[PATCH 0/3] ipu3-cio2 format fixes

[PATCH 0/3] ipu3-cio2 format fixes

[Sakari Ailus]

Hello all,

This set addresses most notable subdev format related issues, namely the
sub-device sink format being unaccessible. The result of accessing it
varied from oopses to crashes.

Sakari Ailus (3):
  ipu3-cio2: Return actual subdev format
  ipu3-cio2: Serialise access to pad format
  ipu3-cio2: Only allow setting valid mbus codes

 drivers/media/pci/intel/ipu3/ipu3-cio2.c | 42 ++++++++++++------------
 drivers/media/pci/intel/ipu3/ipu3-cio2.h |  1 +
 2 files changed, 22 insertions(+), 21 deletions(-)

-- 
2.27.0

[PATCH 1/3] ipu3-cio2: Return actual subdev format

[Sakari Ailus]

Return actual subdev format on ipu3-cio2 subdev pads. The earlier
implementation was based on an infinite recursion that exhausted the
stack.

Reported-by: Tsuchiya Yuto <kitakar@gmail.com>
Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 drivers/media/pci/intel/ipu3/ipu3-cio2.c | 24 +++---------------------
 1 file changed, 3 insertions(+), 21 deletions(-)

diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
index 4e598e937dfe..afa472026ba4 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
@@ -1232,29 +1232,11 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
 			       struct v4l2_subdev_format *fmt)
 {
 	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
-	struct v4l2_subdev_format format;
-	int ret;
 
-	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
+	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
 		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
-		return 0;
-	}
-
-	if (fmt->pad == CIO2_PAD_SINK) {
-		format.which = V4L2_SUBDEV_FORMAT_ACTIVE;
-		ret = v4l2_subdev_call(sd, pad, get_fmt, NULL,
-				       &format);
-
-		if (ret)
-			return ret;
-		/* update colorspace etc */
-		q->subdev_fmt.colorspace = format.format.colorspace;
-		q->subdev_fmt.ycbcr_enc = format.format.ycbcr_enc;
-		q->subdev_fmt.quantization = format.format.quantization;
-		q->subdev_fmt.xfer_func = format.format.xfer_func;
-	}
-
-	fmt->format = q->subdev_fmt;
+	else
+		fmt->format = q->subdev_fmt;
 
 	return 0;
 }
-- 
2.27.0

[PATCH 2/3] ipu3-cio2: Serialise access to pad format

[Sakari Ailus]

Pad format can be accessed from user space. Serialise access to it.

Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 drivers/media/pci/intel/ipu3/ipu3-cio2.c | 11 +++++++++++
 drivers/media/pci/intel/ipu3/ipu3-cio2.h |  1 +
 2 files changed, 12 insertions(+)

diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
index afa472026ba4..9c7b527a8800 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
@@ -1233,11 +1233,15 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
 {
 	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
 
+	mutex_lock(&q->subdev_lock);
+
 	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
 		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
 	else
 		fmt->format = q->subdev_fmt;
 
+	mutex_unlock(&q->subdev_lock);
+
 	return 0;
 }
 
@@ -1261,6 +1265,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
 	if (fmt->pad == CIO2_PAD_SOURCE)
 		return cio2_subdev_get_fmt(sd, cfg, fmt);
 
+	mutex_lock(&q->subdev_lock);
+
 	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
 		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
 	} else {
@@ -1271,6 +1277,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
 		fmt->format = q->subdev_fmt;
 	}
 
+	mutex_unlock(&q->subdev_lock);
+
 	return 0;
 }
 
@@ -1529,6 +1537,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
 
 	/* Initialize miscellaneous variables */
 	mutex_init(&q->lock);
+	mutex_init(&q->subdev_lock);
 
