native_client_sdk/src/tools/lib/get_shared_deps.py

   1 # Copyright 2014 The Chromium Authors. All rights reserved.
   2 # Use of this source code is governed by a BSD-style license that can be
   3 # found in the LICENSE file.
   4
   5 """Helper script to close over all transitive dependencies of a given .nexe
   6 executable.
   7
   8 e.g. Given
   9 A -> B
  10 B -> C
  11 B -> D
  12 C -> E
  13
  14 where "A -> B" means A depends on B, then GetNeeded(A) will return A, B, C, D
  15 and E.
  16 """
  17
  18 import os
  19 import re
  20 import subprocess
  21
  22 import elf
  23
  24 SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
  25 SDK_DIR = os.path.dirname(os.path.dirname(SCRIPT_DIR))
  26
  27 NeededMatcher = re.compile('^ *NEEDED *([^ ]+)\n$')
  28 FormatMatcher = re.compile('^(.+):\\s*file format (.+)\n$')
  29
  30 RUNNABLE_LD = 'runnable-ld.so'  # Name of the dynamic loader
  31
  32 OBJDUMP_ARCH_MAP = {
  33     # Names returned by Linux's objdump:
  34     'elf64-x86-64': 'x86-64',
  35     'elf32-i386': 'x86-32',
  36     'elf32-little': 'arm',
  37     'elf32-littlearm': 'arm',
  38     # Names returned by old x86_64-nacl-objdump:
  39     'elf64-nacl': 'x86-64',
  40     'elf32-nacl': 'x86-32',
  41     # Names returned by new x86_64-nacl-objdump:
  42     'elf64-x86-64-nacl': 'x86-64',
  43     'elf32-x86-64-nacl': 'x86-64',
  44     'elf32-i386-nacl': 'x86-32',
  45     'elf32-littlearm-nacl': 'arm',
  46 }
  47
  48 # The proper name of the dynamic linker, as kept in the IRT.  This is
  49 # excluded from the nmf file by convention.
  50 LD_NACL_MAP = {
  51     'x86-32': 'ld-nacl-x86-32.so.1',
  52     'x86-64': 'ld-nacl-x86-64.so.1',
  53     'arm': None,
  54 }
  55
  56
  57 class Error(Exception):
  58   '''Local Error class for this file.'''
  59   pass
  60
  61
  62 class NoObjdumpError(Error):
  63   '''Error raised when objdump is needed but not found'''
  64   pass
  65
  66
  67 def GetNeeded(main_files, objdump, lib_path):
  68   '''Collect the list of dependencies for the main_files
  69
  70   Args:
  71     main_files: A list of files to find dependencies of.
  72     objdump: Path to the objdump executable.
  73     lib_path: A list of paths to search for shared libraries.
  74
  75   Returns:
  76     A dict with key=filename and value=architecture. The architecture will be
  77     one of ('x86_32', 'x86_64', 'arm').
  78   '''
  79
  80   dynamic = any(elf.ParseElfHeader(f)[1] for f in main_files)
  81
  82   if dynamic:
  83     return _GetNeededDynamic(main_files, objdump, lib_path)
  84   else:
  85     return _GetNeededStatic(main_files)
  86
  87
  88 def _GetNeededDynamic(main_files, objdump, lib_path):
  89   examined = set()
  90   all_files, unexamined = GleanFromObjdump(main_files, None, objdump, lib_path)
  91   for arch in all_files.itervalues():
  92     if unexamined:
  93       unexamined.add((RUNNABLE_LD, arch))
  94
  95   while unexamined:
  96     files_to_examine = {}
  97
  98     # Take all the currently unexamined files and group them
  99     # by architecture.
 100     for name, arch in unexamined:
 101       files_to_examine.setdefault(arch, []).append(name)
 102
 103     # Call GleanFromObjdump() for each architecture.
 104     needed = set()
 105     for arch, files in files_to_examine.iteritems():
 106       new_files, new_needed = GleanFromObjdump(files, arch, objdump, lib_path)
 107       all_files.update(new_files)
 108       needed |= new_needed
 109
 110     examined |= unexamined
 111     unexamined = needed - examined
 112
 113   # With the runnable-ld.so scheme we have today, the proper name of
 114   # the dynamic linker should be excluded from the list of files.
