[vlc-devel] [rms at gnu.org: VLC and V4L2]

[Luis]

I'm pretty paranoid when I do a free, and I always ask first if the pointer
points to null or not before freeing it.

Maybe the driver is not correct, but I would try to check first the
destination of the pointer before freeing it.
Can you put further information about the error shown in your computer?


On 10/4/07, Richard Hosking <richard at hovis.net> wrote:
>
> The capture example is in the public domain... here
> is the working url!:-
>
> http://v4l2spec.bytesex.org/spec/capture-example.html
>
> this is linked from the home page:-
>
> http://v4l2spec.bytesex.org/
>
> I'm pretty sure the line of code calling free is
> correct in the v4l2.c file as the manufacturer
> of my v4l2 device has admitted they have an issue
> inside their driver when closing and are fast
> working on a fix for us.
>
> So I would like to wait at least a week until this
> is done so I can verify that the code in v4l2.c is
> calling the closing operations correctly.
>
> Richard.
>
> Rafaël Carré wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > Richard Hosking a écrit :
> >> I have more or less completed this.  Only mjpeg support
> >> and testing with a device which has a tuner I cannot do
> >> as my v4l2 device does not support this.
> >
> > Maybe the FSF can help here, providing you such a device ?
> > I have no idea how much it costs.
> >
> >> Currently the v4l2 device I have has a bug in its drivers
> >> which crashes when the device is closed. I am waiting for
> >> the manufacturer to fix this fault (hopefully the end of
> >> next week).  When they do this I will be able to test
> >> that the calling code works properly before submitting my
> >> v4l2.c file for hopeful inclusion into the trunk :-)
> >
> > :)
> >
> >> About 50% of the additions are taken from the reference
> >> v4l2 capture program at http://v4l2spec.bytesex.org/spec/a14009.htm
> >> (which I have cited), and about 30% is from the v4l.c
> >> program, and the rest is me.
> >
> > The reference capture program is 404 for me, is it in the public domain,
> > or has a special licence ?
> >
> >> I would rather wait until I can verify that the bug I
> >> have is fixed before submitting my v4l2.c patch, but if
> >> anyone wishes to see the current code it is here:-
> >>
> >> http://hovis.net/v4l2.c
> >>
> >> (line 651 is the call which is causing my device to
> >> segfault at the moment)
> >
> > Seems odd, did you verify you were not overflowing ?
> >
> >> Richard.
> >
> > Thanks
> >
> >> Benjamin Pracht wrote:
> >>> I would welcome any comment on this. I guess what we really lack here
> is
> >>> a developer willing to spend some time on it. Does anybody see
> anything
> >>> concrete the FSF can do to help? I remember some emails on this
> mailing
> >>> list about some people working on the v4l2 module. Did we get any
> feedback?
> >>>
> >>> Writing a full support for all the v4l2 compatible devices might be a
> >>> rather big work, but even a basic support of some devices would be a
> >>> very nice first step.
> >>>
> >>> Currently, the v4l2.c file only contains probing code. I'm not even
> sure
> >>> we really want to keep it. The pvr module probably also have useful
> >>> code for the access part.
> >>>
> >>> What remains to be done is mostly:
> >>>
> >>> - to split the module in an access and an access_demux
> >>> - complete the probing code
> >>> - write device selection, device configuration code (source, tuner
> >>>   properties, v4l chroma to vlc FOURCC currespondance, etc)
> >>> - write the data reading routines (some devices use read calls on the
> >>>   device node, some use ioctls)
> >>> - callbacks usable from a gui to get the device list (see direct show)
> >>> - probably lots of other things I forget. I worked on those more than
> 1
> >>>   year ago, not exactly sure anymore about how all this works.
> >>>
> >>> I'm afraid I won't be able to take this over. I can try to help
> anybody
> >>> having questions on the existing code, but again, it's not much and I
> >>> don't think there is a lot to say about it.
> >>>
> >>>
> >>>
> >>>
> ------------------------------------------------------------------------
> >>>
> >>> Subject:
> >>> VLC and V4L2
> >>> From:
> >>> Richard Stallman <rms at gnu.org>
> >>> Date:
> >>> Sat, 29 Sep 2007 23:04:08 -0400
> >>> To:
> >>> bigben at videolan.org
> >>>
> >>> To:
> >>> bigben at videolan.org
> >>>
> >>>
> >>> I would really like VLC to work with V4L2, so I wonder: is there
> >>> anything the FSF could do to help you make progress on this?  Please
> >>> tell me what would be useful.
> >
> >
> > - --
> > Rafaël Carré
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.4.6 (GNU/Linux)
> > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> >
> > iD8DBQFHBNNnYWCeGMCv8Q8RAp6NAJ4sgPOcz+p0kbzuXpRjs/uS2f0pUQCgsS5y
> > 4VTdO+vLqRdTfcCVc/vKSX8=
> > =4rCp
> > -----END PGP SIGNATURE-----
> > _______________________________________________
> > vlc-devel mailing list
> > To unsubscribe or modify your subscription options:
> > http://mailman.videolan.org/listinfo/vlc-devel
> >
> >
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20071004/697bfc12/attachment.html>