indy has quit [Read error: Connection reset by peer]
indy has joined #tegra
indy has quit [Ping timeout: 260 seconds]
indy has joined #tegra
torez has joined #tegra
<kwizart>
Hi, are nouveau devices (from tk1 to aarch64 t210+) are supposed to be wayland capable ?
<kwizart>
Seems like I'm experiencing issue with current fedora with wayland and I would have expected them to work
<kwizart>
(unless X11 as fallback would requires xorg server from master branch)
<cyndis>
it should theoretically work, but the last time I tested wayland on tk1 (or any other chip for that matter) was years ago
<kwizart>
do you mean you have tested it using xorg or don't test it graphically at all ?
<kwizart>
I don't think the error was very obvious, it's just that something has crashed in the wayland (gdm/gnome) stack and so wayland is discard and xorg is attempted
<kwizart>
I will try to make a better report next time
<cyndis>
i tested it by running weston with chromium over xwayland and ioquake
<cyndis>
this was back when TK1 was the new hotness
torez has quit [Ping timeout: 245 seconds]
torez has joined #tegra
gouchi has joined #tegra
hexdump0815 has joined #tegra
<hexdump0815>
kwizart: in my experience it is not really useable with wayland - i tried it on a nyan chromebook (tk1) with wayfire and it was so full of graphics glitches that it was unuseable ... not sure if i tested tx1 (nano) too back then (= maybe a year ago)
<hexdump0815>
kwizart: i think it is simply bit rotting as nobody is really testing or using it anymore and on top of that it is quite slow - in no way comparable to the state of panfrost or lima nowadays
<kwizart>
hexdump0815, yep the "modifier" xorg feature requires to use 1.21 xorg-server branch
<kwizart>
hexdump0815, do you mean panfrost/lima are way better ? (I have "lafrite" board)
<kwizart>
I mean to test mainline kernel (5.15rc) + backported patches to deal with the tearing issue I'm experiencing with Xorg (1.21rc)
<kwizart>
I expect this is because the missing power management feature
digetx has quit [Ping timeout: 252 seconds]
<hexdump0815>
kwizart: in my experience panfrost/lima are working quite well now - but the latest kernel i dealt with was 5.10 and a bit of 5.13 ... things might always break a bit from time to time :)
digetx has joined #tegra
<hexdump0815>
kwizart: on tk1 and tx1 i did not get any kernel working well anymore after 5.10 due to lots of drm trouble - maybe i'm missing some kernel config changes and on the tk1 side it might be a nyan only problem
<hexdump0815>
kwizart: so in case you get anything working with 5.15 on those i would be interested in your config and potential patches
<tagr>
hexdump0815: when things do break, it'd be great if you could report it, otherwise we may just not notice
<tagr>
it's true that things sometimes bitrot because there are gaps in the test coverage, so we do rely on users to report when things do break
<tagr>
if we don't know that things are broken, we can't fix them =P
<kwizart>
hexdump0815, current fedora kernel are working (I'm also assuming an updated modern mesa) but one needs xorg-server-1.21
<hexdump0815>
kwizart: yes i got it working in the end somehow with newer xorg but even then it does not work well (reported in the issue mentioned above): tk1 seems to run out of memory and tx1 cannot raise the freq of the gpu at all
<hexdump0815>
tagr: yes - i planned to report the problems (beyond what i reported for nouveau) but did not get to the point that i could clearly define a reproducable state ... i think a problem is that the user base of mainline on tk1 and tx1 is close to zero and thus problems are unnoticed for a long time and new ones come up unnoticed
<hexdump0815>
tagr: btw. i reported some of the problems i ran into in this channel there was no real solution for them so far
<tagr>
yeah, IRC is not a great place for issue tracking... there is a drm/tegra project on freedesktop's gitlab now that could be used for this kind of thing
<hexdump0815>
tagr: thanks - good to know - i'll try to open issues there the next time i give it a try
azkali has joined #tegra
<azkali>
Hi, I am trying to
<azkali>
boot mainline on a Jetson TX1 but I get stuck at sd initialization in kernel but there is no crash, it just hang
<azkali>
I am trying to use the latest drm abi as well as nvdec ( I am using googlesource/tegra's linux fork )
<tagr>
azkali: you might have better luck using linux-next
<azkali>
Ah thank you, just a question, since I you are here I have a slight issue understanding how to compose my rootfs to use the new drm abi.
<azkali>
As far as I understood
<azkali>
I need nouveau in kernel, your libdrm fork on the drm-tegra-uabi-v6 branch and mesa, right ?
torez has quit [Remote host closed the connection]
gouchi has quit [Quit: Quitte]
<azkali>
And also, I was told that nouveau + nvidia userspace drivers was a working scenario for the Pixel C, but does this remain true ?
crabbedhaloablut has quit [Remote host closed the connection]
crabbedhaloablut has joined #tegra
<DavidHeidelberg>
digetx: Hey, thanks for review, I updated the bindings + fixed some additional my overlooks (this commit was from earlier era, so more bugs in it 😉 )
<DavidHeidelberg>
s/digetx: Hey, thanks for review, I updated the bindings + fixed some additional my overlooks (this commit was from earlier era, so more bugs in it 😉 )/digetx: Hey, thanks for review, I updated the bindings + fixed some additional stuff I overlooked (this commit was from earlier era, so more bugs in it 😉 )/
dormito has joined #tegra
<DavidHeidelberg>
digetx: not sure if I was explicit, the patch isn't conversion, but replacement of old txt binding in grate tree
<DavidHeidelberg>
best would be, if someone with Glide could upstream it together with mcs driver enhancements