<megi>
it's really not good if PMIC driver thinks pmic registers look like this and then does actually write the garbage into the PMIC, causing havoc :)
<megi>
libva-v4l2-requests doesn't run due to a missing symbol (easily fixable) and then some ioctls (like VIDIOC_S_EXT_CTRLS) are returning EFAULT (at least in mpv)
<fALSO>
yap
<fALSO>
libva-v4l2-requests sucks to build
<fALSO>
someone should update it to be able to build with recent kernels
<fALSO>
when you go down that rabbit hole it sucks
<megi>
it builds for me, (I think necessary kernel headers are included)
<megi>
but maybe they changed since
<fALSO>
nah, i tried it with 5.6
<fALSO>
needs stuff that doesnt exist in the original v4l2 that is in the kernel
<megi>
I already have ffmpeg cross-compiling :) now I'm just figuring out how to write a simple app that will use libavcodec and get v4l buffer references from it for further processing (rotation via rotation engine)
<fALSO>
megi, if you ever find a fork of the v4l2request that builds with non-hacked-kernels
<megi>
though I'm still not sure how DRM device would accelerate video decoding
<fALSO>
the rockpro generates different devices one for encodign and other for decoding i think
<megi>
and that branch is all about adding support for v4l devices, not drm
<fALSO>
yap, i told you - i dont know if that one works on allwinners
<fALSO>
but on rockpro64 with the hantro patches
<fALSO>
it correctly decodes h264 via HW
<fALSO>
like ... vdpau
<megi>
well, it still uses v4l for decoding :) the paths are just hardcoded in the code
<fALSO>
i think its probably because of being impossible to compile v4l2-request
<megi>
anyway, back to hacking :)
<fALSO>
=) do it!
<fALSO>
im gonna update my jenkins instance
reinforce has quit [Quit: Leaving.]
JohnDoe_71Rus has joined #linux-sunxi
zhovner has quit [Quit: bye]
yann has joined #linux-sunxi
tnovotny has quit [Quit: Leaving]
selfbg has quit [Remote host closed the connection]
zhovner has joined #linux-sunxi
bjne has joined #linux-sunxi
<jernej>
megi: I have few minutes if you want some answers regarding HW video decoding
<jernej>
if you want up to date vaapi library, check pull requests
<jernej>
ph5 worked on improving it
<jernej>
but I don't use it, so I don't know how good it works
<jernej>
fALSO: pure ffmpeg is not meant for video playback, so it's normal that video is not the right speed with the line you provided
<fALSO>
i know!
<fALSO>
thats why i wanted to use vlc
<jernej>
although mpv should work nicely with modified ffmpeg (maybe one simple patch is needed, I forgot)
<jernej>
just be aware that ffmpeg contains few modifications which didn't make it in mainline yet, so LibreELEC Cedrus patches are also needed
<jernej>
but allows to decode virtually all videos
<jernej>
while in mainline several features are missing, like interlaced H264 support, HEVC scaling matrix, HEVC multi slice videos (I think) and few more HEVC stuff
<megi>
jernej: thank you for the information
maccraft has quit [Ping timeout: 240 seconds]
reinforce has joined #linux-sunxi
OnkelUlla has quit [Remote host closed the connection]
florian has quit [Quit: Leaving]
tllim has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
vagrantc has joined #linux-sunxi
ganbold has quit [Remote host closed the connection]
ganbold has joined #linux-sunxi
<MoeIcenowy>
fALSO: v4l2request seems to build when you have newest kernel api headers
<MoeIcenowy>
and do some hacks on the tiled_yuv.S source (which is an ARMv7 only assembly)
<fALSO>
well i tried with 5.5
<MoeIcenowy>
buy I didn't get it work
<MoeIcenowy>
but *
<fALSO>
it doesnt have the *newest* kernel api headers
<fALSO>
;-)
<MoeIcenowy>
ah 5.4 works
<megi>
it builds on 5.5 for me, but errors out
<megi>
during runtime
<megi>
with vlc
ldevulder has quit [Remote host closed the connection]
<karlp>
people have been saying that for a whiel now right? :)
<megi>
I see no indications on the media mailing list of that happening
maccraft has quit [Read error: Connection reset by peer]
maccraft has joined #linux-sunxi
ldevulder has quit [Remote host closed the connection]
ldevulder has joined #linux-sunxi
<willmore>
I feel this: <megi> hmm, maybe that's my feelings debuging this issue
yann has quit [Ping timeout: 255 seconds]
<martinayotte>
megi: will you merge your sound_hdmi commits from your 5.5.y branch into 5.6.y branch soon or should I add patch into Armbian DEV ? (I've started the work of switching DEV to 5.6.y and discovered those discrepancies)
matthias_bgg has quit [Ping timeout: 260 seconds]
florian has joined #linux-sunxi
florian has quit [Ping timeout: 240 seconds]
RichardG867 has joined #linux-sunxi
florian has joined #linux-sunxi
tl_lim has joined #linux-sunxi
florian has quit [Ping timeout: 240 seconds]
yann has joined #linux-sunxi
tllim has quit [Ping timeout: 265 seconds]
tl_lim has quit [Ping timeout: 240 seconds]
<megi>
which sound_hdmi commits?
<megi>
probably not, I don't think they even work as is
<martinayotte>
megi: in fact, it is a DT patch from Jernej that you merged into your 5.5.y branch on Jan 20th, but is not present in your 5.6.y branch.
<megi>
I probably left it out because I never got to testing it, and also it's missing some pre-requisite patches I think
gsz has quit [Quit: Konversation terminated!]
<martinayotte>
Ah ! Ok ! Then, I will prepare an Armbian patch, since another one in 5.5.y was depending on it. But, me neither I've not tested it ...
yann has quit [Ping timeout: 240 seconds]
florian has joined #linux-sunxi
gaston1980 has quit [Quit: Konversation terminated!]
RichardG867 has quit [Ping timeout: 258 seconds]
ldevulder has quit [Remote host closed the connection]