[OpenAFS] OpenAFS 1.4.2 on latest FC5 kernel

[Marcus Watts]

Matthew Andrews <matt@slackers.net> writes:
> > looking at the relevant section in osi_machdep.h I see:
> > 
> > #elif defined (STRUCT_TASK_STRUCT_HAS_SIGNAL_RLIM)
> > #define TASK_STRUCT_RLIM signal->rlim
> > #else
> > #error Not sure what to do about rlim (should be in the Linux task
> > struct somewhere....)
> > #endif

This is a red herring.  Smell the fish.

> > 
> > which makes me think that yes that's what should be being seen, but
> > something in the autoconf detection of this is broken.

Yes.  Most likely if you check out what "configure" said, you'll find
*every* kernel test case failed.  Fix that and your rlim problem will
go away.  Unfortunate, "config.log" probably won't tell you what's
broken.

I've got a version of the kernel configuration logic that's smart
enough to see if it can build a guaranteed success test case for the
kernel before it tests actual kernel features, and works much harder in
any case to log what failed.  I don't know when that might show up for
real in openafs.  If you want to try hacking this in yourself, you
might want to look at

/afs/umich.edu/group/itd/build/mdw/openafs/patches/openafs-1.5.8-kconfig.diff

				-Marcus Watts