dest = {'value': 'eax', 'repr': 'eax', 'r': False, 'w': True} 80 | src = {'value': 'ax', 'repr': 'eax', 'r': True, 'w': False} 81 | return {'op': 'mov', 'dest': dest, 'src': src} 82 | elif ins[0] == 'leave': 83 | return nop 84 | elif ins[0] == 'lea': 85 | src = parse_addr(ins[2]) 86 | return {'op': 'mov', 'dest': arg(0), 'src': src} 87 | elif ins[0] == 'push': 88 | return nop 89 | elif ins[0] == 'imul': 90 | # libdisassemble bug fix :( 91 | w[0] = True 92 | return translate('mul') 93 | elif ins[0] == 'idiv': 94 | return translate('div') 95 | elif ins[0] == 'cmp': 96 | cmp = {'value': 'cmp', 'repr': 'cmp', 'r': False, 'w': True} 97 | src = {'op': 'sub', 'dest': arg(0), 'src': arg(1)} 98 | return {'op': 'mov', 'dest': cmp, 'src': src} 99 | elif ins[0] == 'test': 100 | cmp = {'value': 'cmp', 'repr': 'cmp', 'r': False, 'w': True} 101 | src = {'op': 'and', 'dest': arg(0), 'src': arg(1)} 102 | return {'op': 'mov', 'dest': cmp, 'src': src} 103 | elif ins[0] == 'ret': 104 | dest = {'value': 'eax', 'repr': 'eax', 'r': True, 'w': False} 105 | return {'op': 'return', 'src': dest} 106 | elif len(ins) == 1: 107 | return {'op': ins[0]} 108 | elif len(ins) == 2: 109 | return {'op': ins[0], 'dest': arg(0)} 110 | elif len(ins) == 3: 111 | return {'op': ins[0], 'dest': arg(0), 'src': arg(1)} 112 | elif len(ins) == 4: 113 | return {'op': ins[0], 'dest': arg(0), 'src': arg(1), 'arg3': arg(2)} 114 | 115 | raise Exception('Bad x64 instruction: '+str(ins)) 116 | 117 | def disassemble(buf, virt, sections, binary): 118 | ''' 119 | Disassemble a block of x64 code. 120 | ''' 121 | 122 | virts = [(sections[sect]['virt'], sections[sect]['length']) for sect in sections if sections[sect]['virt']] 123 | objmin = min(virt[0] for virt in virts) 124 | objmax = max(virt[0]+virt[1] for virt in virts) 125 | objbounds = (objmin, objmax) 126 | 127 | FORMAT="INTEL" 128 | entries = [virt] 129 | 130 | result = {} 131 | while entries: 132 | addr = entries.pop() 133 | off = addr-virt 134 | while off < len(buf): 135 | p = Opcode(buf[off:], mode=64) 136 | pre = p.getPrefix() 137 | length = p.getSize() 138 | try: 139 | ins, r, w = p.getOpcode(FORMAT) 140 | except ValueError: 141 | break 142 | ins = repr_ins(ins, r, w, objbounds, sections, binary) 143 | 144 | if debug.check('asm_rw'): 145 | print(ins, r, w) 146 | 147 | debug_dis = { 148 | 'prefix': pre, 149 | 'binary': buf[off:off+length] 150 | } 151 | result[addr] = {'ins': ins, 'loc': addr, 'length': length, 'debug': debug_dis, 'display': True} 152 | if ins['op'] == 'return': 153 | break 154 | # if ins['op'] == 'call': 155 | # j_addr = addr+length+ins['dest']['repr'] 156 | # if j_addr not in result: 157 | # entries.append(j_addr) 158 | if ins['op'] == 'jump': 159 | j_addr = addr+length+ins['dest']['repr'] 160 | if j_addr not in result: 161 | entries.append(j_addr) 162 | result[addr]['display'] = False 163 | if ins['cond'] == 'true': 164 | break 165 | off += length 166 | addr += length 167 | 168 | return [result[key] for key in sorted(result.keys())] 169 | -------------------------------------------------------------------------------- /src/function_calls.py: -------------------------------------------------------------------------------- 1 | from disassemblers.libdisassemble.opcode86 import regs 2 | from copy import deepcopy 3 | 4 | legal_integers = ['rdi', 'rsi', 'rdx', 'rcx', 'r8', 'r9'] 5 | legal_sse = ['xmm0', 'xmm1', 'xmm2', 'xmm3', 'xmm4', 'xmm5', 'xmm6', 'xmm7'] 6 | legal_other = ['rax'] 7 | 8 | def reg_normalize(reg): 9 | ''' 10 | Normalize a register to a form independent of its size. 11 | ''' 12 | idx = list(map(lambda x: x[0], regs)).index(reg) 13 | return regs[idx&0xF][0] 14 | 15 | class Params: 16 | ''' 17 | A data structure that holds the current list of parameters used 18 | and is able to let you know when it ends using some heuristics. 