Subject: Re: smbfs panic
To: Patrick Welche <prlw1@newn.cam.ac.uk>
From: Antti Kantee <pooka@cs.hut.fi>
List: current-users
Date: 12/03/2007 19:32:20
On Mon Dec 03 2007 at 17:26:49 +0000, Patrick Welche wrote:
> On Fri, Nov 30, 2007 at 01:25:35PM +0200, Antti Kantee wrote:
> > The actual panic message is also generally helpful, it's not printed
> > just to make your screen look cool ;)
>
> In fact my screen was a frozen xdm login screen :-/ Best I could do was
> plug in a serial console (yup, xconsole disabled here, and console on
> tty00), so I could get a trace - any idea how you would read the panic
> message from ddb?
I don't have a panic at hand to test (slow day ;), but I think "dmesg"
should cover it. Or "dmesg 100" or so for less output.
> > But luckily it's easy guessing where the thing is going wrong.
> > Try smbfs_node.c rev 1.34. You should be able to trigger the condition
> > very quickly by setting kern.maxvnodes low and doing e.g. ls -lR in
> > your smbfs. If this patch doesn't help, let's start investigations with
> > the panic message.
>
> and you guessed right. Having panicked it first on purpose, the panic was
> Reader / writer lock error: pool_do_put: allocation contains active lock
> ;-) So all is well...
Thanks for the thorough testing!
--
Antti Kantee <pooka@iki.fi> Of course he runs NetBSD
http://www.iki.fi/pooka/ http://www.NetBSD.org/
"la qualité la plus indispensable du cuisinier est l'exactitude"