Unable to Start Krun workload

[ryne007]

Hi,

I am getting errors like unable to start workload while creating vm, however it works without raw image but is not useful anyways.
I assume from QEMU docs that raw images for hardisk/root partition or user partition will be created on vm init if it is not existing.. by VMMs

Since there is very limited documentation on krun and I have not seen anyone in mac community actively using this, I am stuck here.

My use case is to boot a lightwight VM to run waydroid(Android in Containers) which is GPU accelerated and I believe Krun is?
Since libkrun can emulate 16k pages to 4k pages which is requirement to run android apps I am trying to use this..

I have already tried with vmware fusion pro to boot ARM ubuntu with 3d accl but waydroid fails to boot there too which I am assuming is due to this page size issue(not sure?)
I have also tried with UTM with Apple's Virtualization framework but I don't think it has 3d acceleration and assuming 4k page size is causing issues there too..

[slp]

I am getting errors like unable to start workload while creating vm, however it works without raw image but is not useful anyways.
I assume from QEMU docs that raw images for hardisk/root partition or user partition will be created on vm init if it is not existing.. by VMMs

Which version of krunkit are you using? In 0.1.4 we added a fix for an error that could manifest in the same way as you're seeing.

My use case is to boot a lightwight VM to run waydroid(Android in Containers) which is GPU accelerated and I believe Krun is?
Since libkrun can emulate 16k pages to 4k pages which is requirement to run android apps I am trying to use this..

The macOS version of libkrun supports GPU acceleration, but only for compute workloads, not for graphics.

[ryne007]

Which version of krunkit are you using? In 0.1.4 we added a fix for an error that could manifest in the same way as you're seeing.

krunkit 0.1.4

The macOS version of libkrun supports GPU acceleration, but only for compute workloads, not for graphics.

ohh I see, is there any plan to support that via virtualization framework?