19:47 <mntmn> compiled etnaviv on LS1028A

19:47 <mntmn> [ 1.832443] etnaviv etnaviv: bound f0c0000.gpu (ops gpu_ops)

19:47 <mntmn> [ 1.838139] etnaviv-gpu f0c0000.gpu: model: GC7000, revision: 6202

20:38 <cphealy> wohoo, nice work!

20:38 <cphealy> Are you using the NXP kernel on the LS10248A?

22:07 <mntmn> cphealy: yep, from LSDK20

22:08 <mntmn> cphealy: so it's kernel 5.4.something

22:08 <mntmn> cphealy: i also have the mali-dp and mhdp running. they use the same from i.MX8

22:09 <mntmn> cphealy: but etnaviv/GC7000 doesn't work yet, all job submits time out. also, for layerscape there are no clocks given to the vivante driver, which is odd

22:09 <mntmn> cphealy: this is the dts entry https://i.imgur.com/ewDe8T0.png

22:09 <mntmn> also they directly specify a phys mem address there and a CMA address, which i find odd / don't know how to interpret.

22:10 <mntmn> the vivante gpu driver for ls1028a is a copypasta of the i.MX8 version with a mini patch for layerscape added.

22:11 <mntmn> https://i.imgur.com/YF9q3l6.png cphealy this is the layerscape specific code

22:14 <mntmn> https://i.imgur.com/w6E9cOn.png this is how far i get with kmscube on ls1028a

22:14 <mntmn> cc austriancoder

22:14 <mntmn> similar with weston, all the jobs time out.