Open onimatrix81 opened 9 years ago
Hi @onimatrix81,
Thank you for your report!
It looks like vdevd is doing all the right things--it's creating the device nodes and links, and all of its helpers appear to be succeeding. If you want, you can remove the async=true
line from zzz-event-push.act
and zzz-udev-compat.act
to cause them to be more verbose, but they are only concerned with pumping events to libudev-compat clients.
The error you're seeing is due to udevd reacting to the device as well, and broadcasting it to all netlink listeners on the host. Vdevd detects and ignores this in vdev_os_next_device
by failing with -EAGAIN
, which tells vdev_os_main
to try again to receive the event. You'll notice that vdevd also logs that it detected the udev message just above the error message :)
The fact that your mouse stops working at all is surprising, given that udev appears to be running. Is libudev-compat installed on your system? If so, did you restart all libudev-linked programs so they'll load libudev-compat instead of libudev? This includes Xorg, for example.
Thanks, Jude
http://pastebin.com/xYDcS0b0
This is what gets logged when i run vdevd (along udev with a fake vdev-test/) and disconnect an usb mouse. This is with daemonlet=false in both zzz-udev-compat.act and zzz-event-push.act. With daemonlet=true for either of them I can reproduce more errors too, but this is now as far as I've got trying to get rid of them :). As a matter of fact, mouse devs do get created allright in /dev/input and I can cat them to console just fine.
Hopefully the log was cut at the right position to be of use.