xref: /openbsd-src/gnu/llvm/lldb/examples/python/symbolication.py (revision f6aab3d83b51b91c24247ad2c2573574de475a82)
1be691f3bSpatrick#!/usr/bin/env python
2061da546Spatrick
3061da546Spatrick#----------------------------------------------------------------------
4061da546Spatrick# Be sure to add the python path that points to the LLDB shared library.
5061da546Spatrick#
6061da546Spatrick# To use this in the embedded python interpreter using "lldb":
7061da546Spatrick#
8061da546Spatrick#   cd /path/containing/crashlog.py
9061da546Spatrick#   lldb
10061da546Spatrick#   (lldb) script import crashlog
11061da546Spatrick#   "crashlog" command installed, type "crashlog --help" for detailed help
12061da546Spatrick#   (lldb) crashlog ~/Library/Logs/DiagnosticReports/a.crash
13061da546Spatrick#
14061da546Spatrick# The benefit of running the crashlog command inside lldb in the
15061da546Spatrick# embedded python interpreter is when the command completes, there
16061da546Spatrick# will be a target with all of the files loaded at the locations
17061da546Spatrick# described in the crash log. Only the files that have stack frames
18061da546Spatrick# in the backtrace will be loaded unless the "--load-all" option
19061da546Spatrick# has been specified. This allows users to explore the program in the
20061da546Spatrick# state it was in right at crash time.
21061da546Spatrick#
22061da546Spatrick# On MacOSX csh, tcsh:
23061da546Spatrick#   ( setenv PYTHONPATH /path/to/LLDB.framework/Resources/Python ; ./crashlog.py ~/Library/Logs/DiagnosticReports/a.crash )
24061da546Spatrick#
25061da546Spatrick# On MacOSX sh, bash:
26061da546Spatrick#   PYTHONPATH=/path/to/LLDB.framework/Resources/Python ./crashlog.py ~/Library/Logs/DiagnosticReports/a.crash
27061da546Spatrick#----------------------------------------------------------------------
28061da546Spatrick
29061da546Spatrickimport lldb
30061da546Spatrickimport optparse
31061da546Spatrickimport os
32061da546Spatrickimport plistlib
33061da546Spatrickimport re
34061da546Spatrickimport shlex
35061da546Spatrickimport sys
36061da546Spatrickimport time
37061da546Spatrickimport uuid
38061da546Spatrick
39061da546Spatrick
40061da546Spatrickclass Address:
41061da546Spatrick    """Class that represents an address that will be symbolicated"""
42061da546Spatrick
43061da546Spatrick    def __init__(self, target, load_addr):
44061da546Spatrick        self.target = target
45061da546Spatrick        self.load_addr = load_addr  # The load address that this object represents
46061da546Spatrick        # the resolved lldb.SBAddress (if any), named so_addr for
47061da546Spatrick        # section/offset address
48061da546Spatrick        self.so_addr = None
49061da546Spatrick        self.sym_ctx = None  # The cached symbol context for this address
50061da546Spatrick        # Any original textual description of this address to be used as a
51061da546Spatrick        # backup in case symbolication fails
52061da546Spatrick        self.description = None
53061da546Spatrick        self.symbolication = None  # The cached symbolicated string that describes this address
54061da546Spatrick        self.inlined = False
55061da546Spatrick
56061da546Spatrick    def __str__(self):
57061da546Spatrick        s = "%#16.16x" % (self.load_addr)
58061da546Spatrick        if self.symbolication:
59061da546Spatrick            s += " %s" % (self.symbolication)
60061da546Spatrick        elif self.description:
61061da546Spatrick            s += " %s" % (self.description)
62061da546Spatrick        elif self.so_addr:
63061da546Spatrick            s += " %s" % (self.so_addr)
64061da546Spatrick        return s
65061da546Spatrick
66061da546Spatrick    def resolve_addr(self):
67061da546Spatrick        if self.so_addr is None:
68061da546Spatrick            self.so_addr = self.target.ResolveLoadAddress(self.load_addr)
69061da546Spatrick        return self.so_addr
70061da546Spatrick
71061da546Spatrick    def is_inlined(self):
72061da546Spatrick        return self.inlined
73061da546Spatrick
74061da546Spatrick    def get_symbol_context(self):
75061da546Spatrick        if self.sym_ctx is None:
76061da546Spatrick            sb_addr = self.resolve_addr()
77061da546Spatrick            if sb_addr:
78061da546Spatrick                self.sym_ctx = self.target.ResolveSymbolContextForAddress(
79061da546Spatrick                    sb_addr, lldb.eSymbolContextEverything)
80061da546Spatrick            else:
81061da546Spatrick                self.sym_ctx = lldb.SBSymbolContext()
82061da546Spatrick        return self.sym_ctx
83061da546Spatrick
84061da546Spatrick    def get_instructions(self):
85061da546Spatrick        sym_ctx = self.get_symbol_context()
86061da546Spatrick        if sym_ctx:
87061da546Spatrick            function = sym_ctx.GetFunction()
88061da546Spatrick            if function:
89061da546Spatrick                return function.GetInstructions(self.target)
