01:18 <digetx> timing should be specified in kernel's panel driver, you need to know exact panel model name, which can be found in panel's edid

01:25 <digetx> chromeos has things like https://chromium-review.googlesource.com/c/chromiumos/third_party/kernel-next/+/200545/ which should be a part of panel description in the kernel driver

07:40 <hexdump0815> digetx: thanks a lot for this pointer - i'll have a look into this direction over the weekend

11:08 <kwizart> tagr, do you have room to test the v3 of the xhci missing interrupt at http://patchwork.ozlabs.org/project/linux-tegra/patch/20211107224455.10359-1-digetx@gmail.com/ fedora distro kernel maintainer are waiting for your ack , thanks in advance

11:09 <tagr> kwizart: it's already got my Acked-by and Tested-by, what more do you want? =)

11:10 <tagr> the only thing that changed between v2 and v3 was a local variable name and a debug message, none of which impact the functionality of the patch, so that Tested-by is still valid

11:11 <kwizart> ha, so what's going on next ? waiting for the next merge windows or can be sent to stable branches as bugfix ?

11:13 <kwizart> anyway thanks for clarification

14:42 <digetx> hexdump0815: you only need to find the panel model name, you can use any of available tools in linux to read edid or run `find /sys/devices | grep edid | xargs cat` that will show the panel name

14:47 <digetx> it's also not a problem to search internet for the panel model, seems should be NV133FHM-N43

17:55 <hexdump0815> digetx: the panel is a b133htn01.1 - in the dts b133xtn01 is referenced

18:04 <DavidHeidelberg> Btw. `make dtbs_check` is showing warning for some time, surface and Ideatab need few fixes... It's very quiet around these devices 😅

18:28 <digetx> hexdump0815: does only 4gb model have 1080 panel?

18:29 <digetx> DavidHeidelberg: can you fix them yourself?

18:31 <hexdump0815> digetx: i'm not sure - i have 4gb with full hd and 2gb with 1366x768 and mostly noticed them in this combination, but there seem to have been other nyan bigs in other markets with 2gb and 1366x768 - so maybe all combinations are possible in the end

18:31 <hexdump0815> sorry i ment other ones with 4gb and 1366x768

18:41 <DavidHeidelberg> without knowledge of HW and ability to test, hardly

18:46 <digetx> hexdump0815: I added tegra124-nyan-big-fhd.dts to grate kernel, please try it

19:08 <hexdump0815> digetx: thanks - will try it and let you know

19:20 <digetx> DavidHeidelberg: which errors you can't fix?

19:33 <hexdump0815> digetx: perfect - setting the proper display in the dts indeed fixes the problem completely and reliably as it seems

19:34 <hexdump0815> i think you can omit the memory entry as its defined like this already in tegra124-nyan.dtsi

19:36 <hexdump0815> thanks a lot for this - i think with this (and the other fixes) the nyan big chromebooks are somewhat useable again with v5.15 :)

19:36 <digetx> 1080p model comes only with 4GB, correct?

19:37 <digetx> yw, I'll add yours tested-by to the DT patch

19:38 <hexdump0815> so far i did never see any 1080p model with only 2gb, but it could be that something like this exsits - on the other hand it might make sense that the full hd version also has the larger mem

19:38 <hexdump0815> there seem to be 2gb and 4gb models with the normal hd screen

19:50 <hexdump0815> btw. - someone seems to have gotten suspend to ram kind of working on nyan-big in lp1 mode - i so far only got s2idle working reliable, but power usage with it was not much lower than just leaving the chromebook running :)

19:51 <hexdump0815> how much more power savings i might expect with lp1 mode?

19:51 <digetx> I'll remove memory node; perhaps there should be two memory regions for lpae, not sure

19:52 <digetx> have you tried lp2?

19:52 <digetx> was lp1 ever working before?

19:52 <hexdump0815> i'm currently handling the 2gb/4gb topic by using two different chainloaded u-boot versions for both - for the 4gb one with lpae enabled and for the other without

19:54 <hexdump0815> without lpae in u-boot the 4gb only sees 2gb and enabling lpae on the 2gb model did not boot - but maybe this is due to the default 4gb mem definition in the dt? so without lpae it can only see 2gb anyway and if lpae enabled it crashes as sdt says 4gb and wardware is only 2gb ...

