• 1
  • 149
  • 150
  • 151(current)
  • 152
  • 153
  • 342
Intel VAAPI howto with Leia v18 nightly based on Ubuntu 18.04 server
-> bugs.freedesktop.org it's a radeon kernel issue.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
(2015-11-29, 21:08)fritsch Wrote: -> bugs.freedesktop.org it's a radeon kernel issue.

Thank you Fritsch but why radeon when I don't have any gpu besides the igpu on the G1610?

Best
Frank
Reply
That's a very, very good question - see the callstack - radeon is involved? No idea how.

Edit: Looking again. It happens very, very early on bootup. OE builds radeon and i915 fixed into the kernel - so it seems they fight against each other. A bug report is good.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
Thank you for your reply.

I rebooted the device with the addon disabled and started a 24p MKV movie.

For some kind of reason kodi was reproducibly not starting the movie at a 24 Hz refresh but at a 50 Hz refresh. I manually ran xrandr once via ssh, which apparently triggered the detection of the other supported modes (http://paste.ubuntu.com/13562377/) with kodi already running. After that kodi changed to 24 Hz with no problems.

For this movie, the error did not occur - sync was fine until the end. I will try to obtain a log next time the sync problem is happing - the log is still attached for the mode change issue.

The computer running kodi is running 24/7. I wonder, if the sync errors might happen due to the TV being switched off and on repeatedly in between watching videos?

kodi log:
http://paste.ubuntu.com/13565156/
Reply
Yes - both possible. For the non existing modes, please see the second post in this thread - it's a driver issue fixed with the driver provided there.

Whenever you turn off the TV the output stops - if that's your usecase consider overwriting the EDID information to force the display on.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
@fsonneli:

Quote:22:04 < airlied> fritsch: two gpus, and modules are probed in parallel
22:04 < fritsch> airlied: there is only one gpu in that box - but i think as
this distro builds radeon and i915 not as modules
22:04 < airlied> ah yes if they are built-in that will happen
22:04 < fritsch> but directly into the kernel - this gets an issue
22:07 < fritsch> airlied: so it can be ignored?
22:08 < airlied> fritsch: yes

airlied is the DRM maintainer - so all fine :-)
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
(2015-11-29, 20:19)fritsch Wrote: If someone is toying with kernel 4.4-rc2 or later, I reworked the Video passthrough patch: http://sprunge.us/eNAW

Kernel: http://fritsch.fruehberger.net/kernel/?C=M;O=D - this should also fix the issue with "switching" to the mode
Thanks for new patch - i was applying modifed 4.3 patch but i see that yours is little bit different (but mine also works Smile

Your kernel has other OE patch or only passthrough patch as previously?
Reply
Yours won't work for sure on 4.4-rc2 ... as you easily see it won't even compile as fields have gone.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
That's why i've applied MODIFIED patch ;-) (modified by myself to be applicable to 4.4 kernel - i learn quickly Wink.
Anyway yours is probably better so right now i compile 4.4-rc2 kernel with passthrough and all other OE patches i've applied to 4.3 kernel (all of them can be applied to 4.4 without changes).
Reply
(2015-11-29, 23:08)fritsch Wrote: @fsonneli:

Quote:22:04 < airlied> fritsch: two gpus, and modules are probed in parallel
22:04 < fritsch> airlied: there is only one gpu in that box - but i think as
this distro builds radeon and i915 not as modules
22:04 < airlied> ah yes if they are built-in that will happen
22:04 < fritsch> but directly into the kernel - this gets an issue
22:07 < fritsch> airlied: so it can be ignored?
22:08 < airlied> fritsch: yes

airlied is the DRM maintainer - so all fine :-)

Thanks a lot!!! Final question: why it doesn't happen in my j1900 nor my 1037u boards?
Anyway thanks again.
Reply
Race - special timing :-)
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
http://lists.freedesktop.org/archives/in...81516.html <- now let's press thumbs
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
So far so good - only variables names to change Wink

There is a hope for inclusion in 4.4 kernel? I hope so.
Reply
@fritsch

Strange: with you previous patch applied by me to 4.4 kernel everything was as expected (as it was with 4.3 kernel) - with my TV (limited range without manual switch) i had good colours with Limited Range set in Kodi (and also 8-bit dithering). On calibration patterns blacks were ok (everything below 16 was black) and whites were "over range" (blinking bars up to 253, and 254 was 100% white).

With you new patch for 4.4 kernel applied to 4.4-rc3 it's first time i get colours how they should be "by the book" for limited range. On calibration patterns everything below 16 is black and for whites everything over 235 is 100% white. But it's with Full Range set in Kodi, with Limited Range all colours are totally washed-out.
Reply
I don't understand your second sentence ...
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
  • 1
  • 149
  • 150
  • 151(current)
  • 152
  • 153
  • 342

Logout Mark Read Team Forum Stats Members Help
Intel VAAPI howto with Leia v18 nightly based on Ubuntu 18.04 server18