90061da546Spatrick            return sym_ctx.GetSymbol().GetInstructions(self.target)
91061da546Spatrick        return None
92061da546Spatrick
93061da546Spatrick    def symbolicate(self, verbose=False):
94061da546Spatrick        if self.symbolication is None:
95061da546Spatrick            self.symbolication = ''
96061da546Spatrick            self.inlined = False
97061da546Spatrick            sym_ctx = self.get_symbol_context()
98061da546Spatrick            if sym_ctx:
99061da546Spatrick                module = sym_ctx.GetModule()
100061da546Spatrick                if module:
101061da546Spatrick                    # Print full source file path in verbose mode
102061da546Spatrick                    if verbose:
103061da546Spatrick                        self.symbolication += str(module.GetFileSpec()) + '`'
104061da546Spatrick                    else:
105061da546Spatrick                        self.symbolication += module.GetFileSpec().GetFilename() + '`'
106061da546Spatrick                    function_start_load_addr = -1
107061da546Spatrick                    function = sym_ctx.GetFunction()
108061da546Spatrick                    block = sym_ctx.GetBlock()
109061da546Spatrick                    line_entry = sym_ctx.GetLineEntry()
110061da546Spatrick                    symbol = sym_ctx.GetSymbol()
111061da546Spatrick                    inlined_block = block.GetContainingInlinedBlock()
112061da546Spatrick                    if function:
113061da546Spatrick                        self.symbolication += function.GetName()
114061da546Spatrick
115061da546Spatrick                        if inlined_block:
116061da546Spatrick                            self.inlined = True
117061da546Spatrick                            self.symbolication += ' [inlined] ' + \
118061da546Spatrick                                inlined_block.GetInlinedName()
119061da546Spatrick                            block_range_idx = inlined_block.GetRangeIndexForBlockAddress(
120061da546Spatrick                                self.so_addr)
121061da546Spatrick                            if block_range_idx < lldb.UINT32_MAX:
122061da546Spatrick                                block_range_start_addr = inlined_block.GetRangeStartAddress(
123061da546Spatrick                                    block_range_idx)
124061da546Spatrick                                function_start_load_addr = block_range_start_addr.GetLoadAddress(
125061da546Spatrick                                    self.target)
126061da546Spatrick                        if function_start_load_addr == -1:
127061da546Spatrick                            function_start_load_addr = function.GetStartAddress().GetLoadAddress(self.target)
128061da546Spatrick                    elif symbol:
129061da546Spatrick                        self.symbolication += symbol.GetName()
130061da546Spatrick                        function_start_load_addr = symbol.GetStartAddress().GetLoadAddress(self.target)
131061da546Spatrick                    else:
132061da546Spatrick                        self.symbolication = ''
133061da546Spatrick                        return False
134061da546Spatrick
135061da546Spatrick                    # Dump the offset from the current function or symbol if it
136061da546Spatrick                    # is non zero
137061da546Spatrick                    function_offset = self.load_addr - function_start_load_addr
138061da546Spatrick                    if function_offset > 0:
139061da546Spatrick                        self.symbolication += " + %u" % (function_offset)
140061da546Spatrick                    elif function_offset < 0:
141061da546Spatrick                        self.symbolication += " %i (invalid negative offset, file a bug) " % function_offset
142061da546Spatrick
143061da546Spatrick                    # Print out any line information if any is available
144061da546Spatrick                    if line_entry.GetFileSpec():
145061da546Spatrick                        # Print full source file path in verbose mode
146061da546Spatrick                        if verbose:
147061da546Spatrick                            self.symbolication += ' at %s' % line_entry.GetFileSpec()
148061da546Spatrick                        else:
149061da546Spatrick                            self.symbolication += ' at %s' % line_entry.GetFileSpec().GetFilename()
150061da546Spatrick                        self.symbolication += ':%u' % line_entry.GetLine()
151061da546Spatrick                        column = line_entry.GetColumn()
152061da546Spatrick                        if column > 0:
153061da546Spatrick                            self.symbolication += ':%u' % column
154061da546Spatrick                    return True
155061da546Spatrick        return False
156061da546Spatrick
157061da546Spatrick
158061da546Spatrickclass Section:
159061da546Spatrick    """Class that represents an load address range"""
160061da546Spatrick    sect_info_regex = re.compile('(?P<name>[^=]+)=(?P<range>.*)')
