[adding Florian to the thread, he is also affected by this] On Tue, Feb 17, 2015 at 9:01 PM, Andrew Duggan <aduggan@xxxxxxxxxxxxx> wrote: > On 02/17/2015 04:30 AM, Jiri Kosina wrote: >> >> On Thu, 29 Jan 2015, linux@xxxxxxxxxxxxxxxx wrote: >> >>> From: Simon Wörner <mail@xxxxxxxxxxxxxxxx> >>> >>> 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@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