19 | ''' 20 | def __init__(self): 21 | self.memory = [] 22 | self.integers = [] 23 | self.sse = [] 24 | self.other = [] 25 | self.args = [] 26 | 27 | def add(self, reg, arg): 28 | ''' 29 | Try to add a register to the list of params. 30 | 31 | If its the next legal register in a list of parameters, it's added 32 | and True is returned. If it isn't, False is returned so that the 33 | function can be wrapped. 34 | ''' 35 | arg = deepcopy(arg) 36 | if 'w' in arg: 37 | arg['w'] = False 38 | arg['r'] = True 39 | try: 40 | param = reg_normalize(reg) 41 | except ValueError: 42 | return False 43 | try: 44 | if legal_integers[len(self.integers)] == param: 45 | self.integers.append(reg) 46 | self.args.append(arg) 47 | return True 48 | elif legal_sse[len(self.sse)] == reg: 49 | #fix normalization here 50 | self.sse.append(reg) 51 | self.args.append(arg) 52 | return True 53 | elif legal_other[len(self.other)] == param: 54 | self.other.append(reg) 55 | return True 56 | else: 57 | return False 58 | except IndexError: 59 | return False 60 | 61 | def fold(cfg, symbols): 62 | ''' 63 | Fold as many function calls as its possible, infering arguments lists 64 | along the way. 65 | ''' 66 | for block, depth in cfg.iterblocks(): 67 | inside_call = False 68 | for n, line in enumerate(reversed(block)): 69 | if inside_call: 70 | if 'dest' not in line['ins']: 71 | continue 72 | dest = line['ins']['dest'] 73 | param = dest['value'] 74 | if call_params.add(param, dest): 75 | pass 76 | else: 77 | apply_ins = {'op': 'apply', 'function': function_name, 'args': call_params.args} 78 | eax = {'value': 'eax', 'repr': 'eax', 'r': False, 'w': True} 79 | mov_ins = {'op': 'mov', 'src': apply_ins, 'dest': eax} 80 | block[len(block)-call_n-1]['ins'] = mov_ins 81 | inside_call = False 82 | call_params = None 83 | if line['ins']['op'] == 'call': 84 | inside_call = True 85 | call_n = n 86 | call_params = Params() 87 | function_name = 'unknown_function' 88 | if 'repr' in line['ins']['dest']: 89 | if type(line['ins']['dest']['repr']) == int: 90 | addr = line['loc']+line['length']+line['ins']['dest']['repr'] 91 | if addr in symbols: 92 | function_name = symbols[addr] 93 | 94 | -------------------------------------------------------------------------------- /src/objdump.py: -------------------------------------------------------------------------------- 1 | from subprocess import Popen, PIPE 2 | 3 | def sections(filename): 4 | sections = {} 5 | p = Popen(["objdump", "-h", filename], stdout=PIPE) 6 | p_out = p.communicate()[0].decode() 7 | 8 | for line in p_out.split('\n'): 9 | words = line.split() 10 | if len(words) > 0 and words[0].isdigit(): 11 | sections[words[1]] = {'start': int(words[5], 16), 'length': int(words[2], 16), 'virt': int(words[3],16)} 12 | return sections 13 | 14 | def symbols(filename): 15 | symbols = {} 16 | p = Popen(["objdump", "-t", filename], stdout=PIPE) 17 | p_out = p.communicate()[0].decode() 18 | 19 | for line in p_out.split('\n'): 20 | words = line.split() 21 | if len(words) > 0 and words[1] == 'g' and not words[-1].startswith('_'): 22 | symbols[words[-1]] = {'start': int(words[0], 16), 'length': int(words[-2], 16), 'type': words[2]} 23 | 24 | return symbols 25 | 26 | def objdump(filename): 27 | ''' 28 | Find sections and symbols in an object file. 29 | ''' 30 | obj_sections = sections(filename) 31 | obj_symbols = symbols(filename) 32 | 33 | return obj_sections, obj_symbols 34 | -------------------------------------------------------------------------------- /src/ocd.