02:10 <anarsoul> hey folks, I'm trying to get my BTT HDMI5 working with BTT CB2 (rk3566) and I'm getting NOCLOCK for all the modes

02:11 <anarsoul> it's a 800x480 screen

02:11 <anarsoul> https://gist.github.com/anarsoul/33599df50aac5f8bab55baac67a5b493

02:12 <anarsoul> any ideas how to fix it up? :)

03:44 <anarsoul> I presume I need to patch clk-rk3568.c to add necessary clocks?

06:35 <Kwiboo> anarsoul: dedicating hpll for hdmi and using a custom pll table should work, tried to use something like https://github.com/LibreELEC/LibreELEC.tv/blob/d961b36d6ac273fbbff7e3661967a24b78f68798/projects/Rockchip/patches/linux/default/linux-3000-RK3568.patch in an early revision for LibreELEC

06:36 <Kwiboo> that seemed to help with lower uncommon resolutions but was not really upstreamable so eventually dropped all custom rk3568 patches

06:40 <anarsoul> 640x480@60 should be pretty common though?

06:45 <anarsoul> *sigh* I really wonder why HDMI is always broken with rockchip

06:50 <anarsoul> btw, it rejects 800x600@60 and 1024x768@60 as well

07:16 <Kwiboo> anarsoul: yep, I had to use a custom hpll rate table exacly for those low resolutions, because default pll table would not allow it, and I did not want to try and adjust the shared pll table, we where already using custom pll table for hdmi on rk3328/rk3399 so why not also on rk3568 ;-)

07:16 <Kwiboo> mmind00: look like your "clk: rockchip: rk3576: define clk_otp_phy_g" has not been applied to any tree for next yet? I am getting "rockchip-otp 2a580000.otp: error -ENOENT: Failed to get clk 'phy'" without it on latest linux-next

07:51 <anarsoul> Kwiboo: I just added two more rates to common table and display seems to work now

09:06 <CounterPillow> Kwiboo: yeah I noticed that patch missing as well, I assumed it was just not in linux-next yet but if it's not in any tree then the nvmem maintainer did an oopsie by dropping it after saying they applied it

09:10 <mmind00> Kwiboo: it looks like mmind00 is blind and overlooked that there was a clock patch in that series ;-)

09:10 <mmind00> CounterPillow: ah right ... the applied message from the nvmem maintainer reads they took the clk patch ... so I'm not blind afterall

09:11 <CounterPillow> I assume they realised their mistake of picking a patch they shouldn't have picked, dropped it, but then... didn't tell anyone :D

09:13 <mmind00> I'll ask the nvmem maintainer what's up and pick the patch then ... this is small enough that it can also fit in as fix if need be