[Lldb-commits] [lldb] r328990 - Added a BSD archive tool.

Greg Clayton via lldb-commits lldb-commits at lists.llvm.org
Mon Apr 2 10:20:21 PDT 2018


Author: gclayton
Date: Mon Apr  2 10:20:21 2018
New Revision: 328990

URL: http://llvm.org/viewvc/llvm-project?rev=328990&view=rev
Log:
Added a BSD archive tool.

This is a combination stand alone BSD archive tool that can dump BSD archives:

% bsd.py /path/to/foo.a

Search archives for an object file:

% bsd.py --object foo.o bar.a

Dump the symbol definitions found in the __.SYMDEF objects:

% bsd.py --symdef bar.a

Find symbols by name that are listed in the __.SYMDEF objects:

% bsd.py --symbol _Z123 bar.a

Extract objects from BSD archives:

% bsd.py --object foo.o bar.a --extract
% bsd.py --object foo.o bar.a --extract --outfile /tmp/foo.o
% bsd.py --object foo.o bar.a --extract --mtime 0x1234556

It also has installs a new LLDB command line command when imported into LLDB:

(lldb) command script import ~/Dropbox/bin/bsd.py
The "verify-debug-map-objects" command has been installed, type "help verify-debug-map-objects" for detailed help.
(lldb) verify-debug-map-objects a.out

This will iterate through all object files and verify the modification times match for any .o files, it will verify any .o files from BSD archives are found and have matching modification times and print out errors if any are found.


Added:
    lldb/trunk/examples/python/bsd.py   (with props)

