[Intel-gfx] [Alsa-user] intel-hda: sound via HDMI only when using interlaced modes

[Jani Nikula]

On Thu, 10 Oct 2013, Jani Nikula <jani.nikula at linux.intel.com> wrote:
> On Thu, 10 Oct 2013, Jasper Smet <josbeir at gmail.com> wrote:
>> I managed do this quickly before i got off to work:
>>
>> Dmesg output with directly connected to the tv: http://sprunge.us/EhJD
>>
>> Neither the VSX-928 nor the TV (passtrough) pick up sound when
>> connected to the AV receiver.
>>
>> I noticed that when i set the resolution to 1080i at 30 sound works on
>> the receiver but not on the TV, everything lower works fine in both
>> passtrough and with the receiver powered on.
>>
>> Only when using 1080p24 or lower everything then works fine on both
>> devices..
>
> At a glance, the main difference seems to be:
>
>  [drm:drm_detect_monitor_audio], Monitor has basic audio support
> -[drm:drm_edid_to_eld], ELD monitor Panasonic-TV
> +[drm:drm_edid_to_eld], ELD monitor VSX-923
>  [drm:parse_hdmi_vsdb], HDMI: DVI dual 0, max TMDS clock 190, latency present 0 0, video latency 0 0, audio latency 0 0
> -[drm:drm_edid_to_eld], ELD size 9, SAD count 1
> +[drm:drm_edid_to_eld], ELD size 13, SAD count 8
>
> This is getting pretty much to unknown territory for me, so if anyone
> knows anything better, please chime in!

So my completely uneducated hunch was that the VSX-923 advertizes audio
capabilities for better quality than the TV, and together the video and
audio oversubscribe the link, unless you degrade video quality. And my
idea below is to use whatever the TV uses. I don't know if we have any
knobs to choose the audio quality somewhere, or whether the hda driver
has something for that. And then again I might be completely off here...

Jani.


>
> I think I'd try using the TV's EDID with the VSX-923 to see if the
> ELD/SAD difference makes, uh, a difference. Something like this:
>
> 1. connect directly to TV
> 2. copy /sys/class/drm/card0-HDMI-A-1/edid under /lib/firmware with some
>    sensible name, e.g. panasonic-edid. the exact source path may vary
>    depending on HDMI port etc.
> 3. make sure you have CONFIG_DRM_LOAD_EDID_FIRMWARE=y
> 4. use drm.edid_firmware=HDMI-A-1:panasonic-edid module parameter to
>    tell DRM to load the edid from /lib/firmware. again, the connector
>    name HDMI-A-1 may very depending on HDMI port you use, do check.
> 5. connect to VSX-923; this should now use the EDID (and consequently
>    ELD/SAD) from the TV
>
> I haven't actually tried this myself. *grin*.
>
> Let us know what happens. Thanks.
>
> BR,
> Jani.
>
>
>
>>
>> Hope this helps.
>>
>> On Thu, Oct 10, 2013 at 7:57 AM, Jani Nikula
>> <jani.nikula at linux.intel.com> wrote:
>>> On Wed, 09 Oct 2013, Jasper Smet <josbeir at gmail.com> wrote:
>>>> As promissed i added the params to the kernel boot and here's the
>>>> dmesg output with debugging enabled:
>>>>
>>>> http://sprunge.us/iEQR
>>>
>>> Similar dmesg for when you connect directly to the tv (and audio works)
>>> might prove useful.
>>>
>>> Is your expectation that the VSX-928 picks up the sound, or pass through
>>> to the tv? Is it possible to try both (I honestly don't know)?
>>>
>>> It might prove useful to file a bug on DRM/Intel at [1] to track
>>> this. Then it won't get lost even if this thread quiets down.
>>>
>>>
>>> Thanks,
>>> Jani.
>>>
>>>
>>> [1] https://bugs.freedesktop.org/enter_bug.cgi?product=DRI
>>>
>>>
>>>>
>>>> Hope this helps!
>>>>
>>>> On Wed, Oct 9, 2013 at 1:28 PM, Jasper Smet <josbeir at gmail.com> wrote:
>>>>> Ok will pass the info when i get home later this day.
>>>>>
>>>>> On Wed, Oct 9, 2013 at 1:01 PM, Daniel Vetter <daniel at ffwll.ch> wrote:
>>>>>> On Wed, Oct 09, 2013 at 12:44:10PM +0200, Jasper Smet wrote:
>>>>>>> Sorry,
>>>>>>>
>>>>>>> Intel NUC NUC Kit DC3217BY (Ivy bridge HD4000) Connected via HDMI to
>>>>>>> my Pioneer VSX-928 which is then connected to my Panasonic tv
>>>>>>>
>>>>>>> Running latest stable Openelec build (3.2.2). (tried different builds
>>>>>>> using stable/unstable drivers)
>>>>>>
>>>>>> That's really old ... Can you please test on something more modern? We
>>>>>> made tons of fixes to the hdmi infoframe code.
>>>>>>>
>>>>>>> Here are some logs i pulled while i was debugging the issue with one
>>>>>>> of the OpenElec developers.
>>>>>>>
>>>>>>> xbmc log (shows ALSA enumeration): http://sprunge.us/YJHc
>>>>>>> alsa playback devices: http://sprunge.us/KZKX
>>>>>>> xrandr modes: http://sprunge.us/DhFK
>>>>>>> dmesg: http://sprunge.us/XRGP
>>>>>>
>>>>>> When you have a new kernel please boot with drm.debug=0xe added to your
>>>>>> kernel cmdline and then grab the dmesg. Otherwise all the useful gfx debug
>>>>>> noise isn't in there.
>>>>>> -Daniel
>>>>>> --
>>>>>> Daniel Vetter
>>>>>> Software Engineer, Intel Corporation
>>>>>> +41 (0) 79 365 57 48 - http://blog.ffwll.ch
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Met Vriendelijke Groeten
>>>>>
>>>>> Jasper Smet
>>>>> Developer
>>>>>
>>>>> Twitter: josbeir
>>>>> E-mail: josbeir at gmail.com
>>>>> Mobile: 0486/41.75.45
>>>>
>>>>
>>>>
>>>> --
>>>> Met Vriendelijke Groeten
>>>>
>>>> Jasper Smet
>>>> Developer
>>>>
>>>> Twitter: josbeir
>>>> E-mail: josbeir at gmail.com
>>>> Mobile: 0486/41.75.45
>>>> _______________________________________________
>>>> Intel-gfx mailing list
>>>> Intel-gfx at lists.freedesktop.org
>>>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>>
>>> --
>>> Jani Nikula, Intel Open Source Technology Center
>>
>>
>>
>> -- 
>> Met Vriendelijke Groeten
>>
>> Jasper Smet
>> Developer
>>
>> Twitter: josbeir
>> E-mail: josbeir at gmail.com
>> Mobile: 0486/41.75.45
>
> -- 
> Jani Nikula, Intel Open Source Technology Center

-- 
Jani Nikula, Intel Open Source Technology Center