[PATCH v7] drm/msm/dp: reuse generic HDMI codec implementation

[Dmitry Baryshkov]

On Tue, Jun 03, 2025 at 10:16:14PM +0800, Xilin Wu wrote:
> On 2025/6/3 22:06:36, Dmitry Baryshkov wrote:
> > On Thu, May 29, 2025 at 10:40:12AM +0800, Xilin Wu wrote:
> > > On 2025/4/24 01:52:45, Dmitry Baryshkov wrote:
> > > > From: Dmitry Baryshkov <lumag at kernel.org>
> > > > 
> > > > The MSM DisplayPort driver implements several HDMI codec functions
> > > > in the driver, e.g. it manually manages HDMI codec device registration,
> > > > returning ELD and plugged_cb support. In order to reduce code
> > > > duplication reuse drm_hdmi_audio_* helpers and drm_bridge_connector
> > > > integration.
> > > > 
> > > > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at linaro.org>
> > > > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at oss.qualcomm.com>
> > > > ---
> > > > A lot of DisplayPort bridges use HDMI Codec in order to provide audio
> > > > support. Present DRM HDMI Audio support has been written with the HDMI
> > > > and in particular DRM HDMI Connector framework support, however those
> > > > audio helpers can be easily reused for DisplayPort drivers too.
> > > > 
> > > > Patches by Hermes Wu that targeted implementing HDMI Audio support in
> > > > the iTE IT6506 driver pointed out the necessity of allowing one to use
> > > > generic audio helpers for DisplayPort drivers, as otherwise each driver
> > > > has to manually (and correctly) implement the get_eld() and plugged_cb
> > > > support.
> > > > 
> > > > Implement necessary integration in drm_bridge_connector and provide an
> > > > example implementation in the msm/dp driver.
> > > > ---
> > > > Changes in v7:
> > > > - Dropped applied patches
> > > > - Link to v6: https://lore.kernel.org/r/20250314-dp-hdmi-audio-v6-0-dbd228fa73d7@oss.qualcomm.com
> > > > 
> > > > Changes in v6:
> > > > - Added DRM_BRIDGE_OP_DP_AUDIO and separate set of DisplayPort audio
> > > >     callbacks to the drm_bridge interface (Maxime)
> > > > - Link to v5: https://lore.kernel.org/r/20250307-dp-hdmi-audio-v5-0-f3be215fdb78@linaro.org
> > > > 
> > > > Changes in v5:
> > > > - Rebased on top of linux-next, also handling HDMI audio piece of the
> > > >     MSM HDMI driver.
> > > > - Link to v4: https://lore.kernel.org/r/20250301-dp-hdmi-audio-v4-0-82739daf28cc@linaro.org
> > > > 
> > > > Changes in v4:
> > > > - Rebased on linux-next, adding DRM_BRIDGE_OP_HDMI_AUDIO to Synopsys QP
> > > >     HDMI driver.
> > > > - Drop outdated comment regarding subconnector from the commit message.
> > > > - Link to v3: https://lore.kernel.org/r/20250219-dp-hdmi-audio-v3-0-42900f034b40@linaro.org
> > > > 
> > > > Changes in v3:
> > > > - Dropped DRM_BRIDGE_OP_DisplayPort, added DRM_BRIDGE_OP_HDMI_AUDIO
> > > >     (Laurent, Maxime)
> > > > - Dropped the subconnector patch (again)
> > > > - Link to v2: https://lore.kernel.org/r/20250209-dp-hdmi-audio-v2-0-16db6ebf22ff@linaro.org
> > > > 
> > > > Changes in v2:
> > > > - Added drm_connector_attach_dp_subconnector_property() patches
> > > > - Link to v1: https://lore.kernel.org/r/20250206-dp-hdmi-audio-v1-0-8aa14a8c0d4d@linaro.org
> > > > ---
> > > >    drivers/gpu/drm/msm/Kconfig         |   1 +
> > > >    drivers/gpu/drm/msm/dp/dp_audio.c   | 131 ++++--------------------------------
> > > >    drivers/gpu/drm/msm/dp/dp_audio.h   |  27 ++------
> > > >    drivers/gpu/drm/msm/dp/dp_display.c |  28 ++------
> > > >    drivers/gpu/drm/msm/dp/dp_display.h |   6 --
> > > >    drivers/gpu/drm/msm/dp/dp_drm.c     |   8 +++
> > > >    6 files changed, 31 insertions(+), 170 deletions(-)
> > > > 
> > > 
> > > This change breaks DP audio on the qcs6490 platform, tested on kernel
> > > next-20250528.
> > 
> > I can not confirm this issue here (though I tested it on a different
> > hardware). Do you have any patches on top of linux-next?
> > 
> 
> I have this patch series applied, but I don't think it could be relevant:
> 
> [PATCH v4 0/8] Enable audio on qcs6490-RB3Gen2 and qcm6490-idp boards
> https://lore.kernel.org/all/20250527111227.2318021-1-quic_pkumpatl@quicinc.com/
> 
> > > 
> > > [    0.368035] [drm:dpu_kms_hw_init:1173] dpu hardware revision:0x70020000
> > > [    0.369359] hdmi-audio-codec hdmi-audio-codec.0.auto: hdmi_codec_probe:
> > > dai_count 0
> > > [    0.369362] hdmi-audio-codec hdmi-audio-codec.0.auto: hdmi_codec_probe:
> > > Missing hw_params
> > > [    0.369364] hdmi-audio-codec hdmi-audio-codec.0.auto: hdmi_codec_probe:
> > > Invalid parameters
> > > [    0.369366] hdmi-audio-codec hdmi-audio-codec.0.auto: probe with driver
> > > hdmi-audio-codec failed with error -22
> > > [    0.370536] [drm] Initialized msm 1.12.0 for ae01000.display-controller
> > > on minor 0
> > > 
> > > Manually reverting this change solves the problem.
> > 
> > It is suspicious, since dai_count can not be 0. We set
> > hdmi_audio_max_i2s_playback_channels to 8, which in turn should set the
> > hdmi_codec_pdata.i2s to 1.
> > 
> 
> It suddenly comes to my mind that I'm using a kernel with everything
> compiled as builtin. Could that be a possible issue?

What kernel args are you using? Do you have any kernel debug options
enabled in the .config? I've tested the kernel on RB3 Gen2 and I still
can not confirm the issue (I'm also using an all-in kernel)

I've verified that on a running system I'm getting three HDMI audio
codecs (one from LT9611UXC and two from DP controllers). Each of them
binds immediately to the driver with no issues observed.

-- 
With best wishes
Dmitry