#10856 NORM 11.2.0-: Idle suspend fails on os18 in Sugar. OK in Gnome

[Zarro Boogs per Child]

#10856: Idle suspend fails on os18 in Sugar. OK in Gnome
----------------------------------------------+-----------------------------
           Reporter:  mavrothal               |       Owner:  pgf                              
               Type:  defect                  |      Status:  new                              
           Priority:  normal                  |   Milestone:  11.2.0-final                     
          Component:  power manager (powerd)  |     Version:  Development build as of this date
         Resolution:                          |    Keywords:                                   
        Next_action:  diagnose                |    Verified:  0                                
Deployment_affected:                          |   Blockedby:                                   
           Blocking:                          |  
----------------------------------------------+-----------------------------

Comment(by pgf):

 i see.  thanks for the trace.

 wlan packet arrivals are treated exactly the same as keystrokes -- i.e.,
 they count as "activity".  that's what this does, at line 1449:
 {{{
                 selfinject fake_useractive $lastwakeup "$wakeupsource"
 }}}

 (this is in contrast to, say, battery wakeups, which simply roll around
 and go back to sleep, or the dim or blank rtc wakeups, which perform their
 action and sleep immediately.)

 when the main event loop gets the fake_useractive event, it does take care
 to not brighten the screen for certain wakeup reasons, including
 wlanpacket.  but this does cause the time-to-dim or time-to-blank to
 stretch indefinitely in the presence of many packets.  avoiding the call
 to reset_idlecounters at line 2243 (if the source was wlanpacket) would
 change this behavior.

 but there's a problem.  if we don't call reset_idlecounters, then
 eventually the system will dim, and then blank, the screen, and this will
 disable wake-on-wlan (assuming our default config).  but the point of
 wake-on-wlan is to allow wlan traffic to keep the system awake, right?

 i suspect you're right that it's enabling multicast which is causing the
 problem.  this could also explain the gnome/sugar difference, if there's
 no telepathy/salut running under gnome.  (line 1069 in powerd.)

-- 
Ticket URL: <http://dev.laptop.org/ticket/10856#comment:5>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system