xfs
[Top] [All Lists]

Re: status of Linux on Alpha?

To: Ivan Kokshaysky <ink@xxxxxxxxxxxxxxxxxxxx>
Subject: Re: status of Linux on Alpha?
From: Marc Giger <gigerstyle@xxxxxx>
Date: Tue, 27 Apr 2004 22:42:31 +0200
Cc: Dru <andru@xxxxxxxxxxx>, linux-xfs@xxxxxxxxxxx, Måns Rullgård <mru@xxxxxx>, linux-kernel@xxxxxxxxxxxxxxx
In-reply-to: <20040428002018.A827@xxxxxxxxxxxxxxx>
References: <yw1xr7vcn1z2.fsf@xxxxxxxxxx> <20040329205233.5b7905aa@xxxxxxxxxxxxxxxxxx> <20040404121032.7bb42b35@xxxxxxxxxxxxxxxxxx> <20040409134534.67805dfd@xxxxxxxxxxxxxxxxxx> <20040409134828.0e2984e5@xxxxxxxxxxxxxxxxxx> <20040409230651.A727@xxxxxxxxxxxxxxx> <20040413194907.7ce8ceb7@xxxxxxxxxxxxxxxxxx> <20040427185124.134073cd@xxxxxxxxxxxxxxxxxx> <20040427215514.A651@xxxxxxxxxxxxxxx> <20040427200830.3f485a54@xxxxxxxxxxxxxxxxxx> <20040428002018.A827@xxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Wed, 28 Apr 2004 00:20:18 +0400
Ivan Kokshaysky <ink@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Tue, Apr 27, 2004 at 08:08:30PM +0200, Marc Giger wrote:
> > I wonder why it happens only with the XFS code. What I saw
> > rw_sem is used all over the place in the kernel.
> 
> Dru says it happens with ext3 as well. XFS folks used their own
> locking code (which hasn't suffered from that bug) until 2.6.4,
> that's why you noticed the difference...

Yes, as I saw that the patch uses the semaphore code in "arch" I was
not sure any longer if it is really a XFS related bug.

> In either case, you need _really_ heavy write IO activity to
> trigger the bug.

I noticed that. The best way to trigger it was "make -j20 vmlinux"
so that the pdflushd comes strongly into action.

Perhaps Dru can show me his "easy way" to reproduce the problem so that
I can test it more easily.

Thanks again

Marc


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