Touchpad testing.

[Jim Gettys]

Zephaniah was unable to get the PS/2 port to work correctly under Insyde
at all; that does not mean, of course, that the PS/2 port is set up
properly on LinuxBIOS.  We have pretty definitively ruled out the
touchpad as a possible problem, as we see similar behavior with
conventional mice, and correct behavior from the earlier touchpad sample
on other machines as well.

On Thu, 2006-09-21 at 09:53 +0800, Ray.Tseng at quantatw.com wrote:
> Jim,
> 	Two possiblities:
> 	1)EC: as you mention EC slow down the rate
> 	2)BIOS: interrupt (serial IRQ) is not programmed correctly (Continuous or Quiet, level or edge,...) and the OS does not handle properly

Also a real possibility: maybe even more likely than EC code being
wrong.  I think you are correct to check that first.  Tom Sylla
suggested to me in private mail to similarly check the interrupt lines
on a logic analyzer.

Note that the PRS code is now available at:
http://dev.laptop.org/ticket/109 and may be able to shed light on
whether we've missed a register setup.  We are going to try to get that
into the tinderbox build testing cycle, so that it will be run and
checked continuously. 

Jordan Crouse also came up with a theory that keyboard emulation code
might be a problem, and was investigating that possibility as well this
afternoon.  I asked him to post any results, but I guess he forgot (and
had nothing to report).
                                           Best Regards,
                                               - Jim


> 	
> 	We are shooting 2nd.
> 
> Ray Tseng 9/21/06 
> 
> -----Original Message-----
> From: devel-bounces at laptop.org [mailto:devel-bounces at laptop.org] On Behalf Of Jim Gettys
> Sent: Thursday, September 21, 2006 3:48 AM
> To: Ray Tseng (曾文瑞); Zephaniah E. Hull
> Cc: OLPC Developer's List
> Subject: Touchpad testing.
> 
> As suggested last night, we (Andres and I) took the older, 5V touch pad sample (the samples from early August; not the new 3.3 V touch pad we received in September) and plugged it into a standard PS/2 port on a conventional desktop.
> 
> It behaves exactly as you might expect for normal operation: bursts of information are at regular 10ms intervals  on the oscilloscope (the default rate for PS/2 is supposed to be 100hz).
> 
> Zephaniah reported that that older touch pad version on an unmodified (e.g. 5V) OLPC ATest board was also running very slowly, along with a standard PS/2 mouse.
> 
> So unless Zephaniah managed to break the Linux kernel driver in some unique way, the theory that the EC is somehow slowing the rate down tremendously seems very likely.
> 
> Zephaniah, does your kernel driver work well with a standard mouse on a conventional system.
>                                     Best regards,
>                                          - Jim Gettys
> --
> Jim Gettys
> One Laptop Per Child
> 
> 
> _______________________________________________
> Devel mailing list
> Devel at laptop.org
> http://mailman.laptop.org/mailman/listinfo/devel
-- 
Jim Gettys
One Laptop Per Child