Emulator freezes after moving mouse

[HunterBoy344]

Strangely, this issue does NOT occur on copy.sh/v86, only on my own copy of v86 with a custom theme (Source code here). Whenever I try to load a disk image, it will work perfectly fine as long as I keep my mouse cursor away from the screen. However, as soon as I move my mouse into the screen canvas area, the VM will completely freeze after a few seconds. The disk image I've been testing with is here, but I also tried a Windows XP disk image that previously worked fine and experienced the exact same issue.

Is this an issue with my configuration? Something to do with the structure of my index.html? I tried updating the network adapter and screen in the config to their new object versions (screen and net_device instead of screen_container and network_relay_url), but it didn't fix anything, so I reverted those changes since they seemed to have no impact on the bug. The version of libv86.js and v86.wasm I'm using is the latest as of 12/31/2024.

I created a new copy of the emulator with libv86-debug.js and the associated wasm file and tested it there, but no error appeared in the console. After moving my mouse, the VM simply froze and the console stopped updating.

No idea why this is happening. This doesn't happen on the official site.

Edit: I should probably specify that this issue did not ever occur before I updated to the latest libv86.js and v86.wasm today. The version I was using prior to updating was from July(?) 2024.

[SuperMaxusa]

For me, Windows 3.1 works properly, but RetrOS-32 is stuck on the login screen.
Can you try to build v86 or use this artefact from commit 298da23 and test it?

This doesn't happen on the official site.

copy.sh/v86 uses an older version than on GitHub

[HunterBoy344]

For me, Windows 3.1 works properly, but RetrOS-32 is stuck on the login screen. Can you try to build v86 or use this artefact from commit 298da23 and test it?

This doesn't happen on the official site.

copy.sh/v86 uses an older version than on GitHub

That artifact works like a charm! I've replaced the version in my repository with it, but in case it's necessary for testing, I made a branch that has the latest (broken) version.

[SuperMaxusa]

After my small research, I can assume that PIC changes in commit 9beaccc may cause this bug on some OSes.
If you are using Linux and Windows XP as a guest OS, can you try enabling acpi in your V86 constructor?

[copy]

Thanks, I can reproduce this (albeit slightly differently: it happens on network activity somewhere in the boot process, and it doesn't completely freeze the VM, e.g. the cursor is still blinking).

Could you try the wip branch to see if it fixes the issue for you?