From owner-freebsd-arch@FreeBSD.ORG  Fri Dec  3 21:22:57 2004
Return-Path: <owner-freebsd-arch@FreeBSD.ORG>
Delivered-To: freebsd-arch@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id A968316A4CE
	for <arch@freebsd.org>; Fri,  3 Dec 2004 21:22:57 +0000 (GMT)
Received: from critter.freebsd.dk (critter.freebsd.dk [212.242.86.163])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 039C543D55
	for <arch@freebsd.org>; Fri,  3 Dec 2004 21:22:57 +0000 (GMT)
	(envelope-from phk@critter.freebsd.dk)
Received: from critter.freebsd.dk (localhost [127.0.0.1])
	by critter.freebsd.dk (8.13.1/8.13.1) with ESMTP id iB3LMuqV026128
	for <arch@freebsd.org>; Fri, 3 Dec 2004 22:22:56 +0100 (CET)
	(envelope-from phk@critter.freebsd.dk)
To: arch@freebsd.org
From: "Poul-Henning Kamp" <phk@phk.freebsd.dk>
Date: Fri, 03 Dec 2004 22:22:56 +0100
Message-ID: <26127.1102108976@critter.freebsd.dk>
Sender: phk@critter.freebsd.dk
Subject: [HEADSUP] omount/nmount migration and rootfs mounting
X-BeenThere: freebsd-arch@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Discussion related to FreeBSD architecture
	<freebsd-arch.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-arch>,
	<mailto:freebsd-arch-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-arch>
List-Post: <mailto:freebsd-arch@freebsd.org>
List-Help: <mailto:freebsd-arch-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-arch>,
	<mailto:freebsd-arch-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Dec 2004 21:22:57 -0000


Ok, here is the big picture:

All filesystems get a ->vfs_mount() which does the job.

Filesystems which have a ->vfs_omount() looses it and gets
a ->vfs_cmount() instead which after pulling compat args in
calls the nmount() stuff.

Here is my prototype for ffs:

+static int
+ffs_cmount(char *path, void *data, int flags, struct thread *td)
+{
+       struct ufs_args args;
+       struct iovec v[10];
+       int error;
+
+       if (data == NULL)
+               return (EINVAL);
+       error = copyin(data, &args, sizeof args);
+       if (error)
+               return (error);
+
+       v[0].iov_base = "fstype";
+       v[0].iov_len = strlen(v[0].iov_base) + 1;
+       v[1].iov_base = "ufs";
+       v[1].iov_len = strlen(v[1].iov_base) + 1;
+
+       v[2].iov_base = "fspath";
+       v[2].iov_len = strlen(v[2].iov_base) + 1;
+       v[3].iov_base = path;
+       v[3].iov_len = strlen(v[3].iov_base) + 1;
+
+       v[4].iov_base = "from";
+       v[4].iov_len = strlen(v[4].iov_base) + 1;
+       v[5].iov_base = args.fspec;
+       v[5].iov_len = strlen(v[5].iov_base) + 1;
+
+       v[6].iov_base = "export";
+       v[6].iov_len = strlen(v[6].iov_base) + 1;
+       v[7].iov_base = &args.export;
+       v[7].iov_len = sizeof(args.export);
+
+       v[8].iov_base = NULL;
+       v[8].iov_len = 0;
+
+       error = kernel_mount(v, 8, flags);
+
+       return (error);
+}

Root mounting will happen the following way:

	create a mountpoint.
	devfs->vfs_mount() on it.
	set rootdir for proc0
	create symlink:  /dev -> /

	Mount real root filesystem as / using normal ->vfs_mount() method.
		(filesystem contains no magic for this)

	Surgically move the devfs mount from / to /dev in the new rootfs.

Keep an eye on p4:phk_bufwork, it works for ffs and nfs already.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.