Re: D state processes in 2.4.18-pre7-ac3

From: Andres Salomon (dilinger@mp3revolution.net)
Date: Thu Feb 07 2002 - 14:02:50 EST


It seems reproducible, but I'm not sure yet. I first saw it (it being
D-state processes similar to what I see now; I didn't have sysrq
compiled into the kernel then) happen when I upgraded to -ac2; at the
time, I was using reiserfs on that partition instead of ext3. I thought
it might be a reiser bug, so I converted it to ext3, and upgraded to -ac3.
The conversion required a lot of LVM util running; I don't remember how
much time lapsed between finishing the conversion, and processes
hanging.

This leads me to believe it's not ext3, but either lvm or pdc202xx. Or
something common to ext3(jbd)/reiserfs..

On Thu, Feb 07, 2002 at 06:29:25PM +0000, Stephen C. Tweedie wrote:
>
> Hi,
>
> On Thu, Feb 07, 2002 at 09:53:56AM -0500, Andres Salomon wrote:
> > No messages at all, actually. I noticed my sysrq-t made it into my logs
> > (although it doesn't look all too correct..), so I've included it
> > below. The D-state processes themselves seem to be the only indication of
> > something gone wrong.
>
> >From this:
>
> > Feb 7 02:30:01 incandescent kernel: bash D C0079EB0 0 14718 27389 3494 27391 (NOTLB)
> > Feb 7 02:30:01 incandescent kernel: Call Trace: [__wait_on_buffer+106/144] [ipt_MASQUERADE:__insmod_ipt_MASQUERADE_O/lib/modules/2.4.18-pre7-ac3/kerne+-599850/96] [ipt_MASQUERADE:__insmod_ipt_MASQUERADE_O/lib/modules/2.4.18-pre7-ac3/kerne+-608890/96] [do_page_fault+355/1172] [do_page_fault+0/1172]
> > Feb 7 02:30:01 incandescent kernel: [vfs_readdir+97/144] [filldir64+0/368] [sys_getdents64+79/259] [filldir64+0/368] [sys_fcntl64+128/144] [system_call+51/56]
>
> it looks as if there are other processes stalled on IO too.
> Unfortunately the modules decoding of the call traces looks broken, so
> it's hard from this to really work out what might be the source of
> that particular stall.
>
> Is this at all reproducible? Had you been running any LVM utils
> recently?
>
> --Stephen
>
>

-- 
"I think a lot of the basis of the open source movement comes from
  procrastinating students..."
	-- Andrew Tridgell <http://www.linux-mag.com/2001-07/tridgell_04.html>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Thu Feb 07 2002 - 21:01:05 EST