[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lock up in ext2fs?

From: David Walter
Subject: Re: lock up in ext2fs?
Date: Sun, 09 Jun 2002 07:46:18 -0400
User-agent: Gnus/5.090007 (Oort Gnus v0.07) XEmacs/21.4 (Common Lisp, i386-debian-linux)

Roland McGrath <roland@frob.com> writes:

>> I  think that   I  have  found  a   repeatable assertion   failure, 
> But you haven't said how to repeat it.

One way is to run something similar.

I am not sure if this is a process creation / threading problem, or if
it is something to do with the number of queued requests for writes.

while [ $i -lt 200 ] ; 
        mail -s TEST david@localhost < .emacs ; 
        i=`expr $i + 1`; 
        echo $i iterations; 

This  is creating  several pending processes   as I  have procmail and
spamassassin installed. I am trying  to get a  system working with the
same tools in the Hurd and Linux by installing the same programs.

>> libc/hurd/intr-msg.c That is I have had two  hangs with this assertion
>> printed on the console. 
>> ext2fs.static: intr-msg.c:304:_hurd_intr_rpc_mach_msg:Assertion
>>                `m->header.msgh_id == msgid + 100` failed.
> See if you can produce this failure in a non-bootstrap filesystem or in the
> bootstrap filesystem of a sub-hurd.  Then you can use gdb on it and we can
> see the full details.

I have created   a separate boot  partition/slice and   home directory
mirroring the programs and data where the problem was created. 

I boot using the following:

settrans -fg /mnt /hurd/ext2fs --readonly /dev/hd0s12
boot -d -D /mnt -I /mnt/boot/servers.boot /dev/hd0s12

hd0s12 has a translator for the duplicate /home slice.

I ran it several times.

First I am getting an  invalid ipc message (sorry  I have to duplicate
this again to get the correct message, at first I was managing to lock
up the machine completely still).

When  attaching with  gdb to  the process  for ext2fs,  the processing
stops unless I do continue.

There are a  lot of threads  with  seemingly the same  state(the where
command gives the  same stack), but I  haven't found out how to  trace
this without stopping the process.

When  I  issue continue I  can't  get back control  of  gdb before the
process hangs.

I will try again this evening (US/EST).

Suggestions welcome.


Hope springs eternal!

/ \

reply via email to

[Prev in Thread] Current Thread [Next in Thread]