#8768 NORM Not Tri: recurring "SLOWS" after trying View Slides

[Zarro Boogs per Child]

#8768: recurring "SLOWS" after trying View Slides
-----------------------------------+----------------------------------------
 Reporter:  thomaswamm             |         Owner:  dsd                              
     Type:  defect                 |        Status:  new                              
 Priority:  normal                 |     Milestone:  Not Triaged                      
Component:  stability-reliability  |       Version:  Development build as of this date
 Keywords:                         |   Next_action:  diagnose                         
 Verified:  0                      |     Blockedby:                                   
 Blocking:                         |  
-----------------------------------+----------------------------------------
 After randomly testing lots of stuff over several hours, in build 8.2-767,
 with surprising stability, I tried running the View Slides activity
 (version 0 ?) with a 100MB zipfile on a USB memory stick, as per what the
 Journal offered.  The XO slowed and froze, then after a few minutes View
 Slides vanished and Journal re-appeared.  I unmounted my USB stick and
 unplugged it, to fiddle with other things.

 Several things I tried after this all led to slowing and freezing for
 minutes at a time.  If not for patient curiousity I would have just
 rebooted, but I'm trying to collect clues about what makes the XO
 unstable.  'uptime' shows load average peaking over 10.0 (when I am able
 to run it in console).  /var/log/messages shows a few instances of oom-
 killer.  olpc-log attached.

 The "SLOWS" always seems to take at least several minutes to pass, until
 oom-killer is invoked and does its killing thing and the XO seems to get
 back to normal.  I am curious if the glacial delay can be shortened so the
 user (me) can get on with other activities sooner.  Now when the SLOWS
 happens, I either force reboot with the power button, or just wait a long
 time for the XO to self-recover.  Either way it's a slow pain in the neck.
 If it can't be prevented, can the pain at least be made to pass faster?

 Maybe a fast 'abort' key sequence could be implemented to kill off the
 foreground activity?  But during the SLOWS, the keyboard seems to be
 ignored for minutes.  Even the power-button is ignored unless held down
 for many seconds.

-- 
Ticket URL: <http://dev.laptop.org/ticket/8768>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system