Added: lldb/trunk/examples/python/bsd.py
URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/examples/python/bsd.py?rev=328990&view=auto
==============================================================================
--- lldb/trunk/examples/python/bsd.py (added)
+++ lldb/trunk/examples/python/bsd.py Mon Apr  2 10:20:21 2018
@@ -0,0 +1,481 @@
+#!/usr/bin/python
+
+import optparse
+import os
+import shlex
+import struct
+import sys
+
+ARMAG = "!<arch>\n"
+SARMAG = 8
+ARFMAG = "`\n"
+AR_EFMT1 = "#1/"
+
+
+def memdump(src, bytes_per_line=16, address=0):
+    FILTER = ''.join([(len(repr(chr(x))) == 3) and chr(x) or '.'
+                     for x in range(256)])
+    for i in range(0, len(src), bytes_per_line):
+        s = src[i:i+bytes_per_line]
+        hex_bytes = ' '.join(["%02x" % (ord(x)) for x in s])
+        ascii = s.translate(FILTER)
+        print("%#08.8x: %-*s %s" % (address+i, bytes_per_line*3, hex_bytes,
+                                    ascii))
+
+
+class Object(object):
+    def __init__(self, file):
+        def read_str(file, str_len):
+            return file.read(str_len).rstrip('\0 ')
+
+        def read_int(file, str_len, base):
+            return int(read_str(file, str_len), base)
+
+        self.offset = file.tell()
+        self.file = file
+        self.name = read_str(file, 16)
+        self.date = read_int(file, 12, 10)
+        self.uid = read_int(file, 6, 10)
+        self.gid = read_int(file, 6, 10)
+        self.mode = read_int(file, 8, 8)
+        self.size = read_int(file, 10, 10)
+        if file.read(2) != ARFMAG:
+            raise ValueError('invalid BSD object at offset %#08.8x' % (
+                             self.offset))
+        # If we have an extended name read it. Extended names start with
+        name_len = 0
+        if self.name.startswith(AR_EFMT1):
+            name_len = int(self.name[len(AR_EFMT1):], 10)
+            self.name = read_str(file, name_len)
+        self.obj_offset = file.tell()
+        self.obj_size = self.size - name_len
+        file.seek(self.obj_size, 1)
+
+    def dump(self, f=sys.stdout, flat=True):
+        if flat:
+            f.write('%#08.8x: %#08.8x %5u %5u %6o %#08.8x %s\n' % (self.offset,
+                    self.date, self.uid, self.gid, self.mode, self.size,
+                    self.name))
+        else:
+            f.write('%#08.8x: \n' % self.offset)
+            f.write(' name = "%s"\n' % self.name)
+            f.write(' date = %#08.8x\n' % self.date)
+            f.write('  uid = %i\n' % self.uid)
+            f.write('  gid = %i\n' % self.gid)
+            f.write(' mode = %o\n' % self.mode)
+            f.write(' size = %#08.8x\n' % (self.size))
+            self.file.seek(self.obj_offset, 0)
+            first_bytes = self.file.read(4)
+            f.write('bytes = ')
+            memdump(first_bytes)
+
+    def get_bytes(self):
+        saved_pos = self.file.tell()
+        self.file.seek(self.obj_offset, 0)
+        bytes = self.file.read(self.obj_size)
+        self.file.seek(saved_pos, 0)
+        return bytes
+
+
+class StringTable(object):
+    def __init__(self, bytes):
+        self.bytes = bytes
+
+    def get_string(self, offset):
+        length = len(self.bytes)
+        if offset >= length:
+            return None
+        return self.bytes[offset:self.bytes.find('\0', offset)]
+
+
+class Archive(object):
+    def __init__(self, path):
+        self.path = path
+        self.file = open(path, 'r')
+        self.objects = []
+        self.offset_to_object = {}
+        if self.file.read(SARMAG) != ARMAG:
+            print("error: file isn't a BSD archive")
+        while True:
+            try:
+                self.objects.append(Object(self.file))
+            except ValueError:
+                break
+
+    def get_object_at_offset(self, offset):
+        if offset in self.offset_to_object:
+            return self.offset_to_object[offset]
+        for obj in self.objects:
+            if obj.offset == offset:
+                self.offset_to_object[offset] = obj
+                return obj
+        return None
+
+    def find(self, name, mtime=None, f=sys.stdout):
+        '''
+            Find an object(s) by name with optional modification time. There
+            can be multple objects with the same name inside and possibly with
+            the same modification time within a BSD archive so clients must be
+            prepared to get multiple results.
+        '''
+        matches = []
+        for obj in self.objects:
+            if obj.name == name and (mtime is None or mtime == obj.date):
+                matches.append(obj)
+        return matches
+
+    @classmethod
+    def dump_header(self, f=sys.stdout):
+        f.write('            DATE       UID   GID   MODE   SIZE       NAME\n')
+        f.write('            ---------- ----- ----- ------ ---------- '
+                '--------------\n')
+
+    def get_symdef(self):
+        def get_uint32(file):
+            '''Extract a uint32_t from the current file position.'''
+            v, = struct.unpack('=I', file.read(4))
+            return v
+
+        for obj in self.objects:
+            symdef = []
+            if obj.name.startswith("__.SYMDEF"):
+                self.file.seek(obj.obj_offset, 0)
+                ranlib_byte_size = get_uint32(self.file)
+                num_ranlib_structs = ranlib_byte_size/8
+                str_offset_pairs = []
+                for _ in range(num_ranlib_structs):
+                    strx = get_uint32(self.file)
+                    offset = get_uint32(self.file)
+                    str_offset_pairs.append((strx, offset))
+                strtab_len = get_uint32(self.file)
+                strtab = StringTable(self.file.read(strtab_len))
+                for s in str_offset_pairs:
+                    symdef.append((strtab.get_string(s[0]), s[1]))
+            return symdef
+
+    def get_object_dicts(self):
+        '''
+            Returns an array of object dictionaries that contain they following
+            keys:
+                'object': the actual bsd.Object instance
+                'symdefs': an array of symbol names that the object contains
+                           as found in the "__.SYMDEF" item in the archive
+        '''
+        symdefs = self.get_symdef()
+        symdef_dict = {}
+        if symdefs:
+            for (name, offset) in symdefs:
+                if offset in symdef_dict:
+                    object_dict = symdef_dict[offset]
+                else:
