[Intel-gfx] [PATCH v5] drm/i915: pass ELD to HDMI/DP audio driver

[Wu Fengguang]

Christopher,

On Thu, Nov 10, 2011 at 12:10:42PM +0800, Christopher White wrote:
> On 11/10/11 4:27 AM, Wu Fengguang wrote:
> > Christopher,
> >
> > Did you enabled CONFIG_SND_DEBUG and CONFIG_SND_DEBUG_VERBOSE in
> > kconfig? I've been looking for the error messages related to zeroed
> > ELD but never managed to find any in your dmesg.
> No, I was unaware of these options, this is the first time you've 
> mentioned them. ;-)

Yeah, never mind :-)

> However, I am unable to reproduce the 0-byte EDID again. Even after 
> multiple reboots, turning the projector on/off, etc. Hmm. We'll have to 
> forget that issue since it's not reproducible anymore. It's always odd 
> when code does one thing one time and another thing every other time, 
> but hopefully it was just some weird fluke, race condition, 
> who-knows-what...
> 
> Heck it could even be the TX-SR607 receiver that did something odd this 
> one time.

OK. We still have the time :-) Enable CONFIG_SND_DEBUG and use it for
months. If it occurs again, we at least have more error messages.
Otherwise it at least demonstrates itself to be some less disturbing
bug :-)

> I looked at the latest dmesg after yet another reboot and see only a 
> single "[drm:intel_hdmi_mode_set], Enabling HDMI audio on pipe A" event, 
> and no misreads. I'll attach that file too just for completeness.
> 
> Anyway, we know that it reads ELD correctly and writes it to the audio 
> register correctly, as we see from the logs. However, my previous 
> message ended in a theory that the problem is that the ELD is written to 
> the register for port 1 but my device is on port 2. Re-read the bottom 
> of the message I sent previously, and it'll detail the idea. It's a 
> possibility worth checking out.

The theory still cannot explain it well.. see the direct reply to the
email. For now let's treat it as some hardware mysteriousness...

Thanks,
Fengguang