Summary: | external numpad gets unresponsive when it's simultaneously used with another keyboard | ||
---|---|---|---|
Product: | Wayland | Reporter: | Kamil Páral <kparal> |
Component: | libinput | Assignee: | Wayland bug list <wayland-bugs> |
Status: | RESOLVED NOTOURBUG | QA Contact: | |
Severity: | normal | ||
Priority: | medium | CC: | benjamin.tissoires, peter.hutterer |
Version: | unspecified | ||
Hardware: | Other | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: |
system journal
evemu record - no extra events evemu record - extra numlock events evemu record - extra number events hid-recorder output |
Description
Kamil Páral
2016-01-08 14:01:09 UTC
Oh, and I'm not actually using Wayland (this place just seemed like the proper upstream place to report libreport bugs, regardless of Wayland/X11 usage). Created attachment 120897 [details]
system journal
System journal after connecting the keypad (nothing gets printed out when it gets unresponsive).
Created attachment 120898 [details]
evemu record - no extra events
evemu record of the numpad (event18) when simultaneously typing on a keyboard and the numpad. The numpad stops producing numbers in the text editor, and no new events are printed out in evemu record (the last event in the log is the last number correctly printed out, then it died).
Created attachment 120899 [details]
evemu record - extra numlock events
Another evemu record, but this time after numpad gets unresponsive, the evemu log contains an infinite stream of numlock events.
Created attachment 120900 [details]
evemu record - extra number events
Another evemu record, but this time after numpad gets unresponsive, the evemu log contains an infinite stream of KEY_KP6 events (I assume that was the last button pressed before it died).
value 2 for key events is the kernel's autorepeat code. which we ignore, because we do software repeat in X. This is a kernel issue, though I don't know what could cause this. Benjamin? Please attach a hid recording here, this may help identify some issues. See https://bentiss.github.io/hid-replay-docs/ Created attachment 121195 [details]
hid-recorder output
Here it is, but I'm afraid it won't show much. This is the output where I pressed some buttons quickly on keypad (no issues), and then combined it with mashing buttons on my keyboard (and the keypad got stuck). But even though evemu-record shows a neverending stream of these events
E: 2.243998 0000 0000 0001 # ------------ SYN_REPORT (1) ---------- +33ms
E: 2.276997 0001 004c 0002 # EV_KEY / KEY_KP5 2
E: 2.276997 0000 0000 0001 # ------------ SYN_REPORT (1) ---------- +33ms
E: 2.309997 0001 004c 0002 # EV_KEY / KEY_KP5 2
the hid-recorder output has shown no such thing, it simply stopped printing anything.
I have to add that I've seen several times today the keypad to get stuck (print the neverending stream of value 2 events, or stop sending any signals) just by pressing its buttons alone, not combined with the keyboard. So this error can be reproduced with just the keypad, but it's just not that likely. When I use my keyboard simultaneously, it makes reproducing this almost instant. One further thing, I'm not sure if this is related, but it seems weird to me, that if I listen to just the keypad using evemu-record, and I press NumLock on keyboard (different device, not listening to that), it prints these events: E: 3.352170 0011 0000 0000 # EV_LED / LED_NUML 0 E: 3.352170 0000 0000 0000 # ------------ SYN_REPORT (0) ---------- +2641ms E: 4.520290 0011 0000 0001 # EV_LED / LED_NUML 1 E: 4.520290 0000 0000 0000 # ------------ SYN_REPORT (0) ---------- +1168ms However, if I press NumLock on they keypad itself, it generates no events. Maybe unrelated, maybe even expected. Just mentioning. ok, this is a hardware or kernel bug then. as said above, value 2 for key events is a software-generated key repeat event so looks like the kernel never sees the key up event either. since you said the device works under windows, I'd say it's a kernel issue. Benjamin, any ideas based on the hid recording? |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.