From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ryusuke Konishi Subject: Re: grub nilfs2 module Date: Thu, 04 Feb 2010 18:03:43 +0900 (JST) Message-ID: <20100204.180343.168036764.ryusuke@osrg.net> References: <20100204.100228.02802641.ryusuke@osrg.net> <20100204.165456.217843065.ryusuke@osrg.net> <87d40l9zyb.wl%jir@sekiba.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <87d40l9zyb.wl%jir-27yqGEOhnJbQT0dZR+AlfA@public.gmane.org> Sender: linux-nilfs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: Text/Plain; charset="us-ascii" To: jir-hfpbi5WX9J54Eiagz67IpQ@public.gmane.org Cc: linux-nilfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org On Thu, 04 Feb 2010 17:18:04 +0900, Jiro SEKIBA wrote: > Hi, > > At Thu, 04 Feb 2010 16:54:56 +0900 (JST), > Ryusuke Konishi wrote: > > > > Hi, > > On Thu, 04 Feb 2010 10:02:28 +0900 (JST), Ryusuke Konishi wrote: > > > Hi, > > > On Wed, 03 Feb 2010 21:43:46 +0900, Jiro SEKIBA wrote: > > > > Hi, > > > > > > > > I've written nilfs2 module for grub2 boot loader. > > > > > > > > Here is the patch and mail I'v sent to grub-devel ML. > > > > http://lists.gnu.org/archive/html/grub-devel/2010-02/msg00025.html > > > > > > Great! > > > > > > The patch looks fine to me except some work, for instance, > > > rollforwarding and checksum based log validation is needed. > > > > > > But meanwhile, this prototype seems enough to load kernels and > > > initrds. > > > > > > I'd like to try it for my laptop. > > > > > > > It may not be mature yet, but I was able to boot kernel > > > > from nilfs2 partition and mount it as rootfs. > > > > So you are able to boot the system that has only nilfs2 partition. > > > > Let me ask you a question. > > > > One thing which is important in NILFS is that every block in the > > partition can be moved by the garbage collection (except two super > > blocks). > > > > So, we cannot assume that file blocks stay in a locked position even > > for unmodified files. > > > > Doesn't this cause problems for the boot sequence of GRUB2 ? > > That's a good question. > > It's basically OK because there usually are room for additional code > besides MBR in PC disk format. The room size is 32KiB followed by MBR. > The room is not required for linux, but historically for DOS compat. > > See the link how modern disk layout is. > http://www.pixelbeat.org/docs/disk/ > > You can check it with the dd by following > # dd if=/dev/hda of=hda bs=512 count=3 skip=65 > # hexdump -C hda > hda.hex > # dd if=/dev/hda1 of=hda1 bs=512 count=3 > # hexdump -C hda1 > hda1.hex > > NOTE, the differences between two dumps are, I think, mtime in a super block. > > So if you install grub2 with root directory in nilfs2, > grub-install will install nilfs2.mod into that additional 32KiB space > The space is out of nilfs2 patition, so GC won't move the blocks. > > thanks Thanks! That cleared up my doubts on GC. The reference you cited is informative. Regards, Ryusuke Konishi -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html