From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from metis.ext.pengutronix.de ([2001:67c:670:201:290:27ff:fe1d:cc33]) by bombadil.infradead.org with esmtps (Exim 4.80.1 #2 (Red Hat Linux)) id 1ZldVf-0003Uh-Nm for barebox@lists.infradead.org; Mon, 12 Oct 2015 13:51:28 +0000 Date: Mon, 12 Oct 2015 15:51:05 +0200 From: Sascha Hauer Message-ID: <20151012135105.GW7858@pengutronix.de> References: <1444241036-23622-1-git-send-email-pmamonov@gmail.com> <20151009080624.GF7858@pengutronix.de> <20151009154037.389bb1ab@berta> <20151009161143.GG7858@pengutronix.de> <20151012132735.5f1a30ff@berta> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <20151012132735.5f1a30ff@berta> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: "barebox" Errors-To: barebox-bounces+u.kleine-koenig=pengutronix.de@lists.infradead.org Subject: Re: [RFC] common: filetype: is_fat_or_mbr() considered harmful To: Peter Mamonov Cc: barebox@lists.infradead.org On Mon, Oct 12, 2015 at 01:27:35PM +0300, Peter Mamonov wrote: > On Fri, 9 Oct 2015 18:11:44 +0200 > Sascha Hauer wrote: > > > On Fri, Oct 09, 2015 at 03:40:37PM +0300, Peter Mamonov wrote: > > > On Fri, 9 Oct 2015 10:06:24 +0200 > > > Sascha Hauer wrote: > > > > > > > Hi Peter, > > > > > > > > On Wed, Oct 07, 2015 at 09:03:56PM +0300, Peter Mamonov wrote: > > > > > Deleted pieces of code detect MBR-containig device as a FAT-type > > > > > device, if it's first partition contains a FAT filesystem. So, > > > > > one can mount the first partition of a hard drive containing > > > > > FAT FS using the following command: barebox: > > > > > mount /dev/ata0.0 /mnt/0 as well as this one: > > > > > barebox: mount /dev/ata0 /mnt/1 > > > > > Both commands mount the same FS. > > > > > > > > > > This behaviour causes automount (mount -a) to mount FAT FS > > > > > on a first partition twice: > > > > > barebox: mount > > > > > none on / type ramfs > > > > > none on /dev type devfs > > > > > /dev/ata0 on /mnt/ata0 type fat > > > > > /dev/ata0.0 on /mnt/ata0.0 type fat > > > > > /dev/ata0.1 on /mnt/ata0.1 type ext4 > > > > > > > > This is_fat_or_mbr mechanism never worked very well and had funny > > > > side effects. Would be nice to get rid of it. > > > > Simply removing this option is not a solution though, we have to > > > > find a proper way to keep the current feature and make it more > > > > sane. > > > > > > Ok, the patch comment is misleading a bit. I do not propose to get > > > rid of the is_fat_or_mbr() completely. However, I do not see the > > > point to check for a FAT FS, after the device was correctly > > > detected as an MBR-type device: > > > > > > enum filetype file_name_detect_type(const char *filename) > > > ... > > > type = file_detect_type(buf, ret); > > > > > > if (type == filetype_mbr) { > > > /* > > > * Get the first partition start sector > > > * and check for FAT in it > > > */ > > > is_fat_or_mbr(buf, &bootsec); > > > ret = lseek(fd, (bootsec) * 512, SEEK_SET); > > > if (ret < 0) > > > goto err_out; > > > ret = read(fd, buf, 512); > > > if (ret < 0) > > > goto err_out; > > > type = is_fat_or_mbr((u8 *)buf, NULL); > > > } > > > > > > > > > The deleted code snippet was introduced by this patch: > > > > > > commit 010ee209b75c5732ae4144e3ee9ce14158193c1f > > > Author: Franck Jullien > > > Date: Wed Sep 19 13:09:01 2012 +0200 > > > > > > filetype: Improve FAT detection > > > > > > We may have some disk with MBR as a first sector. In this case, > > > the current FAT check returns an error. However, the FAT sector > > > exist and the MBR can tell us where it is. > > > > > > This patch add to file_name_detect_type function the ability to > > > find the FAT boot sector on the first sector of the first partition > > > in case it is not on sector 0. > > > > > > It also introduce is_fat_or_mbr to check if a buffer is a FAT > > > boot or MBR sector > > > > > > Signed-off-by: Franck Jullien > > > Signed-off-by: Sascha Hauer > > > > > > According to the patch message it was introduced to workaround FAT > > > detection. However, after deletion of the code I'm still able to > > > detect and mount FAT-containig partiotions. > > > > But can you mount /dev/disk0 if this disk contains a partition table > > and the FAT is on /dev/disk0.0? > > No. This is actually the purpose of my patch, since I don't want > "mount -a" to mount the same partition (FAT on /dev/disk0.0) twice. I know, and this is valid. It just conflicts with what Franck wants. He just wants to mount a USB device without having to know if the FAT is on the raw device or on the first partition. > > > This is what the patch is about. The > > problem the patches solved is that when you plug in a USB drive then > > you don't know whether a FAT is directly on the device or if the > > device is partitioned. You want to be able to mount both ways with > > the same command, so no matter if the FAT is on /dev/disk0 > > or /dev/disk0.0 you can mount both using /dev/disk0. > > Ok. So what is the preferred way to prevent "mount -a" from mounting > /dev/disk0 and /dev/disk0.0 at the same time? Sorry, I do not have a solution currently. I'll have a look into it. Sascha -- Pengutronix e.K. | | Industrial Linux Solutions | http://www.pengutronix.de/ | Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | _______________________________________________ barebox mailing list barebox@lists.infradead.org http://lists.infradead.org/mailman/listinfo/barebox