[[PATCH v3]] videobuf2: Add missing lock held on vb2_fop_relase

[Sylwester Nawrocki]

Hi Ricardo,

On 10/31/2013 09:54 PM, Ricardo Ribalda Delgado wrote:
> From: Ricardo Ribalda<ricardo.ribalda at gmail.com>
>
> vb2_fop_relase does not held the lock although it is modifying the
> queue->owner field.
>
> This could lead to race conditions on the vb2_perform_io function
> when multiple applications are accessing the video device via
> read/write API:
[...]
> v2: Add bug found by Sylvester Nawrocki

"v2: Add fix for a bug found..." ? :)

> fimc-capture and fimc-lite where calling vb2_fop_release with the lock held.
> Therefore a new __vb2_fop_release function has been created to be used by
> drivers that overload the release function.
>
> v3: Comments by Sylvester Nawrocki and Mauro Carvalho Chehab
>
> Use vb2_fop_release_locked instead of __vb2_fop_release

Such notes normally go after the scissors line ("---") after Signed-off-by
lines.

> Signed-off-by: Ricardo Ribalda<ricardo.ribalda at gmail.com>
> Signed-off-by: Ricardo Ribalda Delgado<ricardo.ribalda at gmail.com>

Is this duplication really needed ?

> ---

>   drivers/media/platform/exynos4-is/fimc-capture.c |  2 +-
>   drivers/media/platform/exynos4-is/fimc-lite.c    |  2 +-
>   drivers/media/v4l2-core/videobuf2-core.c         | 24 +++++++++++++++++++++++-
>   include/media/videobuf2-core.h                   |  1 +
>   4 files changed, 26 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/media/platform/exynos4-is/fimc-capture.c b/drivers/media/platform/exynos4-is/fimc-capture.c
> index fb27ff7..c3c3b3b 100644
> --- a/drivers/media/platform/exynos4-is/fimc-capture.c
> +++ b/drivers/media/platform/exynos4-is/fimc-capture.c
> @@ -549,7 +549,7 @@ static int fimc_capture_release(struct file *file)
>   		vc->streaming = false;
>   	}
>
> -	ret = vb2_fop_release(file);
> +	ret = vb2_fop_release_locked(file);

I'm personally not happy with such a change. It is still not obvious
if "locked" means that this function takes the lock internally or it
should be called with the lock held. How about sticking to the common
practice and instead naming it __vb2_fop_release() ?

>   	if (close) {
>   		clear_bit(ST_CAPT_BUSY,&fimc->state);
> diff --git a/drivers/media/platform/exynos4-is/fimc-lite.c b/drivers/media/platform/exynos4-is/fimc-lite.c
> index e5798f7..b8d417f 100644
> --- a/drivers/media/platform/exynos4-is/fimc-lite.c
> +++ b/drivers/media/platform/exynos4-is/fimc-lite.c
> @@ -546,7 +546,7 @@ static int fimc_lite_release(struct file *file)
>   		mutex_unlock(&entity->parent->graph_mutex);
>   	}
>
> -	vb2_fop_release(file);
> +	vb2_fop_release_locked(file);
>   	pm_runtime_put(&fimc->pdev->dev);
>   	clear_bit(ST_FLITE_SUSPENDED,&fimc->state);
>
> diff --git a/drivers/media/v4l2-core/videobuf2-core.c b/drivers/media/v4l2-core/videobuf2-core.c
> index 594c75e..06e6dbd 100644
> --- a/drivers/media/v4l2-core/videobuf2-core.c
> +++ b/drivers/media/v4l2-core/videobuf2-core.c
> @@ -2619,18 +2619,40 @@ int vb2_fop_mmap(struct file *file, struct vm_area_struct *vma)
>   }
>   EXPORT_SYMBOL_GPL(vb2_fop_mmap);
>
> -int vb2_fop_release(struct file *file)
> +int __vb2_fop_release(struct file *file, bool lock_is_held)
>   {
>   	struct video_device *vdev = video_devdata(file);
> +	struct mutex *lock;
>
>   	if (file->private_data == vdev->queue->owner) {
> +		if (lock_is_held)
> +			lock = NULL;
> +		else
> +			lock = vdev->queue->lock ?
> +				vdev->queue->lock : vdev->lock;
> +		if (lock)
> +			mutex_lock(lock);
>   		vb2_queue_release(vdev->queue);
>   		vdev->queue->owner = NULL;
> +		if (lock)
> +			mutex_unlock(lock);
>   	}
>   	return v4l2_fh_release(file);
>   }
> +EXPORT_SYMBOL_GPL(__vb2_fop_release);

We don't need to export this function, do we ?

> +int vb2_fop_release(struct file *file)
> +{
> +	return __vb2_fop_release(file, false);
> +}
>   EXPORT_SYMBOL_GPL(vb2_fop_release);
>
> +int vb2_fop_release_locked(struct file *file)
> +{
> +	return __vb2_fop_release(file, true);
> +}
> +EXPORT_SYMBOL_GPL(vb2_fop_release_locked);

--
Thanks,
Sylwester