From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263207AbTDIMQK (for ); Wed, 9 Apr 2003 08:16:10 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263286AbTDIMQK (for ); Wed, 9 Apr 2003 08:16:10 -0400 Received: from rwcrmhc53.attbi.com ([204.127.198.39]:44246 "EHLO rwcrmhc53.attbi.com") by vger.kernel.org with ESMTP id S263207AbTDIMQG (for ); Wed, 9 Apr 2003 08:16:06 -0400 From: Arun Dharankar To: Piet Delaney Subject: Re: Linux kernel crash dumps (LKCD) and PowerPC ports. Date: Wed, 9 Apr 2003 08:26:36 -0400 User-Agent: KMail/1.5 Cc: piet , "Matt D. Robinson" , linux-kernel@vger.kernel.org, Dave Anderson References: <200304081647.32146.ADharankar@ATTBI.Com> <200304082349.27844.ADharankar@ATTBI.Com> <1049875702.22212.6634.camel@www.piet.net> In-Reply-To: <1049875702.22212.6634.camel@www.piet.net> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200304090826.36304.ADharankar@ATTBI.Com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Piet, thanks for the reply! Some of the points that I found attractive (among other points): - Overall a smaller time to reboot and save the dump, and bring the system back to working state. - Does not require local disk to save the dump on the way down during a panic - works on a diskless system. - No need for a dump/swap device. Looking at the MCLX site, the last patch is on 2.4.17 only. Also, the PowerPC port is a bit broken. Is anyone maintaining this part of the MCLX scheme, and is available else where? Best regards, -Arun. > > Looking at those sites, it appears that the development at > > "http://lists.insecure.org/lists/linux-kernel/2003/Feb/0987.html." > > which I had pointed is based on the SGI's dump scheme. Same > > for the one you pointed to. > > > > The other scheme I poinited to (from Mission Critical Linux/MCLX) > > seems to have some strong points too. Any pointers to discussions > > about why the LKCD work seems to more active than the > > MCLX one? > > I don't think there has been such a discussion. The Mission Critical > hackers were hired by RedHat and Dave Anderson continues to work on > crash and making it available at: > > ftp://people.redhat.com/anderson > > He's had about a release per month. I thought MCLX crash has some > strong points also. After I enhanced LKCD to support the ia64 I > fixed MCLX crash so it could read the new LKCD crash dumps which > are no longer monotonically increasing in memory. I did this to > support NUMA systems which can have more memory than swap space. > This way the most important memory can be saved first. > > If you read the lkcd-devel mailing list you can read our discussion > on maintaining MLCX crash on the lkcd cvs tree. > > > "And imnsho, debugging the kernel on a source > level is the way to do it." > Linus > > -piet > > > Best regards, > > -Arun. > > > > On Tuesday 08 April 2003 07:14 pm, Matt D. Robinson wrote: > > > Please look at the lkcd-devel mailing archives. There is > > > at least one group working on a PPC port of LKCD > > > > > > http://sourceforge.net/mail/?group_id=2726 > > > > > > --Matt