[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1239947737.23397.4018.camel@laptop>
Date: Fri, 17 Apr 2009 07:55:37 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Stephen Hemminger <shemminger@...tta.com>
Cc: paulmck@...ux.vnet.ibm.com, David Miller <davem@...emloft.net>,
kaber@...sh.net, torvalds@...ux-foundation.org,
dada1@...mosbay.com, jeff.chua.linux@...il.com, paulus@...ba.org,
mingo@...e.hu, laijs@...fujitsu.com, jengelh@...ozas.de,
r000n@...0n.net, linux-kernel@...r.kernel.org,
netfilter-devel@...r.kernel.org, netdev@...r.kernel.org,
benh@...nel.crashing.org
Subject: Re: [PATCH] netfilter: per-cpu spin-lock with recursion (v0.8)
On Thu, 2009-04-16 at 16:52 -0700, Stephen Hemminger wrote:
> - Lockdep doesn't really handle this well
> +/**
> + * xt_table_info_lock_all - lock xt table info for update
> + *
> + * Locks out all readers, and blocks bottom half
> + */
> +void xt_table_info_lock_all(void)
> +{
> + int i;
> +
> + local_bh_disable();
> + for_each_possible_cpu(i) {
> + struct xt_lock *lock = &per_cpu(xt_info_locks, i);
> + spin_lock(&lock->lock);
> + BUG_ON(lock->depth != -1);
> + }
> +}
> +EXPORT_SYMBOL_GPL(xt_table_info_lock_all);
Quite so, this is the old MAX_LOCK_DEPTH < NR_CPUS issue for large
systems.
Last time this came up David found another way of solving the problem.
Not having fully read this thread, I cannot suggest one myself -- except
that RCU domains as suggested by David sound good.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists