Adding untrusted crash dump / stack trace tests.

tests/untrusted_crash_dump/decode_dump.py

+#!/usr/bin/python
+# Copyright (c) 2012 The Native Client Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+"""Utility to decode a crash dump generated by untrusted_crash_dump.[ch]
+
+Currently this produces a simple stack trace.
+"""
+
+import json
+import optparse
+import os
+import posixpath
+import subprocess
+import sys
+
+
+def SelectModulePath(options, filename):
+  """Select which path to get a module from.
+
+  Args:
+    options: option object.
+    filename: filename of a module (as appears in phdrs).
+  Returns:
+    Full local path to the file.
+    Derived by consulting the manifest.
+  """
+  # For some names try the main nexe.
+  if options.main_nexe and filename in ['NaClMain', '(null)']:

[Mark Seaborn, 2012/02/13 19:04:08]

You could comment that "NaClMain" is the argv[0] value filled out by sel_main.c.

[bradn, 2012/02/13 23:42:03]

Done.

+    return options.main_nexe
+  filepart = posixpath.basename(filename)
+  nmf_entry = options.nmf_data.get('files', {}).get(filepart, {})
+  # TODO(bradnelson): support x86-64 + arm.
+  nmf_url = nmf_entry.get('x86-32', {}).get('url')
+  # Try filename directly if not in manifest.
+  if not nmf_url:

[Mark Seaborn, 2012/02/13 19:04:08]

"if nmf_url is not None"

[bradn, 2012/02/13 23:42:03]

Done (you mean is None).

+    return filename
+  # Look for the module relative to the manifest, then toolchain.
+  for path in [os.path.dirname(options.nmf), options.toolchain_libs]:
+    if path:

[Mark Seaborn, 2012/02/13 19:04:08]

"if path is not None"

[bradn, 2012/02/13 23:42:03]

Done and changed round to make correct!

+      pfilename = os.path.join(path, nmf_url)
+      if os.path.exists(pfilename):
+        return pfilename
+  # If nothing else, try the path directly.
+  return filename
+
+
+def Addr2Line(options, filename, addr):
+  """Use addr2line to decode a code address.
+
+  Args:
+    options: option object.
+    filename: filename of module that the address is relative to.
+    addr: address.
+  Returns:
+    A list of dicts containing: function, filename, lineno.
+  """
+  filename = SelectModulePath(options, filename)
+  if not os.path.exists(filename):
+    return [{
+        'function': 'Unknown_function',
+        'filename': 'unknown_file',
+        'lineno': -1,
+    }]
+  cmd = [
+      options.addr2line, '-f', '--inlines', '-e', filename, ('0x%08x' % addr),

[Mark Seaborn, 2012/02/13 19:04:08]

Nit: Don't need brackets around '0x%08x' % addr

[bradn, 2012/02/13 23:42:03]

Done.

+  ]
+  p = subprocess.Popen(cmd, stdout=subprocess.PIPE)

[Mark Seaborn, 2012/02/13 19:04:08]

Try to avoid one-char variable names.  e.g. 'p' -> 'proc'

[bradn, 2012/02/13 23:42:03]

Done.

+  p_stdout, p_stderr = p.communicate()
+  assert p.returncode == 0
+  lines = p_stdout.splitlines()
+  assert len(lines) % 2 == 0
+  results = []
+  for index in range(len(lines) / 2):
+    func = lines[index * 2]
+    afilename, lineno = lines[index * 2 + 1].split(':')
+    results.append({
+        'function': func,
+        'filename': afilename,
+        'lineno': int(lineno),
+    })
+  return results
+
+
+def LoadAndDecode(options, core_path):
+  """Given a core.json file, load and embellish with decoded addresses.
+
+  Args:
+    options: options object.
+    core_path: source file containing a dump.
+  Returns:
+
+  """
+  if options.nmf:
+    options.nmf_data = json.loads(open(options.nmf, 'r').read())

[Mark Seaborn, 2012/02/13 19:04:08]

You can do json.load(open(...)) instead.

[bradn, 2012/02/13 23:42:03]

Done.

+  else:
+    options.nmf_data = {}
+  core_text = open(core_path, 'r').read()
+  core = json.loads(core_text)

[Mark Seaborn, 2012/02/13 19:04:08]

Ditto.  json.load()

[bradn, 2012/02/13 23:42:03]

Done.

+  for frame in core['frames']:
+    ip_mapped = frame['ip_mapped']
+    ip_mapped['scopes'] = Addr2Line(
+        options, ip_mapped['file'], ip_mapped['addr'])
+  return core
+
+
+def StackTrace(info):
+  """Convert a decoded core.json dump to a simple stack trace.
+
+  Args:
+    info: core.json info with decoded code addresses.
+  Returns:
+    A list of dicts with filename, lineno, function (deepest first).
+  """
+  trace = []
+  for frame in info['frames']:
+    ip_mapped = frame['ip_mapped']
+    for info in ip_mapped['scopes']:

[Mark Seaborn, 2012/02/13 19:04:08]

Just "frame['ip_mapped']['scopes']"?

[bradn, 2012/02/13 23:42:03]

Done.

+      trace.append(info)
+  return trace
+
+
+def PrintTrace(trace, out):
+  """Print a trace to a file like object.
+
+  Args:
+    trace: A list of [filename, lineno, function] (deepest first).
+    out: file like object to output the trace to.
+  """
+  for scope in trace:
+    out.write('%s at %s:%d\n' % (
+        scope['function'],
+        scope['filename'],
+        scope['lineno']))
+
+
+def Main(args):
+  parser = optparse.OptionParser(
+      usage='USAGE: %prog [options] <core.json> <trace>')

[Mark Seaborn, 2012/02/13 19:04:08]

You show two args here but check for one later.

[bradn, 2012/02/13 23:42:03]

Fixed (changed at some point to one)

+  parser.add_option('-m', '--main-nexe', dest='main_nexe',
+                    help='nexe to resolve NaClMain references from')
+  parser.add_option('-n', '--nmf', dest='nmf',
+                    help='nmf to resolve references from')
+  parser.add_option('-a', '--addr2line', dest='addr2line',
+                    help='path to appropriate addr2line')
+  parser.add_option('-l', '--toolchain-libs', dest='toolchain_libs',
+                    help='path to the toolchain libraries')
+  options, args = parser.parse_args(args)
+  if len(args) != 1:
+    parser.print_help()
+    sys.exit(1)
+  info = LoadAndDecode(options, args[0])
+  trace = StackTrace(info)
+  PrintTrace(trace, sys.stdout)
+
+
+if __name__ == '__main__':
+  Main(sys.argv[1:])