On Fri, Jul 12, 2019 at 6:48 PM Russell King - ARM Linux admin
<li...@armlinux.org.uk> wrote:
>
> On Fri, Jul 12, 2019 at 06:04:40PM +0800, Cheng-Yi Chiang wrote:
> > Allow codec driver register callback function for plug event.
> >
> > The callback registration flow:
> > dw-hdmi <--- hw-hdmi-i2s-audio <--- hdmi-codec
> >
> > dw-hdmi-i2s-audio implements hook_plugged_cb op
> > so codec driver can register the callback.
> >
> > dw-hdmi implements set_plugged_cb op so platform device can register the
> > callback.
> >
> > When connector plug/unplug event happens, report this event using the
> > callback.
> >
> > Make sure that audio and drm are using the single source of truth for
> > connector status.
> >
> > Signed-off-by: Cheng-Yi Chiang <cychi...@chromium.org>
> > ---
> > .../gpu/drm/bridge/synopsys/dw-hdmi-audio.h | 3 +
> > .../drm/bridge/synopsys/dw-hdmi-i2s-audio.c | 10 ++++
> > drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 55 ++++++++++++++++++-
> > 3 files changed, 67 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-audio.h
> > b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-audio.h
> > index 63b5756f463b..f523c590984e 100644
> > --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-audio.h
> > +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-audio.h
> > @@ -2,6 +2,8 @@
> > #ifndef DW_HDMI_AUDIO_H
> > #define DW_HDMI_AUDIO_H
> >
> > +#include <sound/hdmi-codec.h>
> > +
> > struct dw_hdmi;
> >
> > struct dw_hdmi_audio_data {
> > @@ -17,6 +19,7 @@ struct dw_hdmi_i2s_audio_data {
> >
> > void (*write)(struct dw_hdmi *hdmi, u8 val, int offset);
> > u8 (*read)(struct dw_hdmi *hdmi, int offset);
> > + int (*set_plugged_cb)(struct dw_hdmi *hdmi, hdmi_codec_plugged_cb fn);
> > };
> >
> > #endif
> > diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-i2s-audio.c
> > b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-i2s-audio.c
> > index 5cbb71a866d5..7b93cf05c985 100644
> > --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-i2s-audio.c
> > +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-i2s-audio.c
> > @@ -104,10 +104,20 @@ static int dw_hdmi_i2s_get_dai_id(struct
> > snd_soc_component *component,
> > return -EINVAL;
> > }
> >
> > +static int dw_hdmi_i2s_hook_plugged_cb(struct device *dev, void *data,
> > + hdmi_codec_plugged_cb fn)
> > +{
> > + struct dw_hdmi_i2s_audio_data *audio = data;
> > + struct dw_hdmi *hdmi = audio->hdmi;
> > +
> > + return audio->set_plugged_cb(hdmi, fn);
> > +}
> > +
> > static struct hdmi_codec_ops dw_hdmi_i2s_ops = {
> > .hw_params = dw_hdmi_i2s_hw_params,
> > .audio_shutdown = dw_hdmi_i2s_audio_shutdown,
> > .get_dai_id = dw_hdmi_i2s_get_dai_id,
> > + .hook_plugged_cb = dw_hdmi_i2s_hook_plugged_cb,
> > };
> >
> > static int snd_dw_hdmi_probe(struct platform_device *pdev)
> > diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > index 045b1b13fd0e..ce6646067472 100644
> > --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> > @@ -26,6 +26,8 @@
> > #include <drm/drm_probe_helper.h>
> > #include <drm/bridge/dw_hdmi.h>
> >
> > +#include <sound/hdmi-codec.h>
> > +
> > #include <uapi/linux/media-bus-format.h>
> > #include <uapi/linux/videodev2.h>
> >
> > @@ -185,6 +187,9 @@ struct dw_hdmi {
> > void (*disable_audio)(struct dw_hdmi *hdmi);
> >
> > struct cec_notifier *cec_notifier;
> > +
> > + hdmi_codec_plugged_cb plugged_cb;
> > + enum drm_connector_status last_connector_result;
> > };
> >
> > #define HDMI_IH_PHY_STAT0_RX_SENSE \
> > @@ -209,6 +214,40 @@ static inline u8 hdmi_readb(struct dw_hdmi *hdmi, int
> > offset)
> > return val;
> > }
> >
> > +static void handle_plugged_change(struct dw_hdmi *hdmi, bool plugged)
> > +{
> > + struct platform_device *codec_pdev;
> > +
> > + if (!hdmi->audio || IS_ERR(hdmi->audio))
> > + return;
> > + codec_pdev = platform_get_drvdata(hdmi->audio);
> > + if (!codec_pdev || IS_ERR(codec_pdev))
> > + return;
>
> This looks fragile to me, poking about in another device's driver data
> from another driver is really not a good design decision. I think this
> can be simplified if the registration function took the function
> pointer and the struct device pointer, and then you only need one test
> below:
>
Hi Russell, Thank you for the detailed review.
