#12400 NORM 13.1.0: Determine MMP3 audio buffer parameters (was: XO-4 etoys audio is bad)

[Zarro Boogs per Child]

#12400: Determine MMP3 audio buffer parameters
--------------------------------+-------------------------------------------
           Reporter:  dsd       |       Owner:  saadia       
               Type:  defect    |      Status:  new          
           Priority:  normal    |   Milestone:  13.1.0       
          Component:  kernel    |     Version:  not specified
         Resolution:            |    Keywords:               
        Next_action:  diagnose  |    Verified:  0            
Deployment_affected:            |   Blockedby:               
           Blocking:            |  
--------------------------------+-------------------------------------------
Changes (by dsd):

  * component:  etoys-activity => kernel


Comment:

 The default values in mmp_pcm_probe come from Marvell code - maybe there
 is a platform where memory is so low that these values make sense. If not,
 they could perhaps be increased.

 In the MMP3 case, SRAM is shared between audio and other components, so
 the max size used by audio depends on various factors.

 As an initial strategy I tried to follow qseven alpha4's mmp3-squ-zsp
 driver, which uses 64kb (presumably 64kb for playback and 64kb for
 capture), but this broke `speaker-test -c 2 -t wav` (no sound output,
 process seems to hang).

 So I dropped to 64kb (32kb+32kb) instead, this still works and sound in
 etoys/scratch is fixed. This workaround is pushed as
 62abd444014393c039f76e258 (it should reverted and fixed in the DT at a
 later date).

 Now we need to investigate why the 128kb setting didn't work, and once
 we've made a decision, get the chosen values exposed from the firmware.

-- 
Ticket URL: <http://dev.laptop.org/ticket/12400#comment:7>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system