[PATCH v2 RESEND] drm/bridge: dw-hdmi: Restore audio when setting a mode

[PATCH v2 RESEND] drm/bridge: dw-hdmi: Restore audio when setting a mode

[Cheng-Yi Chiang]

From: Daniel Kurtz <djkurtz@chromium.org>

When setting a new display mode, dw_hdmi_setup() calls
dw_hdmi_enable_video_path(), which disables all hdmi clocks, including
the audio clock.

We should only (re-)enable the audio clock if audio was already enabled
when setting the new mode.

Without this patch, on RK3288, there will be HDMI audio on some monitors
if i2s was played to headphone when the monitor was plugged.
ACER H277HU and ASUS PB278 are two of the monitors showing this issue.

Signed-off-by: Cheng-Yi Chiang <cychiang@chromium.org>
Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
---
 Change from v1 to v2:
  - Use audio_lock to protect audio clock.
  - Fix the patch title.

 drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
index aa7efd4da1c8..749d8e4c535b 100644
--- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
+++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
@@ -1982,6 +1982,17 @@ static void hdmi_disable_overflow_interrupts(struct dw_hdmi *hdmi)
 		    HDMI_IH_MUTE_FC_STAT2);
 }
 
+static void dw_hdmi_audio_restore(struct dw_hdmi *hdmi)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&hdmi->audio_lock, flags);
+
+	hdmi_enable_audio_clk(hdmi, hdmi->audio_enable);
+
+	spin_unlock_irqrestore(&hdmi->audio_lock, flags);
+}
+
 static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
 {
 	int ret;
@@ -2045,7 +2056,7 @@ static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
 
 		/* HDMI Initialization Step E - Configure audio */
 		hdmi_clk_regenerator_update_pixel_clock(hdmi);
-		hdmi_enable_audio_clk(hdmi, true);
+		dw_hdmi_audio_restore(hdmi);
 	}
 
 	/* not for DVI mode */
-- 
2.23.0.444.g18eeb5a265-goog

Re: [PATCH v2 RESEND] drm/bridge: dw-hdmi: Restore audio when setting a mode

[Neil Armstrong]

Hi,

On 03/10/2019 06:14, Cheng-Yi Chiang wrote:
> From: Daniel Kurtz <djkurtz@chromium.org>
> 
> When setting a new display mode, dw_hdmi_setup() calls
> dw_hdmi_enable_video_path(), which disables all hdmi clocks, including
> the audio clock.
> 
> We should only (re-)enable the audio clock if audio was already enabled
> when setting the new mode.
> 
> Without this patch, on RK3288, there will be HDMI audio on some monitors
> if i2s was played to headphone when the monitor was plugged.
> ACER H277HU and ASUS PB278 are two of the monitors showing this issue.
> 
> Signed-off-by: Cheng-Yi Chiang <cychiang@chromium.org>
> Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
> Signed-off-by: Yakir Yang <ykk@rock-chips.com>
> ---
>  Change from v1 to v2:
>   - Use audio_lock to protect audio clock.
>   - Fix the patch title.
> 
>  drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 13 ++++++++++++-
>  1 file changed, 12 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> index aa7efd4da1c8..749d8e4c535b 100644
> --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> @@ -1982,6 +1982,17 @@ static void hdmi_disable_overflow_interrupts(struct dw_hdmi *hdmi)
>  		    HDMI_IH_MUTE_FC_STAT2);
>  }
>  
> +static void dw_hdmi_audio_restore(struct dw_hdmi *hdmi)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&hdmi->audio_lock, flags);
> +
> +	hdmi_enable_audio_clk(hdmi, hdmi->audio_enable);
> +
> +	spin_unlock_irqrestore(&hdmi->audio_lock, flags);

Dumb question, why is this protected by a spinlock ?

Neil

> +}
> +
>  static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
>  {
>  	int ret;
> @@ -2045,7 +2056,7 @@ static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
>  
>  		/* HDMI Initialization Step E - Configure audio */
>  		hdmi_clk_regenerator_update_pixel_clock(hdmi);
> -		hdmi_enable_audio_clk(hdmi, true);
> +		dw_hdmi_audio_restore(hdmi);
>  	}
>  
>  	/* not for DVI mode */
> 

Re: [PATCH v2 RESEND] drm/bridge: dw-hdmi: Restore audio when setting a mode

[Cheng-yi Chiang]

On Mon, Oct 7, 2019 at 7:57 PM Neil Armstrong <narmstrong@baylibre.com> wrote:
>
> Hi,
>
> On 03/10/2019 06:14, Cheng-Yi Chiang wrote:
> > From: Daniel Kurtz <djkurtz@chromium.org>
> >
> > When setting a new display mode, dw_hdmi_setup() calls
> > dw_hdmi_enable_video_path(), which disables all hdmi clocks, including
> > the audio clock.
> >
> > We should only (re-)enable the audio clock if audio was already enabled
> > when setting the new mode.
> >
> > Without this patch, on RK3288, there will be HDMI audio on some monitors
> > if i2s was played to headphone when the monitor was plugged.
> > ACER H277HU and ASUS PB278 are two of the monitors showing this issue.
> >
> > Signed-off-by: Cheng-Yi Chiang <cychiang@chromium.org>
> > Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
> > Signed-off-by: Yakir Yang <ykk@rock-chips.com>
> > ---
> >  Change from v1 to v2:
> >   - Use audio_lock to protect audio clock.
> >   - Fix the patch title.
> >
> >  drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > index aa7efd4da1c8..749d8e4c535b 100644
> > --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > @@ -1982,6 +1982,17 @@ static void hdmi_disable_overflow_interrupts(struct dw_hdmi *hdmi)
> >                   HDMI_IH_MUTE_FC_STAT2);
> >  }
> >
> > +static void dw_hdmi_audio_restore(struct dw_hdmi *hdmi)
> > +{
> > +     unsigned long flags;
> > +
> > +     spin_lock_irqsave(&hdmi->audio_lock, flags);
> > +
> > +     hdmi_enable_audio_clk(hdmi, hdmi->audio_enable);
> > +
> > +     spin_unlock_irqrestore(&hdmi->audio_lock, flags);
>
> Dumb question, why is this protected by a spinlock ?
>
> Neil
>

Hi Neil,
Thanks for the review.
Good catch. I found that the spinlock audio_lock was introduced in

b90120a96608 drm: bridge/dw_hdmi: introduce interfaces to enable and
disable audio

to protect N/CTS value.
Actually it was not used to protect audio clock.
So we don't need this spinlock.

Hi Daniel Kurtz,
If this rings any bell in your memory as for why this is protected,
please let me know.
I would like to remove this spinlock and simplify this patch in v3.

Thanks!


> > +}
> > +
> >  static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
> >  {
> >       int ret;
> > @@ -2045,7 +2056,7 @@ static int dw_hdmi_setup(struct dw_hdmi *hdmi, struct drm_display_mode *mode)
> >
> >               /* HDMI Initialization Step E - Configure audio */
> >               hdmi_clk_regenerator_update_pixel_clock(hdmi);
> > -             hdmi_enable_audio_clk(hdmi, true);
> > +             dw_hdmi_audio_restore(hdmi);
> >       }
> >
> >       /* not for DVI mode */
> >
>