narmstrong changed the topic of #linux-amlogic to: Amlogic mainline kernel development discussion - our wiki http://linux-meson.com/ - ml linux-amlogic@lists.infradead.org - official channel moved from Freenode - publicly logged on https://libera.irclog.whitequark.org/linux-amlogic
kenny has joined #linux-amlogic
naoki has joined #linux-amlogic
camus has joined #linux-amlogic
camus has quit [Read error: Connection reset by peer]
camus has joined #linux-amlogic
camus1 has joined #linux-amlogic
camus has quit [Remote host closed the connection]
camus1 is now known as camus
jernej has joined #linux-amlogic
jernej has quit [Ping timeout: 268 seconds]
ldevulder has joined #linux-amlogic
<narmstrong> There's definitely an issue here, and it was not the case when the support was added, so we would need to find when that appeared
<narmstrong> if someone can run a git bisect to determine when it appeared, it could held...
elastic_dog has quit [Ping timeout: 260 seconds]
<lvrp16> narmstrong: I don't have the 59.94hz frequency 1080p displays, on my regular 60hz 1080p displays, I don't get any issue.
<lvrp16> I will try with the 1280x400 display on older version and do a bisect
elastic_dog has joined #linux-amlogic
tortoisedoc has joined #linux-amlogic
Daaanct12 is now known as Danct12
<tortoisedoc> hello, trying to do dmabuf-based  zero-copy on latest kernel here; but it appears meson headers are not public yet. could this be correct or am I just approaching this from the wrong angle? narmstrong pinging you as found you talking about zero-copy in archives, apologies in advance :)
<tortoisedoc> (meson headers meaning for meson-drm / drm-meson (whichever), that is)
<tortoisedoc> found some headers from endlessmobile kernel, but that appears to be UMP-based..
<tortoisedoc> found also this example : https://github.com/anholt/drm-v4l2-test/blob/master/dmabuf-sharing.c but its not for multiplanar devices..
Ballerburg9005 has joined #linux-amlogic
camus has quit [Ping timeout: 244 seconds]
camus has joined #linux-amlogic
Ballerburg9005 has quit [Ping timeout: 265 seconds]
fmiz has joined #linux-amlogic
fmiz has quit [Client Quit]
fmiz has joined #linux-amlogic
jernej has joined #linux-amlogic
fmiz has quit [Remote host closed the connection]
Cheaterman has joined #linux-amlogic
<Cheaterman> Hey narmstrong :-) just wanted to let you know, I updated my kernel so I tried your way and it worked, to add RGB888_1X24 to output_fmts before yuv420/444
<lvrp16> Cheaterman: do you have the patch?
kenny has quit [Ping timeout: 252 seconds]
vagrantc has joined #linux-amlogic
fmiz has joined #linux-amlogic
tortoisedoc has quit [Quit: Client closed]
<Cheaterman> lvrp16: Oh absolutely
<Cheaterman> There you go
<Cheaterman> It's literally no more and no less than narmstrong suggested :-) screens apparently always say they support yuv420p because it's supposed to be the _real_ lowest common denominator according to the specification (or something along those lines)
<Cheaterman> but sometimes they don't, and the _actual_ lowest common denominator is rgb
<Cheaterman> at least that's my understanding of the issue - the facts are, screen says it supports yuv420 but doesn't
<Cheaterman> (I say "screen" but... it's been observed on all the monitors I tried, that's 4 of them)
<lvrp16> Cheaterman: thanks, edid is really a SS...
<Cheaterman> tell me about it :-(
<Cheaterman> I also updated my issue on tobetter's fork for posterity: https://github.com/tobetter/linux/issues/37
<Cheaterman> (hoping tobetter might add it to the patchset)
Ballerburg9005 has joined #linux-amlogic
<lvrp16> Cheaterman: I have to redo the patch for upstream :D doesn't apply
tortoisedoc has joined #linux-amlogic
fmiz has quit [Remote host closed the connection]
<Cheaterman> lvrp16: sorry haha, I'm using that tobetter fork because mainline didn't work for me last time I tried
<lvrp16> i reworked it, testing it now
<Cheaterman> If you reckon I should give it another shot at some point I will, but the Odroid community seems to think the tobetter fork is a better bet for me right now
<Cheaterman> (I'm on N2+)
<lvrp16> *sigh*
<Cheaterman> Did it work? :)
<lvrp16> oh it probably will, it's a simple enough fix
<Cheaterman> yup
<lvrp16> the sigh was more of the general attitude
<lvrp16> for these off-upstream forks
<Cheaterman> tell me about it
<Cheaterman> I just want my s*** to work, not spending days finding the right fork
<Cheaterman> and it's not just kernels btw
<Cheaterman> rtl8821cu has like 3 drivers
<tortoisedoc> oh kernel is a kernel is a kernel..
<Cheaterman> one in kernel and 2 forks
<Cheaterman> only one of the forks works... lost a day on that
<Cheaterman> some things are just uuuuuuuuuugh
<Cheaterman> feels like 2005 all over again
<Cheaterman> ndiswrapper soon? -_-
<lvrp16> mediatek is your friend
<Cheaterman> yeah :-) i already use their socs on my phones
<Cheaterman> realtek has been extremely dodgy lately for me
<Cheaterman> even standard ethernet on most recent machines is out-of-tree
<lvrp16> yikes
<Cheaterman> yupu
<Cheaterman> no firmware and no module? good luck using a usb stick to get network
<lvrp16> hardkernel has a bunch of patches they need to upstream
<Cheaterman> Yes :-(
<lvrp16> but *crickets*
<lvrp16> i steal from their tree time to time
<lvrp16> for upstream
<Cheaterman> I'm using their machines (and therefore the S922X I think it is?) because I need the oomph
<Cheaterman> and TBF broadcom has their own issues (with regards to Pi that I was using before...)
<Cheaterman> wifi is awful for example
<tortoisedoc> wifi + bt on rpi is dreadful
<Cheaterman> and general perf is miserable
<Cheaterman> yes
<lvrp16> haha their PCB cavity
<tortoisedoc> especially on the zero
<Cheaterman> even on pi4
<Cheaterman> it's awful
<lvrp16> the PCB cavity stuff is not reliable
<Cheaterman> at least wifi
<Cheaterman> and pi4 has terrible gpu perf compared to N2+
<tortoisedoc> yeah well, one / two radio usb sticks and byebye badnwidth
<tortoisedoc> was it mali on the pi?
<Cheaterman> comes down to those terrible broadcom videocore gpus vs the malis in general I suppose
<Cheaterman> no
<Cheaterman> it's broadcom crap
<tortoisedoc> ah right
<Cheaterman> and it's really bad
tortoisedoc has quit [Quit: Client closed]
tortoisedoc has joined #linux-amlogic
<Cheaterman> Wb haha
<tortoisedoc> that was a heist
<lvrp16> Pi's antenna relies heavily on ground plane. when ground plane changes, so does your signal.
<Cheaterman> lvrp16: that's fair in principle
<tortoisedoc> oh thats interesting to hear @Ivrp16
* tortoisedoc gotta get rid of the bloody discord syntax
<Cheaterman> the problem is dodgy firmware and whatnot
<Cheaterman> and driver even
<Cheaterman> sometimes the thing just locks up
<Cheaterman> unable to scan or reconnect
<tortoisedoc> pi has been improving on that front I believe (tho they still ship crapton of blogs)
<tortoisedoc> *blobs
<lvrp16> IMHO they should have stuck with chip antennas like the Pi 3 Model B and earlier
<lvrp16> the B+ kind of screwed everything, their hardware tech always seems to be 2 step forward and 1 step back
<tortoisedoc> power constraints perhaps
<Cheaterman> if their GPUs did a bit more for a bit less....
<Cheaterman> that would help...
<tortoisedoc> though the 5v charger on the pi is really a life saver
<tortoisedoc> 0.0000001 v less and good luck running barely anything
<tortoisedoc> anyways
<tortoisedoc> gonna get back to my dmabuf exercise
<Cheaterman> on N2+ recently I did a project where I had to do transparent video playback on top of a webcam feed, record it, then re-encode the recorded bits to append them to each other
<Cheaterman> a Pi4 would have been soooooooooooo dead
<Cheaterman> N2+ handled it like a champ
<tortoisedoc> grr bloody ioctls
tortoisedoc has quit [Quit: Client closed]
<Cheaterman> speaking of controlling io, I had to do some basic stuff with N2+ GPIO for some projects, buttons and potentiometer, it was pretty fun
<Cheaterman> potentiometer easy sysfs thing for the adc, and gpio with gpiod thingy
tortoisedoc has joined #linux-amlogic
<Cheaterman> i had to add a udev rule for the gpiod thing
<Cheaterman> but it was fun
<tortoisedoc> oh adding udev rules is always fun (thank you systemd!)
<Cheaterman> hahahaha
<Cheaterman> I'm not even using systemd
<Cheaterman> and TBF I could do it the ugly way, making a service that calls chgrp or something
<Cheaterman> but udev is neat-ish for that purpose
<Cheaterman> the one that makes me angry usually is polkit
<Cheaterman> this one is really a spawn from hell
<tortoisedoc> hmm you built udev without systemd?
<tortoisedoc> thats something I need to try at some point
<Cheaterman> oh yeah. it's the way to go with openrc nowadays
<tortoisedoc> ( I got bored staring at system once I noticed they hard-linked into libusb)
<tortoisedoc> s/system/systemd
<tortoisedoc> (no time for cheap hacks :) )
<tortoisedoc> openrc hmm
<tortoisedoc> ah I see
<tortoisedoc> "We don't support building a static OpenRC with PAM." lol
<Cheaterman> hahahah
<tortoisedoc> really nobody likes pam it seems
<Cheaterman> doesn't mean that it doesn't work mind you
<Cheaterman> some people do try crazy experiments and PR fixes
<tortoisedoc> heh
<Cheaterman> just, upstream wouldn't want to support that configuration
<Cheaterman> but yeah GEntoo is like that, people experiment with weird stuff for fun
<tortoisedoc> upstream of openrc?
<Cheaterman> yes
<tortoisedoc> yeah gentoo is for crazies
<Cheaterman> well call me crazy then xd
<tortoisedoc> and passionate adventurers :D
<Cheaterman> or a passionate adventurer xD
<tortoisedoc> lol
<Cheaterman> but I use it for work too
<tortoisedoc> ive got way too many deadlines to enjoy gentoo
<tortoisedoc> so I stick with debian :)
<tortoisedoc> (or lately devuan)
<Cheaterman> I got deadlines too haha
naoki has quit [Quit: naoki]
<Cheaterman> just was lucky enough to learn Gentoo before that I guess
<lvrp16> imagine Gentoo on Cortex A7...
<tortoisedoc> its like you first fight to get your system up and running neately, perfectly tuned, then IT swipes in and brings new monitors, new docking station, a new extension card...
<Cheaterman> then again wouldn't have deadlines if it weren't for me learning gentoo
<Cheaterman> lvrp16: been there done that
<tortoisedoc> yeah thats a net combination
<Cheaterman> it's OK
<tortoisedoc> s/net/neat
<lvrp16> that's masochism
<Cheaterman> meh
<Cheaterman> i was cross compiling the kernel
<Cheaterman> the rest of the system is fine
<Cheaterman> try qemu-user
<Cheaterman> feel some REAL sluggishness
<lvrp16> haha try compiling Chrome or Firefox XD
<tortoisedoc> qemu-user on a ryzer :D
<Cheaterman> (did a whole system on it, 0/10 would not recommend)
<lvrp16> 2 years later...
<Cheaterman> Chrome yeah
<Cheaterman> Firefox naaah it's pretty quick
<Cheaterman> merge time: 19 minutes and 55 seconds.
<Cheaterman> on my current system
<Cheaterman> for 104.0.2
<tortoisedoc> weeel, one could still cross-compile first...then only RUN stuff on cortex a7
<tortoisedoc> but that wouldnt be gentoo, would it
<Cheaterman> that would be totally fine
<Cheaterman> to each their own
<Cheaterman> ther's no "gentoo way", the "gentoo way" is the one you choose
<Cheaterman> gentoo is about choice
<tortoisedoc> thats the aspect I like the most really
<Cheaterman> there's even dist-kernels in the tree nowadays
<tortoisedoc> (of gentoo)
<Cheaterman> gentoo-kernel-bin:
<Cheaterman> Description: Pre-built Linux kernel with Gentoo patches
<Cheaterman> Because some people CBA and that too is fine (took us long enough to realize, admittedly)
<Cheaterman> anyway gentoo is not nearly as insane as people make it out to be
<Cheaterman> and doing anything serious on any other platform is seriously painful to me
Cheaterman has quit [Ping timeout: 248 seconds]
Cheaterman has joined #linux-amlogic
buzzmarshall has joined #linux-amlogic
jernej has quit [Ping timeout: 252 seconds]
Ballerburg9005 has quit [Ping timeout: 265 seconds]
naoki has joined #linux-amlogic