diff mbox series

[v2,2/2] drm/ingenic: Fix driver not probing when IPU port is missing

Message ID 20200827114404.36748-2-paul@crapouillou.net (mailing list archive)
State New, archived
Headers show
Series [v2,1/2] drm/ingenic: Fix leak of device_node pointer | expand

Commit Message

Paul Cercueil Aug. 27, 2020, 11:44 a.m. UTC 
Even if support for the IPU was compiled in, we may run on a device
(e.g. the Qi LB60) where the IPU is not available, or simply with an old
devicetree without the IPU node. In that case the ingenic-drm refused to
probe.

Fix the driver so that it will probe even if the IPU node is not present
in devicetree (but then IPU support is disabled of course).

v2: Take a different approach

Fixes: fc1acf317b01 ("drm/ingenic: Add support for the IPU")
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
---
 drivers/gpu/drm/ingenic/ingenic-drm-drv.c | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

Comments

Sam Ravnborg Aug. 29, 2020, 9:08 p.m. UTC | #1 
On Thu, Aug 27, 2020 at 01:44:04PM +0200, Paul Cercueil wrote:
> Even if support for the IPU was compiled in, we may run on a device
> (e.g. the Qi LB60) where the IPU is not available, or simply with an old
> devicetree without the IPU node. In that case the ingenic-drm refused to
> probe.
> 
> Fix the driver so that it will probe even if the IPU node is not present
> in devicetree (but then IPU support is disabled of course).
> 
> v2: Take a different approach
> 
> Fixes: fc1acf317b01 ("drm/ingenic: Add support for the IPU")
> Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
> ---
>  drivers/gpu/drm/ingenic/ingenic-drm-drv.c | 19 +++++++++++--------
>  1 file changed, 11 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> index c1bcb93aed2d..b7074161ccf0 100644
> --- a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> +++ b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> @@ -673,7 +673,7 @@ static void ingenic_drm_unbind_all(void *d)
>  	component_unbind_all(priv->dev, &priv->drm);
>  }
>  
> -static int ingenic_drm_bind(struct device *dev)
> +static int ingenic_drm_bind(struct device *dev, bool has_components)
>  {
>  	struct platform_device *pdev = to_platform_device(dev);
>  	const struct jz_soc_info *soc_info;
> @@ -808,7 +808,7 @@ static int ingenic_drm_bind(struct device *dev)
>  			return ret;
>  		}
>  
> -		if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU)) {
> +		if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && has_components) {
>  			ret = component_bind_all(dev, drm);
>  			if (ret) {
>  				if (ret != -EPROBE_DEFER)
> @@ -939,6 +939,11 @@ static int ingenic_drm_bind(struct device *dev)
>  	return ret;
>  }
>  
> +static int ingenic_drm_bind_with_components(struct device *dev)
> +{
> +	return ingenic_drm_bind(dev, true);
> +}
> +
>  static int compare_of(struct device *dev, void *data)
>  {
>  	return dev->of_node == data;
> @@ -957,7 +962,7 @@ static void ingenic_drm_unbind(struct device *dev)
>  }
>  
>  static const struct component_master_ops ingenic_master_ops = {
> -	.bind = ingenic_drm_bind,
> +	.bind = ingenic_drm_bind_with_components,
>  	.unbind = ingenic_drm_unbind,
>  };
>  
> @@ -968,14 +973,12 @@ static int ingenic_drm_probe(struct platform_device *pdev)
>  	struct device_node *np;
>  
>  	if (!IS_ENABLED(CONFIG_DRM_INGENIC_IPU))
> -		return ingenic_drm_bind(dev);
> +		return ingenic_drm_bind(dev, false);
>  
>  	/* IPU is at port address 8 */
>  	np = of_graph_get_remote_node(dev->of_node, 8, 0);
> -	if (!np) {
> -		dev_err(dev, "Unable to get IPU node\n");
> -		return -EINVAL;
> -	}
> +	if (!np)
> +		return ingenic_drm_bind(dev, false);
>  
>  	drm_of_component_match_add(dev, &match, compare_of, np);
>  	of_node_put(np);
> -- 
> 2.28.0
Ezequiel Garcia Aug. 31, 2020, 12:21 a.m. UTC | #2 
Hi Paul,

On Thu, 27 Aug 2020 at 09:04, Paul Cercueil <paul@crapouillou.net> wrote:
>
> Even if support for the IPU was compiled in, we may run on a device
> (e.g. the Qi LB60) where the IPU is not available, or simply with an old
> devicetree without the IPU node. In that case the ingenic-drm refused to
> probe.
>
> Fix the driver so that it will probe even if the IPU node is not present
> in devicetree (but then IPU support is disabled of course).
>
> v2: Take a different approach
>
> Fixes: fc1acf317b01 ("drm/ingenic: Add support for the IPU")
> Signed-off-by: Paul Cercueil <paul@crapouillou.net>
> ---
>  drivers/gpu/drm/ingenic/ingenic-drm-drv.c | 19 +++++++++++--------
>  1 file changed, 11 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> index c1bcb93aed2d..b7074161ccf0 100644
> --- a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> +++ b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
> @@ -673,7 +673,7 @@ static void ingenic_drm_unbind_all(void *d)
>         component_unbind_all(priv->dev, &priv->drm);
>  }
>
> -static int ingenic_drm_bind(struct device *dev)
> +static int ingenic_drm_bind(struct device *dev, bool has_components)
>  {
>         struct platform_device *pdev = to_platform_device(dev);
>         const struct jz_soc_info *soc_info;
> @@ -808,7 +808,7 @@ static int ingenic_drm_bind(struct device *dev)
>                         return ret;
>                 }
>
> -               if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU)) {
> +               if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && has_components) {
>                         ret = component_bind_all(dev, drm);
>                         if (ret) {
>                                 if (ret != -EPROBE_DEFER)
> @@ -939,6 +939,11 @@ static int ingenic_drm_bind(struct device *dev)
>         return ret;
>  }
>
> +static int ingenic_drm_bind_with_components(struct device *dev)
> +{
> +       return ingenic_drm_bind(dev, true);
> +}
> +
>  static int compare_of(struct device *dev, void *data)
>  {
>         return dev->of_node == data;
> @@ -957,7 +962,7 @@ static void ingenic_drm_unbind(struct device *dev)
>  }
>
>  static const struct component_master_ops ingenic_master_ops = {
> -       .bind = ingenic_drm_bind,
> +       .bind = ingenic_drm_bind_with_components,
>         .unbind = ingenic_drm_unbind,
>  };
>
> @@ -968,14 +973,12 @@ static int ingenic_drm_probe(struct platform_device *pdev)
>         struct device_node *np;
>
>         if (!IS_ENABLED(CONFIG_DRM_INGENIC_IPU))
> -               return ingenic_drm_bind(dev);
> +               return ingenic_drm_bind(dev, false);
>
>         /* IPU is at port address 8 */
>         np = of_graph_get_remote_node(dev->of_node, 8, 0);

How about we get rid of this (seems a bit odd to rely on port address) ?
Rockchip-drm driver has a nice approach, and I think we might need
something like that going forward, to support dw-hdmi.

