Small fixes as found in staging.

Small fixes as found in staging.

- Fix UninstallAPK(), it was broken.
- Fix WaitForDevice() to not crash when the device is not accessible.
- Switch WaitUntilFullyBooted() from sys.boot_completed to init.svc.bootanim.
  sys.boot_completed is not available on all kernels.
- Improve resiliency when a device reboots. It's been observed on the fleet that
  a device can be offline for a surprising amount of time (>10s) while
  rebooting. Change the code to cope with this.
- Better handle race conditions when reading output from Reboot(), Root() and
  Unroot(). It would have been better if adbd didn't have the race condition in
  the first place but we have to live with it.
- Add logging in high.py.
- Remove CLSE logging noise in adb_protocol.

R=stip@chromium.org
BUG=chromium:525873
Committed: https://github.com/luci/luci-py/commit/efb6315aa8359276e5d10a50406457cb64bd3d59

[ghost stip (do not use), 2015-11-03 18:50:47]

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
File appengine/third_party/python-adb/adb/adb_commands_safe.py (right):

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
appengine/third_party/python-adb/adb/adb_commands_safe.py:638: self.Close()
*sigh* context manager, man

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
File appengine/third_party/python-adb/adb/common.py (left):

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
appengine/third_party/python-adb/adb/common.py:143:
self._HANDLE_CACHE[port_path] = (self._handle, stack)
that seems wrong, wouldn't you still want to add the ._handle to the cache?

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
File appengine/third_party/python-adb/adb/high.py (right):

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
appengine/third_party/python-adb/adb/high.py:621: _LOG.info('%s: %s', cmd, out)
might be cool to abstract this into a helper function somewhere

"run and log if not successful"

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
File appengine/third_party/python-adb/adb/parallel.py (right):

https://chromiumcodereview.appspot.com/1424923006/diff/20001/appengine/third_...
appengine/third_party/python-adb/adb/parallel.py:73: e = result
and I guess we'll just clobber repeated errors

[M-A Ruel, 2015-11-03 22:02:34]

Patchset #1 (id:1) has been deleted

[M-A Ruel, 2015-11-04 18:29:36]

Patchset #2 (id:40001) has been deleted

[M-A Ruel, 2015-11-04 18:39:10]

Forked this CL into 4 CLs. This is the remaining so some comments do not apply
here anymore.

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
File appengine/third_party/python-adb/adb/adb_commands_safe.py (right):

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
appengine/third_party/python-adb/adb/adb_commands_safe.py:638: self.Close()
On 2015/11/03 18:50:46, stip wrote:
> *sigh* context manager, man

It's in a loop, I changed the break at line 623 to a return True to make it
easier to parse the code.

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
File appengine/third_party/python-adb/adb/common.py (left):

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
appengine/third_party/python-adb/adb/common.py:143:
self._HANDLE_CACHE[port_path] = (self._handle, stack)
On 2015/11/03 18:50:46, stip wrote:
> that seems wrong, wouldn't you still want to add the ._handle to the cache?

It's wrong too, I'll get rid of this once I figured out all the cases where this
can happen, which is always related to %"\&@!!! python exceptions.

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
File appengine/third_party/python-adb/adb/high.py (right):

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
appengine/third_party/python-adb/adb/high.py:621: _LOG.info('%s: %s', cmd, out)
On 2015/11/03 18:50:47, stip wrote:
> might be cool to abstract this into a helper function somewhere
> 
> "run and log if not successful"

Yes thinking about that but will leave out for now.

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
File appengine/third_party/python-adb/adb/parallel.py (right):

https://codereview.chromium.org/1424923006/diff/20001/appengine/third_party/p...
appengine/third_party/python-adb/adb/parallel.py:73: e = result
On 2015/11/03 18:50:47, stip wrote:
> and I guess we'll just clobber repeated errors

yes

[M-A Ruel, 2015-11-05 00:57:52]

Fixed .serial -> .serial_number when enumerating devices.

[ghost stip (do not use), 2015-11-05 20:20:26]

lgtm

[M-A Ruel, 2015-11-05 20:27:48]

The CQ bit was checked by maruel@chromium.org

[commit-bot: I haz the power, 2015-11-05 20:27:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1424923006/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1424923006/80001

[commit-bot: I haz the power, 2015-11-05 20:28:41]

Committed patchset #3 (id:80001) as
https://github.com/luci/luci-py/commit/efb6315aa8359276e5d10a50406457cb64bd3d59