[PATCH] v4l: omap3isp: Move code out of mutex-protected section

[PATCH] v4l: omap3isp: Move code out of mutex-protected section

[Laurent Pinchart]

The pad::get_fmt call must be protected by a mutex, but preparing its
arguments doesn't need to be. Move the non-critical code out of the
mutex-protected section.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
---
 drivers/media/platform/omap3isp/ispvideo.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/drivers/media/platform/omap3isp/ispvideo.c b/drivers/media/platform/omap3isp/ispvideo.c
index a908d00..f6304bb 100644
--- a/drivers/media/platform/omap3isp/ispvideo.c
+++ b/drivers/media/platform/omap3isp/ispvideo.c
@@ -339,14 +339,11 @@ __isp_video_get_format(struct isp_video *video, struct v4l2_format *format)
 	if (subdev == NULL)
 		return -EINVAL;
 
-	mutex_lock(&video->mutex);
-
 	fmt.pad = pad;
 	fmt.which = V4L2_SUBDEV_FORMAT_ACTIVE;
-	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
-	if (ret == -ENOIOCTLCMD)
-		ret = -EINVAL;
 
+	mutex_lock(&video->mutex);
+	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
 	mutex_unlock(&video->mutex);
 
 	if (ret)
-- 
Regards,

Laurent Pinchart

Re: [PATCH] v4l: omap3isp: Move code out of mutex-protected section

[Sakari Ailus]

Hi Laurent,

Thanks for the patch.

On Mon, Nov 04, 2013 at 11:07:48AM +0100, Laurent Pinchart wrote:
> The pad::get_fmt call must be protected by a mutex, but preparing its
> arguments doesn't need to be. Move the non-critical code out of the
> mutex-protected section.
> 
> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> ---
>  drivers/media/platform/omap3isp/ispvideo.c | 7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/media/platform/omap3isp/ispvideo.c b/drivers/media/platform/omap3isp/ispvideo.c
> index a908d00..f6304bb 100644
> --- a/drivers/media/platform/omap3isp/ispvideo.c
> +++ b/drivers/media/platform/omap3isp/ispvideo.c
> @@ -339,14 +339,11 @@ __isp_video_get_format(struct isp_video *video, struct v4l2_format *format)
>  	if (subdev == NULL)
>  		return -EINVAL;
>  
> -	mutex_lock(&video->mutex);
> -
>  	fmt.pad = pad;
>  	fmt.which = V4L2_SUBDEV_FORMAT_ACTIVE;
> -	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
> -	if (ret == -ENOIOCTLCMD)
> -		ret = -EINVAL;

By removing these lines, you're also returning -ENOIOCTLCMD to the caller.
Is this intentional?

That return value will end up to at least one place which seems to be
isp_video_streamon() and, unless I'm mistaken, will cause
ioctl(VIDIOC_STREAMON) also return ENOTTY.

> +	mutex_lock(&video->mutex);
> +	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
>  	mutex_unlock(&video->mutex);
>  
>  	if (ret)

-- 
Regards,

Sakari Ailus
e-mail: sakari.ailus@iki.fi	XMPP: sailus@retiisi.org.uk

Re: [PATCH] v4l: omap3isp: Move code out of mutex-protected section

[Laurent Pinchart]

Hi Sakari,

On Monday 04 November 2013 13:20:11 Sakari Ailus wrote:
> Hi Laurent,
> 
> Thanks for the patch.
> 
> On Mon, Nov 04, 2013 at 11:07:48AM +0100, Laurent Pinchart wrote:
> > The pad::get_fmt call must be protected by a mutex, but preparing its
> > arguments doesn't need to be. Move the non-critical code out of the
> > mutex-protected section.
> > 
> > Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > ---
> > 
> >  drivers/media/platform/omap3isp/ispvideo.c | 7 ++-----
> >  1 file changed, 2 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/media/platform/omap3isp/ispvideo.c
> > b/drivers/media/platform/omap3isp/ispvideo.c index a908d00..f6304bb
> > 100644
> > --- a/drivers/media/platform/omap3isp/ispvideo.c
> > +++ b/drivers/media/platform/omap3isp/ispvideo.c
> > @@ -339,14 +339,11 @@ __isp_video_get_format(struct isp_video *video,
> > struct v4l2_format *format)> 
> >  	if (subdev == NULL)
> >  	
> >  		return -EINVAL;
> > 
> > -	mutex_lock(&video->mutex);
> > -
> > 
> >  	fmt.pad = pad;
> >  	fmt.which = V4L2_SUBDEV_FORMAT_ACTIVE;
> > 
> > -	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
> > -	if (ret == -ENOIOCTLCMD)
> > -		ret = -EINVAL;
> 
> By removing these lines, you're also returning -ENOIOCTLCMD to the caller.
> Is this intentional?
> 
> That return value will end up to at least one place which seems to be
> isp_video_streamon() and, unless I'm mistaken, will cause
> ioctl(VIDIOC_STREAMON) also return ENOTTY.

I should have split this in two patches, or at least explained the rationale 
in the commit message. The remote subdev is always an internal ISP subdev, the 
pad::get_fmt operation is thus guaranteed to be implemented. There's no need 
to check for ENOIOCTLCMD.

> > +	mutex_lock(&video->mutex);
> > +	ret = v4l2_subdev_call(subdev, pad, get_fmt, NULL, &fmt);
> > 
> >  	mutex_unlock(&video->mutex);
> >  	
> >  	if (ret)
-- 
Regards,

Laurent Pinchart

Re: [PATCH] v4l: omap3isp: Move code out of mutex-protected section

[Sakari Ailus]

On Mon, Nov 04, 2013 at 02:17:53PM +0100, Laurent Pinchart wrote:
> > That return value will end up to at least one place which seems to be
> > isp_video_streamon() and, unless I'm mistaken, will cause
> > ioctl(VIDIOC_STREAMON) also return ENOTTY.
> 
> I should have split this in two patches, or at least explained the rationale 
> in the commit message. The remote subdev is always an internal ISP subdev, the 
> pad::get_fmt operation is thus guaranteed to be implemented. There's no need 
> to check for ENOIOCTLCMD.

Right; true.

If you add an explanation on that to the commit message (which I think is
much less self-evident than access to the local struct not requiring
serialisation),

Acked-by: Sakari Ailus <sakari.ailus@iki.fi>

-- 
Regards,

Sakari Ailus
e-mail: sakari.ailus@iki.fi	XMPP: sailus@retiisi.org.uk