07:08 <bbrezillon> icecream95: okay, so nothing redirecting panfrost ioctls to kbase? If not, I might give it a try (I'd like to be able to test the mesa driver in chromeos so I don't have to reboot every time I need a ddk dump).

07:11 <icecream95> bbrezillon: Not yet, I only wrote a shim for running the ddk on panfrost.ko

07:16 <icecream95> "I might give it a try". I already rebooted into Chrome OS to start implementing it myself, we could have a race...

07:18 <HdkR> How soon until the Mali blob just uses Panfrost? :>

07:20 <icecream95> HdkR: Would't they instead want to use panvk on VKD3D on the d3d12 backend?

07:21 <bbrezillon> icecream95: oh, cool

07:22 <bbrezillon> I didn't start writing code, so I'll just wait for your version ;)

07:23 <HdkR> icecream95: Hopefully the userspace blob wouldn't want to rely on userspace panvk :P

09:59 <icecream95> bbrezillon: I haven't figured out synchronisation yet so I just do a sleep(1) after every job, but otherwise this works for at least simple dEQP tests on Chrome OS on duet:

09:59 <icecream95> https://gitlab.freedesktop.org/icecream95/mesa/-/commits/kbase-shim-only/

10:05 <bbrezillon> nice

13:43 <alyssa> icecream95: bbrezillon: when I used kbase years ago, I had to use a loop with read()

13:43 <alyssa> and that also fixed the sync things

14:15 <bbrezillon> alyssa: makes sense after reading kbase_read()'s code

14:17 <alyssa> bbrezillon: meanwhile from leave

14:17 <alyssa> "modetest ... tying to open device 'apple'...done"

14:18 <bbrezillon> congrats

14:41 <alyssa> heh, thanks

14:48 <macc24> :O

15:06 <macc24> alyssa: btw, do you remember what i said about tripping on power cables?

15:07 <macc24> yeaaaaa i yanked out my phone's usb cord when walking next to it

20:50 <alyssa> macc24: yoof!