Messages in this thread | | | Date | Mon, 2 Jun 2008 15:34:20 -0300 | From | Mauro Carvalho Chehab <> | Subject | Re: v4l regression: V4L2_CID_[VH]CENTER disappeared without notice. |
| |
On Mon, 2 Jun 2008 11:10:35 -0700 Brandon Philips <bphilips@suse.de> wrote:
> On 16:56 Mon 02 Jun 2008, Alan Cox wrote: > > > Please can we revert the addition of _DEPRECATED to these ioctl > > > definitions. Perhaps we can add a runtime warning if they actually get > > > used? Or a compile-time warning if we can manage that?
A runtime warning won't probably work, since the control doesn't appear at the ioctls that enumerates the supported controls, since no kernel driver uses those controls.
> > Can you clarify the problem here ? Is this compile breakage as it appears > > (See 'do not include kernel headers in user space apps') or runtime ?
The breakage appears only if you compile an userspace driver that uses those defined symbols. No in-kernel driver uses such controls. I'm not sure if there is any closed source or out-of-tree driver using they.
> It is compile breakage and was also discussed on > fedora-devel-list@redhat.com with Mauro and myself CC'd. The control > number will be reserved forever to avoid binary breakage. > > However, as Mauro noted in that conversation: "Those controls haven't > been used by any Kernel drivers for a long time (I suspect that they > were never used)." Since no driver had ever implemented the control I > figured user space application wouldn't be using it either. > > If David wants to revert the _DEPRECATED tags that is fine; we can add a > feature removal for a year from now if it gets merged.
I'm also ok on reverting the _DEPRECATED tags, but I would schedule its removal at Documentation/feature-removal-schedule.txt to a shorter time. It doesn't make much sense on keeping those unused controls for more than one kernel version. So, I think we may schedule such removal for October/2008.
I'll apply your patch on my tree and send Linus on my next pull request.
Cheers, Mauro
| |