Android: kill all adb instances before trying to restart usb.

Android: kill all adb instances before trying to restart usb.

On some bots, there are adb instances left over from previous runs.
They don't play well with restarting the usb, and devices eventually "disappear".
This tries to kill all adb instances before restarting usb.

BUG=311058
TBR=navabi@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=231035

[bulach, 2013-10-24 10:58:29]

ptal

[tonyg, 2013-10-24 14:23:53]

lgtm

[navabi, 2013-10-24 18:48:23]

lgtm

[commit-bot: I haz the power, 2013-10-25 09:19:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bulach@chromium.org/39833003/1

[commit-bot: I haz the power, 2013-10-25 09:36:22]

Retried try job too often on chromium_presubmit for step(s) presubmit
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=chromium_p...

[bulach, 2013-10-25 10:09:32]

presubmit fail because the lgtm came from @google :)
trying to CQ with a TBR to navabi@chromium

[commit-bot: I haz the power, 2013-10-25 10:10:51]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bulach@chromium.org/39833003/1

[commit-bot: I haz the power, 2013-10-25 12:07:15]

Retried try job too often on win_rel for step(s) unit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-10-25 12:23:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bulach@chromium.org/39833003/1

[commit-bot: I haz the power, 2013-10-25 16:32:45]

Change committed as 231035

[Isaac (away), 2013-11-01 19:12:31]

This code it sends all three signals without delay, so everything will get a
kill -9.  AFAICT, this CL is equivalent to "pkilll -9 adb".

If we want to be more considerate, you'd need to add a sleep and poll to
determine whether proc is alive.  Something like:

30	def pid_exists(pid):
31	  """Returns True if there is a process in the system with given |pid|."""
32	  try:
33	    os.kill(pid, 0)
34	  except OSError as error:
35	    if error.errno == errno.EPERM:
36	      return True
37	    elif error.errno == errno.ESRCH:
38	      return False
39	    raise
40	  return True

def KillAll(match_str):
  procs = // get list of pids with match_str (using pgrep or psutils)
  for p in procs:
    os.kill(p)
  start_time = time.time()
  while procs and time.time() - start_time < 5:
    time.sleep(.1)
    procs = [p for p in procs if pid_exists(p)]
  for p in procs:
    os.kill(p, signal.SIGKILL)
  while procs and time.time() - start_time < 2:
    time.sleep(.1)
    procs = [p for p in procs if pid_exists(p)]


Based on https://codereview.chromium.org/56063002/, the current string match may
be overzealous.

[bulach, 2013-11-04 17:39:14]

I'll take on the improvements, thanks!