[Top] [All Lists]

Re: again xfstests - now test 076

To: Roman Vasylyev <romis@xxxxxxxxxxxxx>
Subject: Re: again xfstests - now test 076
From: Nathan Scott <nathans@xxxxxxx>
Date: Sat, 25 Oct 2003 14:37:59 +1000
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <3F98FD53.9090407@xxxxxxxxxxxxx>; from romis@xxxxxxxxxxxxx on Fri, Oct 24, 2003 at 02:22:11PM +0400
References: <3F98FD53.9090407@xxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
On Fri, Oct 24, 2003 at 02:22:11PM +0400, Roman Vasylyev wrote:
> 10/13/2003 i wrote about 013 kernel warnings.
> After turning off debug option in XFS module kernel warnings gone. But 
> 013 test sometimes is hangs on. And i find another test which users 
> fsstress - 076(Test blockdev reads in parallel with file system 
> reads/writes).
> But if 013 test is passed 076 always fails. In attached file result log. 
> Can some who say is that critical for using XFS.

Yes, 076 is a known issue (hence its not in "auto" group - see the
-g option to the check script).  I have a bunch of pagebuf locking
changes which address this issue for small block size filesystems,
but afaik its OK with page-sized block sizes... what architecture 
are you testing here? (what is your page size?).

It is a good idea to enable kdb (and debug) while running the tests,
when you see the 013 hang again I would be interested in seeing your
kdb trace.  You do have a relatively small device for you're testing,
so you may be hitting the out-of-space hangs some other people have
seen recently..



<Prev in Thread] Current Thread [Next in Thread>