Suspend/resume erratic on feisty

[Derek Broughton]

Ryan Nowakowski wrote:

> I found the bug reports for scenario #1.  Do you think step #7 in
> scenario #2 warrants a *new* bug report?  I haven't seen anything in
> launchpad with regards to the "double" suspend that I'm seeing.

Double suspends are not uncommon - and if the powermanager's method of
suspending is to actually send the ACPI event from each session, that would
be predictable.  I generally suspend by pressing my power button - then I
have two options to resume.  Repressing the power button starts the
wakeup - but then (generally? always?) triggers another hibernate.  If I
touch any of the multimedia buttons, it wakes up cleanly.  I guess for your
particular case, powermanager needs to be smart enough not to trigger a
hibernate event if one has already been triggered.  Alternatively, I really
think it should be possible for ACPI to clear its event queue on resume - I
doubt any outstanding ACPI events could be valid.  Fan, temperature,
battery, lid, power and processor statuses will likely _all_ be changed by
or during hibernation.
-- 
derek