Bugzilla – Bug 98947
hwinfo detects alps touchpad and stick as same device
Last modified: 2007-06-22 15:22:05 UTC
hwinfo --mouse on this Dell D410 returns two mice, the touchpad and the stick, with the same vendor and device id. This results in two alps mice being configured, which results in the stick and later attached USB mice not working at all: 19: PS/2 00.0: 10500 PS/2 Mouse [Created at input.140] Unique ID: AH6Q.fU1aO9YHBB4 Hardware Class: mouse Model: "PS/2 Mouse" Vendor: int 0x0212 Device: int 0x0002 "PS/2 Mouse" Compatible to: int 0x0210 0x0003 Device File: /dev/input/mice (/dev/input/mouse0) Device Number: char 13:63 (char 13:32) Driver Info #0: Buttons: 3 Wheels: 0 XFree86 Protocol: explorerps/2 GPM Protocol: exps2 Config Status: cfg=new, avail=yes, need=no, active=unknown 20: PS/2 00.0: 10500 PS/2 Mouse [Created at input.140] Unique ID: AH6Q.fU1aO9YHBB4 Hardware Class: mouse Model: "AlpsPS/2 ALPS GlidePoint" Vendor: int 0x0212 Device: int 0x0002 "AlpsPS/2 ALPS GlidePoint" Compatible to: int 0x0210 0x0003 Device File: /dev/input/mice (/dev/input/mouse1) Device Number: char 13:63 (char 13:33) Driver Info #0: Buttons: 3 Wheels: 0 XFree86 Protocol: explorerps/2 GPM Protocol: exps2 Config Status: cfg=new, avail=yes, need=no, active=unknown
It reports exactly what's in /proc/bus/input/devices.
And it's also correct. They're driven by the same chip, hence they have the same IDs (the version part differs, though). If something relies on the fact that there can't be two devices with the same ID, it's utterly wrong - think USB mice and keyboards. Also, the glidepoint should get a 'synaptics' driver assigned, not a regular 'explorerps/2' driver, which, too, seems to be a bug.
Matthias, please run 'getsysinfo' and attach the result.
@Vojtech: This is only partly correct. Both devices are connected to the same chip, but only the touchpad speaks the protocol of the synaptics driver. The stick can only be addressed by a regular 'mouse' driver. Personally, I don't care about the solution, but the way it is now is suboptimal and a regression. getsysinfo follows. There is also a minimal difference between the two devices. Both share vendor and product id, but name, version, EV, KEY, ABS differ.
Created attachment 45601 [details] getsysinfo output
Created attachment 45602 [details] /proc/bus/input/devices
The decision which device is a touchpad shouldn't be done based on the name or IDs, those can change as new devices appear. It should be based on the KEY and ABS bitmaps - a touchpad will always include ABS_X, ABS_Y, ABS_PRESSURE and BTN_TOOL_FINGER.
I still don't see what I can do here. Marcus, is there anything to change in hwinfo?
The problem here is that I cannot distinguish between a synaptics/alps touchpad and a "normal" mouse. To be able to setup a device as synaptics touchpad using the synaptics driver a profile is assigned. This profile is bound to the vendor and device ID reported by libhd. Well the experts say that kind of mapping is wrong. Ok if this is the case libhd should provide a unique flag which enables me to assign a profile. This is currently not possible
Ok, checking the 4 flags mentioned in comment 7 now.
Just verified, works very well in Beta4 now :) Thanks!