> hi,
>
> i just got a new toy - 400gb hardware raid. so for the fun of testing it i
> made a partition of rougly 90gb and created an xfs on this partition which
> work nicely and much faster than mke2fs ;-)
> tho only problem is i cannot mount it. trying to do so just gives an ugly
> oops:
>
> Unable to handle kernel NULL pointer dereference at virtual address 00000000
> *pde = 00000000
> Oops: 0002
> CPU: 0
> EIP: 0010:[<c88d6442>]
> EFLAGS: 00010246
> eax: 00000000 ebx: 00035c00 ecx: 0000d700 edx: c024d05c
> esi: 00000000 edi: 00000000 ebp: 00000001 esp: c2d75a84
> ds: 0018 es: 0018 ss: 0018
> Process mount (pid: 184, stackpage = c2d75000)
> Stack: 000005eb c8779000 c9779344 c88af5d5 00035c00 00000000 000005eb c88e31e
c
> c88bd340 c0799700 00000811 00000000 c612ee69 c0779000 00000001 0000000
0
> 0000000a ffffffff 00000000 00000002 00001000 c889e2cc c028d4c0 c077917
c
> Call Trace: [<c88af5d5>] [<c88e31ec>] [<c88bd340>] [<c889e2cc>] [<c88de83e>]
> [<c88de83d>] [<c88bd893>] [<c88c5a24>] [<c88c5c59>] [<c88e9640>]
> [<c88e9640>] [<c88c5d37>] [<c88e9640>] [<c88c5d14>] [<c88d6897>]
> [<c88e9640>] [<c0135628>] [<c0206a47>] [<c0142a6e>] [<c0137564>]
> [<c88e3768>] [<c013aa48>] [<c0135628>] [<c0135c54>] [<c88e3768>]
> [<c88e3768>] [<c88e3768>] [<c0136145>] [<c88e23f0>] [<c01361e6>]
> [<c0109650>]
> Code: f3 ab f6 c3 02 74 02 66 ab f6 c3 01 74 01 aa 89 f0 5b 5e 5f
> Segmentation fault.
>
> this happened using the latest sources from the cvs tree.
>
> do we have a limit for the filesystem size?
>
> bernd
> --
> Death is happy, death is cool
> Death can be a useful tool
> Use a gun or use a knife
> Death can really change your life
> _________________________________________________________________________
> Bernd Markgraf Otto-von-Guericke-Universitae
t
> markgraf@xxxxxxxxxxxxxxxxxxxxxxxx Magdeburg
> http://www.cs.uni-magdeburg.de/~markgraf Germany
Hmm, any chance of at least a ksymoops conversion of that stack, or
can you build kdb into the kernel and get a real stack back trace out of
it?
There should not be a limitation in size - at least not that 400 Gbytes
is capable of hitting. We will trip up at 2 Tbytes on linux due to
inode numbers getting bigger than 32 bit.
Thanks
Steve
|