hexdump01 has joined #aarch64-laptops
hexdump0815 has quit [Ping timeout: 480 seconds]
hexdump01 has quit []
hexdump0815 has joined #aarch64-laptops
<hexdump0815> macc24: re nyan/tegra display - i plan to recheck when 5.13 is out - so far it only got worse with newer releases: 5.4 still useable, 5.10 worse, 5.12 even more worse ...
<hexdump0815> macc24: i played around with the 5.12 tree from eballetbo and with 5.13-rc6 on duet and at least got the display partially working on both - with panfrost and xorg i did not yet get it working again
<hexdump0815> macc24: i think you got there too - right?
<hexdump0815> macc24: on both cpu dvfs does not seem to work - https://pastebin.com/raw/Ntd27cHW - is it the same for you?
<macc24> hexdump0815: do issues with nyan display look remotely like those on duet on 5.13?
<macc24> 1i'm seeing super weird issues on duet on anything newer than my 5.10 with hacks on hacks on it
<hexdump0815> macc24: i think the issues are different
<hexdump0815> macc24: on nyan it seems to be a probe order problem of the drm drivers - sometimes they end up in the right order and sometimes in the wrong one
<hexdump0815> macc24: on duet the mainlining is still ongoing and i think there are still some hacks required most probably on top of v5.13 to make it really work
<hexdump0815> macc24: but at least there is something on the display now :) ... the usb bandwidth issues seem to be still there as well and i'm not sure if dvfs was ever working on mainline yet
<macc24> hexdump0815: ack
<macc24> iirc cpufreq on mt8183 has some shenanigans that kernel has to work around, same with gpu devfreq
<macc24> hexdump0815: after spending too much time with mediatek drm driver i just think ill complain to linux-mediatek mailing list
neobrain has joined #aarch64-laptops
FizzBuzz has joined #aarch64-laptops
<macc24> has anyone tested how big of an init ramdisk can i shove into arm chromebook?
<robclark> hmm, needs to be (uncompressed) smaller than RAM size ;-)
<macc24> i think i can fit debian + rootfs into 1gb
<robclark> I think at one point we were using initramfs for rootfs in mesa CI (but later switched to nfsroot since that was faster)
<macc24> now whether i can make a 1gb kernel partition and have depthcharge not shit itself is other topic