sunshavi has quit [Read error: Connection reset by peer]
sunshavi has joined #linux-sunxi
tnovotny has joined #linux-sunxi
Mangy_Dog has joined #linux-sunxi
Putti has quit [Ping timeout: 276 seconds]
atsampson has quit [Ping timeout: 258 seconds]
Perlovka has joined #linux-sunxi
atsampson has joined #linux-sunxi
mi11k1 has joined #linux-sunxi
<mi11k1>
yes i looked at wiki .
<mi11k1>
im going to wait and then ask.
lkcl has quit [Ping timeout: 276 seconds]
camus has joined #linux-sunxi
kaspter has quit [Ping timeout: 264 seconds]
camus is now known as kaspter
Putti has joined #linux-sunxi
freemangordon1 has joined #linux-sunxi
freemangordon has quit [Read error: Connection reset by peer]
Kwiboo has quit [Quit: .]
lkcl_ has joined #linux-sunxi
Kwiboo has joined #linux-sunxi
<libv>
mi11k1: wait for what?
<libv>
the order is ask and wait.
<libv>
as over the years there have been far too many people who have asked a question, and then ran off 5 minutes later
<mi11k1>
thats why im waiting for when i have time, lol
<mi11k1>
i was just being silly...i was in #debian and #arduino , didnt know u guys existed here
<mi11k1>
or ask questions that they could have found on google in 10 secs...
<mi11k1>
ill be back later to ask, cause i gotta leave in 5 mins..
<mi11k1>
15mins ago
mmarc__ has quit [Remote host closed the connection]
mmarc__ has joined #linux-sunxi
lkcl_ has quit [Ping timeout: 264 seconds]
<mi11k1>
ok, long story short. i have opi zero v 1.0 i think is printed on the pcb. running armbian buster. going back like 3 years i have this screen. https://github.com/waveshare/LCD-show
<mi11k1>
screen was for an old rpi3 that burnt up...i did ge th escreen working on rpi. ive already takemn a few swings at trying to get it to work on the opi
<mi11k1>
pinout is the same....originally i used the script LCD4-show from the git repo
<mi11k1>
made mods to try and make it work, /boot dir structure is different....tried cp the stuff manually. thats where i stopped
<mi11k1>
cause i thought.....is this even possible. i dont do this everyday, and its been a few years. i dont relly understand the overlays and stuff.
<mi11k1>
not asking for somebody to do it for me. just point me ...please
<mi11k1>
and, prob the distro i used on rpi was stretch
<mi11k1>
ill be in and out, but i will follow up today if theres replies, so no rush...thankyou
<KotCzarny>
possible, yes, but which step are you failing at? enabling spi?
<KotCzarny>
and i would suggest you to try armbian www forums
<mi11k1>
got that enabled, just the stuff in /boot
<KotCzarny>
you would get much better focused answers for armbian
<KotCzarny>
and most likely ready examples too
<mi11k1>
ive looked . ive been picking this up periodically for months and taking hacks at it.
<KotCzarny>
have you tried creating a thread there?
<mi11k1>
i dont use forums.
<mi11k1>
i usually figure it out myself eventually.
<KotCzarny>
if you like reinventing wheel, sure
<mi11k1>
as long as i know its possible, is good enough for me
<KotCzarny>
but what does 'just the stuff in /boot' means
<mi11k1>
like, the overlays..are they the same across boards?
<mi11k1>
i guess is how id put it.
<KotCzarny>
nope, not the same
<mi11k1>
di u look at the git repo?
<KotCzarny>
most likely overlays from the github are specific to spi
<KotCzarny>
*rpi
<KotCzarny>
yes, and strings like 'brcm,pins' inside are rpi specific
<KotCzarny>
thats why i suggest you to either look in armbian forums or linux-sunxi wiki
<mi11k1>
ok ok...so thats why i stopped where i did.
<mi11k1>
no, im done withit. just thought id try before i toss it.
<KotCzarny>
screen will work if you configure it right
<KotCzarny>
thats it
<mi11k1>
ya, which i did on rpi, so i just thought it would crossover easily.
<mi11k1>
at least i hoped it would.
lkcl_ has joined #linux-sunxi
<mi11k1>
all i know is a new rpi is not happening. i think they are rubbish. this opi ive had for years. im just going to keep using it to run dump1090
<mi11k1>
spi is slow anyways. 3 years ago it costed like $15. its a rightoff. thanks though. seriously.
<KotCzarny>
its good enough for ui
<mi11k1>
ya, that kinda did clear up something ive wondered for a long time...when hese companies sell stuff that is compatible, its is and its not..
<mi11k1>
i didnt even think about the broadcom thing..