161061da546Spatrick    addr_regex = re.compile('^\s*(?P<start>0x[0-9A-Fa-f]+)\s*$')
162061da546Spatrick    range_regex = re.compile(
163061da546Spatrick        '^\s*(?P<start>0x[0-9A-Fa-f]+)\s*(?P<op>[-+])\s*(?P<end>0x[0-9A-Fa-f]+)\s*$')
164061da546Spatrick
165061da546Spatrick    def __init__(self, start_addr=None, end_addr=None, name=None):
166061da546Spatrick        self.start_addr = start_addr
167061da546Spatrick        self.end_addr = end_addr
168061da546Spatrick        self.name = name
169061da546Spatrick
170061da546Spatrick    @classmethod
171061da546Spatrick    def InitWithSBTargetAndSBSection(cls, target, section):
172061da546Spatrick        sect_load_addr = section.GetLoadAddress(target)
173061da546Spatrick        if sect_load_addr != lldb.LLDB_INVALID_ADDRESS:
174061da546Spatrick            obj = cls(
175061da546Spatrick                sect_load_addr,
176061da546Spatrick                sect_load_addr +
177061da546Spatrick                section.size,
178061da546Spatrick                section.name)
179061da546Spatrick            return obj
180061da546Spatrick        else:
181061da546Spatrick            return None
182061da546Spatrick
183061da546Spatrick    def contains(self, addr):
184061da546Spatrick        return self.start_addr <= addr and addr < self.end_addr
185061da546Spatrick
186061da546Spatrick    def set_from_string(self, s):
187061da546Spatrick        match = self.sect_info_regex.match(s)
188061da546Spatrick        if match:
189061da546Spatrick            self.name = match.group('name')
190061da546Spatrick            range_str = match.group('range')
191061da546Spatrick            addr_match = self.addr_regex.match(range_str)
192061da546Spatrick            if addr_match:
193061da546Spatrick                self.start_addr = int(addr_match.group('start'), 16)
194061da546Spatrick                self.end_addr = None
195061da546Spatrick                return True
196061da546Spatrick
197061da546Spatrick            range_match = self.range_regex.match(range_str)
198061da546Spatrick            if range_match:
199061da546Spatrick                self.start_addr = int(range_match.group('start'), 16)
200061da546Spatrick                self.end_addr = int(range_match.group('end'), 16)
201061da546Spatrick                op = range_match.group('op')
202061da546Spatrick                if op == '+':
203061da546Spatrick                    self.end_addr += self.start_addr
204061da546Spatrick                return True
205061da546Spatrick        print('error: invalid section info string "%s"' % s)
206061da546Spatrick        print('Valid section info formats are:')
207061da546Spatrick        print('Format                Example                    Description')
208061da546Spatrick        print('--------------------- -----------------------------------------------')
209061da546Spatrick        print('<name>=<base>        __TEXT=0x123000             Section from base address only')
210061da546Spatrick        print('<name>=<base>-<end>  __TEXT=0x123000-0x124000    Section from base address and end address')
211061da546Spatrick        print('<name>=<base>+<size> __TEXT=0x123000+0x1000      Section from base address and size')
212061da546Spatrick        return False
213061da546Spatrick
214061da546Spatrick    def __str__(self):
215061da546Spatrick        if self.name:
216061da546Spatrick            if self.end_addr is not None:
217061da546Spatrick                if self.start_addr is not None:
218061da546Spatrick                    return "%s=[0x%16.16x - 0x%16.16x)" % (
219061da546Spatrick                        self.name, self.start_addr, self.end_addr)
220061da546Spatrick            else:
221061da546Spatrick                if self.start_addr is not None:
222061da546Spatrick                    return "%s=0x%16.16x" % (self.name, self.start_addr)
223061da546Spatrick            return self.name
224061da546Spatrick        return "<invalid>"
225061da546Spatrick
226061da546Spatrick
227061da546Spatrickclass Image:
228061da546Spatrick    """A class that represents an executable image and any associated data"""
229061da546Spatrick
230061da546Spatrick    def __init__(self, path, uuid=None):
231061da546Spatrick        self.path = path
232061da546Spatrick        self.resolved_path = None
233061da546Spatrick        self.resolved = False
234061da546Spatrick        self.unavailable = False
235061da546Spatrick        self.uuid = uuid
236061da546Spatrick        self.section_infos = list()
237061da546Spatrick        self.identifier = None
238061da546Spatrick        self.version = None
239061da546Spatrick        self.arch = None
240061da546Spatrick        self.module = None
241061da546Spatrick        self.symfile = None
242061da546Spatrick        self.slide = None
243061da546Spatrick
244061da546Spatrick    @classmethod
245061da546Spatrick    def InitWithSBTargetAndSBModule(cls, target, module):
246061da546Spatrick        '''Initialize this Image object with a module from a target.'''
247061da546Spatrick        obj = cls(module.file.fullpath, module.uuid)
248061da546Spatrick        obj.resolved_path = module.platform_file.fullpath
249061da546Spatrick        obj.resolved = True
250061da546Spatrick        for section in module.sections:
251061da546Spatrick            symb_section = Section.InitWithSBTargetAndSBSection(
252061da546Spatrick                target, section)
253061da546Spatrick            if symb_section:
254061da546Spatrick                obj.section_infos.append(symb_section)
255061da546Spatrick        obj.arch = module.triple
256061da546Spatrick        obj.module = module
257061da546Spatrick        obj.symfile = None
258061da546Spatrick        obj.slide = None
259061da546Spatrick        return obj
260061da546Spatrick
261061da546Spatrick    def dump(self, prefix):
262061da546Spatrick        print("%s%s" % (prefix, self))
263061da546Spatrick
264061da546Spatrick    def debug_dump(self):
265061da546Spatrick        print('path = "%s"' % (self.path))
266061da546Spatrick        print('resolved_path = "%s"' % (self.resolved_path))
267061da546Spatrick        print('resolved = %i' % (self.resolved))
268061da546Spatrick        print('unavailable = %i' % (self.unavailable))
269061da546Spatrick        print('uuid = %s' % (self.uuid))
270061da546Spatrick        print('section_infos = %s' % (self.section_infos))
271061da546Spatrick        print('identifier = "%s"' % (self.identifier))
272061da546Spatrick        print('version = %s' % (self.version))
273061da546Spatrick        print('arch = %s' % (self.arch))
274061da546Spatrick        print('module = %s' % (self.module))
275061da546Spatrick        print('symfile = "%s"' % (self.symfile))
276061da546Spatrick        print('slide = %i (0x%x)' % (self.slide, self.slide))
277061da546Spatrick
278061da546Spatrick    def __str__(self):
279061da546Spatrick        s = ''
280061da546Spatrick        if self.uuid:
281061da546Spatrick            s += "%s " % (self.get_uuid())
282061da546Spatrick        if self.arch:
283061da546Spatrick            s += "%s " % (self.arch)
284061da546Spatrick        if self.version:
285061da546Spatrick            s += "%s " % (self.version)
286061da546Spatrick        resolved_path = self.get_resolved_path()
287061da546Spatrick        if resolved_path:
288061da546Spatrick            s += "%s " % (resolved_path)
289061da546Spatrick        for section_info in self.section_infos:
290061da546Spatrick            s += ", %s" % (section_info)
291061da546Spatrick        if self.slide is not None:
292061da546Spatrick            s += ', slide = 0x%16.16x' % self.slide
293061da546Spatrick        return s
294061da546Spatrick
295061da546Spatrick    def add_section(self, section):
296061da546Spatrick        # print "added '%s' to '%s'" % (section, self.path)
297061da546Spatrick        self.section_infos.append(section)
298061da546Spatrick
299061da546Spatrick    def get_section_containing_load_addr(self, load_addr):
300061da546Spatrick        for section_info in self.section_infos:
301061da546Spatrick            if section_info.contains(load_addr):
302061da546Spatrick                return section_info
303061da546Spatrick        return None
304061da546Spatrick
305061da546Spatrick    def get_resolved_path(self):
306061da546Spatrick        if self.resolved_path:
307061da546Spatrick            return self.resolved_path
308061da546Spatrick        elif self.path:
309061da546Spatrick            return self.path
310061da546Spatrick        return None
311061da546Spatrick
312061da546Spatrick    def get_resolved_path_basename(self):
313061da546Spatrick        path = self.get_resolved_path()
314061da546Spatrick        if path:
315061da546Spatrick            return os.path.basename(path)
316061da546Spatrick        return None
317061da546Spatrick
318061da546Spatrick    def symfile_basename(self):
319061da546Spatrick        if self.symfile:
320061da546Spatrick            return os.path.basename(self.symfile)
321061da546Spatrick        return None
322061da546Spatrick
323061da546Spatrick    def has_section_load_info(self):
324061da546Spatrick        return self.section_infos or self.slide is not None
325061da546Spatrick
326061da546Spatrick    def load_module(self, target):
327061da546Spatrick        if self.unavailable:
328061da546Spatrick            return None  # We already warned that we couldn't find this module, so don't return an error string
329061da546Spatrick        # Load this module into "target" using the section infos to
330061da546Spatrick        # set the section load addresses
331061da546Spatrick        if self.has_section_load_info():
332061da546Spatrick            if target:
333061da546Spatrick                if self.module:
334061da546Spatrick                    if self.section_infos:
335061da546Spatrick                        num_sections_loaded = 0
336061da546Spatrick                        for section_info in self.section_infos:
337061da546Spatrick                            if section_info.name:
338061da546Spatrick                                section = self.module.FindSection(
339061da546Spatrick                                    section_info.name)
340061da546Spatrick                                if section:
341061da546Spatrick                                    error = target.SetSectionLoadAddress(
342061da546Spatrick                                        section, section_info.start_addr)
343061da546Spatrick                                    if error.Success():
344061da546Spatrick                                        num_sections_loaded += 1
345061da546Spatrick                                    else:
346061da546Spatrick                                        return 'error: %s' % error.GetCString()
347061da546Spatrick                                else:
348061da546Spatrick                                    return 'error: unable to find the section named "%s"' % section_info.name
349061da546Spatrick                            else:
350061da546Spatrick                                return 'error: unable to find "%s" section in "%s"' % (
351061da546Spatrick                                    range.name, self.get_resolved_path())
352061da546Spatrick                        if num_sections_loaded == 0:
353061da546Spatrick                            return 'error: no sections were successfully loaded'
354061da546Spatrick                    else:
355061da546Spatrick                        err = target.SetModuleLoadAddress(
356061da546Spatrick                            self.module, self.slide)
357061da546Spatrick                        if err.Fail():
358061da546Spatrick                            return err.GetCString()
359061da546Spatrick                    return None
360061da546Spatrick                else:
361061da546Spatrick                    return 'error: invalid module'
362061da546Spatrick            else:
363061da546Spatrick                return 'error: invalid target'
364061da546Spatrick        else:
365061da546Spatrick            return 'error: no section infos'
366061da546Spatrick
367061da546Spatrick    def add_module(self, target):
368061da546Spatrick        '''Add the Image described in this object to "target" and load the sections if "load" is True.'''
