Bug: Quickly connected and disconnected device causes crash (failed read: No such device (os error 19) #1683
Description
Requirements
- I've searched platform-specific issues, issues and discussions to see if this has been reported before.
- My issue does not involve multiple simultaneous key presses, OR it does but I've confirmed it is not key rollover or ghosting.
Describe the bug
When running kanata, quickly disconnecting and reconnecting an input device sometimes causes it to crash with the error failed read: No such device (os error 19). This is very timings dependent and usually takes a few tries.
Relevant kanata config
You can reproduce it with a minimal layout:
(defsrc)
(deflayer "base")
To Reproduce
- Check out main and
cdinto it (I used dd02c3e, but I've also confirmed with 1.8.0). - Build using
cargo build. - Run kanata using
sudo ./target/debug/kanata --cfg path/to/layout.kbd(thesudois optional, but it becomes much harder to trigger the bug for me without it). - Repeatedly disconnect and reconnect a USB input device until the error appears. (This is pretty finicky and is very timings dependent.)
Expected behavior
Kanata not crashing under these circumstances.
Kanata version
1.8.0 to main (haven't tried lower)
Debug logs
17:40:48.0341 [INFO] kanata v1.9.0 starting
17:40:48.0347 [WARN] No defcfg is defined. Consider whether the process-unmapped-keys defcfg option should be yes vs. no. Adding defcfg with process-unmapped-keys defined will remove this warning.
17:40:48.0348 [INFO] process unmapped keys: false
17:40:48.0349 [INFO] NOTE: kanata was compiled to never allow cmd
17:40:48.0353 [INFO] config file is valid
17:40:48.0364 [INFO] Created device "/dev/input/event27"
17:40:48.0365 [INFO] Sleeping for 2s. Please release all keys and don't press additional ones. Run kanata with --help to see how understand more and how to disable this sleep.
17:40:50.0367 [INFO] entering the processing loop
17:40:50.0371 [INFO] entering the event loop
17:40:50.0371 [INFO] looking for devices in /dev/input
17:40:50.0372 [INFO] Init: catching only releases and sending immediately
17:40:50.3581 [INFO] registering /dev/input/event9: "ZSA Technology Labs Moonlander Mark I Keyboard"
17:40:50.3880 [INFO] registering /dev/input/event7: "ZSA Technology Labs Moonlander Mark I Consumer Control"
17:40:50.4101 [INFO] registering /dev/input/event5: "ZSA Technology Labs Moonlander Mark I System Control"
17:40:50.4390 [INFO] registering /dev/input/event3: "ZSA Technology Labs Moonlander Mark I"
17:40:50.4601 [INFO] registering /dev/input/event25: "Corsair CORSAIR VIRTUOSO SE Wireless Gaming Headset Consumer Control"
17:40:50.4951 [INFO] registering /dev/input/event23: "DP-5"
17:40:50.5221 [INFO] registering /dev/input/event21: "Framework Laptop 16 Keyboard Module - ISO Keyboard"
17:40:50.5472 [INFO] registering /dev/input/event19: "Framework Laptop 16 Keyboard Module - ISO Consumer Control"
17:40:50.5705 [INFO] Starting kanata proper
17:40:50.5706 [INFO] You may forcefully exit kanata by pressing lctl+spc+esc at any time. These keys refer to defsrc input, meaning BEFORE kanata remaps keys.
17:40:50.5750 [INFO] registering /dev/input/event18: "Framework Laptop 16 Keyboard Module - ISO System Control"
17:40:50.6011 [INFO] registering /dev/input/event17: "Framework Laptop 16 Keyboard Module - ISO"
17:40:54.0363 [WARN] removing kbd device: /dev/input/event3
17:40:54.1824 [WARN] removing kbd device: /dev/input/event5
17:40:54.2122 [WARN] removing kbd device: /dev/input/event7
17:40:54.2855 [WARN] removing kbd device: /dev/input/event9
17:40:54.9951 [INFO] watch found file changes, looking for new devices
17:40:54.9951 [INFO] sleeping for a moment to let devices become ready
17:40:55.1952 [INFO] looking for devices in /dev/input
17:40:55.5950 [INFO] registering /dev/input/event9: "ZSA Technology Labs Moonlander Mark I Keyboard"
17:40:55.6270 [INFO] registering /dev/input/event7: "ZSA Technology Labs Moonlander Mark I Consumer Control"
17:40:55.6502 [INFO] registering /dev/input/event5: "ZSA Technology Labs Moonlander Mark I System Control"
17:40:55.6740 [INFO] registering /dev/input/event3: "ZSA Technology Labs Moonlander Mark I"
17:40:55.7960 [INFO] watch found file changes, looking for new devices
17:40:55.7960 [INFO] sleeping for a moment to let devices become ready
17:40:55.9961 [INFO] looking for devices in /dev/input
17:41:02.2271 [WARN] removing kbd device: /dev/input/event3
17:41:02.3555 [WARN] removing kbd device: /dev/input/event5
17:41:02.3913 [WARN] removing kbd device: /dev/input/event7
17:41:02.4553 [WARN] removing kbd device: /dev/input/event9
17:41:04.3070 [INFO] watch found file changes, looking for new devices
17:41:04.3071 [INFO] sleeping for a moment to let devices become ready
17:41:04.5071 [INFO] looking for devices in /dev/input
17:41:04.8940 [INFO] registering /dev/input/event9: "ZSA Technology Labs Moonlander Mark I Keyboard"
17:41:05.2442 [ERROR] failed read: No such device (os error 19)
Press enter to exit
17:41:05.2443 [ERROR] channel disconnected
Note: Running with --debug didn't seem to give any additional relevant information, so I decided to not add it here to keep this log shorter. But if you think it may be relevant I can add it.
Operating system
Linux running NixOS
Additional context
As I said this is pretty timings dependent, and (based on the logs) seems to occur when the device is connected and removed quicker than the interval between the watcher detecting the changed input device and actually trying to register the new device (since it prints something about sleeping for a short while after detecting changed files).
For some reason it's much easier to reproduce the bug when running with sudo. When running with sudo I'm able to trigger it within 3 tries of replugging my keyboard, while it takes 10+ without it... For context, I've created a group uinput which my user is a part of. I have the following udev rule, which I think is what which assigns input devices to the group.
KERNEL=="uinput", MODE="0660", GROUP="uinput", OPTIONS+="static_node=uinput"