 115   ldso = [LD_NACL_MAP[arch] for arch in set(OBJDUMP_ARCH_MAP.values())]
 116   for filename, arch in all_files.items():
 117     name = os.path.basename(filename)
 118     if name in ldso:
 119       del all_files[filename]
 120
 121   return all_files
 122
 123
 124 def GleanFromObjdump(files, arch, objdump, lib_path):
 125   '''Get architecture and dependency information for given files
 126
 127   Args:
 128     files: A list of files to examine.
 129         [ '/path/to/my.nexe',
 130           '/path/to/lib64/libmy.so',
 131           '/path/to/mydata.so',
 132           '/path/to/my.data' ]
 133     arch: The architecure we are looking for, or None to accept any
 134           architecture.
 135     objdump: Path to the objdump executable.
 136     lib_path: A list of paths to search for shared libraries.
 137
 138   Returns: A tuple with the following members:
 139     input_info: A dict with key=filename and value=architecture. The
 140         architecture will be one of ('x86_32', 'x86_64', 'arm').
 141     needed: A set of strings formatted as "arch/name".  Example:
 142         set(['x86-32/libc.so', 'x86-64/libgcc.so'])
 143   '''
 144   if not objdump:
 145     raise NoObjdumpError('No objdump executable found!')
 146
 147   full_paths = set()
 148   for filename in files:
 149     if os.path.exists(filename):
 150       full_paths.add(filename)
 151     else:
 152       for path in _FindLibsInPath(filename, lib_path):
 153         full_paths.add(path)
 154
 155   cmd = [objdump, '-p'] + list(full_paths)
 156   env = {'LANG': 'en_US.UTF-8'}
 157   proc = subprocess.Popen(cmd, stdout=subprocess.PIPE,
 158                           stderr=subprocess.PIPE, bufsize=-1,
 159                           env=env)
 160
 161   input_info = {}
 162   found_basenames = set()
 163   needed = set()
 164   output, err_output = proc.communicate()
 165   if proc.returncode:
 166     raise Error('%s\nStdError=%s\nobjdump failed with error code: %d' %
 167                 (output, err_output, proc.returncode))
 168
 169   file_arch = None
 170   for line in output.splitlines(True):
 171     # Objdump should display the architecture first and then the dependencies
 172     # second for each file in the list.
 173     matched = FormatMatcher.match(line)
 174     if matched:
 175       filename = matched.group(1)
 176       file_arch = OBJDUMP_ARCH_MAP[matched.group(2)]
 177       if arch and file_arch != arch:
 178         continue
 179       name = os.path.basename(filename)
 180       found_basenames.add(name)
 181       input_info[filename] = file_arch
 182     matched = NeededMatcher.match(line)
 183     if matched:
 184       if arch and file_arch != arch:
 185         continue
 186       filename = matched.group(1)
 187       new_needed = (filename, file_arch)
 188       needed.add(new_needed)
 189
 190   for filename in files:
 191     if os.path.basename(filename) not in found_basenames:
 192       raise Error('Library not found [%s]: %s' % (arch, filename))
 193
 194   return input_info, needed
 195
 196
 197 def _FindLibsInPath(name, lib_path):
 198   '''Finds the set of libraries matching |name| within lib_path
 199
 200   Args:
 201     name: name of library to find
 202     lib_path: A list of paths to search for shared libraries.
 203
 204   Returns:
 205     A list of system paths that match the given name within the lib_path'''
 206   files = []
 207   for dirname in lib_path:
 208     # The libc.so files in the the glibc toolchain is actually a linker
 209     # script which references libc.so.<SHA1>.  This means the lib.so itself
 210     # does not end up in the NEEDED section for glibc.  However with bionic
 211     # the SONAME is actually libc.so.  If we pass glibc's libc.so to objdump
 212     # if fails to parse it, os this filters out libc.so expept for within
 213     # the bionic toolchain.
 214     # TODO(bradnelson): Remove this once the SONAME in bionic is made to be
 215     # unique in the same it is under glibc:
 216     # https://code.google.com/p/nativeclient/issues/detail?id=3833
 217     rel_dirname = os.path.relpath(dirname, SDK_DIR)
 218     if name == 'libc.so' and 'bionic' not in rel_dirname:
 219       continue
 220     filename = os.path.join(dirname, name)
 221     if os.path.exists(filename):
 222       files.append(filename)
 223   if not files:
 224     raise Error('cannot find library %s' % name)
 225   return files
 226
 227
 228 def _GetNeededStatic(main_files):
 229   needed = {}
 230   for filename in main_files:
 231     arch = elf.ParseElfHeader(filename)[0]
 232     needed[filename] = arch
 233   return needed