[adding Florian to the thread, he is also affected by this]

On Tue, Feb 17, 2015 at 9:01 PM, Andrew Duggan [off-list ref] wrote:

On 02/17/2015 04:30 AM, Jiri Kosina wrote:

quoted

On Thu, 29 Jan 2015, linux@simon-woerner.de wrote:

quoted

From: Simon Wörner <redacted>

HID: Add driver for acer keybard with broken rdesc

Hi Simon,

to make sure proper device <-> driver binding is performed, you also have
to add device ID to the hid_have_special_driver[] array.

To the cost of an error in the kernel log, the proper binding will be
done even if if there is no entry in hid_have_special_driver :)

See Florian's log:
[    3.153892] hid-generic 0003:06CB:2991.0003: usage index exceeded
[    3.153896] hid-generic 0003:06CB:2991.0003: item 0 2 2 2 parsing failed
[    3.153921] hid-generic: probe of 0003:06CB:2991.0003 failed with error -22

So hid-generic will fail to bind the device, so normally hid-acer
should bind it, fix the report descriptor and the keyboard should be
working.

Not very clean, but it should work :)

I have been meaning to respond to this patch. Unfortunately, simply adding
this driver to the hid_have_special_driver[] list will prevent
hid-multitouch from binding to the touchpad which shares the same vid and
pid. My suggestion would be to create a new scan_flag for GD_KEYBOARD and to
add to the code in hid_scan_collection() which scans the GENDESK usages.
Similar to the code which is searching for HID_GD_POINTER. Then in
hid_scan_report() we could assign a group for this driver based on the pid
and the GD_KEYBOARD scan flag in the vendor handling code.

This could work, but that means that the list of special quirks after
the parsing is going to explode :(

Plus we have to be sure that the scan_report doesn't fail or we will
not be able to tag the device correctly.

I can help out with implementing this part of the patch if there aren't any
other suggestions. Adding a new group and more code to the hid_scan_report()
vendor handling code is definately not ideal. But, I am not sure of a better
way of binding different sub drivers to devices with the same vid and pid.

An alternative would be to have hid-rmi handle all Synaptics touchpads, even
the ones which currently use hid-multitouch. Then the keyboard report
descriptor fixup could just be handled in hid-rmi. Accessing the finger data
via rmi mode would also have the benefit of being able to report events
which are currently not supported by the PTP collection (ie ABS_MT_PRESSURE,
ABS_MT_TOUCH_MAJOR/MINOR). The touchpads which currently use hid-multitouch
report finger data a little differently then the ones currently using
hid-rmi so there is some work which would need to be done to add support for
them in hid-rmi.

I am not sure we want to do that. Because we don't want hid-rmi to fix
all crappy keyboards around when the OEM reuses the synaptics PID.

Maybe a better way of handling such situation is to provide a generic
mechanism to overwrite/patch the report descriptor so that we could
get rid of the drivers which just fix the report descriptor.
I have on a branch a version where I look into /lib/firmware/hid if
there is a matching device and a matching report descriptor. Then, I
read this firmware dynamically and patch the report descriptor.
This however requires that the hid modules are not included directly
in the kernel, or the /lib/firmware dir is not available and the
device doesn't bind.

Cheers,
Benjamin
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html