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env python3 2 | from disassemble import disassemble 3 | from decompile import decompile_functions 4 | from objdump import objdump 5 | import control_flow 6 | import debug 7 | import representation 8 | 9 | if __name__=="__main__": 10 | import sys 11 | from optparse import OptionParser 12 | 13 | usage = "usage: %prog [options] file" 14 | parser = OptionParser(usage=usage) 15 | parser.add_option("-d", "--debug", dest="debug", action="append", 16 | help="turn debug option on", default=[], choices=debug.options, 17 | metavar="OPTION") 18 | parser.add_option("-g", "--graph", action="store", dest="graphfile", 19 | metavar="FILE", type="string", help="output a control flow graph") 20 | 21 | options, args = parser.parse_args() 22 | 23 | for option in options.debug: 24 | debug.set(option) 25 | 26 | if len(args) < 1: 27 | parser.print_help() 28 | 29 | sys.exit(0) 30 | 31 | filename = args[0] 32 | 33 | f = open(filename, "rb") 34 | binary = f.read() 35 | f.close() 36 | 37 | sections, symbols = objdump(filename) 38 | text = sections['.text'] 39 | 40 | if options.graphfile: 41 | gf = open(options.graphfile, 'w') 42 | gf.write("digraph cfg {\n") 43 | control_flow.graphfile = gf 44 | 45 | functions = {} 46 | if symbols: 47 | for name, symbol in symbols.items(): 48 | if not name.startswith('_') and symbol['type'] == 'F': 49 | start = symbol['start']-text['virt']+text['start'] 50 | length = symbol['length'] 51 | functions[name] = disassemble(binary[start:start+length], symbol['start'], sections, binary) 52 | else: 53 | functions['start'] = disassemble(binary[text['start']:text['start']+text['length']], text['virt'], sections, binary) 54 | symbols['start'] = {'start': text['virt'], 'length': text['length']} 55 | 56 | decompiled_functions = decompile_functions(functions, symbols) 57 | 58 | print(representation.output_functions(decompiled_functions)) 59 | 60 | if options.graphfile: 61 | gf.write("}\n") 62 | gf.close() 63 | -------------------------------------------------------------------------------- /src/output/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/drx/ocd/bbab05bcd4e6ec83fbc4e8c30f52babd140cc2dc/src/output/__init__.py -------------------------------------------------------------------------------- /src/output/c.py: -------------------------------------------------------------------------------- 1 | from output.indent import Indent 2 | from output.conditions import conditions, condition_negs 3 | import debug 4 | import zlib 5 | 6 | def operator(op, precedence): 7 | ''' 8 | Helper function for C operators. 9 | ''' 10 | return ('{i[dest]} = ', '{i[dest]} '+op+' {i[src]}', precedence) 11 | 12 | ins_table = { 13 | 'add': operator('+', 12), 14 | 'and': operator('&', 8), 15 | 'div': operator('/', 13), 16 | 'mul': operator('*', 13), 17 | 'mov': ('{i[dest]} = ', '{i[src]}', 2), 18 | 'nop': ('', '', 0), 19 | 'return': ('return {i[src]}', '', 0), 20 | 'sar': operator('>>', 11), 21 | 'sal': operator('<<', 11), 22 | 'sub': operator('-', 12), 23 | 'xor': operator('^', 7), 24 | } 25 | 26 | def condition(cond, var='cmp'): 27 | ''' 28 | Output a jump condition. 29 | ''' 30 | if cond[0] == '!': 31 | cond = condition_negs[cond[1:]] 32 | return conditions[cond].format(cond=var) 33 | 34 | def striplines(s): 35 | ''' 36 | Strip empty lines from string. 37 | ''' 38 | return '\n'.join(line for line in s.split('\n') if line.strip()) 39 | 40 | def repr_int(n): 41 | ''' 42 | Get the representation of an integer. 