 	/* Initialize formats to default values */
 	fmt = &q->subdev_fmt;
@@ -1646,6 +1655,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
 	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
 fail_fbpt:
 	mutex_destroy(&q->lock);
+	mutex_destroy(&q->subdev_lock);
 
 	return r;
 }
@@ -1658,6 +1668,7 @@ static void cio2_queue_exit(struct cio2_device *cio2, struct cio2_queue *q)
 	media_entity_cleanup(&q->subdev.entity);
 	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
 	mutex_destroy(&q->lock);
+	mutex_destroy(&q->subdev_lock);
 }
 
 static int cio2_queues_init(struct cio2_device *cio2)
diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.h b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
index 549b08f88f0c..df0247326a1d 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2.h
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
@@ -335,6 +335,7 @@ struct cio2_queue {
 
 	/* Subdev, /dev/v4l-subdevX */
 	struct v4l2_subdev subdev;
+	struct mutex subdev_lock;
 	struct media_pad subdev_pads[CIO2_PADS];
 	struct v4l2_mbus_framefmt subdev_fmt;
 	atomic_t frame_sequence;
-- 
2.27.0

[PATCH 3/3] ipu3-cio2: Only allow setting valid mbus codes

[Sakari Ailus]

Check the mbus code provided by the user is one of those the driver
supports. Ignore the code in set_fmt otherwise.

Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 drivers/media/pci/intel/ipu3/ipu3-cio2.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
index 9c7b527a8800..2ea6313e00b0 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
@@ -1270,10 +1270,17 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
 	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
 		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
 	} else {
+		unsigned int i;
+
 		/* It's the sink, allow changing frame size */
 		q->subdev_fmt.width = fmt->format.width;
 		q->subdev_fmt.height = fmt->format.height;
-		q->subdev_fmt.code = fmt->format.code;
+		for (i = 0; i < ARRAY_SIZE(formats); i++) {
+			if (formats[i].mbus_code == fmt->format.code) {
+				q->subdev_fmt.code = fmt->format.code;
+				break;
+			}
+		}
 		fmt->format = q->subdev_fmt;
 	}
 
-- 
2.27.0

Re: [PATCH 1/3] ipu3-cio2: Return actual subdev format

[Laurent Pinchart]

Hi Sakari,

Thank you for the patch.

On Thu, Oct 08, 2020 at 11:47:45PM +0300, Sakari Ailus wrote:
> Return actual subdev format on ipu3-cio2 subdev pads. The earlier
> implementation was based on an infinite recursion that exhausted the
> stack.

A bad idea indeed :-)

> Reported-by: Tsuchiya Yuto <kitakar@gmail.com>
> Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> ---
>  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 24 +++---------------------
>  1 file changed, 3 insertions(+), 21 deletions(-)
> 
> diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> index 4e598e937dfe..afa472026ba4 100644
> --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> @@ -1232,29 +1232,11 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
>  			       struct v4l2_subdev_format *fmt)
>  {
>  	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
> -	struct v4l2_subdev_format format;
> -	int ret;
>  
> -	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
> +	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
>  		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
> -		return 0;
> -	}
> -
> -	if (fmt->pad == CIO2_PAD_SINK) {
> -		format.which = V4L2_SUBDEV_FORMAT_ACTIVE;
> -		ret = v4l2_subdev_call(sd, pad, get_fmt, NULL,
> -				       &format);
> -
> -		if (ret)
> -			return ret;
> -		/* update colorspace etc */
> -		q->subdev_fmt.colorspace = format.format.colorspace;
> -		q->subdev_fmt.ycbcr_enc = format.format.ycbcr_enc;
> -		q->subdev_fmt.quantization = format.format.quantization;
> -		q->subdev_fmt.xfer_func = format.format.xfer_func;
> -	}
> -
> -	fmt->format = q->subdev_fmt;

I'm pretty speechless. All I can say is

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

and let's forget this has ever existed :-)

> +	else
> +		fmt->format = q->subdev_fmt;
>  
>  	return 0;
>  }

-- 
Regards,

Laurent Pinchart

Re: [PATCH 2/3] ipu3-cio2: Serialise access to pad format

[Laurent Pinchart]

Hi Sakari,

Thank you for the patch.