+                    object_dict = {
+                        'object': self.get_object_at_offset(offset),
+                        'symdefs': []
+                    }
+                    symdef_dict[offset] = object_dict
+                object_dict['symdefs'].append(name)
+        object_dicts = []
+        for offset in sorted(symdef_dict):
+            object_dicts.append(symdef_dict[offset])
+        return object_dicts
+
+    def dump(self, f=sys.stdout, flat=True):
+        f.write('%s:\n' % self.path)
+        if flat:
+            self.dump_header(f=f)
+        for obj in self.objects:
+            obj.dump(f=f, flat=flat)
+
+
+def main():
+    parser = optparse.OptionParser(
+        prog='bsd',
+        description='Utility for BSD archives')
+    parser.add_option(
+        '--object',
+        type='string',
+        dest='object_name',
+        default=None,
+        help=('Specify the name of a object within the BSD archive to get '
+              'information on'))
+    parser.add_option(
+        '-s', '--symbol',
+        type='string',
+        dest='find_symbol',
+        default=None,
+        help=('Specify the name of a symbol within the BSD archive to get '
+              'information on from SYMDEF'))
+    parser.add_option(
+        '--symdef',
+        action='store_true',
+        dest='symdef',
+        default=False,
+        help=('Dump the information in the SYMDEF.'))
+    parser.add_option(
+        '-v', '--verbose',
+        action='store_true',
+        dest='verbose',
+        default=False,
+        help='Enable verbose output')
+    parser.add_option(
+        '-e', '--extract',
+        action='store_true',
+        dest='extract',
+        default=False,
+        help=('Specify this to extract the object specified with the --object '
+              'option. There must be only one object with a matching name or '
+              'the --mtime option must be specified to uniquely identify a '
+              'single object.'))
+    parser.add_option(
+        '-m', '--mtime',
+        type='int',
+        dest='mtime',
+        default=None,
+        help=('Specify the modification time of the object an object. This '
+              'option is used with either the --object or --extract options.'))
+    parser.add_option(
+        '-o', '--outfile',
+        type='string',
+        dest='outfile',
+        default=None,
+        help=('Specify a different name or path for the file to extract when '
+              'using the --extract option. If this option isn\'t specified, '
+              'then the extracted object file will be extracted into the '
+              'current working directory if a file doesn\'t already exist '
+              'with that name.'))
+
+    (options, args) = parser.parse_args(sys.argv[1:])
+
+    for path in args:
+        archive = Archive(path)
+        if options.object_name:
+            print('%s:\n' % (path))
+            matches = archive.find(options.object_name, options.mtime)
+            if matches:
+                dump_all = True
+                if options.extract:
+                    if len(matches) == 1:
+                        dump_all = False
+                        if options.outfile is None:
+                            outfile_path = matches[0].name
+                        else:
+                            outfile_path = options.outfile
+                        if os.path.exists(outfile_path):
+                            print('error: outfile "%s" already exists' % (
+                              outfile_path))
+                        else:
+                            print('Saving file to "%s"...' % (outfile_path))
+                            with open(outfile_path, 'w') as outfile:
+                                outfile.write(matches[0].get_bytes())
+                    else:
+                        print('error: multiple objects match "%s". Specify '
+                              'the modification time using --mtime.' % (
+                                options.object_name))
+                if dump_all:
+                    for obj in matches:
+                        obj.dump(flat=False)
+            else:
+                print('error: object "%s" not found in archive' % (
+                      options.object_name))
+        elif options.find_symbol:
+            symdefs = archive.get_symdef()
+            if symdefs:
+                success = False
+                for (name, offset) in symdefs:
+                    obj = archive.get_object_at_offset(offset)
+                    if name == options.find_symbol:
+                        print('Found "%s" in:' % (options.find_symbol))
+                        obj.dump(flat=False)
+                        success = True
+                if not success:
+                    print('Didn\'t find "%s" in any objects' % (
+                          options.find_symbol))
+            else:
+                print("error: no __.SYMDEF was found")
+        elif options.symdef:
+            object_dicts = archive.get_object_dicts()
+            for object_dict in object_dicts:
+                object_dict['object'].dump(flat=False)
+                print("symbols:")
+                for name in object_dict['symdefs']:
+                    print("  %s" % (name))
+        else:
+            archive.dump(flat=not options.verbose)
+
+
+if __name__ == '__main__':
+    main()
+
+
+def print_mtime_error(result, dmap_mtime, actual_mtime):
+    print >>result, ("error: modification time in debug map (%#08.8x) doesn't "
+                     "match the .o file modification time (%#08.8x)" % (
+                        dmap_mtime, actual_mtime))
+
+
+def print_file_missing_error(result, path):
+    print >>result, "error: file \"%s\" doesn't exist" % (path)
+
+
+def print_multiple_object_matches(result, object_name, mtime, matches):
+    print >>result, ("error: multiple matches for object '%s' with with "
+                     "modification time %#08.8x:" % (object_name, mtime))
+    Archive.dump_header(f=result)
+    for match in matches:
+        match.dump(f=result, flat=True)
+
+
+def print_archive_object_error(result, object_name, mtime, archive):
+    matches = archive.find(object_name, f=result)
+    if len(matches) > 0:
+        print >>result, ("error: no objects have a modification time that "
+                         "matches %#08.8x for '%s'. Potential matches:" % (
+                            mtime, object_name))
+        Archive.dump_header(f=result)
+        for match in matches:
