#10061 HIGH 1.5-sof: Bad interrupt when accessing viafb-camera with XVideo acceleration

Zarro Boogs per Child bugtracker at laptop.org
Fri Mar 12 16:57:08 EST 2010


#10061: Bad interrupt when accessing viafb-camera with XVideo acceleration
-------------------------------------+--------------------------------------
           Reporter:  jon.nettleton  |       Owner:  dsaxena            
               Type:  defect         |      Status:  new                
           Priority:  high           |   Milestone:  1.5-software-update
          Component:  kernel         |     Version:  not specified      
         Resolution:                 |    Keywords:  viafb-camera       
        Next_action:  never set      |    Verified:  0                  
Deployment_affected:                 |   Blockedby:                     
           Blocking:                 |  
-------------------------------------+--------------------------------------

Comment(by corbet):

 OK, I see what is going on.  The Xv stuff is enabling the HQV0 ("high
 quality video") interrupt, but there is no handler installed for that
 interrupt.  So once it asserts, it hammers the CPU forever until the core
 kernel runs out of patience and shuts it off.

 I'll attach a little patch which makes the problem go away by disabling
 the interrupt when it gets in the way.  This is *not* a real fix, it's a
 piece of paper with unprintable message written on it which has been
 placed over the problem.  It allows the camera to operate - albeit with
 the occasional pause that I don't understand - but it's not a permanent
 solution.

 Said permanent solution could get a bit messy.  This code is mucking
 around in the same set of control registers (from user space) that the
 camera, i2c, and gpio drivers are using.  They are bound to step on each
 others' toes.  The *real* fix for that is to turn this stuff into a proper
 KMS driver - probably not going to happen next week.  So we're going to
 have to come up with something a little more practical.  Will investigate.

 In the mean time, though, if the openchrome code could avoid messing
 around with the interrupt control register, life would be a little better.

 (As an aside, has anybody ever noted that the cheery "Sorry, can not save
 your changes. This ticket has been modified by someone else since you
 started" message from this tracker is annoying as hell?)

-- 
Ticket URL: <http://dev.laptop.org/ticket/10061#comment:8>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system


More information about the Bugs mailing list