From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from stargate.chelsio.com ([12.32.117.8]:11146 "EHLO stargate.chelsio.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752716AbeBSMes (ORCPT ); Mon, 19 Feb 2018 07:34:48 -0500 Date: Mon, 19 Feb 2018 18:04:17 +0530 From: Rahul Lakkireddy To: David Miller Cc: "netdev@vger.kernel.org" , Ganesh GR , Nirranjan Kirubaharan , Indranil Choudhury Subject: Re: [PATCH net-next] cxgb4: append firmware dump to vmcore in kernel panic Message-ID: <20180219123416.GA7737@chelsio.com> References: <1518702882-29688-1-git-send-email-rahul.lakkireddy@chelsio.com> <20180216.154101.1707041533181015167.davem@davemloft.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20180216.154101.1707041533181015167.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Saturday, February 02/17/18, 2018 at 02:11:01 +0530, David Miller wrote: > From: Rahul Lakkireddy > Date: Thu, 15 Feb 2018 19:24:42 +0530 > > > Register callback to panic_notifier_list. Invoke dump collect routine > > to append dump to vmcore. > > > > Signed-off-by: Rahul Lakkireddy > > Signed-off-by: Ganesh Goudar > > There is absolutely no precedence for a networking driver dumping > things into the vmcore image on a panic. > > And I don't think this is a good idea. > > Really, this commit message should have explained why this is desired > and in what context it is legitimate for this driver in particular > to do it. > > A very detailed, long, complete commit message is especially important > when you are deciding to blaze you own trail and do something no other > networking driver has done before. > My mistake. Will add more info in the commit message in v2. > I get really upset when I see changes like this, because you give me > no preparation for what I'm about to read in the patch and therefore > I have to go into this routine asking you to explain things properly. > > But as-is, I see this panic notifier as a really bad idea. > Our requirement is to analyze the state of firmware/hardware at the time of kernel panic. The dump will be written to pre allocated buffer during kernel panic, which can be extracted later from the vmcore, for post-analysis. Panic notifier seemed to meet our requirement, as we are able to collect dump during kernel panic and then extract it from vmcore. Please suggest, if this can be done in a better way. Thanks, Rahul