[PATCH 04/11] drm/exynos: hdmi: convert to struct drm_edid

[PATCH 04/11] drm/exynos: hdmi: convert to struct drm_edid

[Jani Nikula]

Prefer the struct drm_edid based functions for reading the EDID and
updating the connector.

The functional change is that the CEC physical address gets invalidated
when the EDID could not be read.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

---

Cc: Inki Dae <inki.dae@samsung.com>
Cc: Seung-Woo Kim <sw0312.kim@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Alim Akhtar <alim.akhtar@samsung.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-samsung-soc@vger.kernel.org
---
 drivers/gpu/drm/exynos/exynos_hdmi.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/exynos/exynos_hdmi.c b/drivers/gpu/drm/exynos/exynos_hdmi.c
index e968824a4c72..9033e8b66816 100644
--- a/drivers/gpu/drm/exynos/exynos_hdmi.c
+++ b/drivers/gpu/drm/exynos/exynos_hdmi.c
@@ -883,27 +883,30 @@ static const struct drm_connector_funcs hdmi_connector_funcs = {
 static int hdmi_get_modes(struct drm_connector *connector)
 {
 	struct hdmi_context *hdata = connector_to_hdmi(connector);
-	struct edid *edid;
+	const struct drm_display_info *info = &connector->display_info;
+	const struct drm_edid *drm_edid;
 	int ret;
 
 	if (!hdata->ddc_adpt)
 		return 0;
 
-	edid = drm_get_edid(connector, hdata->ddc_adpt);
-	if (!edid)
+	drm_edid = drm_edid_read_ddc(connector, hdata->ddc_adpt);
+
+	drm_edid_connector_update(connector, drm_edid);
+
+	cec_notifier_set_phys_addr(hdata->notifier, info->source_physical_address);
+
+	if (!drm_edid)
 		return 0;
 
-	hdata->dvi_mode = !connector->display_info.is_hdmi;
+	hdata->dvi_mode = !info->is_hdmi;
 	DRM_DEV_DEBUG_KMS(hdata->dev, "%s : width[%d] x height[%d]\n",
 			  (hdata->dvi_mode ? "dvi monitor" : "hdmi monitor"),
-			  edid->width_cm, edid->height_cm);
-
-	drm_connector_update_edid_property(connector, edid);
-	cec_notifier_set_phys_addr_from_edid(hdata->notifier, edid);
+			  info->width_mm / 10, info->height_mm / 10);
 
-	ret = drm_add_edid_modes(connector, edid);
+	ret = drm_edid_connector_add_modes(connector);
 
-	kfree(edid);
+	drm_edid_free(drm_edid);
 
 	return ret;
 }
-- 
2.39.2

Re: [PATCH 04/11] drm/exynos: hdmi: convert to struct drm_edid

[Inki Dae]

Hi, Jani Nikula,

Thanks for your contribution and sorry for being late. Below are my comments.

2024년 5월 14일 (화) 오후 9:57, Jani Nikula <jani.nikula@intel.com>님이 작성:
>
> Prefer the struct drm_edid based functions for reading the EDID and
> updating the connector.
>
> The functional change is that the CEC physical address gets invalidated
> when the EDID could not be read.
>
> Signed-off-by: Jani Nikula <jani.nikula@intel.com>
>
> ---
>
> Cc: Inki Dae <inki.dae@samsung.com>
> Cc: Seung-Woo Kim <sw0312.kim@samsung.com>
> Cc: Kyungmin Park <kyungmin.park@samsung.com>
> Cc: Krzysztof Kozlowski <krzk@kernel.org>
> Cc: Alim Akhtar <alim.akhtar@samsung.com>
> Cc: linux-arm-kernel@lists.infradead.org
> Cc: linux-samsung-soc@vger.kernel.org
> ---
>  drivers/gpu/drm/exynos/exynos_hdmi.c | 23 +++++++++++++----------
>  1 file changed, 13 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/gpu/drm/exynos/exynos_hdmi.c b/drivers/gpu/drm/exynos/exynos_hdmi.c
> index e968824a4c72..9033e8b66816 100644
> --- a/drivers/gpu/drm/exynos/exynos_hdmi.c
> +++ b/drivers/gpu/drm/exynos/exynos_hdmi.c
> @@ -883,27 +883,30 @@ static const struct drm_connector_funcs hdmi_connector_funcs = {
>  static int hdmi_get_modes(struct drm_connector *connector)
>  {
>         struct hdmi_context *hdata = connector_to_hdmi(connector);
> -       struct edid *edid;
> +       const struct drm_display_info *info = &connector->display_info;
> +       const struct drm_edid *drm_edid;
>         int ret;
>
>         if (!hdata->ddc_adpt)
>                 return 0;
>
> -       edid = drm_get_edid(connector, hdata->ddc_adpt);
> -       if (!edid)
> +       drm_edid = drm_edid_read_ddc(connector, hdata->ddc_adpt);

drm_edid_read_ddc function can return NULL for an error. Could you add
an exception handling?

> +
> +       drm_edid_connector_update(connector, drm_edid);

Ditto. drm_edid_connector_update function can return a negative value
for an error.

> +
> +       cec_notifier_set_phys_addr(hdata->notifier, info->source_physical_address);
> +
> +       if (!drm_edid)
>                 return 0;
>
> -       hdata->dvi_mode = !connector->display_info.is_hdmi;
> +       hdata->dvi_mode = !info->is_hdmi;

Above change wouldn't be related to this patch.

