Re: [PATCH/RFC] videodev.[ch] redesign

[Gerd Knorr <kraxel@bytesex.org>]

> > It also provides a ioctl wrapper function which handles copying the
> > ioctl args from/to userspace, so we have this at one place can drop all
> > the copy_from/to_user calls within the v4l device driver ioctl handlers.
>  
>  That is a large improvement.
>  But you don't include a lock against reentry, which is bad.

I don't want to handle the wrapper function too much.  IMHO it is the
job of the driver to do locking if needed.  For some read-only ioctls
like VIDIOCGCAP you don't need locking at all.

> > Comments?
>  
>  Could you make a helper for open like for ioctl ?

video_open does call video_device[minor]->fops->open(), isn't that
enought?

>  And please don't use a pointer to the device descriptor
>  in the file structure. It makes live for USB devices much harder.

Sorry, I don't understand.  What exactly do you mean?
file->private_data?  videodev.c doesn't touch it ...

  Gerd

-- 
#define	ENOCLUE 125 /* userland programmer induced race condition */