ACK to this suggestion.
I have updated the registration function following your suggestion in v4.
It looks much cleaner.
> > + if (!hdmi->plugged_cb)
> > + return;
> > +
> > + hdmi->plugged_cb(&codec_pdev->dev, plugged);
> > +}
> > +
> > +static int hdmi_set_plugged_cb(struct dw_hdmi *hdmi, hdmi_codec_plugged_cb
> > fn)
> > +{
> > + bool plugged;
> > + struct platform_device *codec_pdev;
> > +
> > + if (!hdmi->audio || IS_ERR(hdmi->audio))
> > + return -EINVAL;
>
> Given the current code structure, how can this ever be true when the
> function is called?
>
ACK
Removed in v4.
> > + codec_pdev = platform_get_drvdata(hdmi->audio);
> > + if (!codec_pdev || IS_ERR(codec_pdev))
> > + return -EINVAL;
>
> This doesn't seem like a good idea as I've pointed out above.
>
ACK
Fixed in v4.
> > +
> > + mutex_lock(&hdmi->mutex);
> > + hdmi->plugged_cb = fn;
> > + plugged = hdmi->last_connector_result == connector_status_connected;
> > + handle_plugged_change(hdmi, plugged);
> > + mutex_unlock(&hdmi->mutex);
>
> Should be a blank line here for readability.
>
ACK
Fixed in v4.
> > + return 0;
> > +}
> > +
> > static void hdmi_modb(struct dw_hdmi *hdmi, u8 data, u8 mask, unsigned reg)
> > {
> > regmap_update_bits(hdmi->regm, reg << hdmi->reg_shift, mask, data);
> > @@ -2044,6 +2083,7 @@ dw_hdmi_connector_detect(struct drm_connector
> > *connector, bool force)
> > {
> > struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi,
> > connector);
> > + enum drm_connector_status result;
> >
> > mutex_lock(&hdmi->mutex);
> > hdmi->force = DRM_FORCE_UNSPECIFIED;
> > @@ -2051,7 +2091,18 @@ dw_hdmi_connector_detect(struct drm_connector
> > *connector, bool force)
> > dw_hdmi_update_phy_mask(hdmi);
> > mutex_unlock(&hdmi->mutex);
> >
> > - return hdmi->phy.ops->read_hpd(hdmi, hdmi->phy.data);
> > + result = hdmi->phy.ops->read_hpd(hdmi, hdmi->phy.data);
> > +
> > + mutex_lock(&hdmi->mutex);
> > + if (result != hdmi->last_connector_result) {
> > + dev_dbg(hdmi->dev, "read_hpd result: %d", result);
> > + handle_plugged_change(hdmi,
> > + result == connector_status_connected);
> > + hdmi->last_connector_result = result;
> > + }
> > + mutex_unlock(&hdmi->mutex);
> > +
> > + return result;
> > }
> >
> > static int dw_hdmi_connector_get_modes(struct drm_connector *connector)
> > @@ -2460,6 +2511,7 @@ __dw_hdmi_probe(struct platform_device *pdev,
> > hdmi->rxsense = true;
> > hdmi->phy_mask = (u8)~(HDMI_PHY_HPD | HDMI_PHY_RX_SENSE);
> > hdmi->mc_clkdis = 0x7f;
> > + hdmi->last_connector_result = connector_status_disconnected;
> >
> > mutex_init(&hdmi->mutex);
> > mutex_init(&hdmi->audio_mutex);
> > @@ -2653,6 +2705,7 @@ __dw_hdmi_probe(struct platform_device *pdev,
> > audio.hdmi = hdmi;
> > audio.write = hdmi_writeb;
> > audio.read = hdmi_readb;
> > + audio.set_plugged_cb = hdmi_set_plugged_cb;
>
> Why is this necessary?
>
> The I2S audio driver already depends on the dw-hdmi module through its
> use of functions already exported. Indirecting this through the
> platform data makes no sense.
>
> Just rename hdmi_set_plugged_cb to dw_hdmi_set_plugged_cb() and export
> it for dw-hdmi-i2s-audio.c to use.
ACK.
Your suggestion makes sense.
Removed in v4.
>
> Thanks.
Thanks so much!
>
> --
> RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
> FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
> According to speedtest.net: 11.9Mbps down 500kbps up
