Skip to contentSkip to conversation

media: mali-c55: Fix NULL dev stream alert for tpg sub-device

PATCH Superseded jempty.liang 4 participants 6 messages Jan 13

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 media: mali-c55: Fix NULL dev stream alert for tpg sub-device

Fix "(NULL device *): collect_streams: sub-device 'mali-c55 tpg' does not support streams" by setting sd->dev = mali_c55->dev for isp/resizer/tpg modules.

---
 drivers/media/platform/arm/mali-c55/mali-c55-isp.c     | 1 +
 drivers/media/platform/arm/mali-c55/mali-c55-resizer.c | 1 +
 drivers/media/platform/arm/mali-c55/mali-c55-tpg.c     | 1 +
 3 files changed, 3 insertions(+)
drivers/media/platform/arm/mali-c55/mali-c55-isp.c +1 -0
index 497f25fbdd13..ce84aad9f1d5 100644
@@ -610,6 +610,7 @@ int mali_c55_register_isp(struct mali_c55 *mali_c55)
 	sd->entity.ops = &mali_c55_isp_media_ops;
 	sd->entity.function = MEDIA_ENT_F_PROC_VIDEO_ISP;
 	sd->internal_ops = &mali_c55_isp_internal_ops;
+	sd->dev = mali_c55->dev;
 	strscpy(sd->name, MALI_C55_DRIVER_NAME " isp", sizeof(sd->name));
 
 	isp->pads[MALI_C55_ISP_PAD_SINK_VIDEO].flags = MEDIA_PAD_FL_SINK |
drivers/media/platform/arm/mali-c55/mali-c55-resizer.c +1 -0
index a8d739af74b6..c4f46651dcee 100644
@@ -1070,6 +1070,7 @@ static int mali_c55_register_resizer(struct mali_c55 *mali_c55,
 	sd->flags |= V4L2_SUBDEV_FL_HAS_DEVNODE | V4L2_SUBDEV_FL_STREAMS;
 	sd->entity.function = MEDIA_ENT_F_PROC_VIDEO_SCALER;
 	sd->internal_ops = &mali_c55_resizer_internal_ops;
+	sd->dev = mali_c55->dev;
 
 	rsz->pads[MALI_C55_RSZ_SINK_PAD].flags = MEDIA_PAD_FL_SINK;
 	rsz->pads[MALI_C55_RSZ_SOURCE_PAD].flags = MEDIA_PAD_FL_SOURCE;
drivers/media/platform/arm/mali-c55/mali-c55-tpg.c +1 -0
index 1af5d2759a83..894f4cf377af 100644
@@ -370,6 +370,7 @@ int mali_c55_register_tpg(struct mali_c55 *mali_c55)
 	sd->flags |= V4L2_SUBDEV_FL_HAS_DEVNODE | V4L2_SUBDEV_FL_HAS_EVENTS;
 	sd->entity.function = MEDIA_ENT_F_CAM_SENSOR;
 	sd->internal_ops = &mali_c55_tpg_internal_ops;
+	sd->dev = mali_c55->dev;
 	strscpy(sd->name, MALI_C55_DRIVER_NAME " tpg", sizeof(sd->name));
 
 	pad->flags = MEDIA_PAD_FL_SOURCE;
Signature · 1 line
2.25.1
maintainer
#1 lore raw

Hi Jempty

Quoted 3 lines · jempty.liang

Fix "(NULL device *): collect_streams: sub-device 'mali-c55 tpg' does not support streams" by setting sd->dev = mali_c55->dev for isp/resizer/tpg modules.

The missing device name in the error message is just a symptom, I think this is a geniunely missing part of the subdevice initalization.

I would re-phrase this as

-------------------------------------------------------------------------------
The subdevices registered by the Mali-C55 driver do not have their 'struct device *dev' member initialized. This is visibile when looking at debug message, as in example:

"(NULL device *): collect_streams: sub-device 'mali-c55 tpg' does not support streams"

Fix this by initializing the *dev field for each subdevice registered by the Mali-C55 driver.
-------------------------------------------------------------------------------

I wonder if v4l2_device_register_subdev() shouldn't WARN or if return an error if sd->dev is not initialized. Laurent, Sakari: opinions ?

Thanks
j

Quoted 47 lines
---
 drivers/media/platform/arm/mali-c55/mali-c55-isp.c     | 1 +
 drivers/media/platform/arm/mali-c55/mali-c55-resizer.c | 1 +
 drivers/media/platform/arm/mali-c55/mali-c55-tpg.c     | 1 +
 3 files changed, 3 insertions(+)
drivers/media/platform/arm/mali-c55/mali-c55-isp.c +1 -0
index 497f25fbdd13..ce84aad9f1d5 100644
@@ -610,6 +610,7 @@ int mali_c55_register_isp(struct mali_c55 *mali_c55)
 	sd->entity.ops = &mali_c55_isp_media_ops;
 	sd->entity.function = MEDIA_ENT_F_PROC_VIDEO_ISP;
 	sd->internal_ops = &mali_c55_isp_internal_ops;
+	sd->dev = mali_c55->dev;
 	strscpy(sd->name, MALI_C55_DRIVER_NAME " isp", sizeof(sd->name));
 