On Thu, Oct 08, 2020 at 11:47:46PM +0300, Sakari Ailus wrote:
> Pad format can be accessed from user space. Serialise access to it.
> 
> Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> ---
>  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 11 +++++++++++
>  drivers/media/pci/intel/ipu3/ipu3-cio2.h |  1 +
>  2 files changed, 12 insertions(+)
> 
> diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> index afa472026ba4..9c7b527a8800 100644
> --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> @@ -1233,11 +1233,15 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
>  {
>  	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
>  
> +	mutex_lock(&q->subdev_lock);
> +
>  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
>  		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
>  	else
>  		fmt->format = q->subdev_fmt;
>  
> +	mutex_unlock(&q->subdev_lock);
> +
>  	return 0;
>  }
>  
> @@ -1261,6 +1265,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
>  	if (fmt->pad == CIO2_PAD_SOURCE)
>  		return cio2_subdev_get_fmt(sd, cfg, fmt);
>  
> +	mutex_lock(&q->subdev_lock);
> +
>  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
>  		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
>  	} else {
> @@ -1271,6 +1277,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
>  		fmt->format = q->subdev_fmt;
>  	}

Not a candidate for this patch, but this should restrict the pixel
format and size to supported values.

>  
> +	mutex_unlock(&q->subdev_lock);
> +
>  	return 0;
>  }
>  
> @@ -1529,6 +1537,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
>  
>  	/* Initialize miscellaneous variables */
>  	mutex_init(&q->lock);
> +	mutex_init(&q->subdev_lock);
>  
>  	/* Initialize formats to default values */
>  	fmt = &q->subdev_fmt;
> @@ -1646,6 +1655,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
>  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
>  fail_fbpt:
>  	mutex_destroy(&q->lock);
> +	mutex_destroy(&q->subdev_lock);
>  
>  	return r;
>  }
> @@ -1658,6 +1668,7 @@ static void cio2_queue_exit(struct cio2_device *cio2, struct cio2_queue *q)
>  	media_entity_cleanup(&q->subdev.entity);
>  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
>  	mutex_destroy(&q->lock);
> +	mutex_destroy(&q->subdev_lock);
>  }
>  
>  static int cio2_queues_init(struct cio2_device *cio2)
> diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.h b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> index 549b08f88f0c..df0247326a1d 100644
> --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> @@ -335,6 +335,7 @@ struct cio2_queue {
>  
>  	/* Subdev, /dev/v4l-subdevX */
>  	struct v4l2_subdev subdev;
> +	struct mutex subdev_lock;

Can you add a small comment to tell which field(s) this lock protects ?

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

>  	struct media_pad subdev_pads[CIO2_PADS];
>  	struct v4l2_mbus_framefmt subdev_fmt;
>  	atomic_t frame_sequence;

-- 
Regards,

Laurent Pinchart

Re: [PATCH 3/3] ipu3-cio2: Only allow setting valid mbus codes

[Laurent Pinchart]

Hi Sakari,

Thank you for the patch.

On Thu, Oct 08, 2020 at 11:47:47PM +0300, Sakari Ailus wrote:
> Check the mbus code provided by the user is one of those the driver
> supports. Ignore the code in set_fmt otherwise.

You're reading my mind :-)

The code shouldn't be ignored though, when an unsupported code is set,
it's best to use a fixed default code instead to make the driver
behaviour as stateless as possible.

> Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> ---
>  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> index 9c7b527a8800..2ea6313e00b0 100644
> --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> @@ -1270,10 +1270,17 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
>  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
>  		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
>  	} else {
> +		unsigned int i;
> +
>  		/* It's the sink, allow changing frame size */
>  		q->subdev_fmt.width = fmt->format.width;
>  		q->subdev_fmt.height = fmt->format.height;
> -		q->subdev_fmt.code = fmt->format.code;
> +		for (i = 0; i < ARRAY_SIZE(formats); i++) {
> +			if (formats[i].mbus_code == fmt->format.code) {
> +				q->subdev_fmt.code = fmt->format.code;
> +				break;
> +			}
> +		}
>  		fmt->format = q->subdev_fmt;

This should equally apply to the TRY format, we should accept an
unsupported code. I'd rework the code as follows:

	v4l2_mbus_framefmt *format;

	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
		format = v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
	else
		format = &q->subdev_fmt;

(this can be done outside of the mutex-protected section) and then
operate on format unconditionally.