369061da546Spatrick        if target:
370061da546Spatrick            # Try and find using UUID only first so that paths need not match
371061da546Spatrick            # up
372061da546Spatrick            uuid_str = self.get_normalized_uuid_string()
373061da546Spatrick            if uuid_str:
374061da546Spatrick                self.module = target.AddModule(None, None, uuid_str)
375061da546Spatrick            if not self.module:
376061da546Spatrick                self.locate_module_and_debug_symbols()
377061da546Spatrick                if self.unavailable:
378061da546Spatrick                    return None
379061da546Spatrick                resolved_path = self.get_resolved_path()
380061da546Spatrick                self.module = target.AddModule(
381*f6aab3d8Srobert                    resolved_path, None, uuid_str, self.symfile)
382061da546Spatrick            if not self.module:
383061da546Spatrick                return 'error: unable to get module for (%s) "%s"' % (
384061da546Spatrick                    self.arch, self.get_resolved_path())
385061da546Spatrick            if self.has_section_load_info():
386061da546Spatrick                return self.load_module(target)
387061da546Spatrick            else:
388061da546Spatrick                return None  # No sections, the module was added to the target, so success
389061da546Spatrick        else:
390061da546Spatrick            return 'error: invalid target'
391061da546Spatrick
392061da546Spatrick    def locate_module_and_debug_symbols(self):
393061da546Spatrick        # By default, just use the paths that were supplied in:
394061da546Spatrick        # self.path
395061da546Spatrick        # self.resolved_path
396061da546Spatrick        # self.module
397061da546Spatrick        # self.symfile
398061da546Spatrick        # Subclasses can inherit from this class and override this function
399061da546Spatrick        self.resolved = True
400061da546Spatrick        return True
401061da546Spatrick
402061da546Spatrick    def get_uuid(self):
403061da546Spatrick        if not self.uuid and self.module:
404061da546Spatrick            self.uuid = uuid.UUID(self.module.GetUUIDString())
405061da546Spatrick        return self.uuid
406061da546Spatrick
407061da546Spatrick    def get_normalized_uuid_string(self):
408061da546Spatrick        if self.uuid:
409061da546Spatrick            return str(self.uuid).upper()
410061da546Spatrick        return None
411061da546Spatrick
412be691f3bSpatrick    def create_target(self, debugger):
413061da546Spatrick        '''Create a target using the information in this Image object.'''
414061da546Spatrick        if self.unavailable:
415061da546Spatrick            return None
416061da546Spatrick
417061da546Spatrick        if self.locate_module_and_debug_symbols():
418061da546Spatrick            resolved_path = self.get_resolved_path()
419061da546Spatrick            path_spec = lldb.SBFileSpec(resolved_path)
420061da546Spatrick            error = lldb.SBError()
421be691f3bSpatrick            target = debugger.CreateTarget(
422061da546Spatrick                resolved_path, self.arch, None, False, error)
423061da546Spatrick            if target:
424061da546Spatrick                self.module = target.FindModule(path_spec)
425061da546Spatrick                if self.has_section_load_info():
426061da546Spatrick                    err = self.load_module(target)
427061da546Spatrick                    if err:
428061da546Spatrick                        print('ERROR: ', err)
429061da546Spatrick                return target
430061da546Spatrick            else:
431061da546Spatrick                print('error: unable to create a valid target for (%s) "%s"' % (self.arch, self.path))
432061da546Spatrick        else:
433061da546Spatrick            print('error: unable to locate main executable (%s) "%s"' % (self.arch, self.path))
434061da546Spatrick        return None
435061da546Spatrick
436061da546Spatrick
437061da546Spatrickclass Symbolicator:
438061da546Spatrick
439be691f3bSpatrick    def __init__(self, debugger=None, target=None, images=list()):
440be691f3bSpatrick        """A class the represents the information needed to symbolicate
441be691f3bSpatrick        addresses in a program.
442be691f3bSpatrick
443be691f3bSpatrick        Do not call this initializer directly, but rather use the factory
444be691f3bSpatrick        methods.
445be691f3bSpatrick        """
446be691f3bSpatrick        self.debugger = debugger
447be691f3bSpatrick        self.target = target
448be691f3bSpatrick        self.images = images  # a list of images to be used when symbolicating
449061da546Spatrick        self.addr_mask = 0xffffffffffffffff
450061da546Spatrick
451061da546Spatrick    @classmethod
452061da546Spatrick    def InitWithSBTarget(cls, target):
453be691f3bSpatrick        """Initialize a new Symbolicator with an existing SBTarget."""
