Date: Mon, 9 Jun 2003 14:08:34 -0700
From: Andrew Morton <akpm@digeo.com>
To: "Martin J. Bligh" <mbligh@aracnet.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: 2.5.70 (virgin) hangs running SDET
Message-Id: <20030609140834.11ad0d63.akpm@digeo.com>
In-Reply-To: <60380000.1055188542@flay>
References: <60380000.1055188542@flay>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1005
Lines: 28

"Martin J. Bligh" <mbligh@aracnet.com> wrote:
>
> I thought this was specific to the -mjb tree for a while, but I can
> reproduce it pretty easily on virgin 2.5.70 (but 2.5.69 didn't do
> this). Andrew looked and pointed out it seems to be an i_sem (ISTR
> some mention of /proc somewhere too?). Got a good sysrq+t trace out
> of the virgin kernel (on 16x NUMA-Q) ... below:

The ps instance which is spinning on kernel_flag in proc_root_lookup is
what's holding things up.

Or is it spinning in proc_lookup() or proc_pid_lookup()?  I have a vague
feeling that I've seen these traces miss out the innermost stack slot...

Suggest:

a) Use CONFIG_SPINLINE, get a new sysrq-T trace

b) Enable spinlock debugging

c) Try disabling the sched_balance_exec() code.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/