Login at Kodi Home

**fritsch**

Okay - this was what I asumed ... 10 years back. The intel driver changed the bit depths it uses for hdmi handshaking. https://bugs.freedesktop.org/show_bug.cgi?id=97442 - now it's also clear why forcing 8 bit fixes it - which is btw. exactly what the AVR does, when you switch it to Standard instead of Enhanced -> see the Denon Manual.

Please file a bug with the intel people and provide them your drm.debug=0xe enabled log. I think the CTS values for 24p on your setup are still not correct and that's causing this said internal audio clock drift.

https://github.com/LibreELEC/LibreELEC.t...6ed5e0a769 <- Yeah, I did not track that stuff in recent years, so have no idea what was upstreamed. I think I would go with the "Standard setting" for the Denon for now ... you don't really loose anything as there is no 10 bit HDR output for intel on linux yet.

**jawilljr** · 2023-03-03, 06:13

(2023-03-02, 23:29)fritsch Wrote: Okay - this was what I asumed ... 10 years back. The intel driver changed the bit depths it uses for hdmi handshaking. https://bugs.freedesktop.org/show_bug.cgi?id=97442 - now it's also clear why forcing 8 bit fixes it - which is btw. exactly what the AVR does, when you switch it to Standard instead of Enhanced -> see the Denon Manual.

Please file a bug with the intel people and provide them your drm.debug=0xe enabled log. I think the CTS values for 24p on your setup are still not correct and that's causing this said internal audio clock drift.

https://github.com/LibreELEC/LibreELEC.t...6ed5e0a769 <- Yeah, I did not track that stuff in recent years, so have no idea what was upstreamed. I think I would go with the "Standard setting" for the Denon for now ... you don't really loose anything as there is no 10 bit HDR output for intel on linux yet.

Does that apply with with these drivers?

Those drivers is what a person needs for the latest Intel GPU's.

Will those work with LibreELEC?

Flachzange · 2023-03-03, 07:58

(2023-03-02, 23:29)fritsch Wrote: Please file a bug with the intel people and provide them your drm.debug=0xe enabled log. I think the CTS values for 24p on your setup are still not correct and that's causing this said internal audio clock drift.

Stupid question: Where do I find those Intel People? Smile

(2023-03-03, 06:13)jawilljr Wrote: Those drivers is what a person needs for the latest Intel GPU's.

My GPU is like 5 years old. It is nothing limited to newer GPUs. And I summe I already tested the latest driver by having that Intel kernel tested.

alfton · 2023-03-03, 08:01

i just need to trow inn on this to, i think this apply on the new gpu to, I have the gen12 intel gpu. And my atmos drop problem did go away with the change to standard on my denon

**fritsch** · 2023-03-03, 08:10

(2023-03-03, 07:58)Flachzange Wrote:
(2023-03-02, 23:29)fritsch Wrote: Please file a bug with the intel people and provide them your drm.debug=0xe enabled log. I think the CTS values for 24p on your setup are still not correct and that's causing this said internal audio clock drift.

Stupid question: Where do I find those Intel People?

(2023-03-03, 06:13)jawilljr Wrote: Those drivers is what a person needs for the latest Intel GPU's.

My GPU is like 5 years old. It is nothing limited to newer GPUs. And I summe I already tested the latest driver by having that Intel kernel tested.

On kernel's bugzilla. Make the LE test first and then kindly prepare your bugreport. People over there won't react at all if something is missing, therefore -> single shot needs to hit.
https://www.kernel.org/doc/html/v4.15/ad...-bugs.html
Reporting: https://bugzilla.kernel.org/

I would propose the following route:

Install the kernel again that I linked you some days ago - it's a quite new one.
Add drm.debug=0xe to the kernel command, boot it with Standard, then boot it with Extended. Always save the dmesg output into a file.
Then go there, describe your problem and say that it seems to work when 8 bpc is used, provide the two logs and ask if audio registers / clocks are differently programmed depending on 8 and 12 bpc if it is e.g. a bandwidth issue. Your issue happens with 192 khz 8 channel 16 bit output of digital formats.

Good luck.

Flachzange · 2023-03-03, 08:19

(2023-03-03, 08:01)alfton Wrote: i just need to trow inn on this to, i think this apply on the new gpu to, I have the gen12 intel gpu.

Yes and in this case, it is actually "good", because we can rule out hardware issues.

(2023-03-03, 08:10)fritsch Wrote: Make the LE test first

What do you mean by that?

And thanks again. Will spend some time at weekend on the bug report prep just to have it ignored then by kernel people Smile

**fritsch** · 2023-03-03, 13:43

What I mean: Try LibreELEC from a USB stick and check if issue is the same. If not -> the different kernel version / patches might give a more detailed hint.

Flachzange · 2023-03-04, 08:41

Ah ok. I did this already and issue is the same. See here

Flachzange · 2023-03-05, 16:07

Hey @fritsch ,

I am currently preparing the bug report. Just to be on the safe side: I should file it to the Kernel's Bugzilla and not as described here https://gitlab.freedesktop.org/drm/intel...-i915-bugs ?

Reason I am asking is that I struggle a little when it comes to the components to be selected in the Bugzilla:

FYI dmesg output

Denon "Standard": https://paste.kodi.tv/emajiceved
Denon "Extended": https://paste.kodi.tv/asopanuvaz

Not too many differences but this is the most suspicious:

Standard:

Quote:[    4.962917] kernel: i915 0000:00:02.0: [drm:drm_parse_cea_ext [drm]] [CONNECTOR:144:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 300000 kHz
[    4.963011] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does deep color 30.
[    4.963089] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does deep color 36.
[    4.963163] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] Assigning HDMI sink color depth as 12 bpc.
[    4.963237] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does YCRCB444 in deep color.
[    4.963314] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_forum_scds [drm]] [CONNECTOR:144:HDMI-A-1] HF-VSDB: max TMDS clock: 0 KHz, HDMI 2.1 support: no, DSC 1.2 support: no

Extended:

Quote:[ 5.025345] kernel: i915 0000:00:02.0: [drm:drm_parse_cea_ext [drm]] [CONNECTOR:144:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 300000 kHz
[ 5.025438] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does deep color 30.
[ 5.025515] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does deep color 36.
[ 5.025589] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] Assigning HDMI sink color depth as 12 bpc.
[ 5.025663] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_deep_color_info [drm]] [CONNECTOR:144:HDMI-A-1] HDMI sink does YCRCB444 in deep color.
[ 5.025737] kernel: i915 0000:00:02.0: [drm:drm_parse_hdmi_forum_scds [drm]] [CONNECTOR:144:HDMI-A-1] HF-VSDB: max TMDS clock: 600000 KHz, HDMI 2.1 support: no, DSC 1.2 support: no

Interesting to me that it even recognises YCRCB444 and 12 Bit in Standard mode.

**fritsch** · 2023-03-05, 17:24

Your link is correct. Do what Jani suggests. The Ubuntu PPA also has DRM tip kernels to test.

Flachzange · 2023-03-07, 19:40

Unfortunately, no binary packages are available in the Ubuntut PPA (at least I have not found them). I built the kernel then myself. Interestingly, the drm-tip kernel has much more debug output then the standard kernel.

Issue with the Intel people has been raised: https://gitlab.freedesktop.org/drm/intel/-/issues/8276