Hi Anand,
in the file uio.c, in the function kissone(), make the following
change:
if (lp->bed != NULL)
{
lp->pid = current->pid;
wake_up(&lp->bed);
/* lp->bed = NULL; */
}
The variable lp->bed is read and modified by the process that is
woken up. It should be left alone.
hope this helps.
Mike
Anand Ranganathan wrote:
>
> Hi,
> I am an M.S. student at Univ of Utah and am using userfs-0.9.6 on
> linux-2.0.36 kernel to develop a research file system (basically linux
> ext2fs running as a user-level process using distributed storage).
> However, when I stress this file system a bit (such as when running the
> Andrew benchmark which does a make on a file tree), the userfs kernel
> module is generating a general protection fault in the sleep_on routine
> causing everything related to userfs to hang. The rest of the system
> works fine except when I do a sync etc. I attach a small piece of debug
> output from the userfs kernel module
>
> I can generate and send more information/trace logs etc if you could
> tell me what would be helpful.
>
> Now can you please answer the following questions for me?
> 1. Has this problem been seen elsewhere? I searched the web but couldn't
> find any.
> 2. What could be happening here?
> 3. Are there any new versions of userfs I can try out? Is any new
> development work happening on userfs anymore? If not, can you suggest
> any alternatives?
>
> Anxiously awaiting your response.
> Anand
>
> userfs kernel module debug output:
> May 11 09:10:34 wren kernel: suckin: pid 706, seq 1701 got packet seq 1701, sz=4
> May 11 09:10:34 wren kernel: kissone: pid 706: kissing seq=-1, why=3
> May 11 09:10:34 wren kernel: suckin: sucker found its own (seq=1701), more=003d9
> df8
> May 11 09:10:34 wren kernel: userfs_doop: more=003d9df8
> May 11 09:10:34 wren kernel: userfs_sleep: sleeping on list, pid=706 seq=1703 be
> d=0
> May 11 09:10:35 wren kernel: general protection: 0000
> May 11 09:10:35 wren kernel: CPU: 0
> May 11 09:10:35 wren kernel: EIP: 0010:[sleep_on+96/124]
> May 11 09:10:35 wren kernel: EFLAGS: 00010096
> May 11 09:10:35 wren kernel: eax: f000ff53 ebx: 07cc1e64 ecx: 07cc1e2c edx
> : f000ff53
> May 11 09:10:35 wren kernel: esi: 00000297 edi: 07cc1e2c ebp: 07cc1e34 esp
> : 07cc1e20
> May 11 09:10:35 wren kernel: ds: 0018 es: 0018 fs: 002b gs: 002b ss: 001
> 8
>
> --
> Anand Ranganathan [EMAIL PROTECTED]
> Random Wodehouse nugget:
> I suppose I'm one of those fellows my father always warned me
> against.
