On 8/19/2010 6:19 PM, Neil Leeder wrote:

Hi Dmitry,

On 8/19/2010 1:33 AM, Dmitry Torokhov wrote:

quoted

On Wed, Aug 18, 2010 at 05:38:38PM -0400, Neil Leeder wrote:

quoted

Using proto=bare gets around the GETINFO failure, but doesn't help
with the more important GETID failure.

Does it help if you change write() to transmit (and read) 1 byte at a
time?

Thanks for the suggestions. Nothing I change seems able to get a
consistent valid response to GETID. I think this is the first time
anyone has written anything other than 0xF4 to the firmware in the EC
and it's just not passing through the other PS/2 commands and responses
reliably.

quoted

One way would be to look for PSMOUSE_CMD_ENABLE/PSMOUSE_CMD_DISABLE
(0xf4/0xf5) in ->write() method to switch between 1 and 3-byte transfers.

If I force my way past GETID and set the receive length to 3 based on
seeing F4/F5 being written then the touchpad does work. However the
serio driver is now a touchpad-specific driver rather than a generic
PS/2 one. I have a keyboard on the same device. I'm not sure if it will
work with serio, but if so it will probably have its own requirements
for the length of data sent on responses, probably not 3.

BTW, using 1 as the receive length for commands has its own problems as
some commands require more than one byte response (GETID, GETINFO). I
can work around it, but it's not particularly clean and getting further
away from a generic driver.

At this point I'm thinking that the interface is close to being able to
work with serio/psmouse, but just not close enough. The unreliability of
responding to basic commands as well as the length of data problems
indicates some custom driver is going to be needed. That could either be
Quanta's original touchpad driver I posted, or a modification of your
serio driver. I'd lean towards not having a serio driver which includes
workarounds for a specific device, but I'd appreciate hearing your opinion.

Thanks.

Hi Dmitry,

Any comment on the above alternatives?

A third suggestion might be to have a fairly general serio driver, and 
rewrite the original Quanta driver to use serio rather than using i2c 
directly. I think that has its own set of problems, but thought I'd 
mention it anyway. The nuvoTon EC doesn't present a fully-working and 
reliable PS/2 interface to the driver so I'd lean toward using the 
Quanta touchpad driver by itself.

Thanks.

--
Neil
-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.