[PATCH 1/3] media: imx: Unregister csc/scaler only if registered

[PATCH 1/3] media: imx: Unregister csc/scaler only if registered

[Ezequiel Garcia]

The csc/scaler device pointer (imxmd->m2m_vdev) is assigned
after the imx media device v4l2-async probe completes,
therefore we need to check if the device is non-NULL
before trying to unregister it.

This can be the case if the non-completed imx media device
is unbinded (or the driver is removed), leading to a kernel oops.

Fixes: a8ef0488cc59 ("media: imx: add csc/scaler mem2mem device")
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
---
 drivers/staging/media/imx/imx-media-dev.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/staging/media/imx/imx-media-dev.c b/drivers/staging/media/imx/imx-media-dev.c
index 6d2205461e56..b6d5f844ad79 100644
--- a/drivers/staging/media/imx/imx-media-dev.c
+++ b/drivers/staging/media/imx/imx-media-dev.c
@@ -107,10 +107,14 @@ static int imx_media_remove(struct platform_device *pdev)
 
 	v4l2_info(&imxmd->v4l2_dev, "Removing imx-media\n");
 
+	if (imxmd->m2m_vdev) {
+		imx_media_csc_scaler_device_unregister(imxmd->m2m_vdev);
+		imxmd->m2m_vdev = NULL;
+	}
+
 	v4l2_async_notifier_unregister(&imxmd->notifier);
 	imx_media_unregister_ipu_internal_subdevs(imxmd);
 	v4l2_async_notifier_cleanup(&imxmd->notifier);
-	imx_media_csc_scaler_device_unregister(imxmd->m2m_vdev);
 	media_device_unregister(&imxmd->md);
 	v4l2_device_unregister(&imxmd->v4l2_dev);
 	media_device_cleanup(&imxmd->md);
-- 
2.29.2

[PATCH 2/3] media: imx: Fix csc/scaler unregister

[Ezequiel Garcia]

The csc/scaler device private struct is released by
ipu_csc_scaler_video_device_release(), which can be called
by video_unregister_device() if there are no users
of the underlying struct video device.

Therefore, the mutex can't be held when calling
video_unregister_device() as its memory may be freed
by it, leading to a kernel oops.

Fortunately, the fix is quite simple as no locking
is needed when calling video_unregister_device(): v4l2-core
already has its own internal locking, and the structures
are also properly refcounted.

Fixes: a8ef0488cc59 ("media: imx: add csc/scaler mem2mem device")
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
---
 drivers/staging/media/imx/imx-media-csc-scaler.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/drivers/staging/media/imx/imx-media-csc-scaler.c b/drivers/staging/media/imx/imx-media-csc-scaler.c
index fab1155a5958..63a0204502a8 100644
--- a/drivers/staging/media/imx/imx-media-csc-scaler.c
+++ b/drivers/staging/media/imx/imx-media-csc-scaler.c
@@ -869,11 +869,7 @@ void imx_media_csc_scaler_device_unregister(struct imx_media_video_dev *vdev)
 	struct ipu_csc_scaler_priv *priv = vdev_to_priv(vdev);
 	struct video_device *vfd = priv->vdev.vfd;
 
-	mutex_lock(&priv->mutex);
-
 	video_unregister_device(vfd);
-
-	mutex_unlock(&priv->mutex);
 }
 
 struct imx_media_video_dev *
-- 
2.29.2

[PATCH 3/3] media: imx: Clean capture unregister

[Ezequiel Garcia]

No locking is needed to call video_unregister_device(). Drop it.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
---
 drivers/staging/media/imx/imx-media-capture.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/staging/media/imx/imx-media-capture.c b/drivers/staging/media/imx/imx-media-capture.c
index c1931eb2540e..e10ce103a5b4 100644
--- a/drivers/staging/media/imx/imx-media-capture.c
+++ b/drivers/staging/media/imx/imx-media-capture.c
@@ -816,14 +816,8 @@ void imx_media_capture_device_unregister(struct imx_media_video_dev *vdev)
 	struct capture_priv *priv = to_capture_priv(vdev);
 	struct video_device *vfd = priv->vdev.vfd;
 
