Bisection debugging helper script

pydir/bisection-tool.py

+#!/usr/bin/env python2
+import argparse
+import os
+import signal
+import subprocess
+
+class Runner(object):
+  def __init__(self, input_cmd, timeout, comma_join = False):
+    self._input_cmd = input_cmd
+    self._timeout = timeout
+    self._num_tries = 0
+    self._comma_join = comma_join
+
+  def Run(self, included_ranges):
+    def timeout_handler(signum, frame):
+      raise RuntimeError("Timeout")
+
+    self._num_tries += 1
+    cmd = self._input_cmd
+    for i in included_ranges:
+      if isinstance(i, int):

[John, 2016/07/20 22:23:05]

I would always pass a range. If you want a single element, then just pass

(n, n+1)

This simplifies the logic, and makes it easier to read the code (fewer code
paths to keep in memory.)

+        range_str = str(i)
+      else:
+        range_str = "{start}:{end}".format(start=i[0], end=i[1])
+      if self._comma_join:
+        cmd += "," + range_str
+      else:
+        cmd += " -i " + range_str
+    print cmd
+    p = subprocess.Popen(cmd, shell = True, cwd = None,
+      stdout = subprocess.PIPE, stderr = subprocess.PIPE, env = None)
+    if self._timeout != -1:
+      signal.signal(signal.SIGALRM, timeout_handler)
+      signal.alarm(self._timeout)
+
+    try:
+      _, _ = p.communicate()
+      if self._timeout != -1:
+        signal.alarm(0)
+    except:
+      try:
+        os.kill(p.pid, signal.SIGKILL)
+      except OSError:
+        pass
+      print "Timeout"
+      return -9
+    print "Return Code: " + str(p.returncode)
+    return p.returncode
+
+def flatten(tree):
+  if isinstance(tree, list):
+    result = []
+    for node in tree:
+      result.extend(flatten(node))
+    return result
+  else:
+    return [tree] # leaf
+
+def find_crashes(runner, current_interval, include_ranges, find_all):
+  if current_interval[0] == current_interval[1]:
+    return []
+  mid = (current_interval[0] + current_interval[1])/2
+
+  first_half = (current_interval[0], mid)
+  second_half = (mid, current_interval[1])
+
+  exit_code_2 = 0
+
+  exit_code_1 = runner.Run([first_half] + include_ranges)
+  if find_all or exit_code_1 == 0:
+    exit_code_2 = runner.Run([second_half] + include_ranges)
+
+  if exit_code_1 == 0 and exit_code_2 == 0:
+    # Whole range fails but both halves pass
+    # So, some conjunction of functions cause a failure, but none individually.
+    partial_result = flatten(find_crashes(runner, first_half, [second_half]

[John, 2016/07/20 22:23:05]

you **REALLY** shouldn't need to flatten your result. You should figure out why
you're returning a list of lists instead, and fix that.

+                             + include_ranges, find_all))
+    # Heavy list concatenation, but this is insignificant compared to the 
+    # process run times
+    partial_result.extend(flatten(find_crashes(runner, second_half,
+                          partial_result + include_ranges, find_all)))
+    return [partial_result]

[John, 2016/07/20 22:23:05]

why do you return partial result wrapped in a list? partial result is a list
already (or something with an extend method...)

+  else:
+    result = []
+    if exit_code_1 != 0:

[John, 2016/07/20 22:23:05]

I don't see you using "find_all" here, so you might find multiple crashes here
anyways...

+      if first_half[1] == first_half[0] + 1:
+        result.append(first_half[0])
+      else:
+        result.extend(find_crashes(runner, first_half,
+                                     include_ranges, find_all))
+    if exit_code_2 != 0:
+      if second_half[1] == second_half[0] + 1:
+        result.append(second_half[0])
+      else:
+        result.extend(find_crashes(runner, second_half,
+                                     include_ranges, find_all))
+    return result
+
+
+def main():
+  desc = 'Bisection debugging helper script'
+  argparser = argparse.ArgumentParser(description=desc)
+  argparser.add_argument('--cmd', required=True,
+                           dest='cmd',
+                           help='Runnable command')
+  argparser.add_argument('--start', dest='start', default=0,
+                           help='Start of initial range')
+  argparser.add_argument('--end', dest='end', default=50000,
+                           help='End of initial range')
+  argparser.add_argument('--timeout', dest='timeout', default=60,
+                           help='Timeout for each invocation of the input')
+
+  argparser.add_argument('--all', dest='all', action='store_true')
+  argparser.add_argument('--no-all', dest='all', action='store_false')
+  argparser.set_defaults(all=True)
+
+  argparser.add_argument('--comma-join', dest='comma_join', action='store_true')
+  argparser.add_argument('--separate', dest='comma_join', action='store_false')
+  argparser.set_defaults(all=True)
+
+  args = argparser.parse_args()
+
+  fail_list = []
+
+  initial_range = (int(args.start), int(args.end))
+  timeout = int(args.timeout)
+  runner = Runner(args.cmd, timeout, args.comma_join)
+  if runner.Run([initial_range]) != 0:

[John, 2016/07/20 22:23:05]

why not

crashes = find_crashes(runner, initial_range, [], args.all)
if not crashes:
  print "Pass"
else:
  for c in crashes:
    print c

(i.e., you don't need to invoke run here...)

+    fail_list = find_crashes(runner, initial_range, [], args.all)
+  else:
+    print "Pass" # The whole input range works, maybe check subzero build flags?
+
+  if fail_list:
+    print "Failing Functions:"
+    for fail in fail_list:
+      print fail
+  print "Number of tries: " + str(runner._num_tries)
+  # TODO(manasijm): Pretty print, associate these numbers with filenames
+
+if __name__ == '__main__':
+    main()