RE: [patch 0/5] [PATCH,RFC] vfs: per-superblock unused dentries list (2nd version)

["Ravinandan Arakali" <ravinandan.arakali@neterion.com>]

Andrew,
This is a known problem and has been fixed in our internal source tree. We
will be submitting the patch soon.

Ravi

-----Original Message-----
From: Andrew Morton [mailto:akpm@osdl.org]
Sent: Monday, June 19, 2006 4:01 AM
To: tglx@linutronix.de; Ravinandan Arakali
Cc: dgc@sgi.com; mingo@elte.hu; neilb@suse.de; jblunck@suse.de;
linux-kernel@vger.kernel.org; linux-fsdevel@vger.kernel.org;
viro@zeniv.linux.org.uk; balbir@in.ibm.com
Subject: Re: [patch 0/5] [PATCH,RFC] vfs: per-superblock unused dentries
list (2nd version)


On Mon, 19 Jun 2006 12:48:44 +0200
Thomas Gleixner <tglx@linutronix.de> wrote:

> On Sun, 2006-06-18 at 23:33 -0700, Andrew Morton wrote:
> > > ....
> > >     eth3      device: S2io Inc. Xframe 10 Gigabit Ethernet PCI-X (rev
03)
> > >     eth3      configuration: eth-id-00:0c:fc:00:02:c8
> > > irq 60, desc: a0000001009a2d00, depth: 1, count: 0, unhandled: 0
> > > ->handle_irq():  0000000000000000, 0x0
> > > ->chip(): a000000100a0fe40, irq_type_sn+0x0/0x80
> > > ->action(): e00000b007471b80
> > > ->action->handler(): a0000002059373d0, s2io_msi_handle+0x1510/0x660
[s2io]    eth3
> > > IP address: 192.168.1.248/24
> > > Unexpected irq vector 0x3c on CPU 3!
> >
> > I guess that's where things start to go bad.  genirq changes?
>
> Hmm, The extra noisy printout is from geirq. The unhandled interrupt
> should be unrelated.
>
> The s2io driver enables interrupts in the card in start_nic() before
> requesting the interrupt itself with request_irq(). So I suspect thats a
> problem which has been there already, just the noisy printout makes it
> more visible

OK, that's not good.  It would be strange for the NIC to be aserting an
interrupt in that window though - the machine would end up taking a zillion
interrupts and would disable the whole IRQ line.

Still.  Ravinandan, could you take a look at fixing that up, please?  Wire
up the IRQ handler before enabling interrupts?

We still don't know why these things happened.