20:05 <hexdump0815> no idea regarding the different lp modes - this is the only thing i found: https://archlinuxarm.org/forum/viewtopic.php?f=49&t=12185&start=230#p64180

20:07 <hexdump0815> i so far only got s2idle working with an ugly hack to bring back the display otherwise: https://github.com/hexdump0815/imagebuilder/blob/main/systems/chromebook_nyan/extra-files/lib/systemd/system-sleep/mrvl-and-edp-reload

20:08 <hexdump0815> otherwise the screen remains black after resume even from s2idle ... maybe this is related to the same i2c edid problems in the end?

20:23 <digetx> hexdump0815: https://dpaste.com/G89VPRXMW

20:25 <hexdump0815> cool - thanks - will give it a try - will first have to create a fresh test sd card to have a reliable base to test with - will let you know the result

20:32 <digetx> ok

20:32 <digetx> I don't know whether 2gb version should work with lpae, lpae should be supported by cpu

20:48 <digetx> if bootloader sets wrong memory size for 2gb version, then it's a bug

21:03 <hexdump0815> not sure about u-boot (will have to check) - i just thought about the 4gb memory definition in the tegra124-nyan.dtsi for the kernel - not sure if that is used anywhere or just there as a reference

21:07 <digetx> tegra124-nyan.dtsi has 2gb

21:07 <hexdump0815> btw. u-boot for the nyan big is quite strange too: in its dts it sets the panel timing fixed for 1366x768 - this is working well for the full hd nyan big (although that is fhd and the screen also seems to be fhd) but it does not work for the hd version (crza screen flickering as it does not seem to be able to initialize the panel properly)

21:07 <hexdump0815> crza = crazy

21:08 <hexdump0815> the nyan big u-boot works as well for nyan blaze (which has normal hd screen) and there the same u-boot is working perfectly fine :)

21:12 <hexdump0815> oh right - its indeed 2gb in tegra124-nyan.dtsi - then i guess its being ignored in the end maybe as the 4gb nyan big is working well with 4gb or useable ram with this setting

21:13 <hexdump0815> but i must admit that i do not fully understand how lpae works, so maybe its ok due to that

21:14 <digetx> which nyan-big u-boot? I don't think that upstream u-boot supports eDP

21:18 <digetx> 1366x768 resolution shouldn't work on 1080p panel as it supports only one fixed mode and DE mode-only

21:21 <hexdump0815> mainline u-boot gets chainloaded by depthcharge, the chromebook bootloader and i guess it takes over the console from it

21:24 <hexdump0815> the u-boot sources have this in: https://source.denx.de/u-boot/u-boot/-/blob/master/arch/arm/dts/tegra124-nyan-big.dts#L35-45

21:24 <hexdump0815> with this in u-boot it flickers like crazy on the hd model and it works on the fhd model (with the screen ending up in fhd mode)

21:26 <hexdump0815> if i remove this timing entry then no more flickering and just a black screen on the hd model and no more working u-boot display and a black screen too on the fhd model

21:26 <hexdump0815> i also played around with replacing the hd timing entry with a proper fdh timing entry, but it did not change anything - fhd model works as before

21:27 <digetx> I don't understand what "with the screen ending up in fhd mode" means

21:28 <hexdump0815> i guess the mainline u-boot code was not primarily tested in chainloading mode, but by running u-boot from spi flash which seems to be possible too but is more risky as it could brick the system worst case

21:29 <digetx> do you mean the 768 display area inside 1080?

21:30 <hexdump0815> screen ending up in fhd mode means: the display output of u-boot is shown in full hd resolution (text is bigger on the normal hd model for comparison) - so in the end it looks like the hd timing in the dts does not seem to be used for the fhd panel timing, but if the timing entry gets removed the display stays black for u-boot

21:31 <hexdump0815> no - no 768 area in 1080 - full 1080 and in full 1080 resoltion

21:32 <digetx> ah, u-boot actually supports eDP

21:33 <digetx> so it should use edid's mode

21:34 <digetx> https://github.com/grate-driver/u-boot/tree/master/drivers/video/tegra124