From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick Schaaf Subject: Re: ip_conntrack_hash() problem Date: Thu, 5 Sep 2002 08:51:03 +0200 Sender: netfilter-devel-admin@lists.netfilter.org Message-ID: <20020905085103.G19551@oknodo.bof.de> References: <20020904152626.A11438@wotan.suse.de> <20020905044436.0772A2C0DF@lists.samba.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andi Kleen , Harald Welte , Netfilter Mailing List , Linux Kernel mailing list , Patrick Schaaf Return-path: To: Rusty Russell Content-Disposition: inline In-Reply-To: <20020905044436.0772A2C0DF@lists.samba.org>; from rusty@rustcorp.com.au on Thu, Sep 05, 2002 at 10:39:40AM +1000 Errors-To: netfilter-devel-admin@lists.netfilter.org List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: List-Id: netfilter-devel.vger.kernel.org Rusty, > This work is already done: > http://www.kernel.org/pub/linux/kernel/people/rusty/patches/Netfilter/conntrack_hashing.patch.gz Regarding the rehash check in ip_conntrack_find_get, wouldn't it be better to do that in the confirm function, where a new conntrack is put into the list? That's called a lot less often than _find_get, and should be logically equivalent. IOW, why wait until we _find_ an overly long list, when we can rehash at the point in time when it _became_ overly long? best regards Patrick