<KotCzarny>
that screen will work without a trouble with allwinner
<KotCzarny>
its just that their github is rpi specific
<mi11k1>
whats a good cheap board ?
<mi11k1>
no the git link is from the screen manufacturer
<KotCzarny>
yes
<KotCzarny>
doesnt change what i wrote
<mi11k1>
ya, i know. but, i mean its not a random repo
<mi11k1>
when i said id toss it, i meant in a drawer. prob can get it working with arduino pretty easily.
<KotCzarny>
give armbian forums a try
<mi11k1>
it might even be cooked from the rpi. i dont know
<KotCzarny>
you dont have to reinvent the wheel
<mi11k1>
i looked in all those places
<mi11k1>
i ended up here.
<mi11k1>
the manufacturer sie, armbian, then debian, arduno, rpi, now here.
<KotCzarny>
anyway, those screens work, the only trouble is getting touch and display work at the same time
<mi11k1>
cause this hardware and this stuff from china is never what its supposed to be.
<mi11k1>
variants of variants of variants.
<mi11k1>
my screen is technically none of those
<mi11k1>
i manage to get it working with the script for the 4, but its a 5 inch
<mi11k1>
sometimes, i just need somebody to tell me that reinventing the wheel is a dumb idea
<mi11k1>
agreed.
<mi11k1>
if i need a screen that badly...ill just spend $15 and get a proper one. this was back in the day when i used to order stuff off ebay without having a clue.
pCactus has joined #linux-sunxi
pCactus has quit [Remote host closed the connection]
damex has joined #linux-sunxi
mi11k1 has left #linux-sunxi ["Leaving"]
damex has quit [Read error: Connection reset by peer]
damex has joined #linux-sunxi
gaston1980 has joined #linux-sunxi
damex has quit [Read error: Connection reset by peer]
damex has joined #linux-sunxi
damex has quit [Read error: Connection reset by peer]
damex has joined #linux-sunxi
<Mangy_Dog>
hi Pi in general question not just allwinner.... but is there any common i2c GT911 direct connection cap touch digitiser drivers going around?
<Mangy_Dog>
for either raspberry or allwinner (if the difference matters at all)
<Mangy_Dog>
or how easy would it be to make one
<Mangy_Dog>
id like to avoid having to use usb HID interfaces for it
<KotCzarny>
i2c is i2c
<KotCzarny>
it's brand neutral
<Mangy_Dog>
yeah but the data being sent isnt :p
<KotCzarny>
and gt9xx driver is in kernel afair
<Mangy_Dog>
who knows how the packets are organised
<Mangy_Dog>
(the datasheet
<Mangy_Dog>
)_
<KotCzarny>
you know i2c is a standard?
<Mangy_Dog>
ahh so should just work as long as i flag it or something?
<Mangy_Dog>
i2c is a standard for serial but its up to the developers how they construct there packets?
<Mangy_Dog>
or do all digitisers stick to a digitiser standard?
<KotCzarny>
that being said, there is a devil in the details
<KotCzarny>
so when you look inside the driver you might notice buttload of ifs
<Mangy_Dog>
nods
<KotCzarny>
for different revisions of gt9xx
<Mangy_Dog>
and as you say its in kernal, its likely in raspberry as well?
<KotCzarny>
and sometimes you have to provide digitizer 'bios' too
<Mangy_Dog>
ok
<KotCzarny>
you can use kernel's driver if it works, or you can reinvent one from scratch
<KotCzarny>
as usual
<KotCzarny>
the kernel one is integrated nicely to input infra
<Mangy_Dog>
i wont lie, id prefer not to :p
<Mangy_Dog>
from scratch that is
<KotCzarny>
then there is a chance gt911 touchscreen would work after some param tweaking
<Mangy_Dog>
can the driver be tweked for dual display?
<KotCzarny>
that's window manager part
<karlp>
Mangy_Dog: what's wrong with usb?
<KotCzarny>
driver only knows some min/max, it's up to you how you interpret the results
<Mangy_Dog>
kalp limited pornts and plenty of gpio available
<Mangy_Dog>
ports
<KotCzarny>
add usb hub chip?
<Mangy_Dog>
also feels wasteful using the usb hids
messpl has quit [Remote host closed the connection]
<Mangy_Dog>
when it can be interfaced directly anyway
<karlp>
sure, if you've got no usb ports left.
<Mangy_Dog>
tbh at this stage im not sure if ill be using one of my allwinner boards i got for this or a raspberry pi
<Mangy_Dog>
maybe even a cm4
<KotCzarny>
and thats where 'standards' part shines
<Mangy_Dog>
hence why i was asking more in gerneral that just allwinner
<Mangy_Dog>
the fun bit is im not even sure ill be using a gt9 based digitiser yet
<Mangy_Dog>
Im going to be taking apart an old tablet
<Mangy_Dog>
using its screen and see if i can use its digitiser for one half
<Mangy_Dog>
but then having a e ink in the other half and might need to buy a gt based digitiser for that
<Mangy_Dog>
though it might also have its own digitiser
<Mangy_Dog>
built in
<Mangy_Dog>
the project idea is an e "book" sleeper
<Mangy_Dog>
lcd based display on one side of the book, e ink on the other
<Mangy_Dog>
with the e ink for reading and on screen keyboard
<Mangy_Dog>
lcd for normal other stuff
<KotCzarny>
einks get faster and faster every year
<Mangy_Dog>
nodnods im not worried about refresh speed
<Mangy_Dog>
for the e ink
<Mangy_Dog>
its going to be used traditionally
<Mangy_Dog>
Though theres a whole load of knowledge gap on how im going to force organise a working UI for it. So the system knows how to handle the e ink and what to display on it
<Mangy_Dog>
I dont want it to work like a traditional dual display desktop
<KotCzarny>
just write custom app for the eink part
<Mangy_Dog>
where you can move windows between them
<Mangy_Dog>
you make it sound easy :D
<Mangy_Dog>
but yes
<Mangy_Dog>
i have a feeling its something along those lines
<ndufresne>
PPA: sorry, wasn't connect over the weekend, if you board supports it, you may be able to get output with changing the z-order, otherwise, perhaps give a shot to force-modesettings=1 property on kmssink, this element lacks the smart stuff that is in let's say Khodi or Compositors in general
bauen1 has quit [Ping timeout: 276 seconds]
bauen1 has joined #linux-sunxi
cnxsoft has quit [Read error: Connection reset by peer]
cnxsoft1 has joined #linux-sunxi
freemangordon1 has quit [Quit: Leaving.]
freemangordon has joined #linux-sunxi
elros1 has joined #linux-sunxi
<PPA>
Unfortunately "! kmssink plane-properties=s,zpos=1" doesn't work either, nor force-modesettings=1
<PPA>
I built gstreamer with: "-Dbad=enabled -Dgst-plugins-bad:v4l2codecs=enabled -Dbase=enabled -Dgst-plugins-base:playback=enabled", maybe I'm missing something?
willmore has quit [Ping timeout: 256 seconds]
cmeerw has joined #linux-sunxi
<PPA>
ndufresne: https://pastebin.com/kvEthQGZ log output, if it helps. The stuff at the end looks like a syntax error on my part, but this worked on older gst-play stuff
tnovotny has quit [Remote host closed the connection]
<ndufresne>
PPA: I think failure by not failing are kind of the worst ;-P
<ndufresne>
PPA: my guess would be that there is no writable zpos for this version of the SoC
<ndufresne>
and then whatever combination kmssink picks, it will always endup drawing under an opaque layer
<ndufresne>
the bad news here is that to use kmssink, you'd need to start modifying kmssink to make its mode setting smarter, giving it the ability to put from transparent buffers on the upper layer
<ndufresne>
e.g. weston will do this magically, because it setup a cursor plane
<ndufresne>
khodi will render on that top layer anyway, so all pretty natural
<jernej>
ndufresne: It's Kodi, at least I think you mean that :)
<PPA>
I'm mainly confused because this worked with the unpatched 5.10 (or 5.11, forgot) + GStreamer 1.18 combo (when it didn't segfault—also no hwaccel)
<jernej>
PPA: which SoC? A64?
<PPA>
H3, Orange Pi PC
<jernej>
ok, so ZPos is settable
<jernej>
can you compile libdrm test programs?
<jernej>
there is one very useful tool for that - modetest
<jernej>
with it, you can easily change zpos
<PPA>
I'll try, thanks
<ndufresne>
PPA: oh really, H3, same as me, I'm confused too
<ndufresne>
PPA: perhaps it would have been as simple as kmssink plane-properties=s,zpos=3
<ndufresne>
perhaps there is more layers somehow
<ndufresne>
the change is permanent per boot btw
<PPA>
ndufresne: I plan to repeat this on another device so I'll try that there
<PPA>
Oh, a third file plays fine
<PPA>
I'm going to try a few different ones
<ndufresne>
cool, on conformance tests, I got lot more failure on Cedrus then on Hantro/RKVDEC, I would need to research, if they play find on ffmpeg, then it's straightforward to fix
<ndufresne>
PPA: what would be extra nice, is to grab a good collection of failing stream you have, and report that as a gstreamer issue
<jernej>
PPA: ndufresne: it plays fine in Kodi, so I guess it's gstreamer issue
<ndufresne>
exactly, I do expect few oops still
<ndufresne>
now, of course that is if there isn't any patches in khodi that we are missing in 5.11
<PPA>
Out of the five sample files here, 2 artifact horribly, 3 are perfectly fine
<PPA>
They're a mifx of different h264 variants, let me ffprobe them again
<ndufresne>
things is that out of few hundreds of parameters, some stream may not use one that was miss-implemented
<ndufresne>
(this is VP9 through ffmpeg sw / VAAPI, gst libvpx / VAAPI / new VA plugin + libvpx reference runner
gaston1980 has quit [Quit: Konversation terminated!]
