2.6.8.1-mm4 oops

2.6.8.1-mm4 oops

[Pierre Etchemaite]

	Hi all,

Just gave kernel 2.6.8.1-mm4's Reiserfs4 support a try on my home box :)

As a test, I tried to import Linux kernel 2.6.0 source tree into a brand new
Subversion repository (Just in case you don't
know about Subversion yet, try http://subversion.tigris.org/, it's a
very nice revision control system).

Importing was much slower than expected, so I straced svn. The problem was
easy to understand: being based on BerkeleyDB4 transactions, it was doing
fsync() all around, and I remember reading in this list that Reiserfs4's
fsync()s [well, was it sync() ?] was currently exiting synchronously, which
is not the traditional behavior of this call.

Ok, nevermind if this was a bad speed test ;) I ^C'ed strace (during a
fsync() I think, since most of svn's time was spent there), and all
the sudden OOPses started scrolling in logs' window. After a few seconds, I
had to reset the system.

Too bad I couldn't find anything in the logs after reboot, but the problem
occured on first try, so I suspect it will be easy to reproduce.

Good luck!
Pierre.

Re: 2.6.8.1-mm4 oops

[Alex Zarochentsev]

On Sat, Aug 28, 2004 at 05:07:21PM +0200, Pierre Etchemaite wrote:
> 
> 	Hi all,
> 
> Just gave kernel 2.6.8.1-mm4's Reiserfs4 support a try on my home box :)
> 
> As a test, I tried to import Linux kernel 2.6.0 source tree into a brand new
> Subversion repository (Just in case you don't
> know about Subversion yet, try http://subversion.tigris.org/, it's a
> very nice revision control system).
> 
> Importing was much slower than expected, so I straced svn. The problem was
> easy to understand: being based on BerkeleyDB4 transactions, it was doing
> fsync() all around, and I remember reading in this list that Reiserfs4's
> fsync()s [well, was it sync() ?] was currently exiting synchronously, which
> is not the traditional behavior of this call.
> 
> Ok, nevermind if this was a bad speed test ;) I ^C'ed strace (during a
> fsync() I think, since most of svn's time was spent there), and all
> the sudden OOPses started scrolling in logs' window. After a few seconds, I
> had to reset the system.

did you enable reiser4 debugging options?

> 
> Too bad I couldn't find anything in the logs after reboot, but the problem
> occured on first try, so I suspect it will be easy to reproduce.
> 
> Good luck!
> Pierre.

thanks for the report.


-- 
Alex.