Suspend- function for site_wifitest

Suspend function for site_wifitest

Suspend for a specified number of seconds.  This makes use a if a newly created python script for using the RTC to trigger a suspend, then running a command.  There is both a foreground and background variant.

BUG=none
TEST=Created a test in the Roaming Suite (not yet approved).
Manually ran command on x201, l13 and Andretti.

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=23cfe94

[Sam Leffler, 2010-09-27 16:28:37]

http://codereview.chromium.org/3471015/diff/1/2
File server/site_system_suspend.py (right):

http://codereview.chromium.org/3471015/diff/1/2#newcode10
server/site_system_suspend.py:10: 
style says 2 blank lines between functions/methods

http://codereview.chromium.org/3471015/diff/1/2#newcode39
server/site_system_suspend.py:39: sts = os.waitpid(p.pid, 0)[1]
does this work w/ the recent synchronous suspend operation? e.g. how do you know
you won't wakeup before we suspend or wakeup early?)

[Paul Stewart, 2010-09-27 16:54:39]

http://codereview.chromium.org/3471015/diff/1/2
File server/site_system_suspend.py (right):

http://codereview.chromium.org/3471015/diff/1/2#newcode10
server/site_system_suspend.py:10: 
On 2010/09/27 16:28:37, Sam Leffler wrote:
> style says 2 blank lines between functions/methods

Done.

http://codereview.chromium.org/3471015/diff/1/2#newcode39
server/site_system_suspend.py:39: sts = os.waitpid(p.pid, 0)[1]
On 2010/09/27 16:28:37, Sam Leffler wrote:
> does this work w/ the recent synchronous suspend operation? e.g. how do you
know
> you won't wakeup before we suspend or wakeup early?)

It should.  Before exit of the powerd_suspend script, it has already sent off a
"we have resumed from suspend" DBus event, so I assume on exit we should have
resumed.  Since this script should be run on a relatively quiescent DUT, I don't
currently consider other activity that would wake us up early (e.g, user
activity).  If I'm missing something here, let me know.

[Sameer Nanda, 2010-09-27 17:00:49]

Drive by comments: there already are utils functions to set wake alarm
(utils.set_wake_alarm) and suspend the machine (utils.suspend_to_ram).  See
power_Resume test for usage.

[Paul Stewart, 2010-09-27 17:29:59]

http://codereview.chromium.org/3471015/diff/1/2
File server/site_system_suspend.py (right):

http://codereview.chromium.org/3471015/diff/1/2#newcode39
server/site_system_suspend.py:39: sts = os.waitpid(p.pid, 0)[1]
Pursuant to our in-person conversation, I've added a comment for this module
that highlights the fact that this module doesn't account for the time it takes
to bring the machine to suspend, and that it is the caller's responsibility to
estimate an upper bound for this.

[Paul Stewart, 2010-09-27 17:35:21]

Thanks Sameer.  I spotted this, but was unsure how to incorporate this into
non-autotest-client code since client/bin/base_utils.py includes autotest
libraries.  The code below is heavily based on that work, though.

[Sam Leffler, 2010-09-27 17:37:53]

LGTM but please resolve Sameer's comments.

[Sameer Nanda, 2010-09-27 20:31:38]

The code as it is looks fine so I am ok with checking it in.

I am still concerned about maintainability since powerd_suspend is an internal
script whose behavior may change over time and we will forget to fix this one
test.  Maybe you want to move set_wake_alarm and suspend_to_ram from client/bin
to client/comon_lib?

[Paul Stewart, 2010-09-28 22:47:16]

On 2010/09/27 20:31:38, Sameer Nanda wrote:
> Maybe you want to move set_wake_alarm and suspend_to_ram from client/bin
> to client/comon_lib?

I've given this a shot, but got dead-ended on the fact that their core these
functions use the "utils.system()" and "utils.write_one_line()" commands. 
Unless you consent to me not using these in common_lib, I still end up with the
problem of having to download all the autotest deps.  Otherwise, since
power_Resume is the only one to use these functions so far, I have a CL that
changes its references to the new ones, as well as moving the code out and
removing the "utils." references.

http://codereview.chromium.org/3516004 should be in your inbox.

[Sameer Nanda, 2010-09-28 22:50:39]

On 2010/09/28 22:47:16, Paul Stewart wrote:
> On 2010/09/27 20:31:38, Sameer Nanda wrote:
> > Maybe you want to move set_wake_alarm and suspend_to_ram from client/bin
> > to client/comon_lib?
> 
> I've given this a shot, but got dead-ended on the fact that their core these
> functions use the "utils.system()" and "utils.write_one_line()" commands. 
> Unless you consent to me not using these in common_lib, I still end up with
the
> problem of having to download all the autotest deps.  Otherwise, since
> power_Resume is the only one to use these functions so far, I have a CL that
> changes its references to the new ones, as well as moving the code out and
> removing the "utils." references.
> 
> http://codereview.chromium.org/3516004 should be in your inbox.

LGTM'ed the other CL.

[Paul Stewart, 2010-09-29 06:26:46]

Sam, Sameer, PTAL.  I've incorporated the change I've done to the the autotest
client common libs, so the suspend code is only in one place, and both
power_Resume and the command line tester I wrote use the same library code.

[Sameer Nanda, 2010-09-29 15:36:06]

LGTM w/ nit

http://codereview.chromium.org/3471015/diff/19001/20002
File server/site_wifitest.py (right):

http://codereview.chromium.org/3471015/diff/19001/20002#newcode776
server/site_wifitest.py:776: 
need 2 blank lines between methods per autotest coding style.