extras/misc/stackhandler.py

   1 #!/usr/bin/python
   2 #####################################################################
   3 #             DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
   4 #                    Version 2, December 2004
   5 #
   6 # Copyright (C) 2011-2012 Ludovic Fauvet <etix@videolan.org>
   7 #                         Jean-Baptiste Kempf <jb@videolan.org>
   8 #
   9 # Everyone is permitted to copy and distribute verbatim or modified
  10 # copies of this license document, and changing it is allowed as long
  11 # as the name is changed.
  12 #
  13 #            DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
  14 #   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  15 #
  16 #  0. You just DO WHAT THE FUCK YOU WANT TO.
  17 #####################################################################
  18 #
  19 # This script can be started in two ways:
  20 # - Without any arguments:
  21 #   The script will search for stacktrace in the WORKDIR, process
  22 #   them and dispatch them in their respective subdirectories.
  23 # - With a stacktrace as only argument:
  24 #   The script will write the output on stdout and exit immediately
  25 #   after the stacktrace has been processed.
  26 #   The input file will stay in place, untouched.
  27 #
  28 # NOTE: Due to a bug in the mingw32-binutils > 2.19 the section
  29 #       .gnu_debuglink in the binary file is trimmed thus preventing
  30 #       gdb to find the associated symbols. This script will
  31 #       work around this issue and rerun gdb for each dbg file.
  32 #
  33 #####################################################################
  34
  35 VLC_VERSION         = "2.0.3"
  36 VLC_BIN             = "/home/videolan/vlc/" + VLC_VERSION + "/vlc-" VLC_VERSION + "/vlc.exe"
  37 VLC_BASE_DIR        = "/home/videolan/vlc/" + VLC_VERSION + "/vlc-" + VLC_VERSION + "/"
  38 VLC_SYMBOLS_DIR     = "/home/videolan/vlc/" + VLC_VERSION + "/symbols-" + VLC_VERSION + "/"
  39 WORKDIR             = "/srv/ftp/crashes-win32"
  40 FILE_MATCH          = r"^\d{14}$"
  41 FILE_MAX_SIZE       = 10000
  42 GDB_CMD             = "gdb --exec=%(VLC_BIN)s --symbols=%(VLC_SYMBOLS_DIR)s%(DBG_FILE)s.dbg --batch -x %(BATCH_FILE)s"
  43
  44 EMAIL_TO            = "bugreporter -- videolan.org"
  45 EMAIL_FROM          = "crashes@crash.videolan.org"
  46 EMAIL_SUBJECT       = "[CRASH] New Win32 crash report"
  47 EMAIL_BODY          = \
  48 """
  49 Dear Bug Squasher,
  50
  51 This crash has been reported automatically and might be incomplete and/or broken.
  52 Windows version: %(WIN32_VERSION)s
  53
  54 %(STACKTRACE)s
  55
  56 Truly yours,
  57 a python script.
  58 """
  59
  60 import os, sys, re, tempfile
  61 import string, shlex, subprocess
  62 import smtplib, datetime, shutil
  63 import traceback
  64 from email.mime.text import MIMEText
  65
  66
  67 def processFile(filename):
  68     print "Processing " + filename
  69     global win32_version
  70
  71     f = open(filename, 'r')
  72     # Read (and repair) the input file
  73     content = "".join(filter(lambda x: x in string.printable, f.read()))
  74     f.close()
  75
  76     if os.path.getsize(filename) < 10:
  77         print("File empty")
  78         os.remove(filename)
  79         return
  80
  81     # Check if VLC version match
  82     if not isValidVersion(content):
  83         print("Invalid VLC version")
  84         moveFile(filename, outdated = True)
  85         return
  86
  87     # Get Windows version
  88     win32_version = getWinVersion(content) or 'unknown'
  89
  90     # Map eip <--> library
  91     mapping = mapLibraries(content)
  92     if not mapping:
  93         print("Stacktrace not found")
  94         os.remove(filename)
  95         return
  96
  97     # Associate all eip to their respective lib
  98     # lib1
  99     #     `- 0x6904f020
 100     #      - 0x6927d37c
 101     # lib2
 102     #     `- 0x7e418734
 103     #      - 0x7e418816
 104     #      - 0x7e42bf15
 105     sortedEIP,delta_libs = sortEIP(content,mapping)
 106     # Compute the stacktrace using GDB
 107     eipmap = findSymbols(sortedEIP)
 108     # Generate the body of the email
 109     body = genEmailBody(mapping, eipmap, delta_libs)
 110     # Send the email
 111     sendEmail(body)
 112     # Print the output
 113     print(body)
 114
 115     # Finally archive the stacktrace
 116     moveFile(filename, outdated = False)
 117
 118 def isValidVersion(content):
 119     pattern = re.compile(r"^VLC=%s " % VLC_VERSION, re.MULTILINE)
 120     res = pattern.search(content)
 121     return True if res else False
 122
 123 def getWinVersion(content):
 124     pattern = re.compile(r"^OS=(.*)$", re.MULTILINE)
 125     res = pattern.search(content)
 126     if res is not None:
 127         return res.group(1)
 128     return None
 129
 130 def getDiffAddress(content, name):
 131     plugin_name_section = content.find(name)
 132     if plugin_name_section < 0:
 133         return None
 134
 135     begin_index = content.rfind("\n", 0, plugin_name_section) + 1
 136     end_index = content.find("|", begin_index)
 137
 138     tmp_index = name.rfind('plugins\\')
 139     libname = name[tmp_index :].replace("\\", "/")
 140     full_path = VLC_BASE_DIR + libname
 141
 142     if not os.path.isfile(full_path):
 143         return None
 144
 145     cmd = "objdump -p " + full_path + " |grep ImageBase -|cut -f2-"
 146     p = subprocess.Popen(cmd, stdout=subprocess.PIPE, shell=True).stdout.read().strip()
 147
 148     diff = int(content[begin_index:end_index], 16) - int(p, 16)
 149     return diff
 150
 151 def mapLibraries(content):
 152     stacktrace_section = content.find("[stacktrace]")
 153     if stacktrace_section < 0:
 154         return None
 155
 156     stacklines = content[stacktrace_section:]
 157     stacklines = stacklines.splitlines()
 158     pattern = re.compile(r"^([0-9a-fA-F]+)\|(.+)$")
 159
 160     mapping = []
 161     for line in stacklines:
 162         m = pattern.match(line)
 163         print(line)
 164         if m is not None:
 165             mapping.append(m.group(1, 2))
 166
 167     if len(mapping) == 0:
 168         return None
 169     return mapping
 170
 171
 172 def sortEIP(content, mapping):
 173     # Merge all EIP mapping to the same library
 174     libs = {}
 175     libs_address = {}
 176     for item in mapping:
 177         # Extract the library name (without the full path)
 178         index = item[1].rfind('\\')
 179         libname = item[1][index + 1:]
 180
 181         # Append the eip to its respective lib
 182         if libname not in libs:
 183             libs[libname] = []
 184             diff = getDiffAddress(content, item[1])
 185             if diff is not None:
 186                 libs_address[libname] = diff
 187             else:
