[Libdlo] DisplayLink devices reporting HID or MSC class, and default configuration ID of 2

[Bernie Thompson]

We now have more complete information on the universe of
DisplayLink-based device configurations that are out there with these
issues (their PIDs) -- including better descriptions of the possible
alternatives for handling them.  Devices like this may (or in the case
of HID, will) get incorrectly matched against those class drivers on
Linux, rather than the displaylink-mod framebuffer driver, causing the
device to not function as expected (e.g. no green screen).

The details are at http://libdlo.freedesktop.org/wiki/DeviceQuirks

Because there are only a very small number of PIDs (5) shipping today
which match against HID incorrectly, it would be easy to blacklist
those (and several are popular devices, so this would be a good win).
Then displaylink-mod could load, set configuration 1, and all would be
well.

Unfortunately, there are many more PIDs for devices which report a
default configuration #2 matching mass storage class (many PIDs, but
not necessarily a large percentage of shipping DisplayLink devices).
The USB mass storage driver could be modified to blacklist these, but
there are problems mentioned in wiki. At this time, the best option
may be to let these be handled in a more ad-hoc fashion by the user
using scripts to set the configuration via udev or other (e.g. a
usb-modeswitch type solution).

On those udev scripts: for multiseat solutions, it's possible to use
udev rules to find groupings of display, keyboard, mouse, etc. in the
USB topology, record that in the /dev filesystem as a numbered seat,
and then launch a (gdmdynamic) login prompt on each of those seats to
get plug and play multiseat.  If a udev script providing that kind of
functionality is appealing enough to get picked up by most/all Linux
distros, then setting configuration 1 for all DisplayLink devices in
that same script might be a good place for a catch-all solution to
these quirks.  But as this isn't happening yet, it's not a short-term
solution.

Any feedback on the best way to deal with these quirks (or correction
on anything that I didn't describe quite correctly) is welcome.
Otherwise, I think Greg is looking at least at adding the quirks for
the 5 affected PIDs to the HID driver. And we still need to get
displaylink-mod updated to set configuration 1 for all devices it is
loaded on.

Thanks,
Bernie