Note that we should also allow changing the field and colorspace
information.

-- 
Regards,

Laurent Pinchart

Re: [PATCH 2/3] ipu3-cio2: Serialise access to pad format

[Sakari Ailus]

Hi Laurent,

On Fri, Oct 09, 2020 at 03:44:12AM +0300, Laurent Pinchart wrote:
> Hi Sakari,
> 
> Thank you for the patch.

Thanks for the review!

> 
> On Thu, Oct 08, 2020 at 11:47:46PM +0300, Sakari Ailus wrote:
> > Pad format can be accessed from user space. Serialise access to it.
> > 
> > Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> > Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> > ---
> >  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 11 +++++++++++
> >  drivers/media/pci/intel/ipu3/ipu3-cio2.h |  1 +
> >  2 files changed, 12 insertions(+)
> > 
> > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > index afa472026ba4..9c7b527a8800 100644
> > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > @@ -1233,11 +1233,15 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
> >  {
> >  	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
> >  
> > +	mutex_lock(&q->subdev_lock);
> > +
> >  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
> >  		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
> >  	else
> >  		fmt->format = q->subdev_fmt;
> >  
> > +	mutex_unlock(&q->subdev_lock);
> > +
> >  	return 0;
> >  }
> >  
> > @@ -1261,6 +1265,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
> >  	if (fmt->pad == CIO2_PAD_SOURCE)
> >  		return cio2_subdev_get_fmt(sd, cfg, fmt);
> >  
> > +	mutex_lock(&q->subdev_lock);
> > +
> >  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
> >  		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
> >  	} else {
> > @@ -1271,6 +1277,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
> >  		fmt->format = q->subdev_fmt;
> >  	}
> 
> Not a candidate for this patch, but this should restrict the pixel
> format and size to supported values.

Yes.

> 
> >  
> > +	mutex_unlock(&q->subdev_lock);
> > +
> >  	return 0;
> >  }
> >  
> > @@ -1529,6 +1537,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
> >  
> >  	/* Initialize miscellaneous variables */
> >  	mutex_init(&q->lock);
> > +	mutex_init(&q->subdev_lock);
> >  
> >  	/* Initialize formats to default values */
> >  	fmt = &q->subdev_fmt;
> > @@ -1646,6 +1655,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
> >  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
> >  fail_fbpt:
> >  	mutex_destroy(&q->lock);
> > +	mutex_destroy(&q->subdev_lock);
> >  
> >  	return r;
> >  }
> > @@ -1658,6 +1668,7 @@ static void cio2_queue_exit(struct cio2_device *cio2, struct cio2_queue *q)
> >  	media_entity_cleanup(&q->subdev.entity);
> >  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
> >  	mutex_destroy(&q->lock);
> > +	mutex_destroy(&q->subdev_lock);
> >  }
> >  
> >  static int cio2_queues_init(struct cio2_device *cio2)
> > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.h b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > index 549b08f88f0c..df0247326a1d 100644
> > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > @@ -335,6 +335,7 @@ struct cio2_queue {
> >  
> >  	/* Subdev, /dev/v4l-subdevX */
> >  	struct v4l2_subdev subdev;
> > +	struct mutex subdev_lock;
> 
> Can you add a small comment to tell which field(s) this lock protects ?

How about:

	/* Serialise access to subdev_fmt field */

Currently it's just that, but I feel locking in this driver would generally
benefit from refactoring. That can wait a little though as it's not an
acute problem.

