[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Red Hat's 2.4 kernel funniness (was:Re: XFS in Fedora Core 2)

On Friday 19 December 2003 07:29 pm, Alan Cox wrote:
> All this will hopefully get easier with both 2.6 and Fedora. The 2.6 kernel
> cuts down a lot on Red Hat patching so its easier to pull in updates
> safely, while the 2.4 RH tree has chunks of 2.6 stuff in it which make life
> more interesting.

Tell me about it.

I have three Linux Media Labs LMLBT44 video capture cards.  The driver for 
these cards is based on bttv 0.9.1, and requires the kraxel v4l2 patchset for 
the 2.4 kernel, which does not apply cleanly to a Red Hatish 2.4.22 kernel. 

The patches for the LMLBT44 aren't very large, but they are very specific to 
bttv 0.9.x.  Now, Axel Thimm has packaged a v4l2 kernel for Fedora Core that 
works quite well; but when I compile bttv 0.9.11 from patched source (since 
the RPM build that Axel uses is kindof interesting in itself, and he didn't 
answer me when I asked about it) I get a driver that oopses the kernel after 
a very few seconds of full motion video capture.  It boiled down to a version 
check in the bttv headers that produces the wrong result for the Red Hatish 
2.4.22 kernel.  You end up with an IRQ stack overflow, which oopses the 
kernel.  By changing this one thing I get more stability, but things still 
aren't quite right with the LML patch and bttv 0.9.11 with Axel's v4l2 kernel 
on my dual PPro 200 testbed.

So, getting closer to an upstream kernel is a very good thing, IMO.
Lamar Owen
Director of Information Technology
Pisgah Astronomical Research Institute
1 PARI Drive
Rosman, NC  28772

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]