-	mutex_lock(&priv->mutex);
-
-	if (video_is_registered(vfd)) {
-		video_unregister_device(vfd);
-		media_entity_cleanup(&vfd->entity);
-	}
-
-	mutex_unlock(&priv->mutex);
+	media_entity_cleanup(&vfd->entity);
+	video_unregister_device(vfd);
 }
 EXPORT_SYMBOL_GPL(imx_media_capture_device_unregister);
 
-- 
2.29.2

Re: [PATCH 1/3] media: imx: Unregister csc/scaler only if registered

[Philipp Zabel]

Hi Ezequiel,

On Mon, 2020-12-28 at 09:21 -0300, Ezequiel Garcia wrote:
> The csc/scaler device pointer (imxmd->m2m_vdev) is assigned
> after the imx media device v4l2-async probe completes,
> therefore we need to check if the device is non-NULL
> before trying to unregister it.
> 
> This can be the case if the non-completed imx media device
> is unbinded (or the driver is removed), leading to a kernel oops.
> 
> Fixes: a8ef0488cc59 ("media: imx: add csc/scaler mem2mem device")
> Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> ---
>  drivers/staging/media/imx/imx-media-dev.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/staging/media/imx/imx-media-dev.c b/drivers/staging/media/imx/imx-media-dev.c
> index 6d2205461e56..b6d5f844ad79 100644
> --- a/drivers/staging/media/imx/imx-media-dev.c
> +++ b/drivers/staging/media/imx/imx-media-dev.c
> @@ -107,10 +107,14 @@ static int imx_media_remove(struct platform_device *pdev)
>  
>  	v4l2_info(&imxmd->v4l2_dev, "Removing imx-media\n");
>  
> +	if (imxmd->m2m_vdev) {

Thank you, it's even worse. If imx_media_csc_scaler_device_init() fails
in imx6_media_probe_complete(), imxmd->m2m_vdev contains an error value.

So either this should check

	if (!IS_ERR_OR_NULL(imxmd->m2m_vdev))

or (probably better) probe_complete should be changed as well.

> +		imx_media_csc_scaler_device_unregister(imxmd->m2m_vdev);
> +		imxmd->m2m_vdev = NULL;
> +	}
> +
>  	v4l2_async_notifier_unregister(&imxmd->notifier);
>  	imx_media_unregister_ipu_internal_subdevs(imxmd);
>  	v4l2_async_notifier_cleanup(&imxmd->notifier);
> -	imx_media_csc_scaler_device_unregister(imxmd->m2m_vdev);
>  	media_device_unregister(&imxmd->md);
>  	v4l2_device_unregister(&imxmd->v4l2_dev);
>  	media_device_cleanup(&imxmd->md);

regards
Philipp

Re: [PATCH 2/3] media: imx: Fix csc/scaler unregister

[Philipp Zabel]

On Mon, 2020-12-28 at 09:21 -0300, Ezequiel Garcia wrote:
> The csc/scaler device private struct is released by
> ipu_csc_scaler_video_device_release(), which can be called
> by video_unregister_device() if there are no users
> of the underlying struct video device.
> 
> Therefore, the mutex can't be held when calling
> video_unregister_device() as its memory may be freed
> by it, leading to a kernel oops.
> 
> Fortunately, the fix is quite simple as no locking
> is needed when calling video_unregister_device(): v4l2-core
> already has its own internal locking, and the structures
> are also properly refcounted.
> 
> Fixes: a8ef0488cc59 ("media: imx: add csc/scaler mem2mem device")
> Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> ---
>  drivers/staging/media/imx/imx-media-csc-scaler.c | 4 ----
>  1 file changed, 4 deletions(-)
> 
> diff --git a/drivers/staging/media/imx/imx-media-csc-scaler.c b/drivers/staging/media/imx/imx-media-csc-scaler.c
> index fab1155a5958..63a0204502a8 100644
> --- a/drivers/staging/media/imx/imx-media-csc-scaler.c
> +++ b/drivers/staging/media/imx/imx-media-csc-scaler.c
> @@ -869,11 +869,7 @@ void imx_media_csc_scaler_device_unregister(struct imx_media_video_dev *vdev)
>  	struct ipu_csc_scaler_priv *priv = vdev_to_priv(vdev);
>  	struct video_device *vfd = priv->vdev.vfd;
>  
> -	mutex_lock(&priv->mutex);
> -
>  	video_unregister_device(vfd);
> -
> -	mutex_unlock(&priv->mutex);
>  }
>  
>  struct imx_media_video_dev *

Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>

regards
Philipp

Re: [PATCH 3/3] media: imx: Clean capture unregister

[Philipp Zabel]

On Mon, 2020-12-28 at 09:21 -0300, Ezequiel Garcia wrote:
> No locking is needed to call video_unregister_device(). Drop it.

The commit message should probably also mention that the superfluous
video_is_registered() check is dropped and media_entity_cleanup() and
video_unregister_device() are reordered.

> Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> ---
>  drivers/staging/media/imx/imx-media-capture.c | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/staging/media/imx/imx-media-capture.c b/drivers/staging/media/imx/imx-media-capture.c
> index c1931eb2540e..e10ce103a5b4 100644
> --- a/drivers/staging/media/imx/imx-media-capture.c
> +++ b/drivers/staging/media/imx/imx-media-capture.c
> @@ -816,14 +816,8 @@ void imx_media_capture_device_unregister(struct imx_media_video_dev *vdev)
>  	struct capture_priv *priv = to_capture_priv(vdev);
>  	struct video_device *vfd = priv->vdev.vfd;
>  
> -	mutex_lock(&priv->mutex);
> -
> -	if (video_is_registered(vfd)) {
> -		video_unregister_device(vfd);
> -		media_entity_cleanup(&vfd->entity);
> -	}
> -
> -	mutex_unlock(&priv->mutex);
> +	media_entity_cleanup(&vfd->entity);
> +	video_unregister_device(vfd);
>  }
>  EXPORT_SYMBOL_GPL(imx_media_capture_device_unregister);

regards
Philipp

Re: [PATCH 1/3] media: imx: Unregister csc/scaler only if registered

[Ezequiel Garcia]

Hello Philipp,

Thanks for the quick review.

On Mon, 2021-01-04 at 18:16 +0100, Philipp Zabel wrote:
> Hi Ezequiel,
> 
> On Mon, 2020-12-28 at 09:21 -0300, Ezequiel Garcia wrote:
> > The csc/scaler device pointer (imxmd->m2m_vdev) is assigned
> > after the imx media device v4l2-async probe completes,
> > therefore we need to check if the device is non-NULL
> > before trying to unregister it.
> > 
> > This can be the case if the non-completed imx media device
> > is unbinded (or the driver is removed), leading to a kernel oops.
> > 
> > Fixes: a8ef0488cc59 ("media: imx: add csc/scaler mem2mem device")
> > Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> > ---
> >  drivers/staging/media/imx/imx-media-dev.c | 6 +++++-
> >  1 file changed, 5 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/staging/media/imx/imx-media-dev.c b/drivers/staging/media/imx/imx-media-dev.c
> > index 6d2205461e56..b6d5f844ad79 100644
> > --- a/drivers/staging/media/imx/imx-media-dev.c
> > +++ b/drivers/staging/media/imx/imx-media-dev.c
> > @@ -107,10 +107,14 @@ static int imx_media_remove(struct platform_device *pdev)
> >  
> >         v4l2_info(&imxmd->v4l2_dev, "Removing imx-media\n");
> >  
> > +       if (imxmd->m2m_vdev) {
> 
> Thank you, it's even worse. If imx_media_csc_scaler_device_init() fails
> in imx6_media_probe_complete(), imxmd->m2m_vdev contains an error value.
> 
> So either this should check
> 
>         if (!IS_ERR_OR_NULL(imxmd->m2m_vdev))
> 
> or (probably better) probe_complete should be changed as well.
> 

Good catch. I guess the simplest solution for that is to
clear the m2m_vdev pointer in imx6_media_probe_complete,
if it fails.

Thanks,
Ezequiel