Thanks,
Ezequiel
Paul Cercueil Aug. 31, 2020, 12:47 a.m. UTC | #3 
Le dim. 30 août 2020 à 21:21, Ezequiel Garcia 
<ezequiel@vanguardiasur.com.ar> a écrit :
> Hi Paul,
> 
> On Thu, 27 Aug 2020 at 09:04, Paul Cercueil <paul@crapouillou.net> 
> wrote:
>> 
>>  Even if support for the IPU was compiled in, we may run on a device
>>  (e.g. the Qi LB60) where the IPU is not available, or simply with 
>> an old
>>  devicetree without the IPU node. In that case the ingenic-drm 
>> refused to
>>  probe.
>> 
>>  Fix the driver so that it will probe even if the IPU node is not 
>> present
>>  in devicetree (but then IPU support is disabled of course).
>> 
>>  v2: Take a different approach
>> 
>>  Fixes: fc1acf317b01 ("drm/ingenic: Add support for the IPU")
>>  Signed-off-by: Paul Cercueil <paul@crapouillou.net>
>>  ---
>>   drivers/gpu/drm/ingenic/ingenic-drm-drv.c | 19 +++++++++++--------
>>   1 file changed, 11 insertions(+), 8 deletions(-)
>> 
>>  diff --git a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c 
>> b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
>>  index c1bcb93aed2d..b7074161ccf0 100644
>>  --- a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
>>  +++ b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
>>  @@ -673,7 +673,7 @@ static void ingenic_drm_unbind_all(void *d)
>>          component_unbind_all(priv->dev, &priv->drm);
>>   }
>> 
>>  -static int ingenic_drm_bind(struct device *dev)
>>  +static int ingenic_drm_bind(struct device *dev, bool 
>> has_components)
>>   {
>>          struct platform_device *pdev = to_platform_device(dev);
>>          const struct jz_soc_info *soc_info;
>>  @@ -808,7 +808,7 @@ static int ingenic_drm_bind(struct device *dev)
>>                          return ret;
>>                  }
>> 
>>  -               if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU)) {
>>  +               if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && 
>> has_components) {
>>                          ret = component_bind_all(dev, drm);
>>                          if (ret) {
>>                                  if (ret != -EPROBE_DEFER)
>>  @@ -939,6 +939,11 @@ static int ingenic_drm_bind(struct device *dev)
>>          return ret;
>>   }
>> 
>>  +static int ingenic_drm_bind_with_components(struct device *dev)
>>  +{
>>  +       return ingenic_drm_bind(dev, true);
>>  +}
>>  +
>>   static int compare_of(struct device *dev, void *data)
>>   {
>>          return dev->of_node == data;
>>  @@ -957,7 +962,7 @@ static void ingenic_drm_unbind(struct device 
>> *dev)
>>   }
>> 
>>   static const struct component_master_ops ingenic_master_ops = {
>>  -       .bind = ingenic_drm_bind,
>>  +       .bind = ingenic_drm_bind_with_components,
>>          .unbind = ingenic_drm_unbind,
>>   };
>> 
>>  @@ -968,14 +973,12 @@ static int ingenic_drm_probe(struct 
>> platform_device *pdev)
>>          struct device_node *np;
>> 
>>          if (!IS_ENABLED(CONFIG_DRM_INGENIC_IPU))
>>  -               return ingenic_drm_bind(dev);
>>  +               return ingenic_drm_bind(dev, false);
>> 
>>          /* IPU is at port address 8 */
>>          np = of_graph_get_remote_node(dev->of_node, 8, 0);
> 
> How about we get rid of this (seems a bit odd to rely on port 
> address) ?
> Rockchip-drm driver has a nice approach, and I think we might need
> something like that going forward, to support dw-hdmi.

The rockchip-drm approach works because all the sub-drivers must probe. 
In the case of ingenic-drm, even if the ingenic-drm driver was compiled 
with the ipu and dw-hdmi sub-drivers, you can't rely on these probing, 
as they are not present on e.g. the JZ4740.

Cheers,
-Paul
diff mbox series

Patch

diff --git a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
index c1bcb93aed2d..b7074161ccf0 100644
--- a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
+++ b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c
@@ -673,7 +673,7 @@  static void ingenic_drm_unbind_all(void *d)
 	component_unbind_all(priv->dev, &priv->drm);
 }
 
-static int ingenic_drm_bind(struct device *dev)
+static int ingenic_drm_bind(struct device *dev, bool has_components)
 {
 	struct platform_device *pdev = to_platform_device(dev);
 	const struct jz_soc_info *soc_info;
@@ -808,7 +808,7 @@  static int ingenic_drm_bind(struct device *dev)
 			return ret;
 		}
 
-		if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU)) {
+		if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && has_components) {
 			ret = component_bind_all(dev, drm);
 			if (ret) {
 				if (ret != -EPROBE_DEFER)
@@ -939,6 +939,11 @@  static int ingenic_drm_bind(struct device *dev)
 	return ret;
 }
 
+static int ingenic_drm_bind_with_components(struct device *dev)
+{
+	return ingenic_drm_bind(dev, true);
+}
+
 static int compare_of(struct device *dev, void *data)
 {
 	return dev->of_node == data;
@@ -957,7 +962,7 @@  static void ingenic_drm_unbind(struct device *dev)
 }
 
 static const struct component_master_ops ingenic_master_ops = {
-	.bind = ingenic_drm_bind,
+	.bind = ingenic_drm_bind_with_components,
 	.unbind = ingenic_drm_unbind,
 };
 
@@ -968,14 +973,12 @@  static int ingenic_drm_probe(struct platform_device *pdev)
 	struct device_node *np;
 
 	if (!IS_ENABLED(CONFIG_DRM_INGENIC_IPU))
-		return ingenic_drm_bind(dev);
+		return ingenic_drm_bind(dev, false);
 
 	/* IPU is at port address 8 */
 	np = of_graph_get_remote_node(dev->of_node, 8, 0);
-	if (!np) {
-		dev_err(dev, "Unable to get IPU node\n");
-		return -EINVAL;
-	}
+	if (!np)
+		return ingenic_drm_bind(dev, false);
 
 	drm_of_component_match_add(dev, &match, compare_of, np);
 	of_node_put(np);