On Thursday 15 November 2007 16:43:05 Mark Paulus wrote:
> Hans Verkuil wrote:
> > Hi all,
> >
> > A long standing problem with ivtv (and v4l drivers in general) is
> > that sometimes i2c devices are misidentified. So the driver thinks
> > that at a certain i2c address a tuner is connected, when in reality
> > it is a audio muxer or whatever. The root cause was a problem in
> > the i2c subsystem of the kernel which made it very hard to tell the
> > i2c subsystem what to expect at which address.
> >
> > A new way of handling the i2c bus was added in kernel 2.6.22.
> > However, in order to actually make use of this (and so finally fix
> > this issue once and for all) all i2c drivers that are used by ivtv
> > had to be converted first and only then could ivtv itself be
> > changed. Since ivtv uses some 15 different i2c devices (each board
> > uses a subset of 1-4/5 devices) this was a substantial amount of
> > work. And it's also a record: no other driver has to support that
> > many i2c devices.
> >
> > I've finally converted all the i2c drivers (they are now part of
> > the v4l-dvb repository) and have finished converting ivtv itself.
> > But before I ask the v4l-dvb maintainer to pull my ivtv changes I
> > want people to test it first. If I've made a mistake here, then
> > devices suddenly won't work anymore, so it's rather important that
> > I let people test first. Especially for non-standard boards like
> > Japanese variants and other non-Hauppauge cards.
> >
> > It is also important that I know I haven't broken support for older
> > pre-2.6.22 kernels, so if you have an older kernel then please
> > still test it!
> >
> > So please help me out by downloading my ivtv-i2c tree here:
> > http://linuxtv.org/hg/~hverkuil/ivtv-i2c/archive/tip.tar.bz2
> >
> > Unpack, run 'make' and 'make install' and see if everything still
> > works after loading the ivtv driver. Just in case, you might want
> > to make a copy of /lib/modules/2.6.XXX first. You never know :-)
> >
> > Whether it works or not, please post (or mail) the ivtv
> > initialization messages from the kernel log so that I know which
> > devices are tested.
> >
> > If nothing breaks, then I can merge the code in the v4l-dvb
> > repository and it will get into 2.6.25.
> >
> > I've CC-ed Trev in the hope that he can test the adaptec cards and
> > tadachi in the hope that he can test (or let other people test) the
> > various Asian boards.
> >
> > Also several people had conflicts between a DVB card and the ivtv
> > driver: these conflicts should now be solved as well.
>
> I was playing with the new drivers last night, and noticed an oddity
> on my Avermedia A180 card (SAA7134 based, used for QAM256). I was
> having issues getting a LOCK on one frequency. So, I booted back to
> the old modules, and got a LOCK right away as expected.
>
> The dmesg output should be the same as what I posted originally. I
> have added the dmesg output from the boot that allows the card to
> work as expected.
>
> What I did was create a copy of my /lib/modules/`uname -r`.orig.
> Then I installed your experimental drivers, and rebooted. I have a
> link in /lib/modules, so I can rapidly switch module base and reboot,
> to see the results of modules.
>
> In addition, I have attached the output from the azap sessions.
Hi Mark,
You should probably try to use the latest v4l-dvb master repository and
see if this problem has been fixed already. If not, you should report
it to the video4linux mailinglist. The ivtv-i2c tree also has bleeding
edge versions of other drivers, so bugs are not impossible. But it's
not something ivtv introduced.
Thanks for testing!
Regards,
Hans
_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel
