On Tue, Dec 31, 2024 at 01:43:11AM +0100, Martin Blumenstingl wrote:
> Hello Dmitry,
>
> this is great work - thanks for your efforts!
>
> To give some context:
> I am working on a HDMI controller driver for the Amlogic Meson8/8b/8m2
> SoCs. Unfortunately the driver is not mature enough for upstream
> submission (all I have is the vendor driver which serves as reference).
> That said, my goal is to upstream the driver at some point. I have
> already switched my driver to use hdmi_{clear,write}_infoframe. Now
> I'm trying this series to simplify my code even further - by using
> your audio helper work!
>
> > --- /dev/null
> > +++ b/drivers/gpu/drm/display/drm_hdmi_audio_helper.c
> > @@ -0,0 +1,190 @@
> > [...]
> > +static const struct hdmi_codec_ops drm_connector_hdmi_audio_ops = {
> > + .audio_startup = drm_connector_hdmi_audio_startup,
> > + .prepare = drm_connector_hdmi_audio_prepare,
> > + .audio_shutdown = drm_connector_hdmi_audio_shutdown,
> > + .mute_stream = drm_connector_hdmi_audio_mute_stream,
> > + .get_eld = drm_connector_hdmi_audio_get_eld,
> > + .get_dai_id = drm_connector_hdmi_audio_get_dai_id,
> > + .hook_plugged_cb = drm_connector_hdmi_audio_hook_plugged_cb,
> > +};
> On my platform drm_connector_hdmi_audio_prepare() is never called. As
> a result of that the audio infoframe is never written to my HDMI
> controller hardware (hdmi_write_infoframe() is never called with type
> HDMI_INFOFRAME_TYPE_AUDIO). My hack to make it work is to add the
> following line to drm_connector_hdmi_audio_ops:
> .hw_params = drm_connector_hdmi_audio_prepare,
>
> I checked all instances of struct hdmi_codec_ops in v6.13-rc3 and it
> seems that there is only a single driver which uses the .prepare
> callback (drivers/gpu/drm/vc4/vc4_hdmi.c). All other drivers seem to
> implement .hw_params instead.
Yes. However .hw_params don't have access to the infoframe contents, so
I had to settle on implementing .prepare.
>
> The audio controller code for my platform is already upstream:
> - sound/soc/meson/aiu-codec-ctrl.c
> - sound/soc/meson/aiu-encoder-i2s.c
> - sound/soc/meson/aiu-fifo-i2s.c
>
> My understanding is that you have a platform with a lontium-lt9611
> HDMI controller available for testing. Can you please help me
> investigate and find out which piece of code is calling
> hdmi_codec_prepare() and therefore lt9611_hdmi_audio_prepare() on
> your board?
Sure, this is the call trace on my platform (Qualcomm SDM845,
sdm845-db845c.dts):
lt9611_hdmi_audio_prepare+0x1c/0xc0 (P)
drm_bridge_connector_audio_prepare+0x2c/0x40 (L)
drm_bridge_connector_audio_prepare+0x2c/0x40
drm_connector_hdmi_audio_prepare+0x24/0x30
hdmi_codec_prepare+0xec/0x144
snd_soc_pcm_dai_prepare+0x78/0x10c
__soc_pcm_prepare+0x54/0x190
dpcm_be_dai_prepare+0x120/0x1d0
dpcm_fe_dai_prepare+0x9c/0x2a0
snd_pcm_do_prepare+0x30/0x50
snd_pcm_action_single+0x48/0xa4
snd_pcm_action_nonatomic+0xa0/0xa8
snd_pcm_prepare+0x90/0xec
snd_pcm_common_ioctl+0xd94/0x1a24
snd_pcm_ioctl+0x30/0x48
__arm64_sys_ioctl+0xb4/0xec
invoke_syscall+0x48/0x110
el0_svc_common.constprop.0+0x40/0xe0
do_el0_svc+0x1c/0x28
el0_svc+0x48/0x110
el0t_64_sync_handler+0x10c/0x138
el0t_64_sync+0x198/0x19c
Hope, this helps.
--
With best wishes
Dmitry
