#11550 NORM 11.3.1: XO-1.5 finger rolling on touchpad causes cursor position jumps

[Zarro Boogs per Child]

#11550: XO-1.5 finger rolling on touchpad causes cursor position jumps
---------------------------------------+------------------------------------
           Reporter:  martin.langhoff  |       Owner:  reuben                                 
               Type:  defect           |      Status:  new                                    
           Priority:  normal           |   Milestone:  11.3.1                                 
          Component:  hardware         |     Version:  1.5/1.0 Software Build os860 aka 10.1.3
         Resolution:                   |    Keywords:                                         
        Next_action:  diagnose         |    Verified:  0                                      
Deployment_affected:                   |   Blockedby:                                         
           Blocking:                   |  
---------------------------------------+------------------------------------

Comment(by avatar):

 Replying to [comment:22 pgf]:
 >
 > avatar -- i'm sure we can get packat logging from OFW.
 >
 > but in the meantime, perhaps you can attach some documentation that
 describes the values that the script is setting/verifying?  what is the
 script doing?  why aren't these register writes in the driver in the first
 place?

 Honestly, I'm also puzzled by those values.  Last time I was told that
 they are algorithm related parameters which will effect the signal
 processing result done by the on-chip DSP. Following are the summary I
 tried to translate from our algorithm guys:[[BR]]
 ----
 The first group of registers(45=c1, 15=64, 17=99, 1b=6f, 30=a0, 42=e7) are
 used to reduce the signal strength received by the on-chip DSP to avoid
 possible cursor drifting when one presses the lid while it's being flipped
 and then closed.[[BR]]

 Where the second group of registers(64=63, 65=63, 6c=a8, 6d=8a) are trying
 to adjust the capacitance threshold such that FSP won't misidentify the
 finger swinging as cursor movement.[[BR]]
 ----
 The script(updated) writes aforementioned parameters and reads them back
 to make sure that they are identical to what they've been written
 to.[[BR]]

 For now I didn't see a better place for doing register write back after
 resumption except the powerd callback.  If I understand the code
 correctly, psmouse should invoke fsp_reconnect() after resumption but I
 didn't see such behaviour on OLPC.  Additionally, this register patch is
 temporary for current released FSP as for newer modules, these parameters
 will be programmed into the chip before they're released to production.

-- 
Ticket URL: <http://dev.laptop.org/ticket/11550#comment:23>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system