From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 27 Jun 2001 11:38:50 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 27 Jun 2001 11:38:30 -0400 Received: from ns.suse.de ([213.95.15.193]:23826 "HELO Cantor.suse.de") by vger.kernel.org with SMTP id ; Wed, 27 Jun 2001 11:38:23 -0400 To: "Jens Hoffrichter" Cc: linux-kernel@vger.kernel.org Subject: Re: printk and sk_buffs In-Reply-To: From: Andi Kleen Date: 27 Jun 2001 17:38:20 +0200 In-Reply-To: "Jens Hoffrichter"'s message of "27 Jun 2001 16:07:46 +0200" Message-ID: User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.7 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org "Jens Hoffrichter" writes: > I do not fully unterstandt the printk code, so perhaps somebody can answer > me this (probably stupid ;) ) question: > > If I do a printk, is there a packet (aka a sk_buff) created? If I turn on > debugging in my code, I see a huge pile of sk_buffs which are allocated but > which do not get in touch with the "essential" parts of the network-code > (e.g. ip_rcv) where I have modified some code. I can't explain it to me > fully, but perhaps someone of yours has a suitable answer. klogd uses unix sockets to talk to syslogd. unix sockets use sk_buffs. You can kill klogd if you're not interested in logging kernel messages to disk. -Andi