<PPA>
Works not: h264 (High), h264 (High 10)
<ndufresne>
PPA: oh, btw, I don't have interlaced support implemented yet, or 10bit
<PPA>
huh
<ndufresne>
10bit is trivial of course, just need to map the formats
<ndufresne>
interlaced is more involved
<jernej>
ndufresne: Cedrus doesn't support H264 10-bit
<ndufresne>
ah ok, that will keep failing then ;-P
<ndufresne>
rkvdec does
<PPA>
Yeah, I put the 10-bit file in there because I read that it doesnÄt work and wanted something to test it
<PPA>
so that failing isn't surprising
<PPA>
How do I check whether a file is interlaced? That might explain the test file linked above failing
<jernej>
one way is to probe file with mediainfo
<jernej>
you'll get "Scan type : Interlaced"
<PPA>
Scan type : Progressive
<jernej>
so there is probably some other issue
<PPA>
I'm currently comparing that file with the ones that work. Codec ID: avc1 isn't it, another file with that works fine
<jernej>
if you have too much time, you can test it in LibreELEC and see if you get same result - that would give good indication where to check for a problem
<PPA>
It's the only one with "Format profile: High@L5.1 of the batch"
<PPA>
Sure, I could do that later this evening
<PPA>
One more thing: the files with audio aren't outputting any (when they did on that broken 5.10/1.18 test from last week)
Mangy_Dog has quit [Remote host closed the connection]
<PPA>
I downloaded another test file in High@L5.0 and it played fine
<PPA>
This broken one is High@5.1, I couldn't find another sample like that that wasn't too big
<ndufresne>
I confirmed I can produce, it's also really slow
<ndufresne>
but I think there is a problem with the network on that board, let me download for the speed
<PPA>
I was playing it off the SD card
<ndufresne>
yeah, but I got some issues on that board, same URI, 1.5MB/s on PC, and half that on the board
<PPA>
ow
<ndufresne>
but still not realtime for me, anyway, I guess the main point is to deal with the explosion
<ndufresne>
ah wait
<ndufresne>
gst-play-1.0 --use-playbin3
<ndufresne>
fixes it
<PPA>
Huh, I've been using --use-playbin3 all this time
<ndufresne>
with the old playbin (called playbin2 amongs dev) the decoder is not aware if downstream can cope with the padded buffer
<ndufresne>
I think it about time to make it the default ...
<ndufresne>
jernej: PPA: which one is the latest ffmpeg (that works on 5.11 +) ?
<PPA>
Good question. I tried building one from kwiboo's recent fork but it's giving me trouble too
<ndufresne>
I'm trying the master-stable branch, a guess
<jernej>
5.11 will be skipped by LE, so it's barely tested
<PPA>
Yeah, that's what I tried
<jernej>
I guess v4l2-request-hwaccel-4.3.1-stable should work, but I didn't work with it
mmarc__ has joined #linux-sunxi
kaspter has quit [Ping timeout: 246 seconds]
camus has joined #linux-sunxi
<PPA>
I tried v4l2-request-hwaccel-master-stable but I can't tell if it works; all my tests max out the CPU (way more than if it was just sw decoding), probably some pixel format or colorspace conversion problem
<PPA>
(same result if I don't specify -pix_fmt, or if I output to the framebuffer with fbdev /dev/fb0)
<jernej>
well, that's doing color conversion
<jernej>
which is quiet expensive
<PPA>
oh
<PPA>
Can I disable that?
<PPA>
(I should reiterate that I have no idea what I'm doing, and I'm just piecing together scattered posts on various forums referring to older versions of stuff)
<jernej>
try replacing "-pix_fmt bgra -f null /dev/null" just with "-f null -"
BenG83 has quit [Quit: Leaving]
<jernej>
and debug output should tell you what's going on
dddddd has quit [Ping timeout: 256 seconds]
mmarc__ has quit [Remote host closed the connection]
<ndufresne>
hmm, rebuild that, while the v4l2_request got build, the accelerators does not show up ...