+            match.dump(f=result, flat=True)
+    else:
+        print >>result, "error: no object named \"%s\" found in archive:" % (
+            object_name)
+        Archive.dump_header(f=result)
+        for match in archive.objects:
+            match.dump(f=result, flat=True)
+        # archive.dump(f=result, flat=True)
+
+
+class VerifyDebugMapCommand:
+    name = "verify-debug-map-objects"
+
+    def create_options(self):
+        usage = "usage: %prog [options]"
+        description = '''This command reports any .o files that are missing
+or whose modification times don't match in the debug map of an executable.'''
+
+        self.parser = optparse.OptionParser(
+            description=description,
+            prog=self.name,
+            usage=usage,
+            add_help_option=False)
+
+        self.parser.add_option(
+            '-e', '--errors',
+            action='store_true',
+            dest='errors',
+            default=False,
+            help="Only show errors")
+
+    def get_short_help(self):
+        return "Verify debug map object files."
+
+    def get_long_help(self):
+        return self.help_string
+
+    def __init__(self, debugger, unused):
+        self.create_options()
+        self.help_string = self.parser.format_help()
+
+    def __call__(self, debugger, command, exe_ctx, result):
+        import lldb
+        # Use the Shell Lexer to properly parse up command options just like a
+        # shell would
+        command_args = shlex.split(command)
+
+        try:
+            (options, args) = self.parser.parse_args(command_args)
+        except:
+            result.SetError("option parsing failed")
+            return
+
+        # Always get program state from the SBExecutionContext passed in
+        target = exe_ctx.GetTarget()
+        if not target.IsValid():
+            result.SetError("invalid target")
+            return
+        archives = {}
+        for module_spec in args:
+            module = target.module[module_spec]
+            if not (module and module.IsValid()):
+                result.SetError('error: invalid module specification: "%s". '
+                                'Specify the full path, basename, or UUID of '
+                                'a module ' % (module_spec))
+                return
+            num_symbols = module.GetNumSymbols()
+            num_errors = 0
+            for i in range(num_symbols):
+                symbol = module.GetSymbolAtIndex(i)
+                if symbol.GetType() != lldb.eSymbolTypeObjectFile:
+                    continue
+                path = symbol.GetName()
+                if not path:
+                    continue
+                # Extract the value of the symbol by dumping the
+                # symbol. The value is the mod time.
+                dmap_mtime = int(str(symbol).split('value = ')
+                                 [1].split(',')[0], 16)
+                if not options.errors:
+                    print >>result, '%s' % (path)
+                if os.path.exists(path):
+                    actual_mtime = int(os.stat(path).st_mtime)
+                    if dmap_mtime != actual_mtime:
+                        num_errors += 1
+                        if options.errors:
+                            print >>result, '%s' % (path),
+                        print_mtime_error(result, dmap_mtime,
+                                          actual_mtime)
+                elif path[-1] == ')':
+                    (archive_path, object_name) = path[0:-1].split('(')
+                    if not archive_path and not object_name:
+                        num_errors += 1
+                        if options.errors:
+                            print >>result, '%s' % (path),
+                        print_file_missing_error(path)
+                        continue
+                    if not os.path.exists(archive_path):
+                        num_errors += 1
+                        if options.errors:
+                            print >>result, '%s' % (path),
+                        print_file_missing_error(archive_path)
+                        continue
+                    if archive_path in archives:
+                        archive = archives[archive_path]
+                    else:
+                        archive = Archive(archive_path)
+                        archives[archive_path] = archive
+                    matches = archive.find(object_name, dmap_mtime)
+                    num_matches = len(matches)
+                    if num_matches == 1:
+                        print >>result, '1 match'
+                        obj = matches[0]
+                        if obj.date != dmap_mtime:
+                            num_errors += 1
+                            if options.errors:
+                                print >>result, '%s' % (path),
+                            print_mtime_error(result, dmap_mtime, obj.date)
+                    elif num_matches == 0:
+                        num_errors += 1
+                        if options.errors:
+                            print >>result, '%s' % (path),
+                        print_archive_object_error(result, object_name,
+                                                   dmap_mtime, archive)
+                    elif num_matches > 1:
+                        num_errors += 1
+                        if options.errors:
+                            print >>result, '%s' % (path),
+                        print_multiple_object_matches(result,
+                                                      object_name,
+                                                      dmap_mtime, matches)
+            if num_errors > 0:
+                print >>result, "%u errors found" % (num_errors)
+            else:
+                print >>result, "No errors detected in debug map"
+
+
+def __lldb_init_module(debugger, dict):
+    # This initializer is being run from LLDB in the embedded command
+    # interpreter.
+    # Add any commands contained in this module to LLDB
+    debugger.HandleCommand(
+        'command script add -c %s.VerifyDebugMapCommand %s' % (
+            __name__, VerifyDebugMapCommand.name))
+    print('The "%s" command has been installed, type "help %s" for detailed '
+          'help.' % (VerifyDebugMapCommand.name, VerifyDebugMapCommand.name))

Propchange: lldb/trunk/examples/python/bsd.py
------------------------------------------------------------------------------
    svn:executable = *




More information about the lldb-commits mailing list