Comment 54 for bug 512192

On Thu, Apr 29, 2010 at 08:55:05PM -0000, Dmitry Torokhov wrote:
> Hmm, I wonder if all firmwares with major 2 and above produce 6-byte
> protocol and the entire "jumpy cursor" workaround for firmware 2.34 is
> not really needed.

I'm fairly sure that'd be wrong. The jumpy_cursor codepath very much
relies on 4-byte packets.

On Thu, Apr 29, 2010 at 08:58:29PM -0000, Dmitry Torokhov wrote:
> ... Or we need to start paying attention to the middle byte of the
> firmware query...

That's more like it, but with the information we have so far, we can't
really tell.

At this point, I feel inclined to just add a list of devices with their
full firmware version query result and their associated quirks to the
driver.

It's apparent we're guessing wrong somehow, but changing the guesses
blindly doesn't really have a good chance of making it better. It's more
likely to make already working devices break.

Having the possibility to turn individual quirks on and off, which would
be easy enough once we have the suggested device table, will probably
allow users to report their firmware version and the necessary quirks
easier.

Once a clear pattern emerges from the firmware -> quirks mapping, we can
still adjust the guessing algorithm without fearing breakage too much.

Would that work for you, Dmitry?

--
BOFH excuse #225:
It's those computer people in X {city of world}. They keep stuffing
things up.