[Openpvrsgx-devgroup] Current Status linux_openpvrsgx

H. Nikolaus Schaller hns at goldelico.com
Mon Apr 6 19:32:34 CEST 2020


> Am 06.04.2020 um 18:38 schrieb Philipp Rossak <embed3d at gmail.com>:
> 
> Hey Nikolaus,
> 
> this time I'm not too stupid, to click the wrong Reply button. ;)
> 
> I got some good progress with the sun8i-a83t chip. The kernel module is loading and based on Jonathan's work.

Wow!

> Right now I'm struggling with the current linux version that we are using as basis for our development. Currently there is a bug in the network system that already got fixed in v5.6. I will create a branch based on the v5.6 mainline kernel and patch our work on to this branch. I will call it pvrsrvkm-v5.6.

Well, I had rebased and pushed the code automatically every week but Tony ran into problems with this model so I now only rebase for every -rc1. If you can wait until Monday, I'll have a 5.7-rc1 based version which should include the networking fixes. But may include new ones...

> 
> Right now the revision of the sun8i-a83t is wrong. When I read out the register $(gpu_base) + 0x24 on my legacy reference linux, I get the following:
> 
> pi at bananapi:~$ sudo ./devmem2 0x01C40024
> /dev/mem opened.
> Memory mapped at address 0xb6f3e000.
> Value at address 0x1C40024 (0xb6f3e024): 0x10105

Good to know!

> So here we have the SGX revision 01.01.05 --> 115.

That is interesting. I have not yet seen a 115 revision.

> Can you please correct this in your patch set for mainline?

Yes, sure.

> 
> One interesting fact about my bananapi-m3:
> I might be able to use dvfs on my gpu since I have a separate dcdc regulator for the gpu. But for now I will keep the clocks and voltage fixed.

Yes, let's start in small steps :)

BR and thanks,
Nikolaus



More information about the openpvrsgx-devgroup mailing list