Unkillable processes/OS lockup using virtio drives

Development OS
1

I’ve been getting back to using Zig on Haiku, and I needed to build llvm 17 from source again. For context, I’m using a QEMU VM managed by libvirt/virt-manager, with my system drive using VirtIO.

After some time Ninja and it’s children will lock up and not respond to signals; I’ve even seen the whole OS lock up. Both of these cases require a hard reset. The only information I can get from Ninja is after pressing Ctrl-C:

waitpid(864): Interrupted system call

I switched from VirtIO to SATA in the VM settings and this problem immediately goes away. I also tried bringing up the kernel debugger but I assume it’s disabled in release builds since it never shows when pressing the key combo.

2

The kernel debugger is not disabled and is always available. If the key combo doesn’t work (probably your virtual machine is intercepting some keys?), try the kernel_debugger command from Terminal.

3

Please check if using virtio-block or virtio-scsi makes a difference.