 	isp->pads[MALI_C55_ISP_PAD_SINK_VIDEO].flags = MEDIA_PAD_FL_SINK |
drivers/media/platform/arm/mali-c55/mali-c55-resizer.c +1 -0
index a8d739af74b6..c4f46651dcee 100644
@@ -1070,6 +1070,7 @@ static int mali_c55_register_resizer(struct mali_c55 *mali_c55,
 	sd->flags |= V4L2_SUBDEV_FL_HAS_DEVNODE | V4L2_SUBDEV_FL_STREAMS;
 	sd->entity.function = MEDIA_ENT_F_PROC_VIDEO_SCALER;
 	sd->internal_ops = &mali_c55_resizer_internal_ops;
+	sd->dev = mali_c55->dev;
 
 	rsz->pads[MALI_C55_RSZ_SINK_PAD].flags = MEDIA_PAD_FL_SINK;
 	rsz->pads[MALI_C55_RSZ_SOURCE_PAD].flags = MEDIA_PAD_FL_SOURCE;
drivers/media/platform/arm/mali-c55/mali-c55-tpg.c +1 -1
index 1af5d2759a83..894f4cf377af 100644
@@ -370,6 +370,7 @@ int mali_c55_register_tpg(struct mali_c55 *mali_c55)
 	sd->flags |= V4L2_SUBDEV_FL_HAS_DEVNODE | V4L2_SUBDEV_FL_HAS_EVENTS;
 	sd->entity.function = MEDIA_ENT_F_CAM_SENSOR;
 	sd->internal_ops = &mali_c55_tpg_internal_ops;
+	sd->dev = mali_c55->dev;
 	strscpy(sd->name, MALI_C55_DRIVER_NAME " tpg", sizeof(sd->name));
 
 	pad->flags = MEDIA_PAD_FL_SOURCE;
--

2.25.1

#2 lore raw

Hi Jacopo,

Quoted 2 lines · Jacopo Mondi

I wonder if v4l2_device_register_subdev() shouldn't WARN or if return an error if sd->dev is not initialized. Laurent, Sakari: opinions ?

I might even return an error but I wonder what would break. Perhaps indeed could start with a warning?

Signature · 1 line
Sakari Ailus
maintainer
#3 lore raw

Hi Sakari

Quoted 8 lines · Sakari Ailus

Hi Jacopo,

Quoted 2 lines · Jacopo Mondi

I wonder if v4l2_device_register_subdev() shouldn't WARN or if return an error if sd->dev is not initialized. Laurent, Sakari: opinions ?

I might even return an error but I wonder what would break. Perhaps indeed could start with a warning?

I would start with:

drivers/media/v4l2-core/v4l2-device.c +6 -0
@@ -117,6 +117,13 @@ int __v4l2_device_register_subdev(struct v4l2_device *v4l2_dev,
        if (!v4l2_dev || !sd || sd->v4l2_dev || !sd->name[0])
                return -EINVAL;
 
+       /*
+        * Initializing sd->dev is mandatory. Warn for now to give faulty
+        * drivers a bit of time to update before actually returning an error.
+        */
+       WARN_ON(!sd->dev);
+
        /*
         * The reason to acquire the module here is to avoid unloading
         * a module of sub-device which is registered to a media

Would you like a patch ?

Quoted 3 lines

--
Sakari Ailus

#4 lore raw

Hi Jacopo,

Quoted 30 lines · Jacopo Mondi

Hi Sakari

Quoted 8 lines · Sakari Ailus

Hi Jacopo,

Quoted 2 lines · Jacopo Mondi

I wonder if v4l2_device_register_subdev() shouldn't WARN or if return an error if sd->dev is not initialized. Laurent, Sakari: opinions ?

I might even return an error but I wonder what would break. Perhaps indeed could start with a warning?

I would start with:

drivers/media/v4l2-core/v4l2-device.c +6 -0
@@ -117,6 +117,13 @@ int __v4l2_device_register_subdev(struct v4l2_device *v4l2_dev,
        if (!v4l2_dev || !sd || sd->v4l2_dev || !sd->name[0])
                return -EINVAL;
 
+       /*
+        * Initializing sd->dev is mandatory. Warn for now to give faulty
+        * drivers a bit of time to update before actually returning an error.
+        */
+       WARN_ON(!sd->dev);
+
        /*
         * The reason to acquire the module here is to avoid unloading
         * a module of sub-device which is registered to a media
Quoted 1 line

Would you like a patch ?

Please. :-)

Signature · 3 lines
Kind regards,

Sakari Ailus
maintainer
#5 lore raw
In reply to Sakari Ailus
Quoted 9 lines · Sakari Ailus

Hi Jacopo,

Quoted 2 lines · Jacopo Mondi

I wonder if v4l2_device_register_subdev() shouldn't WARN or if return an error if sd->dev is not initialized. Laurent, Sakari: opinions ?

I might even return an error but I wonder what would break. Perhaps indeed could start with a warning?

It looks like it's only used for dev_*() printks, unless I'm missing something. Thanks Jempty for spotting the problem!

Dan