454be691f3bSpatrick        obj = cls(target=target)
455061da546Spatrick        triple = target.triple
456061da546Spatrick        if triple:
457061da546Spatrick            arch = triple.split('-')[0]
458061da546Spatrick            if "arm" in arch:
459061da546Spatrick                obj.addr_mask = 0xfffffffffffffffe
460061da546Spatrick
461061da546Spatrick        for module in target.modules:
462061da546Spatrick            image = Image.InitWithSBTargetAndSBModule(target, module)
463061da546Spatrick            obj.images.append(image)
464061da546Spatrick        return obj
465061da546Spatrick
466be691f3bSpatrick    @classmethod
467be691f3bSpatrick    def InitWithSBDebugger(cls, debugger, images):
468be691f3bSpatrick        """Initialize a new Symbolicator with an existing debugger and list of
469be691f3bSpatrick        images. The Symbolicator will create the target."""
470be691f3bSpatrick        obj = cls(debugger=debugger, images=images)
471be691f3bSpatrick        return obj
472be691f3bSpatrick
473061da546Spatrick    def __str__(self):
474061da546Spatrick        s = "Symbolicator:\n"
475061da546Spatrick        if self.target:
476061da546Spatrick            s += "Target = '%s'\n" % (self.target)
477061da546Spatrick            s += "Target modules:\n"
478061da546Spatrick            for m in self.target.modules:
479061da546Spatrick                s += str(m) + "\n"
480061da546Spatrick        s += "Images:\n"
481061da546Spatrick        for image in self.images:
482061da546Spatrick            s += '    %s\n' % (image)
483061da546Spatrick        return s
484061da546Spatrick
485061da546Spatrick    def find_images_with_identifier(self, identifier):
486061da546Spatrick        images = list()
487061da546Spatrick        for image in self.images:
488061da546Spatrick            if image.identifier == identifier:
489061da546Spatrick                images.append(image)
490061da546Spatrick        if len(images) == 0:
491061da546Spatrick            regex_text = '^.*\.%s$' % (re.escape(identifier))
492061da546Spatrick            regex = re.compile(regex_text)
493061da546Spatrick            for image in self.images:
494061da546Spatrick                if regex.match(image.identifier):
495061da546Spatrick                    images.append(image)
496061da546Spatrick        return images
497061da546Spatrick
498061da546Spatrick    def find_image_containing_load_addr(self, load_addr):
499061da546Spatrick        for image in self.images:
500061da546Spatrick            if image.get_section_containing_load_addr(load_addr):
501061da546Spatrick                return image
502061da546Spatrick        return None
503061da546Spatrick
504061da546Spatrick    def create_target(self):
505061da546Spatrick        if self.target:
506061da546Spatrick            return self.target
507061da546Spatrick
508061da546Spatrick        if self.images:
509061da546Spatrick            for image in self.images:
510be691f3bSpatrick                self.target = image.create_target(self.debugger)
511061da546Spatrick                if self.target:
512061da546Spatrick                    if self.target.GetAddressByteSize() == 4:
513061da546Spatrick                        triple = self.target.triple
514061da546Spatrick                        if triple:
515061da546Spatrick                            arch = triple.split('-')[0]
516061da546Spatrick                            if "arm" in arch:
517061da546Spatrick                                self.addr_mask = 0xfffffffffffffffe
518061da546Spatrick                    return self.target
519061da546Spatrick        return None
520061da546Spatrick
521061da546Spatrick    def symbolicate(self, load_addr, verbose=False):
522061da546Spatrick        if not self.target:
523061da546Spatrick            self.create_target()
524061da546Spatrick        if self.target:
525061da546Spatrick            live_process = False
526061da546Spatrick            process = self.target.process
527061da546Spatrick            if process:
528061da546Spatrick                state = process.state
529061da546Spatrick                if state > lldb.eStateUnloaded and state < lldb.eStateDetached:
530061da546Spatrick                    live_process = True
531061da546Spatrick            # If we don't have a live process, we can attempt to find the image
532061da546Spatrick            # that a load address belongs to and lazily load its module in the
533061da546Spatrick            # target, but we shouldn't do any of this if we have a live process
534061da546Spatrick            if not live_process:
535061da546Spatrick                image = self.find_image_containing_load_addr(load_addr)
536061da546Spatrick                if image:
537061da546Spatrick                    image.add_module(self.target)
538061da546Spatrick            symbolicated_address = Address(self.target, load_addr)
539061da546Spatrick            if symbolicated_address.symbolicate(verbose):
540061da546Spatrick                if symbolicated_address.so_addr:
541061da546Spatrick                    symbolicated_addresses = list()
542061da546Spatrick                    symbolicated_addresses.append(symbolicated_address)
543061da546Spatrick                    # See if we were able to reconstruct anything?
