Hello all

A while back there was quite a discussion on this issue and then
specifically "i8042 timing issues".  I refer you to
http://lkml.org/lkml/2005/1/27/11 for more detail.

It turns out that it was the case that the i8042 controller responds too
late on commands, for example, we issue a AUX_LOOP command, whilst
waiting for the response it would time out, however, when issuing
AUX_TEST we will get the response for AUX_LOOP.  It turns out this is
also true for the MUX and KBD parts of the driver.  The attached patch
resolves this issue by providing a configurable option to issue commands
that expects results twice (not sure whether this is a good idea - but
it seems all commands either set some value or query some value).
Between the two commands it calls i8042_flush() to make sure that there
are not any earlier responses in the buffer.

I've built the patch against linux-2.6.11.6 and it should apply cleanly
(I've attached the patch cause it is about 80 lines in length and I fear
mozilla will break the formatting).

The patch fixes the problem for me - not sure how many other people out
there has the same issue.  It provides a configurable option which can
be selected in the kernel configuration to actually enable this (albeit
enabling it should only result in a minor performance hit should it be
enabled needlessly - I think).  Should the option not be enabled the
code remains 100% unchanged (except for one extra level of indentation).

I'm not sure what the policy regarding #ifdefs are, especially
considering they enclose the endpoints of a loop, please review and comment.

Jaco
-- 
There are only 10 kinds of people in this world,
  those that understand binary and those that don't.
http://www.kroon.co.za/