#12541 BLOC 13.1.0: XO-4: Wakeup events during suspend hang the machine. 13.1.0 build 29

[Zarro Boogs per Child]

#12541: XO-4: Wakeup events during suspend hang the machine. 13.1.0 build 29
---------------------------------+------------------------------------------
           Reporter:  rsmith     |       Owner:  pgf                              
               Type:  defect     |      Status:  new                              
           Priority:  blocker    |   Milestone:  13.1.0                           
          Component:  kernel     |     Version:  Development build as of this date
         Resolution:             |    Keywords:                                   
        Next_action:  never set  |    Verified:  0                                
Deployment_affected:             |   Blockedby:                                   
           Blocking:             |  
---------------------------------+------------------------------------------

Comment(by rsmith):

 I'm currently testing this in my testbed.  Since there seemed to be a lot
 of IRC confusion of the test, I shall describe.

 I'm testing build 31 with firmware q7b17 and the included EC code of
 0.3.13 on 1Ghz C1 machines with 8787 wlan cards.  The combination of the
 EC code included in that firmware and a runin version that does not do
 locking around a suspend while scanning channels results in a EC IRQ
 delivered while the machine is suspending.

 This combination was very unstable in my testing a week ago. (2012/03/23)
 It would hang in less than 50 S/R cycles. Today the same setup on the same
 machines was much more stable it took close to 300 cycles for things to
 hang but they eventually did hang.

 I have that same setup running with a kernel containing the commits
 mentioned above.  Its been running now for over an hour.  Still to early
 to tell anything but watching the status I see that the suspend count is
 incrementing much faster than on the kernel included with build 31.  This
 indicates that many of the suspend attempts are getting aborted.  ie The
 patchs are indeed doing something.  The setup will need to run for a while
 to see if it really stops the hangs.

-- 
Ticket URL: <http://dev.laptop.org/ticket/12541#comment:8>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system