>         DRM_DEV_DEBUG_KMS(hdata->dev, "%s : width[%d] x height[%d]\n",
>                           (hdata->dvi_mode ? "dvi monitor" : "hdmi monitor"),
> -                         edid->width_cm, edid->height_cm);
> -
> -       drm_connector_update_edid_property(connector, edid);
> -       cec_notifier_set_phys_addr_from_edid(hdata->notifier, edid);
> +                         info->width_mm / 10, info->height_mm / 10);

The purpose of this patch would be to replace edid with drm_edid so
how about updating the above change like below?
    drm_edid->edid->width_cm, erm_edid->edid->height_cm);

Thanks,
Inki Dae

>
> -       ret = drm_add_edid_modes(connector, edid);
> +       ret = drm_edid_connector_add_modes(connector);
>
> -       kfree(edid);
> +       drm_edid_free(drm_edid);
>
>         return ret;
>  }
> --
> 2.39.2
>
>

Re: [PATCH 04/11] drm/exynos: hdmi: convert to struct drm_edid

[Jani Nikula]

On Mon, 10 Jun 2024, Inki Dae <daeinki@gmail.com> wrote:
> Hi, Jani Nikula,
>
> Thanks for your contribution and sorry for being late. Below are my comments.
>
> 2024년 5월 14일 (화) 오후 9:57, Jani Nikula <jani.nikula@intel.com>님이 작성:
>>
>> Prefer the struct drm_edid based functions for reading the EDID and
>> updating the connector.
>>
>> The functional change is that the CEC physical address gets invalidated
>> when the EDID could not be read.
>>
>> Signed-off-by: Jani Nikula <jani.nikula@intel.com>
>>
>> ---
>>
>> Cc: Inki Dae <inki.dae@samsung.com>
>> Cc: Seung-Woo Kim <sw0312.kim@samsung.com>
>> Cc: Kyungmin Park <kyungmin.park@samsung.com>
>> Cc: Krzysztof Kozlowski <krzk@kernel.org>
>> Cc: Alim Akhtar <alim.akhtar@samsung.com>
>> Cc: linux-arm-kernel@lists.infradead.org
>> Cc: linux-samsung-soc@vger.kernel.org
>> ---
>>  drivers/gpu/drm/exynos/exynos_hdmi.c | 23 +++++++++++++----------
>>  1 file changed, 13 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/exynos/exynos_hdmi.c b/drivers/gpu/drm/exynos/exynos_hdmi.c
>> index e968824a4c72..9033e8b66816 100644
>> --- a/drivers/gpu/drm/exynos/exynos_hdmi.c
>> +++ b/drivers/gpu/drm/exynos/exynos_hdmi.c
>> @@ -883,27 +883,30 @@ static const struct drm_connector_funcs hdmi_connector_funcs = {
>>  static int hdmi_get_modes(struct drm_connector *connector)
>>  {
>>         struct hdmi_context *hdata = connector_to_hdmi(connector);
>> -       struct edid *edid;
>> +       const struct drm_display_info *info = &connector->display_info;
>> +       const struct drm_edid *drm_edid;
>>         int ret;
>>
>>         if (!hdata->ddc_adpt)
>>                 return 0;
>>
>> -       edid = drm_get_edid(connector, hdata->ddc_adpt);
>> -       if (!edid)
>> +       drm_edid = drm_edid_read_ddc(connector, hdata->ddc_adpt);
>
> drm_edid_read_ddc function can return NULL for an error. Could you add
> an exception handling?

If it returns NULL, the below code intentionally passes NULL to
drm_edid_connector_update() to reset the EDID. After that,
cec_notifier_set_phys_addr() also resets the source physical
address. And !drm_edid is handled after that.

>
>> +
>> +       drm_edid_connector_update(connector, drm_edid);
>
> Ditto. drm_edid_connector_update function can return a negative value
> for an error.

Okay.cg

>
>> +
>> +       cec_notifier_set_phys_addr(hdata->notifier, info->source_physical_address);
>> +
>> +       if (!drm_edid)
>>                 return 0;
>>
>> -       hdata->dvi_mode = !connector->display_info.is_hdmi;
>> +       hdata->dvi_mode = !info->is_hdmi;
>
> Above change wouldn't be related to this patch.

Yeah, reuses info which is for info->source_physical_address which would
be too long otherwise. Can add another intermediate patch for that.

>
>>         DRM_DEV_DEBUG_KMS(hdata->dev, "%s : width[%d] x height[%d]\n",
>>                           (hdata->dvi_mode ? "dvi monitor" : "hdmi monitor"),
>> -                         edid->width_cm, edid->height_cm);
>> -
>> -       drm_connector_update_edid_property(connector, edid);
>> -       cec_notifier_set_phys_addr_from_edid(hdata->notifier, edid);
>> +                         info->width_mm / 10, info->height_mm / 10);
>
> The purpose of this patch would be to replace edid with drm_edid so
> how about updating the above change like below?
>     drm_edid->edid->width_cm, erm_edid->edid->height_cm);

struct drm_edid is an opaque type and drm_edid->edid is not available
for drivers to look at.

BR,
Jani.

>
> Thanks,
> Inki Dae
>
>>
>> -       ret = drm_add_edid_modes(connector, edid);
>> +       ret = drm_edid_connector_add_modes(connector);
>>
>> -       kfree(edid);
>> +       drm_edid_free(drm_edid);
>>
>>         return ret;
>>  }
>> --
>> 2.39.2
>>
>>

-- 
Jani Nikula, Intel