1 #===- disassembler.py - Python LLVM Bindings -----------------*- python -*--===#
3 # The LLVM Compiler Infrastructure
5 # This file is distributed under the University of Illinois Open Source
6 # License. See LICENSE.TXT for details.
8 #===------------------------------------------------------------------------===#
10 from ctypes import CFUNCTYPE
11 from ctypes import POINTER
12 from ctypes import addressof
13 from ctypes import c_byte
14 from ctypes import c_char_p
15 from ctypes import c_int
16 from ctypes import c_size_t
17 from ctypes import c_ubyte
18 from ctypes import c_uint64
19 from ctypes import c_void_p
20 from ctypes import cast
22 from .common import LLVMObject
23 from .common import c_object_p
24 from .common import get_library
33 # Constants for set_options
36 class Disassembler(LLVMObject):
37 """Represents a disassembler instance.
39 Disassembler instances are tied to specific "triple," which must be defined
42 Disassembler instances can disassemble instructions from multiple sources.
44 def __init__(self, triple):
45 """Create a new disassembler instance.
47 The triple argument is the triple to create the disassembler for. This
48 is something like 'i386-apple-darwin9'.
50 ptr = lib.LLVMCreateDisasm(c_char_p(triple), c_void_p(None), c_int(0),
51 callbacks['op_info'](0), callbacks['symbol_lookup'](0))
53 raise Exception('Could not obtain disassembler for triple: %s' %
56 LLVMObject.__init__(self, ptr, disposer=lib.LLVMDisasmDispose)
58 def get_instruction(self, source, pc=0):
59 """Obtain the next instruction from an input source.
61 The input source should be a str or bytearray or something that
62 represents a sequence of bytes.
64 This function will start reading bytes from the beginning of the
67 The pc argument specifies the address that the first byte is at.
69 This returns a 2-tuple of:
71 long number of bytes read. 0 if no instruction was read.
72 str representation of instruction. This will be the assembly that
73 represents the instruction.
75 buf = cast(c_char_p(source), POINTER(c_ubyte))
76 out_str = cast((c_byte * 255)(), c_char_p)
78 result = lib.LLVMDisasmInstruction(self, buf, c_uint64(len(source)),
79 c_uint64(pc), out_str, 255)
81 return (result, out_str.value)
83 def get_instructions(self, source, pc=0):
84 """Obtain multiple instructions from an input source.
86 This is like get_instruction() except it is a generator for all
87 instructions within the source. It starts at the beginning of the
88 source and reads instructions until no more can be read.
90 This generator returns 3-tuple of:
92 long address of instruction.
93 long size of instruction, in bytes.
94 str representation of instruction.
96 source_bytes = c_char_p(source)
97 out_str = cast((c_byte * 255)(), c_char_p)
99 # This could probably be written cleaner. But, it does work.
100 buf = cast(source_bytes, POINTER(c_ubyte * len(source))).contents
103 end_address = pc + len(source)
104 while address < end_address:
105 b = cast(addressof(buf) + offset, POINTER(c_ubyte))
106 result = lib.LLVMDisasmInstruction(self, b,
107 c_uint64(len(source) - offset), c_uint64(address),
113 yield (address, result, out_str.value)
118 def set_options(self, options):
119 if not lib.LLVMSetDisasmOptions(self, options):
120 raise Exception('Unable to set all disassembler options in %i' % options)
123 def register_library(library):
124 library.LLVMCreateDisasm.argtypes = [c_char_p, c_void_p, c_int,
125 callbacks['op_info'], callbacks['symbol_lookup']]
126 library.LLVMCreateDisasm.restype = c_object_p
128 library.LLVMDisasmDispose.argtypes = [Disassembler]
130 library.LLVMDisasmInstruction.argtypes = [Disassembler, POINTER(c_ubyte),
131 c_uint64, c_uint64, c_char_p, c_size_t]
132 library.LLVMDisasmInstruction.restype = c_size_t
134 library.LLVMSetDisasmOptions.argtypes = [Disassembler, c_uint64]
135 library.LLVMSetDisasmOptions.restype = c_int
138 callbacks['op_info'] = CFUNCTYPE(c_int, c_void_p, c_uint64, c_uint64, c_uint64,
140 callbacks['symbol_lookup'] = CFUNCTYPE(c_char_p, c_void_p, c_uint64,
141 POINTER(c_uint64), c_uint64,
144 register_library(lib)