 188                 libs_address[libname] = 0
 189
 190         libs[libname].append(int(item[0],16) - libs_address[libname])
 191
 192     return libs,libs_address
 193
 194
 195 def findSymbols(sortedEIP):
 196     eipmap = {}
 197
 198     for k, v in sortedEIP.items():
 199         # Create the gdb batchfile
 200         batchfile = tempfile.NamedTemporaryFile(mode="w")
 201         batchfile.write("set print symbol-filename on\n")
 202
 203         # Append all eip for this lib
 204         for eip in v:
 205             batchfile.write('p/a %s\n' % hex(eip))
 206         batchfile.flush()
 207
 208         # Generate the command line
 209         cmd = GDB_CMD % {"VLC_BIN": VLC_BIN, "VLC_SYMBOLS_DIR": VLC_SYMBOLS_DIR, "DBG_FILE": k, "BATCH_FILE": batchfile.name}
 210         args = shlex.split(cmd)
 211
 212         # Start GDB and get result
 213         p = subprocess.Popen(args, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 214
 215         # Parse result
 216         gdb_pattern = re.compile(r"^\$\d+ = (.+)$")
 217         cnt = 0
 218         while p.poll() == None:
 219             o = p.stdout.readline()
 220             if o != b'':
 221                 o = bytes.decode(o)
 222                 m = gdb_pattern.match(o)
 223                 if m is not None:
 224                     #print("LINE: [%s]" % m.group(1))
 225                     eipmap[v[cnt]] = m.group(1)
 226                     cnt += 1
 227         batchfile.close()
 228     return eipmap
 229
 230
 231 def genEmailBody(mapping, eipmap, delta_libs):
 232     stacktrace = ""
 233     cnt = 0
 234     for item in mapping:
 235         index = item[1].rfind('\\')
 236         libname = item[1][index + 1:]
 237         print(int(item[0],16), delta_libs[libname])
 238         #print(eipmap)
 239         #print(mapping)
 240         stacktrace += "%d. %s [in %s]\n" % (cnt, eipmap[int(item[0],16)-delta_libs[libname]], item[1])
 241         cnt += 1
 242     stacktrace = stacktrace.rstrip('\n')
 243     return EMAIL_BODY % {"STACKTRACE": stacktrace, "WIN32_VERSION": win32_version}
 244
 245
 246 def sendEmail(body):
 247     msg = MIMEText(body)
 248     msg['Subject'] = EMAIL_SUBJECT
 249     msg['From'] = EMAIL_FROM
 250     msg['To'] = EMAIL_TO
 251
 252     # Send the email
 253     s = smtplib.SMTP()
 254     s.connect("127.0.0.1")
 255     s.sendmail(EMAIL_FROM, [EMAIL_TO], msg.as_string())
 256     s.quit()
 257
 258 def moveFile(filename, outdated = False):
 259     today = datetime.datetime.now().strftime("%Y%m%d")
 260     today_path = "%s/%s" % (WORKDIR, today)
 261     if not os.path.isdir(today_path):
 262         os.mkdir(today_path)
 263     if not outdated:
 264         shutil.move(filename, "%s/%s" % (today_path, os.path.basename(filename)))
 265     else:
 266         outdated_path = "%s/outdated/" % today_path
 267         if not os.path.isdir(outdated_path):
 268             os.mkdir(outdated_path)
 269         shutil.move(filename, "%s/%s" % (outdated_path, os.path.basename(filename)))
 270
 271
 272 ### ENTRY POINT ###
 273
 274 batch = len(sys.argv) != 2
 275 if batch:
 276     print("Running in batch mode")
 277
 278 input_files = []
 279 if not batch:
 280     if not os.path.isfile(sys.argv[1]):
 281         exit("file does not exists")
 282     input_files.append(sys.argv[1])
 283 else:
 284     file_pattern = re.compile(FILE_MATCH)
 285     entries = os.listdir(WORKDIR)
 286     for entry in entries:
 287         path_entry = WORKDIR + "/" + entry
 288         if not os.path.isfile(path_entry):
 289             continue
 290         if not file_pattern.match(entry):
 291             print(entry)
 292             os.remove(path_entry)
 293             continue
 294         if os.path.getsize(path_entry) > FILE_MAX_SIZE:
 295             print("%s is too big" % entry)
 296             os.remove(path_entry)
 297             continue
 298         input_files.append(path_entry)
 299
 300 if not len(input_files):
 301     exit("Nothing to process")
 302
 303 # Start processing each file
 304 for input_file in input_files:
 305     try:
 306         processFile(input_file)
 307     except Exception as ex:
 308         print(traceback.format_exc())