544061da546Spatrick                    while True:
545061da546Spatrick                        inlined_parent_so_addr = lldb.SBAddress()
546061da546Spatrick                        inlined_parent_sym_ctx = symbolicated_address.sym_ctx.GetParentOfInlinedScope(
547061da546Spatrick                            symbolicated_address.so_addr, inlined_parent_so_addr)
548061da546Spatrick                        if not inlined_parent_sym_ctx:
549061da546Spatrick                            break
550061da546Spatrick                        if not inlined_parent_so_addr:
551061da546Spatrick                            break
552061da546Spatrick
553061da546Spatrick                        symbolicated_address = Address(
554061da546Spatrick                            self.target, inlined_parent_so_addr.GetLoadAddress(
555061da546Spatrick                                self.target))
556061da546Spatrick                        symbolicated_address.sym_ctx = inlined_parent_sym_ctx
557061da546Spatrick                        symbolicated_address.so_addr = inlined_parent_so_addr
558061da546Spatrick                        symbolicated_address.symbolicate(verbose)
559061da546Spatrick
560061da546Spatrick                        # push the new frame onto the new frame stack
561061da546Spatrick                        symbolicated_addresses.append(symbolicated_address)
562061da546Spatrick
563061da546Spatrick                    if symbolicated_addresses:
564061da546Spatrick                        return symbolicated_addresses
565061da546Spatrick        else:
566061da546Spatrick            print('error: no target in Symbolicator')
567061da546Spatrick        return None
568061da546Spatrick
569061da546Spatrick
570061da546Spatrickdef disassemble_instructions(
571061da546Spatrick        target,
572061da546Spatrick        instructions,
573061da546Spatrick        pc,
574061da546Spatrick        insts_before_pc,
575061da546Spatrick        insts_after_pc,
576061da546Spatrick        non_zeroeth_frame):
577061da546Spatrick    lines = list()
578061da546Spatrick    pc_index = -1
579061da546Spatrick    comment_column = 50
580061da546Spatrick    for inst_idx, inst in enumerate(instructions):
581061da546Spatrick        inst_pc = inst.GetAddress().GetLoadAddress(target)
582061da546Spatrick        if pc == inst_pc:
583061da546Spatrick            pc_index = inst_idx
584061da546Spatrick        mnemonic = inst.GetMnemonic(target)
585061da546Spatrick        operands = inst.GetOperands(target)
586061da546Spatrick        comment = inst.GetComment(target)
587061da546Spatrick        lines.append("%#16.16x: %8s %s" % (inst_pc, mnemonic, operands))
588061da546Spatrick        if comment:
589061da546Spatrick            line_len = len(lines[-1])
590061da546Spatrick            if line_len < comment_column:
591061da546Spatrick                lines[-1] += ' ' * (comment_column - line_len)
592061da546Spatrick                lines[-1] += "; %s" % comment
593061da546Spatrick
594061da546Spatrick    if pc_index >= 0:
595061da546Spatrick        # If we are disassembling the non-zeroeth frame, we need to backup the
596061da546Spatrick        # PC by 1
597061da546Spatrick        if non_zeroeth_frame and pc_index > 0:
598061da546Spatrick            pc_index = pc_index - 1
599061da546Spatrick        if insts_before_pc == -1:
600061da546Spatrick            start_idx = 0
601061da546Spatrick        else:
602061da546Spatrick            start_idx = pc_index - insts_before_pc
603061da546Spatrick        if start_idx < 0:
604061da546Spatrick            start_idx = 0
605061da546Spatrick        if insts_before_pc == -1:
606061da546Spatrick            end_idx = inst_idx
607061da546Spatrick        else:
608061da546Spatrick            end_idx = pc_index + insts_after_pc
609061da546Spatrick        if end_idx > inst_idx:
610061da546Spatrick            end_idx = inst_idx
611061da546Spatrick        for i in range(start_idx, end_idx + 1):
612061da546Spatrick            if i == pc_index:
613061da546Spatrick                print(' -> ', lines[i])
614061da546Spatrick            else:
615061da546Spatrick                print('    ', lines[i])
616061da546Spatrick
617061da546Spatrick
618061da546Spatrickdef print_module_section_data(section):
619061da546Spatrick    print(section)
620061da546Spatrick    section_data = section.GetSectionData()
621061da546Spatrick    if section_data:
622061da546Spatrick        ostream = lldb.SBStream()
623061da546Spatrick        section_data.GetDescription(ostream, section.GetFileAddress())
624061da546Spatrick        print(ostream.GetData())
625061da546Spatrick
626061da546Spatrick
627061da546Spatrickdef print_module_section(section, depth):
628061da546Spatrick    print(section)
629061da546Spatrick    if depth > 0:
630061da546Spatrick        num_sub_sections = section.GetNumSubSections()
631061da546Spatrick        for sect_idx in range(num_sub_sections):
632061da546Spatrick            print_module_section(
633061da546Spatrick                section.GetSubSectionAtIndex(sect_idx), depth - 1)
634061da546Spatrick
635061da546Spatrick
636061da546Spatrickdef print_module_sections(module, depth):
637061da546Spatrick    for sect in module.section_iter():
638061da546Spatrick        print_module_section(sect, depth)
639061da546Spatrick
640061da546Spatrick
641061da546Spatrickdef print_module_symbols(module):
642061da546Spatrick    for sym in module:
643061da546Spatrick        print(sym)
644061da546Spatrick
645061da546Spatrick
646be691f3bSpatrickdef Symbolicate(debugger, command_args):
647061da546Spatrick
648061da546Spatrick    usage = "usage: %prog [options] <addr1> [addr2 ...]"
