Return-path: Received: from ns.suse.de ([195.135.220.2] helo=mx1.suse.de) by canuck.infradead.org with esmtps (Exim 4.63 #1 (Red Hat Linux)) id 1IL4R7-0000bb-4F for kexec@lists.infradead.org; Tue, 14 Aug 2007 18:00:35 -0400 Date: Wed, 15 Aug 2007 00:00:25 +0200 From: Bernhard Walle Subject: Re: [RFC][PATCH 1/3] Embed vmcoreinfo into kernel file Message-ID: <20070814220025.GB16408@suse.de> References: <20070813163212oomichi@mail.jp.nec.com> <20070813163653oomichi@mail.jp.nec.com> <20070813173811.GA14191@localdomain> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <20070813173811.GA14191@localdomain> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: kexec-bounces@lists.infradead.org Errors-To: kexec-bounces+dwmw2=infradead.org+dwmw2=infradead.org@lists.infradead.org To: Dan Aloni Cc: Don Zickus , Ken'ichi Ohmichi , Neil Horman , Vivek Goyal , kexec-ml * Dan Aloni [2007-08-13 19:38]: > On Mon, Aug 13, 2007 at 04:36:53PM +0900, Ken'ichi Ohmichi wrote: > >[..] > > Patches: > > - [1/3] patch for linux-2.6.22. > > Changelog: > > * Rename "mkdfinfo" to "vmcoreinfo". > > * Multi memory models (FLATMEM, DISCONTIGMEM, and SPASEMEM) are > > supported. > > * The elf note typedef for vmcoreinfo is added because the original > > elf note size is limited to 1024. > > * The generation of the elf note is moved to boot time instead of > > crash time. > > This looks very good. Just one suggestion - in order to reduce clutter > from kernel/kexec.c (as to avoid #ifdefs) perhaps it might be better > to have crash_save_vmcoreinfo_init() additionally call an arch-specific > arch_crash_save_vmcoreinfo_init() function and have that append all > the arch-specific stuff. Because the only architecture that is really different is IA64, it may make sense here to provide a default implementation that is used on any architecture and declared as __attribute__((weak)). Then, any architecture (here: IA64) can override that implementation. That would avoid code duplication. The rest looks good from my point of view. I'd like to see that in the mainline kernel. :) Thanks, Bernhard _______________________________________________ kexec mailing list kexec@lists.infradead.org http://lists.infradead.org/mailman/listinfo/kexec