A large retooling of site_wlan_connect.py association code.

A large retooling of site_wifitest.py association code. There are a number of ways the system can fall off the rails during the association process. Many of these are now caught gracefully.

TEST=ran autotests
BUG=

[Sam Leffler, 2010-06-10 21:07:21]

Big improvements; close but please check comments.

http://codereview.chromium.org/2790009/diff/1/2
File server/site_wlan_connect.py (right):

http://codereview.chromium.org/2790009/diff/1/2#newcode12
server/site_wlan_connect.py:12: "org.chromium.flimflam.Manager")
While you're in this code please pull out "org.chromium.flimflam" to a single
place.

http://codereview.chromium.org/2790009/diff/1/2#newcode45
server/site_wlan_connect.py:45: service.ClearProperty('Error')
Need to add a comment explaining that this is a WAR for a flimflam bug (we
think).

http://codereview.chromium.org/2790009/diff/1/2#newcode47
server/site_wlan_connect.py:47: while wait_time < assoc_timeout:
This is re-using the assoc_timeout; does it make sense to define a different
timeout for this case? (can init to assoc_timeout for now)

http://codereview.chromium.org/2790009/diff/1/2#newcode53
server/site_wlan_connect.py:53: print>>sys.stderr, "cleared oruselves out of
'%s' after %3.1f secs" % \
s/oru/our/

http://codereview.chromium.org/2790009/diff/1/2#newcode66
server/site_wlan_connect.py:66: if init_state != 'idle':
maybe move this check into ResetService so it can be called w/o regard for the
current state?

http://codereview.chromium.org/2790009/diff/1/2#newcode86
server/site_wlan_connect.py:86: time.sleep(5)
Why?  This is a flimflam bug /me thinks.

http://codereview.chromium.org/2790009/diff/1/2#newcode103
server/site_wlan_connect.py:103: stdout=subprocess.PIPE).communicate()[0]
Use iw instead of iwlist (trying to stamp out all uses of antiquated tools)

http://codereview.chromium.org/2790009/diff/1/2#newcode109
server/site_wlan_connect.py:109: msgs.seek(0, 2)
So this lets us get all msgs from our run, right?  What about hostap.log?

http://codereview.chromium.org/2790009/diff/1/2#newcode124
server/site_wlan_connect.py:124: 
Do we record attempts that are retried in the keyvals or someplace else visible?
 We should only ever need 1 attempt so we want to know when we need >1 even if
the test passes.

http://codereview.chromium.org/2790009/diff/1/2#newcode148
server/site_wlan_connect.py:148: print "FAIL(config): exited configuration sate
to '%s' %s assoc %3.1f config %3.1f secs" \
80 cols?

[Paul Stewart, 2010-06-10 23:33:59]

http://codereview.chromium.org/2790009/diff/1/2
File server/site_wlan_connect.py (right):

http://codereview.chromium.org/2790009/diff/1/2#newcode45
server/site_wlan_connect.py:45: service.ClearProperty('Error')
FIXED

http://codereview.chromium.org/2790009/diff/1/2#newcode47
server/site_wlan_connect.py:47: while wait_time < assoc_timeout:
FIXED

http://codereview.chromium.org/2790009/diff/1/2#newcode53
server/site_wlan_connect.py:53: print>>sys.stderr, "cleared oruselves out of
'%s' after %3.1f secs" % \
FIXED

http://codereview.chromium.org/2790009/diff/1/2#newcode66
server/site_wlan_connect.py:66: if init_state != 'idle':
FIXED

http://codereview.chromium.org/2790009/diff/1/2#newcode86
server/site_wlan_connect.py:86: time.sleep(5)
Indeed.  I have no idea why.  We do end up in this state though.

http://codereview.chromium.org/2790009/diff/1/2#newcode103
server/site_wlan_connect.py:103: stdout=subprocess.PIPE).communicate()[0]
Fair enough. FIXED.  (See the end of DumpLogs())

http://codereview.chromium.org/2790009/diff/1/2#newcode109
server/site_wlan_connect.py:109: msgs.seek(0, 2)
I can do that too.  Now there is a generic OpenLogs / DumpLogs functionality.

http://codereview.chromium.org/2790009/diff/1/2#newcode124
server/site_wlan_connect.py:124: 
I need more info on where I need to put this extra info.  There is a message
placed in the logs about it.  In general I'm in the dark about what automated
consumers look at in the output.

The current version of the CL now has a a "quirks" string that is put in after
the assoc time and config time.  It looks something like this:

16:29:13 INFO | Check11g: OK 9.0 2.5 ['multiple_attempts', 'clear_error',
'fast_fail'] (assoc and config times in sec, quirks)

http://codereview.chromium.org/2790009/diff/1/2#newcode148
server/site_wlan_connect.py:148: print "FAIL(config): exited configuration sate
to '%s' %s assoc %3.1f config %3.1f secs" \
FIXED

[Sam Leffler, 2010-06-11 14:45:09]

LGTM