Shared memory unaccessible to SingleApplication after gpt4all crash

[autogris]

Bug Report

In linux, after trying to run chat with a model that didn't fit the GPU, it reversed to CPU. By clicking in "delete history" while response is processing, the program crashes. Trying to run the application again I get every single time an immediate crash with these errors (until restarting system):

 [Warning] Could not find the Qt platform plugin "wayland" in ""
[Critical]  SingleApplication: Unable to attach to shared memory block.
[Critical]  SingleApplication:  QSharedMemory::UnknownError "QSharedMemory::attach (shm_open): unknown error: No such device or address"

Searching with ps, gpt4all's binary (bin/chat) is not running, as expected. Looking at "ipcs -a" there is no usage of shared memory by any program.

Steps to Reproduce

1. Run a chat with model larger than GPU's VRAM
2. When chat start to use CPU to provide answer, click delete history, causing a crash
3. Try to run the application again after crash

Expected Behavior

Application should be able to restart normally with a clean memory state.

Your Environment

• GPT4All version: 3.10.0
• Operating System: Linux Slackware, kernel 6.17.8
• Chat model used (if applicable):

[james-see]

I do not have this bug on Mac OS fwiw.