43 | 44 | The representation is chosen based on its Kolmogorov complexity, 45 | i.e. the repr r is chosen which has the minimal value 46 | len(zlib.compress(r)). 47 | ''' 48 | reprs = [('{0}','{0}'), ('{0:x}', '0x{0:x}')] 49 | lengths = [len(zlib.compress(r[0].format(n).encode())) for r in reprs] 50 | return reprs[lengths.index(min(lengths))][1].format(n) 51 | 52 | 53 | def output_op(op): 54 | ''' 55 | Output an instruction according to it's opcode. 56 | ''' 57 | return ins_table[op] 58 | 59 | def output_line(line, indent): 60 | ''' 61 | Output a line of code. A line contains an instruction and 62 | some metainformation about that instruction (its location, 63 | etc.) 64 | 65 | The pseudo-EBNF of lines is as follows: 66 | 67 | line = location, debug, instruction 68 | instruction = op, argument* 69 | argument = value, repr | instruction 70 | ''' 71 | def output_ins(ins): 72 | ''' 73 | Decompile an instruction. Instructions are trees, see 74 | the EBNF for output_line. 75 | ''' 76 | repr = {} 77 | prec = {} 78 | for k, arg in ins.items(): 79 | if k not in ('src', 'dest'): 80 | continue 81 | 82 | if 'op' in ins[k]: 83 | lhs, rhs, prec[k] = output_ins(ins[k]) 84 | repr[k] = rhs 85 | else: 86 | repr[k] = ins[k]['repr'] 87 | prec[k] = 20 88 | 89 | if type(repr[k]) == int: 90 | repr[k] = repr_int(repr[k]) 91 | 92 | if ins['op'] == 'apply': 93 | lhs = '' 94 | args = [] 95 | for arg in ins['args']: 96 | if 'op' in arg: 97 | lhs, rhs, inner_prec = output_ins(arg) 98 | args.append(rhs) 99 | else: 100 | args.append(arg['repr']) 101 | rhs = '{fun}({args})'.format(fun=ins['function'], args=', '.join(args)) 102 | outer_prec = 20 103 | 104 | else: 105 | try: 106 | lhs, rhs, outer_prec = output_op(ins['op']) 107 | except KeyError: 108 | lhs, rhs = '', '/* Unsupported immediate instruction: {ins[op]} */'.format(ins=ins) 109 | outer_prec = 20 110 | 111 | for k in dict(repr): 112 | if outer_prec > prec[k]: 113 | repr[k] = '(' + repr[k] + ')' 114 | 115 | return lhs.format(i=repr), rhs.format(i=repr), outer_prec 116 | 117 | lhs, rhs, prec = output_ins(line['ins']) 118 | line_repr = indent.out() + lhs + rhs 119 | if lhs+rhs != '': 120 | line_repr += ';' 121 | 122 | if not line['display']: 123 | line_repr = '' 124 | 125 | if debug.check('misc'): 126 | from binascii import hexlify 127 | line_repr += '\n{indent}{blue}/* {line} */{black}'.format(indent=indent.out(), line=line, blue='\033[94m', black='\033[0m') 128 | 129 | return line_repr 130 | 131 | def output_vertex(vertex, indent=None): 132 | ''' 133 | Output a CFG vertex. 134 | ''' 135 | key, block = vertex 136 | block_type, block_start = key 137 | 138 | if indent is None: 139 | indent = Indent(1) 140 | 141 | if block_type == 'block': 142 | return '\n'.join(output_line(line, indent) for line in block) 143 | 144 | elif block_type == 'if': 145 | cond, true = block 146 | fmt = ('\n' 147 | '{indent}if ({cond})\n' 148 | '{indent}{{\n' 149 | '{true}\n' 150 | '{indent}}}\n') 151 | return fmt.format(cond=condition(cond), true=output_vertex(true, indent.inc()), indent=indent.out()) 152 | 153 | elif block_type == 'ifelse': 154 | cond, true, false = block 155 | fmt = ('\n' 156 | '{indent}if ({cond})\n' 157 | '{indent}{{\n' 158 | '{true}\n' 159 | '{indent}}}\n' 160 | '{indent}else\n' 161 | '{indent}{{\n' 162 | '{false}\n' 163 | '{indent}}}\n') 164 | return fmt.format(cond=condition(cond), 165 | true=output_vertex(true, indent.inc()), false=output_vertex(false, indent.inc()), indent=indent.out()) 166 | 167 | elif block_type == 'while': 168 | cond, pre, loop = block 169 | fmt = ('\n' 170 | '{indent}while ({cond})\n' 171 | '{indent}{{\n' 172 | '{loop}\n' 173 | '{pre}\n' 174 | '{indent}}}\n') 175 | return fmt.format(indent=indent.out(), cond=condition(cond), 176 | pre=output_vertex(pre, indent.inc()), loop=output_vertex(loop, indent.inc())) 177 | 178 | elif block_type == 'cons': 179 | out = '' 180 | for b in block: 181 | out += output_vertex(b, indent) 182 | return out 183 | 184 | return '/* Error: unsupported block type */' 185 | 186 | def output_signature(signature, name): 187 | ''' 188 | Output the signature of a function. 