#8902 NORM Not Tri: Frequent Suspend/Resume cycles cause CPU leak.

[Zarro Boogs per Child]

#8902: Frequent Suspend/Resume cycles cause CPU leak.
-------------------------+--------------------------------------------------
  Reporter:  thomaswamm  |           Owner:  marco        
      Type:  defect      |          Status:  new          
  Priority:  normal      |       Milestone:  Not Triaged  
 Component:  sugar       |         Version:  not specified
  Keywords:              |     Next_action:  diagnose     
Spec_stage:  unknown     |       Blockedby:               
  Blocking:              |   Spec_reviewer:               
  Verified:  0           |   Spec_reviewed:  0            
-------------------------+--------------------------------------------------
 Observed with  8.2-767   Q2E18  (official release).

 Frequent Suspend/Resume cycles cause CPU leak.

 I was walking around with my XO, doing Wireless Access Reconnaisance (WAR
 Walking). I was doing frequent suspend/resumes because that's the only way
 I know to make the Neighborhood view quickly update the RF status display.
 After a couple dozen suspend/resumes, Sugar became unresponsive, and the
 display would not update.  Suspend/resume was still working though.  I was
 able to ctrl-alt-F1 into console and run # top -c  which showed sugar-
 shell hogging over 90% CPU.

 Later, sitting in one spot, I rebooted, then watched $ top -c  in Terminal
 as I did numerous suspend/resumes (very roughly one cycle per minute).
 After about 20 cycles, top showed sugar-shell hogging about 59% CPU. It
 was a gradual buildup, about 3% more CPU hogging per suspend/resume.  It
 seemed to get worse if I suspended/resumed before Network Manager
 established a wlan connection.  It seemed to improve if I delayed my next
 suspend until NM got connected to my AP.

 This bug is similar or maybe related to #8592, though the method of
 triggering the bug is different.

 Note that a child might encounter this bug when looking for an Access
 Point or other wireless contact.

-- 
Ticket URL: <http://dev.laptop.org/ticket/8902>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system