Re: Does XFS require DevFs?

[Seth Mos <knuffie@xxxxxxxxx>]

At 13:34 15-6-2001 +1000, Keith Owens wrote:
> >or gives me a kernel
> >that produces random segfaults, oopsen, and fs corruption.
>
> Sounds like the build is doing something wrong.

Or some of the hardware is failing. If you ran 2.2 kernels before you may 
be in for a suprise. 2.4 pushes harder against the hardware which will show 
on flaky or not 100 percent hardware.

> >    Often I get oopses in rc.sysinit at about the time it tries to
> >activate quota support -- do I need quota support built in?
>
> Not sure.  I have run XFS without any quota code at all and it works
> but you might have a problem with /sbin/quotan.  OTOH it could be a
> build problem.

I do suspect that.

> >    I'm not exactly new at building kernels, but I'm no coder, and my
> >inability to build a working kernel with XFS support is frustrating me
> >to say the least.  Should I post ksymoops output, am I doing something
> >obviously wrong, or should I provide more [detailed | organized]
> >information?  Any help would be greatly appreciated.
>
> Start with information about the failing build.  I would not even look
> at run time failures until you are sure that the build is correct.

Make sure you uncomment the CC = $(CROSS_COMPILE)kgcc line in Makefile to 
make sure you are using the right compiler.

People have reported that a kernel built with the standard gcc on rh7.x can 
fail in unexpected yet subtle ways. Thinks like fs corruption and crashes 
after trying to use that corrupt data.

Run a xfs_repair on your fs to see if it's still in one piece.
xfs_repair -n can be run on a mounted filesystem to see if it wants to fix 
something.

Good luck.

--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.