lanefu changed the topic of #armbian-rockchip to: Armbian - Linux for ARM development boards | Rockchip SoC | www.armbian.com | This channel is relayed to the equivalent Discord channel | this channel is logged
<DC-IRC>
<lanefu> @microlinux
<DC-IRC>
<lanefu> I guess something new in a 6.6.2
<DC-IRC>
<c0rnelius77> its an easy fix. its just failing because they changed 3 lines.
<DC-IRC>
<lanefu> But what's the patch for and why is it something "we" upstream?
<DC-IRC>
<c0rnelius77> Its a Dynamic Memory Controller driver for RK3328. Not sure where it came from, might have originated from vendor sources? Could have come from lore.kernel.org and never got accepted? I'm not the fella to ask. Looks like there are two diff rk3328-dram-default-timing.dtsi files. One for general units and one for the Renegade.
<DC-IRC>
<c0rnelius77> Its a hack-a-noodle patch 🙂
<DC-IRC>
<Tonymac32> Rk3328 is a processor no one wants to maintain so the ram controller does not get mainline attention. The different timings are due to different board layouts, one for those that copy the Rockchip layout and one for Firefly
<DC-IRC>
<Tonymac32> It speeds up the ram
<DC-IRC>
<lanefu> sounds like it's earned the right to be it's own board family 😛
<DC-IRC>
<lanefu> sorry SOC family
<DC-IRC>
<Tonymac32> Yeah basically. I opened a PR to shitcan the patch so we didn't have to deal with it, but Paolo saved it (I understand, I just don't have the time to fix it repeatedly on random minor kernel updates)
<DC-IRC>
<lanefu> should we can for now, and then he can re-add later if he wants to?
<DC-IRC>
<lanefu> or i guess see if we can figure out the 3 line fix lol
<DC-IRC>
<Tonymac32> Just change name to *.disabled, yes
<DC-IRC>
<Tonymac32> Yeah it murders performance on boards without decent u-boot configs to set the RAM speed
<DC-IRC>
<Tonymac32> It's actually the source of that years old TKaiser ticket about the ram perf going to shit between vendor and mainline
<DC-IRC>
<Tonymac32> I tested it last week? 2 weeks ago? And closed it as fixed because it was working fine
<DC-IRC>
<Tonymac32> 🤷♂️🤦♂️
<DC-IRC>
<lanefu> bwaah fucking `wifi-4003-uwe5622-adjust-for-rockchip ` breaking too
<DC-IRC>
<lanefu> disabling dmc patch is cascading to other failures of dts patches for different rk3328 boards
<DC-IRC>
<Tonymac32> Of course I should have guessed, the whole thing is a card tower
<DC-IRC>
<Tonymac32> The device trees are all patched to use it
<DC-IRC>
<lanefu> i'm upset becuase "Fix the patch" is probably the most apropriate path.. other than breakout rk3328
<DC-IRC>
<igorpec> i just tested my fix, compilation is succesful
<DC-IRC>
<lanefu> -current failed for me with EXTRA_WIFI=no I thought we'd fixed that recently. but maybe i'm mistaken..
<DC-IRC>
<lanefu> building -current w/ wifi and loosk okay so far
<DC-IRC>
<igorpec> aha, that i didn't check
<DC-IRC>
<viraniac> rockchip is supposed to fail with EXTRA_WIFI=no
<DC-IRC>
<viraniac> rockchip patches has a patch for uwe5622 driver. if you are using EXTRA_WIFI=no, then kernel doesn't have uwe5622 driver to get patched. Hence you will have patch application failure
<DC-IRC>
<viraniac> I am no Idea why it was done that way and why that patch is not living in driver harness.
<DC-IRC>
<viraniac> I have no Idea why it was done that way and why that patch is not living in driver harness.
<DC-IRC>
<lanefu> we've had some covnersations before