649061da546Spatrick    description = '''Symbolicate one or more addresses using LLDB's python scripting API..'''
650061da546Spatrick    parser = optparse.OptionParser(
651061da546Spatrick        description=description,
652061da546Spatrick        prog='crashlog.py',
653061da546Spatrick        usage=usage)
654061da546Spatrick    parser.add_option(
655061da546Spatrick        '-v',
656061da546Spatrick        '--verbose',
657061da546Spatrick        action='store_true',
658061da546Spatrick        dest='verbose',
659061da546Spatrick        help='display verbose debug info',
660061da546Spatrick        default=False)
661061da546Spatrick    parser.add_option(
662061da546Spatrick        '-p',
663061da546Spatrick        '--platform',
664061da546Spatrick        type='string',
665061da546Spatrick        metavar='platform',
666061da546Spatrick        dest='platform',
667061da546Spatrick        help='Specify the platform to use when creating the debug target. Valid values include "localhost", "darwin-kernel", "ios-simulator", "remote-freebsd", "remote-macosx", "remote-ios", "remote-linux".')
668061da546Spatrick    parser.add_option(
669061da546Spatrick        '-f',
670061da546Spatrick        '--file',
671061da546Spatrick        type='string',
672061da546Spatrick        metavar='file',
673061da546Spatrick        dest='file',
674061da546Spatrick        help='Specify a file to use when symbolicating')
675061da546Spatrick    parser.add_option(
676061da546Spatrick        '-a',
677061da546Spatrick        '--arch',
678061da546Spatrick        type='string',
679061da546Spatrick        metavar='arch',
680061da546Spatrick        dest='arch',
681061da546Spatrick        help='Specify a architecture to use when symbolicating')
682061da546Spatrick    parser.add_option(
683061da546Spatrick        '-s',
684061da546Spatrick        '--slide',
685061da546Spatrick        type='int',
686061da546Spatrick        metavar='slide',
687061da546Spatrick        dest='slide',
688061da546Spatrick        help='Specify the slide to use on the file specified with the --file option',
689061da546Spatrick        default=None)
690061da546Spatrick    parser.add_option(
691061da546Spatrick        '--section',
692061da546Spatrick        type='string',
693061da546Spatrick        action='append',
694061da546Spatrick        dest='section_strings',
695061da546Spatrick        help='specify <sect-name>=<start-addr> or <sect-name>=<start-addr>-<end-addr>')
696061da546Spatrick    try:
697061da546Spatrick        (options, args) = parser.parse_args(command_args)
698061da546Spatrick    except:
699061da546Spatrick        return
700be691f3bSpatrick    symbolicator = Symbolicator(debugger)
701061da546Spatrick    images = list()
702061da546Spatrick    if options.file:
703061da546Spatrick        image = Image(options.file)
704061da546Spatrick        image.arch = options.arch
705061da546Spatrick        # Add any sections that were specified with one or more --section
706061da546Spatrick        # options
707061da546Spatrick        if options.section_strings:
708061da546Spatrick            for section_str in options.section_strings:
709061da546Spatrick                section = Section()
710061da546Spatrick                if section.set_from_string(section_str):
711061da546Spatrick                    image.add_section(section)
712061da546Spatrick                else:
713061da546Spatrick                    sys.exit(1)
714061da546Spatrick        if options.slide is not None:
715061da546Spatrick            image.slide = options.slide
716061da546Spatrick        symbolicator.images.append(image)
717061da546Spatrick
718061da546Spatrick    target = symbolicator.create_target()
719061da546Spatrick    if options.verbose:
720061da546Spatrick        print(symbolicator)
721061da546Spatrick    if target:
722061da546Spatrick        for addr_str in args:
723061da546Spatrick            addr = int(addr_str, 0)
724061da546Spatrick            symbolicated_addrs = symbolicator.symbolicate(
725061da546Spatrick                addr, options.verbose)
726061da546Spatrick            for symbolicated_addr in symbolicated_addrs:
727061da546Spatrick                print(symbolicated_addr)
728061da546Spatrick            print()
729061da546Spatrick    else:
730061da546Spatrick        print('error: no target for %s' % (symbolicator))
731061da546Spatrick
732061da546Spatrickif __name__ == '__main__':
733061da546Spatrick    # Create a new debugger instance
734be691f3bSpatrick    debugger = lldb.SBDebugger.Create()
735be691f3bSpatrick    Symbolicate(debugger, sys.argv[1:])
736be691f3bSpatrick    SBDebugger.Destroy(debugger)
737