> 
> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> 
> >  	struct media_pad subdev_pads[CIO2_PADS];
> >  	struct v4l2_mbus_framefmt subdev_fmt;
> >  	atomic_t frame_sequence;
> 
> -- 
> Regards,
> 
> Laurent Pinchart

-- 
Sakari Ailus

Re: [PATCH 3/3] ipu3-cio2: Only allow setting valid mbus codes

[Sakari Ailus]

Hi Laurent,

On Fri, Oct 09, 2020 at 03:49:49AM +0300, Laurent Pinchart wrote:
> Hi Sakari,
> 
> Thank you for the patch.
> 
> On Thu, Oct 08, 2020 at 11:47:47PM +0300, Sakari Ailus wrote:
> > Check the mbus code provided by the user is one of those the driver
> > supports. Ignore the code in set_fmt otherwise.
> 
> You're reading my mind :-)
> 
> The code shouldn't be ignored though, when an unsupported code is set,
> it's best to use a fixed default code instead to make the driver
> behaviour as stateless as possible.

I can change it to that, yes.

> 
> > Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> > Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> > ---
> >  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 9 ++++++++-
> >  1 file changed, 8 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > index 9c7b527a8800..2ea6313e00b0 100644
> > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > @@ -1270,10 +1270,17 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
> >  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
> >  		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
> >  	} else {
> > +		unsigned int i;
> > +
> >  		/* It's the sink, allow changing frame size */
> >  		q->subdev_fmt.width = fmt->format.width;
> >  		q->subdev_fmt.height = fmt->format.height;
> > -		q->subdev_fmt.code = fmt->format.code;
> > +		for (i = 0; i < ARRAY_SIZE(formats); i++) {
> > +			if (formats[i].mbus_code == fmt->format.code) {
> > +				q->subdev_fmt.code = fmt->format.code;
> > +				break;
> > +			}
> > +		}
> >  		fmt->format = q->subdev_fmt;
> 
> This should equally apply to the TRY format, we should accept an
> unsupported code. I'd rework the code as follows:
> 
> 	v4l2_mbus_framefmt *format;
> 
> 	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
> 		format = v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
> 	else
> 		format = &q->subdev_fmt;
> 
> (this can be done outside of the mutex-protected section) and then
> operate on format unconditionally.

Agreed.

> Note that we should also allow changing the field and colorspace
> information.

Indeed.

-- 
Sakari Ailus

Re: [PATCH 2/3] ipu3-cio2: Serialise access to pad format

[Laurent Pinchart]

