Joakim Bodin wrote:
>
> I updated xfs cvs just now (1 March 23:30 GMT) and just after rebooting
> kupdate segfaulted on me and resultet this attached Call Trace. I'm no
> kernel expert and I don't know if this sayes much. I could probably get
> more info about this if given the prober pointers. This is a up to date
> Linux-Mandrake cooker system and the kernel was compiled with RH7.0
> kgcc. I'm attaching my dmesg log as well.
>
I strongly suspect you are getting a bh value of NULL here,
in fs/pagebuf/page_buf_io.c:
1066 } else {
1067 __pb_block_commit_write_async(inode, page, NULL, 0);
1068
1069 /*
1070 * Kick-start that last write...
1071 */
1072 atomic_inc(&bh->b_count);
1073 ll_rw_block(WRITE, 1, &bh);
1074 atomic_dec(&bh->b_count);
1075 }
Can you please put a debug line in here, just before the "Kick-start ..."
comment?
if (!bh || page->buffers != bh) {
printk("Inconsistent page buffers 0x%p bh 0x%p\n"
bh, page->buffers);
}
thanks,
ananth.
--------------------------------------------------------------------------
Rajagopal Ananthanarayanan ("ananth")
Member Technical Staff, SGI.
--------------------------------------------------------------------------
|