189 | ''' 190 | fun_type, args = signature 191 | 192 | pre = fun_type + ' ' + name + '(' + ', '.join(args) + ')\n{\n' 193 | post = '\n}' 194 | return pre, post 195 | 196 | def output_function(function, name): 197 | ''' 198 | Output a function. 199 | ''' 200 | cfg, signature = function 201 | 202 | pre, post = output_signature(signature, name) 203 | 204 | innards = '\n/*----*/\n'.join(output_vertex(v) for v in cfg.sortedvertices()) 205 | 206 | return pre + innards + post 207 | 208 | def output(functions): 209 | ''' 210 | Output everything. 211 | ''' 212 | out = '' 213 | for name, function in functions.items(): 214 | out += output_function(function, name) 215 | out += '\n' 216 | 217 | return striplines(out) 218 | -------------------------------------------------------------------------------- /src/output/conditions.py: -------------------------------------------------------------------------------- 1 | condition_negs = { 2 | 'a': 'be', 'ae': 'b', 'b': 'ae', 'be': 'a', 'c': 'a', 3 | 'e': 'ne', 'ne': 'e', 4 | 'g': 'le', 'ge': 'l', 'l': 'ge', 'le': 'g', 5 | 'na': 'a', 'nae': 'ae', 'nb': 'b', 'nbe': 'be', 'nc': 'c', 6 | 'ng': 'g', 'nge': 'ge', 'nl': 'l', 'nle': 'le', 7 | 'no': 'o', 'o': 'no', 's': 'ns', 'ns': 's', 8 | 'p': 'np', 'np': 'p', 'pe': 'np', 'po': 'p', 9 | 'z': 'nz', 'nz': 'z', 'true': 'false', 'false': 'true' 10 | } 11 | 12 | conditions = { 13 | 'a': '{cond} > 0', 14 | 'ae': '{cond} >= 0', 15 | 'b': '{cond} < 0', 16 | 'be': '{cond} <= 0', 17 | 'c': '{cond} <= 0', 18 | 'cxz': '!{cx}', 19 | 'ecxz': '!{ecx}', 20 | 'rcxz': '!{rcx}', 21 | 'e': '!{cond}', 22 | 'g': '{cond} > 0', 23 | 'ge': '{cond} >= 0', 24 | 'l': '{cond} < 0', 25 | 'le': '{cond} <= 0', 26 | 'na': '{cond} <= 0', 27 | 'nae': '{cond} < 0', 28 | 'nb': '{cond} >= 0', 29 | 'nbe': '{cond} > 0', 30 | 'nc': '{cond} >= 0', 31 | 'ne': '!{cond}', 32 | 'ng': '{cond} <= 0', 33 | 'nge': '{cond} < 0', 34 | 'nl': '{cond} >= 0', 35 | 'nle': '{cond} > 0', 36 | 'no': 'no_overflow', 37 | 'np': '{cond} & 1', 38 | 'ns': '({cond}>0)-(cmp<0) == -1', 39 | 'o': 'overflow', 40 | 'p': '!({cond} & 1)', 41 | 'pe': '!({cond} & 1)', 42 | 'po': '{cond} & 1', 43 | 's': '({cond}>0)-(cmp<0) >= 0', 44 | 'nz': '{cond}', 45 | 'z': '!{cond}', 46 | 'true': '1', 47 | 'false': '0', 48 | } 49 | 50 | -------------------------------------------------------------------------------- /src/output/indent.py: -------------------------------------------------------------------------------- 1 | from copy import copy 2 | 3 | class Indent(): 4 | ''' 5 | Auto-indentation. 6 | ''' 7 | def __init__(self, level=0): 8 | ''' 9 | Initialize indentation level. 10 | ''' 11 | self.level = level 12 | 13 | def inc(self): 14 | ''' 15 | Increment indentation level. 16 | ''' 17 | new = copy(self) 18 | new.level += 1 19 | return new 20 | 21 | def out(self): 22 | ''' 23 | Output indentation. 24 | ''' 25 | return '\t'*self.level 26 | -------------------------------------------------------------------------------- /src/representation.py: -------------------------------------------------------------------------------- 1 | import output.c 2 | 3 | def output_functions(functions, lang='C'): 4 | ''' 5 | Output all functions in a selected language. 6 | ''' 7 | langs = { 8 | 'C': output.c 9 | } 10 | return langs[lang].output(functions) 11 | -------------------------------------------------------------------------------- /tests/Makefile: -------------------------------------------------------------------------------- 1 | CC = gcc 2 | CFLAGS = -std=c99 -Wall -W -O0 -m64 3 | 4 | SOURCES := $(wildcard *.c) 5 | BINARIES := $(patsubst %.c, %, $(SOURCES)) 6 | 7 | all: $(BINARIES) 8 | 9 | clean: 10 | rm $(BINARIES) 11 | -------------------------------------------------------------------------------- /tests/test_ack.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | return printf("%d\n", ack(3, 4) ); 4 | } 5 | 6 | int ack(int m, int n) 7 | { 8 | if (m == 0) 9 | { 10 | return n + 1; 11 | } 12 | else 13 | { 14 | if (n == 0) 15 | { 16 | return ack(m - 1, 1); 17 | } 18 | else { 19 | return ack(m - 1, ack(m, n - 1)); 20 | } 21 | } 22 | } 23 | -------------------------------------------------------------------------------- /tests/test_ack_scanf.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a, b, r; 4 | r = scanf("%d %d", &a, &b); 5 | r = printf("%d", ack(a, b) ); 6 | } 7 | 8 | int ack(int m, int n) 9 | { 10 | if( m == 0 ) 11 | { return n + 1; } 12 | else 13 | { 14 | if( n == 0 ) 15 | { return ack(m - 1, 1); } 16 | else { return ack(m - 1, ack(m, n - 1)); } 17 | } 18 | } 19 | -------------------------------------------------------------------------------- /tests/test_add.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | return 4+2; 4 | } 5 | -------------------------------------------------------------------------------- /tests/test_add_var.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4 + 4; 4 | return 0; 5 | } 6 | -------------------------------------------------------------------------------- /tests/test_block.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int sum=0; 4 | int x = 4; 5 | { 6 | sum+=x; 7 | int x = 3; 8 | { 9 | sum+=x; 10 | int x = 2; 11 | sum+=x; 12 | } 13 | } 14 | sum+=x; 15 | } 16 | -------------------------------------------------------------------------------- /tests/test_collapse.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4; 4 | int b = 2; 5 | int c = 3; 6 | int d = 6; 7 | a = (b+c)*d-a; 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_collapse2.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4; 4 | int b = 2; 5 | int c = 3; 6 | int d = 6; 7 | a = (b+c)*d-a; 8 | b = a - c + a * a + d / b; 9 | } 10 | -------------------------------------------------------------------------------- /tests/test_div_zero.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4/0; 4 | } 5 | -------------------------------------------------------------------------------- /tests/test_for.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 0; 4 | for (int i=0; i<1000; i++) 5 | { 6 | a += i; 7 | } 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_for_if.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 0; 4 | for (int i=0; i<1000; i++) 5 | { 6 | if (a > 500) 7 | { 8 | a += i; 9 | } 10 | } 11 | } 12 | -------------------------------------------------------------------------------- /tests/test_mod_zero.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4%0; 4 | } 5 | -------------------------------------------------------------------------------- /tests/test_ptr.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int s = 1; 4 | int *p = &s; 5 | *p = 3; 6 | } 7 | -------------------------------------------------------------------------------- /tests/test_return_0.