#8901 HIGH Not Tri: Kernel/EC loose keyboard data when running in Synaptics protocol mode

[Zarro Boogs per Child]

#8901: Kernel/EC loose keyboard data when running in Synaptics protocol mode
-----------------------------------+----------------------------------------
   Reporter:  dilinger             |       Owner:  rsmith       
       Type:  defect               |      Status:  new          
   Priority:  high                 |   Milestone:  Not Triaged  
  Component:  embedded controller  |     Version:  not specified
 Resolution:                       |    Keywords:               
Next_action:  diagnose             |    Verified:  0            
  Blockedby:                       |    Blocking:  8894         
-----------------------------------+----------------------------------------
Changes (by dsaxena):

  * next_action:  reproduce => diagnose


Comment:

 Replying to [comment:11 rsmith]:
 > While I don't doubt that the EC may be responsible for the lost keyboard
 data I don't think its due to any protocol interpretation because there
 isn't any.  For the touchpad the data stream is just a series of bytes
 that get sent up the chain.  There is some code that inits an an
 Intellimouse which has to know how many packets are in the return data.
 But it appears to be disabled.

 Good to know. I had assumed the EC was doing some sort of protocol
 handling as part of the muxing.


 > What is the difference between standard ps2 protocol and synaptics mode?

 http://www.synaptics.com/sites/default/files/ACF126.pdf is the interfacing
 documentation.
 /me looking through it now to grok.

 > The EC code has debugging that we can enable to print out the data as it
 goes up the lpc buss.  That will change the timing quite a bit but if it
 still happens then we can see where things go wrong.

 I'll add some debug to kernel side to see what we're receiving.

-- 
Ticket URL: <http://dev.laptop.org/ticket/8901#comment:12>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system