<html>
<head>
<base href="https://bugs.freedesktop.org/" />
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - [SKL][Audio][HD-A Display]: Can't detect Display audio codec if not connect HDMI and DP monitor when boot up"
href="https://bugs.freedesktop.org/show_bug.cgi?id=89419#c17">Comment # 17</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - [SKL][Audio][HD-A Display]: Can't detect Display audio codec if not connect HDMI and DP monitor when boot up"
href="https://bugs.freedesktop.org/show_bug.cgi?id=89419">bug 89419</a>
from <span class="vcard"><a class="email" href="mailto:imre.deak@intel.com" title="Imre Deak <imre.deak@intel.com>"> <span class="fn">Imre Deak</span></a>
</span></b>
<pre>Hi,
(In reply to Libin Yang from <a href="show_bug.cgi?id=89419#c15">comment #15</a>)
<span class="quote">> Hi Imre,
>
> Based on my experience, EM4/5 impacts on the playback speed, I don't think
> it will impact on the audio codec detection. (Actually before we detect HDMI
> codec, we will never touch such registers)</span >
What I meant is that the EM4/5 registers do not exist on SKL. Adding simply the
AZX_DCAPS_I915_POWERWELL flag to AZX_DCAPS_INTEL_SKYLAKE would result in
programming these registers also on SKL, which is incorrect, so that
programming needs to be avoided.
<span class="quote">> For the power well, yes, we need set the power well before HDMI codec
> detection. And the audio driver does do it already.</span >
It didn't do it, you need to set AZX_DCAPS_I915_POWERWELL for that.
<span class="quote">> After power well on, we will do the HDMI codec detection. We won't power off
> the power well before HDMI codec detection.
>
> BTW: I have asked the BIOS team about this issue. They told me that we
> should:
> 1. update the BIOS to latest (we have already done)
> 2. update the gfx driver to latest(windows). So I believe that BIOS may have
> changed some interfaces with gfx for this issue. Maybe we missed the change?</span >
I'm not aware of any other interaction between the gfx and sound driver besides
what is done via the i915/audio component framework. For SKL in practice this
means turning on and off power well 2. I still suspect that after turning power
well 2 off then back on (this sequence actually happens based on the attached
dmesg) some of the audio HW state gets reset and is not reinitialized properly.
I would probably start enabling debugging in the sound driver and see where the
detection fails.
(In reply to Libin Yang from <a href="show_bug.cgi?id=89419#c16">comment #16</a>)
<span class="quote">> Hi Imre,
>
> I found SKL doesn't apply the powerwell flag. I will test by adding the flag
> although it shouldn't add the flag here. Will update the test result later.</span >
Not sure what you mean here. The sound driver needs to turn on/off its
dependent power well (power well 2 in case of SKL), and it does this on
platforms with the AZX_DCAPS_I915_POWERWELL flag set. So the flag needs to be
set for SKL.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are the QA Contact for the bug.</li>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>