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | return 0; 4 | } 5 | -------------------------------------------------------------------------------- /tests/test_sum1.c: -------------------------------------------------------------------------------- 1 | int sum(int n1) 2 | { 3 | return n1; 4 | } 5 | int main() 6 | { 7 | return sum(100); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum2.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2) 2 | { 3 | return n1+n2; 4 | } 5 | int main() 6 | { 7 | return sum(1, 10); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum3.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2, int n3) 2 | { 3 | return n1+n2+n3; 4 | } 5 | int main() 6 | { 7 | return sum(1, 2, 4); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum4.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2, int n3, int n4) 2 | { 3 | return n1+n2+n3+n4; 4 | } 5 | int main() 6 | { 7 | return sum(1, 2, 4, 8); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum5.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2, int n3, int n4, int n5) 2 | { 3 | return n1+n2+n3+n4+n5; 4 | } 5 | int main() 6 | { 7 | return sum(1, 2, 4, 8, 16); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum6.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2, int n3, int n4, int n5, int n6) 2 | { 3 | return n1+n2+n3+n4+n5+n6; 4 | } 5 | int main() 6 | { 7 | return sum(1, 2, 4, 8, 16, 32); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sum7.c: -------------------------------------------------------------------------------- 1 | int sum(int n1, int n2, int n3, int n4, int n5, int n6, int n7) 2 | { 3 | return n1+n2+n3+n4+n5+n6+n7; 4 | } 5 | int main() 6 | { 7 | return sum(1, 2, 4, 8, 16, 32, 64); 8 | } 9 | -------------------------------------------------------------------------------- /tests/test_sznurki_alistra.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | #define SIZE 1000001 4 | #define SIZE05 500001 5 | int a[ SIZE ]; 6 | int sum, n, i, count, length, changed, c; 7 | int main() 8 | { 9 | scanf( "%d", &n ); 10 | for( i = 0; i < n; ++i ) 11 | { 12 | scanf("%d %d", &length, &count); 13 | a[ length ] = count; 14 | } 15 | 16 | for( i = 1; i < SIZE05 ; ++i ) 17 | if( a[ i ] ) 18 | { 19 | sum += a[ i ] % 2; 20 | a[ 2*i ] += a[ i ] / 2; 21 | } 22 | 23 | for( i = SIZE05 ; i < SIZE; ++i ) 24 | { 25 | if( a[ i ] ) 26 | { 27 | for (c = 0; a[ i ]; c++) 28 | { 29 | a[ i ] &= a[ i ] - 1; 30 | sum++; 31 | } 32 | } 33 | } 34 | printf( "%d\n", sum ); 35 | } 36 | -------------------------------------------------------------------------------- /tests/test_sznurki_drx.c: -------------------------------------------------------------------------------- 1 | #include 2 | 3 | int sznurki[1000010]; 4 | int main() 5 | { 6 | int d, n_d, m, i, out = 0; 7 | scanf("%d", &m); 8 | for (i=0;i500000) 17 | { 18 | for (; n; out++) 19 | n &= n - 1; 20 | } 21 | else 22 | { 23 | if (n > 1) 24 | { 25 | sznurki[2*i] += n/2; 26 | } 27 | if (n%2 == 1) out++; 28 | } 29 | } 30 | printf("%d\n", out); 31 | } 32 | -------------------------------------------------------------------------------- /tests/test_variables.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 4; 4 | int b = 3; 5 | int c = 2; 6 | } 7 | -------------------------------------------------------------------------------- /tests/test_while.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int n = 10, m; 4 | while(n > 0) 5 | { 6 | m = printf("n = %d\n", n); 7 | n--; 8 | } 9 | } 10 | -------------------------------------------------------------------------------- /tests/test_zero.c: -------------------------------------------------------------------------------- 1 | int main() 2 | { 3 | int a = 0; 4 | } 5 | --------------------------------------------------------------------------------