2.6.4: Oops when mounting HFS on ppc32

2.6.4: Oops when mounting HFS on ppc32

[Michel Roelofs]

Kernel: plain 2.6.4 from ftp.nl.kernel.org

When mounting an hfs filesystem on my ppc32 (powertower pro 250, 604e
cpu) I got the following:

kernel BUG in grow_buffers at fs/buffer.c:1191!
Oops: Exception in kernel mode, sig: 5 [#1]
NIP: C005D21C LR: C005D1DC SP: D8013C50 REGS: d8013ba0 TRAP: 0700    Not
tainted
MSR: 00029032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK = d9ac6a00[1246] 'mount' Last syscall: 21 
GPR00: 000001FF D8013C50 D9AC6A00 00000000 00000000 00000003 00000000
00000000 
GPR08: 0000001C 00000000 00000000 00000200 22004444 100289FC 00000000
00000000 
GPR16: 00000000 00000000 00000000 100DB688 00000000 00000000 7FFFFC56
00000000 
GPR24: 00000000 7FFFFC60 00000000 00000200 00000000 DBFF3520 00000000
DB73A520 
Call trace:
 [c005d880] __getblk+0x5c/0x64
 [c005d8e0] __bread+0x10/0x40
 [c00cfdf8] hfs_mdb_get+0xac/0x6a0
 [c00d1248] hfs_fill_super+0xc4/0x1a8
 [c0062c58] get_sb_bdev+0x120/0x178
 [c00d1344] hfs_get_sb+0x18/0x28
 [c0062f28] do_kern_mount+0x64/0x120
 [c007cbe8] do_add_mount+0x9c/0x1d0
 [c007cf70] do_mount+0x148/0x178
 [c007d480] sys_mount+0xe8/0x15c
 [c000785c] ret_from_syscall+0x0/0x44

The mount then fails, with the message "Trace/breakpoint trap".

Re: 2.6.4: Oops when mounting HFS on ppc32

[Roman Zippel]

Hi,

On Wed, 24 Mar 2004, Michel Roelofs wrote:

> Kernel: plain 2.6.4 from ftp.nl.kernel.org
>
> When mounting an hfs filesystem on my ppc32 (powertower pro 250, 604e
> cpu) I got the following:
>
> kernel BUG in grow_buffers at fs/buffer.c:1191!

The problem is that this happens only on Macs, it doesn't seem to be a big
endian problem, as I've just tested it on a different ppc32 machine and it
works fine. So I need someone who at least tries to figure out what
happens, before I can fix this rather mysterious bug. All I know is that
something gets corrupted, but I have no idea what.

bye, Roman