Clustered read and write are switched at mount-option level.

1. Clustered I/O is switched by the MNT_NOCLUSTERR and MNT_NOCLUSTERW
   bits of the mnt_flag.  The sysctl variables, vfs.foo.doclusterread
   and vfs.foo.doclusterwrite are deleted.  Only mount option can
   control clustered I/O from userland.
2. When foofs_mount mounts block device, foofs_mount checks D_CLUSTERR
   and D_CLUSTERW bits of the d_flags member in the block device switch
   table.  If D_NOCLUSTERR / D_NOCLUSTERW are set, MNT_NOCLUSTERR /
   MNT_NOCLUSTERW bits will be set.  In this case, MNT_NOCLUSTERR and
   MNT_NOCLUSTERW cannot be cleared from userland.
3. Vnode driver disables both clustered read and write.
4. Union filesystem disables clutered write.

Reviewed by:	bde

4 files changed, 18 insertions, 28 deletions

diff --git a/sys/gnu/ext2fs/ext2_extern.h b/sys/gnu/ext2fs/ext2_extern.h
index 277106b..4c32799 100644
--- a/sys/gnu/ext2fs/ext2_extern.h
+++ b/sys/gnu/ext2fs/ext2_extern.h
@@ -42,19 +42,6 @@
 #ifndef _SYS_GNU_EXT2FS_EXT2_EXTERN_H_
 #define	_SYS_GNU_EXT2FS_EXT2_EXTERN_H_
 
-/*
- * Sysctl values for the ext2fs filesystem.
- */
-#define	EXT2FS_CLUSTERREAD		1		/* cluster reading enabled */
-#define	EXT2FS_CLUSTERWRITE		2		/* cluster writing enable */
-#define	EXT2FS_MAXID			3		/* number of valid ext2fs ids */
-
-#define	EXT2FS_NAMES {\
-	{0, 0}, \
-	{ "doclusterread", CTLTYPE_INT }, \
-	{ "doculsterwrite", CTLTYPE_INT }, \
-}
-
 struct dinode;
 struct ext2_inode;
 struct inode;
diff --git a/sys/gnu/ext2fs/ext2_readwrite.c b/sys/gnu/ext2fs/ext2_readwrite.c
index e28cd0a..986bc14 100644
--- a/sys/gnu/ext2fs/ext2_readwrite.c
+++ b/sys/gnu/ext2fs/ext2_readwrite.c
@@ -112,7 +112,7 @@ READ(ap)
 
 		if (lblktosize(fs, nextlbn) >= ip->i_size)
 			error = bread(vp, lbn, size, NOCRED, &bp);
-		else if (ext2_doclusterread)
+		else if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0)
 			error = cluster_read(vp,
 			    ip->i_size, lbn, size, NOCRED,
 				uio->uio_resid, (ap->a_ioflag >> 16), &bp);
@@ -277,7 +277,7 @@ WRITE(ap)
 		if (ioflag & IO_SYNC) {
 			(void)bwrite(bp);
 		} else if (xfersize + blkoffset == fs->s_frag_size) {
-			if (ext2_doclusterwrite) {
+			if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERW) == 0) {
 #if defined(__FreeBSD__)
 				bp->b_flags |= B_CLUSTEROK;
 #endif
diff --git a/sys/gnu/ext2fs/ext2_vfsops.c b/sys/gnu/ext2fs/ext2_vfsops.c
index 7aef5f6..11d1b33 100644
--- a/sys/gnu/ext2fs/ext2_vfsops.c
+++ b/sys/gnu/ext2fs/ext2_vfsops.c
@@ -149,6 +149,10 @@ ext2_mountroot()
 	bzero((char *)mp, (u_long)sizeof(struct mount));
 	mp->mnt_op = &ext2fs_vfsops;
 	mp->mnt_flag = MNT_RDONLY;
+	if (bdevsw[major(rootdev)]->d_flags & D_NOCLUSTERR)
+		mp->mnt_flag |= MNT_NOCLUSTERR;
+	if (bdevsw[major(rootdev)]->d_flags & D_NOCLUSTERW)
+		mp->mnt_flag |= MNT_NOCLUSTERW;
 	if (error = ext2_mountfs(rootvp, mp, p)) {
 		bsd_free(mp, M_MOUNT);
 		return (error);
@@ -206,11 +210,17 @@ ext2_mount(mp, path, data, ndp, p)
 	/*
 	 * If updating, check whether changing from read-only to
 	 * read/write; if there is no device name, that's all we do.
+	 * Disallow clearing MNT_NOCLUSTERR and MNT_NOCLUSTERW flags,
+	 * if block device requests.
 	 */
 	if (mp->mnt_flag & MNT_UPDATE) {
 		ump = VFSTOUFS(mp);
 		fs = ump->um_e2fs;
 		error = 0;
+		if (bdevsw[major(ump->um_dev)]->d_flags & D_NOCLUSTERR)
+			mp->mnt_flag |= MNT_NOCLUSTERR;
+		if (bdevsw[major(ump->um_dev)]->d_flags & D_NOCLUSTERW)
+			mp->mnt_flag |= MNT_NOCLUSTERW;
 		if (fs->s_rd_only == 0 && (mp->mnt_flag & MNT_RDONLY)) {
 			flags = WRITECLOSE;
 			if (mp->mnt_flag & MNT_FORCE)
@@ -255,9 +265,13 @@ ext2_mount(mp, path, data, ndp, p)
 		vrele(devvp);
 		return (ENXIO);
 	}
-	if ((mp->mnt_flag & MNT_UPDATE) == 0)
+	if ((mp->mnt_flag & MNT_UPDATE) == 0) {
+		if (bdevsw[major(devvp->v_rdev)]->d_flags & D_NOCLUSTERR)
+			mp->mnt_flag |= MNT_NOCLUSTERR;
+		if (bdevsw[major(devvp->v_rdev)]->d_flags & D_NOCLUSTERW)
+			mp->mnt_flag |= MNT_NOCLUSTERW;
 		error = ext2_mountfs(devvp, mp, p);
-	else {
+	} else {
 		if (devvp != ump->um_devvp)
 			error = EINVAL;	/* needs translation */
 		else
diff --git a/sys/gnu/ext2fs/ext2_vnops.c b/sys/gnu/ext2fs/ext2_vnops.c
index 583164b..94fe0a7 100644
--- a/sys/gnu/ext2fs/ext2_vnops.c
+++ b/sys/gnu/ext2fs/ext2_vnops.c
@@ -237,17 +237,6 @@ static struct vnodeopv_desc ext2fs_fifoop_opv_desc =
 	VNODEOP_SET(ext2fs_fifoop_opv_desc);
 #endif
 
-/*
- * Enabling cluster read/write operations.
- */
-static int	ext2_doclusterread = 1;
-static int	ext2_doclusterwrite = 1;
-SYSCTL_NODE(_vfs, MOUNT_EXT2FS, ext2fs, CTLFLAG_RW, 0, "EXT2FS filesystem");
-SYSCTL_INT(_vfs_ext2fs, EXT2FS_CLUSTERREAD, doclusterread,
-		   CTLFLAG_RW, &ext2_doclusterread, 0, "");
-SYSCTL_INT(_vfs_ext2fs, EXT2FS_CLUSTERWRITE, doclusterwrite,
-		   CTLFLAG_RW, &ext2_doclusterwrite, 0, "");
-
 #include <gnu/ext2fs/ext2_readwrite.c>
 
 /*