Hi Sakari,

On Fri, Oct 09, 2020 at 12:43:47PM +0300, Sakari Ailus wrote:
> On Fri, Oct 09, 2020 at 03:44:12AM +0300, Laurent Pinchart wrote:
> > On Thu, Oct 08, 2020 at 11:47:46PM +0300, Sakari Ailus wrote:
> > > Pad format can be accessed from user space. Serialise access to it.
> > > 
> > > Fixes: c2a6a07afe4a ("media: intel-ipu3: cio2: add new MIPI-CSI2 driver")
> > > Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> > > ---
> > >  drivers/media/pci/intel/ipu3/ipu3-cio2.c | 11 +++++++++++
> > >  drivers/media/pci/intel/ipu3/ipu3-cio2.h |  1 +
> > >  2 files changed, 12 insertions(+)
> > > 
> > > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.c b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > > index afa472026ba4..9c7b527a8800 100644
> > > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.c
> > > @@ -1233,11 +1233,15 @@ static int cio2_subdev_get_fmt(struct v4l2_subdev *sd,
> > >  {
> > >  	struct cio2_queue *q = container_of(sd, struct cio2_queue, subdev);
> > >  
> > > +	mutex_lock(&q->subdev_lock);
> > > +
> > >  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY)
> > >  		fmt->format = *v4l2_subdev_get_try_format(sd, cfg, fmt->pad);
> > >  	else
> > >  		fmt->format = q->subdev_fmt;
> > >  
> > > +	mutex_unlock(&q->subdev_lock);
> > > +
> > >  	return 0;
> > >  }
> > >  
> > > @@ -1261,6 +1265,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
> > >  	if (fmt->pad == CIO2_PAD_SOURCE)
> > >  		return cio2_subdev_get_fmt(sd, cfg, fmt);
> > >  
> > > +	mutex_lock(&q->subdev_lock);
> > > +
> > >  	if (fmt->which == V4L2_SUBDEV_FORMAT_TRY) {
> > >  		*v4l2_subdev_get_try_format(sd, cfg, fmt->pad) = fmt->format;
> > >  	} else {
> > > @@ -1271,6 +1277,8 @@ static int cio2_subdev_set_fmt(struct v4l2_subdev *sd,
> > >  		fmt->format = q->subdev_fmt;
> > >  	}
> > 
> > Not a candidate for this patch, but this should restrict the pixel
> > format and size to supported values.
> 
> Yes.
> 
> > 
> > >  
> > > +	mutex_unlock(&q->subdev_lock);
> > > +
> > >  	return 0;
> > >  }
> > >  
> > > @@ -1529,6 +1537,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
> > >  
> > >  	/* Initialize miscellaneous variables */
> > >  	mutex_init(&q->lock);
> > > +	mutex_init(&q->subdev_lock);
> > >  
> > >  	/* Initialize formats to default values */
> > >  	fmt = &q->subdev_fmt;
> > > @@ -1646,6 +1655,7 @@ static int cio2_queue_init(struct cio2_device *cio2, struct cio2_queue *q)
> > >  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
> > >  fail_fbpt:
> > >  	mutex_destroy(&q->lock);
> > > +	mutex_destroy(&q->subdev_lock);
> > >  
> > >  	return r;
> > >  }
> > > @@ -1658,6 +1668,7 @@ static void cio2_queue_exit(struct cio2_device *cio2, struct cio2_queue *q)
> > >  	media_entity_cleanup(&q->subdev.entity);
> > >  	cio2_fbpt_exit(q, &cio2->pci_dev->dev);
> > >  	mutex_destroy(&q->lock);
> > > +	mutex_destroy(&q->subdev_lock);
> > >  }
> > >  
> > >  static int cio2_queues_init(struct cio2_device *cio2)
> > > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2.h b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > > index 549b08f88f0c..df0247326a1d 100644
> > > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2.h
> > > @@ -335,6 +335,7 @@ struct cio2_queue {
> > >  
> > >  	/* Subdev, /dev/v4l-subdevX */
> > >  	struct v4l2_subdev subdev;
> > > +	struct mutex subdev_lock;
> > 
> > Can you add a small comment to tell which field(s) this lock protects ?
> 
> How about:
> 
> 	/* Serialise access to subdev_fmt field */

That's perfect :-)

> Currently it's just that, but I feel locking in this driver would generally
> benefit from refactoring. That can wait a little though as it's not an
> acute problem.

Sure.

> > Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > 
> > >  	struct media_pad subdev_pads[CIO2_PADS];
> > >  	struct v4l2_mbus_framefmt subdev_fmt;
> > >  	atomic_t frame_sequence;

-- 
Regards,

Laurent Pinchart

Re: [PATCH 2/3] ipu3-cio2: Serialise access to pad format

[Andy Shevchenko]

On Fri, Oct 9, 2020 at 12:44 AM Sakari Ailus
<sakari.ailus@linux.intel.com> wrote:
>
> Pad format can be accessed from user space. Serialise access to it.

...

>         /* Initialize miscellaneous variables */
>         mutex_init(&q->lock);
> +       mutex_init(&q->subdev_lock);

...

>         mutex_destroy(&q->lock);
> +       mutex_destroy(&q->subdev_lock);

A nit: perhaps reversed order?

>         mutex_destroy(&q->lock);
> +       mutex_destroy(&q->subdev_lock);

Ditto.

-- 
With Best Regards,
Andy Shevchenko