diff options
author | David Howells <dhowells@redhat.com> | 2006-06-23 05:02:57 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@g5.osdl.org> | 2006-06-23 10:42:45 -0400 |
commit | 454e2398be9b9fa30433fccc548db34d19aa9958 (patch) | |
tree | 1f61cb0c3716a33b661cfc8977e9beeb480a322c /fs/nfs | |
parent | 1ad5544098a69d7dc1fa508cbb17e13a7a952fd8 (diff) |
[PATCH] VFS: Permit filesystem to override root dentry on mount
Extend the get_sb() filesystem operation to take an extra argument that
permits the VFS to pass in the target vfsmount that defines the mountpoint.
The filesystem is then required to manually set the superblock and root dentry
pointers. For most filesystems, this should be done with simple_set_mnt()
which will set the superblock pointer and then set the root dentry to the
superblock's s_root (as per the old default behaviour).
The get_sb() op now returns an integer as there's now no need to return the
superblock pointer.
This patch permits a superblock to be implicitly shared amongst several mount
points, such as can be done with NFS to avoid potential inode aliasing. In
such a case, simple_set_mnt() would not be called, and instead the mnt_root
and mnt_sb would be set directly.
The patch also makes the following changes:
(*) the get_sb_*() convenience functions in the core kernel now take a vfsmount
pointer argument and return an integer, so most filesystems have to change
very little.
(*) If one of the convenience function is not used, then get_sb() should
normally call simple_set_mnt() to instantiate the vfsmount. This will
always return 0, and so can be tail-called from get_sb().
(*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the
dcache upon superblock destruction rather than shrink_dcache_anon().
This is required because the superblock may now have multiple trees that
aren't actually bound to s_root, but that still need to be cleaned up. The
currently called functions assume that the whole tree is rooted at s_root,
and that anonymous dentries are not the roots of trees which results in
dentries being left unculled.
However, with the way NFS superblock sharing are currently set to be
implemented, these assumptions are violated: the root of the filesystem is
simply a dummy dentry and inode (the real inode for '/' may well be
inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries
with child trees.
[*] Anonymous until discovered from another tree.
(*) The documentation has been adjusted, including the additional bit of
changing ext2_* into foo_* in the documentation.
[akpm@osdl.org: convert ipath_fs, do other stuff]
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Nathan Scott <nathans@sgi.com>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'fs/nfs')
-rw-r--r-- | fs/nfs/inode.c | 96 |
1 files changed, 58 insertions, 38 deletions
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index d0b991a92327..ff645a961bc8 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c | |||
@@ -1690,8 +1690,8 @@ static int nfs_compare_super(struct super_block *sb, void *data) | |||
1690 | return !nfs_compare_fh(&old->fh, &server->fh); | 1690 | return !nfs_compare_fh(&old->fh, &server->fh); |
1691 | } | 1691 | } |
1692 | 1692 | ||
1693 | static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | 1693 | static int nfs_get_sb(struct file_system_type *fs_type, |
1694 | int flags, const char *dev_name, void *raw_data) | 1694 | int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt) |
1695 | { | 1695 | { |
1696 | int error; | 1696 | int error; |
1697 | struct nfs_server *server = NULL; | 1697 | struct nfs_server *server = NULL; |
@@ -1699,14 +1699,14 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1699 | struct nfs_fh *root; | 1699 | struct nfs_fh *root; |
1700 | struct nfs_mount_data *data = raw_data; | 1700 | struct nfs_mount_data *data = raw_data; |
1701 | 1701 | ||
1702 | s = ERR_PTR(-EINVAL); | 1702 | error = -EINVAL; |
1703 | if (data == NULL) { | 1703 | if (data == NULL) { |
1704 | dprintk("%s: missing data argument\n", __FUNCTION__); | 1704 | dprintk("%s: missing data argument\n", __FUNCTION__); |
1705 | goto out_err; | 1705 | goto out_err_noserver; |
1706 | } | 1706 | } |
1707 | if (data->version <= 0 || data->version > NFS_MOUNT_VERSION) { | 1707 | if (data->version <= 0 || data->version > NFS_MOUNT_VERSION) { |
1708 | dprintk("%s: bad mount version\n", __FUNCTION__); | 1708 | dprintk("%s: bad mount version\n", __FUNCTION__); |
1709 | goto out_err; | 1709 | goto out_err_noserver; |
1710 | } | 1710 | } |
1711 | switch (data->version) { | 1711 | switch (data->version) { |
1712 | case 1: | 1712 | case 1: |
@@ -1718,7 +1718,7 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1718 | dprintk("%s: mount structure version %d does not support NFSv3\n", | 1718 | dprintk("%s: mount structure version %d does not support NFSv3\n", |
1719 | __FUNCTION__, | 1719 | __FUNCTION__, |
1720 | data->version); | 1720 | data->version); |
1721 | goto out_err; | 1721 | goto out_err_noserver; |
1722 | } | 1722 | } |
1723 | data->root.size = NFS2_FHSIZE; | 1723 | data->root.size = NFS2_FHSIZE; |
1724 | memcpy(data->root.data, data->old_root.data, NFS2_FHSIZE); | 1724 | memcpy(data->root.data, data->old_root.data, NFS2_FHSIZE); |
@@ -1727,24 +1727,24 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1727 | dprintk("%s: mount structure version %d does not support strong security\n", | 1727 | dprintk("%s: mount structure version %d does not support strong security\n", |
1728 | __FUNCTION__, | 1728 | __FUNCTION__, |
1729 | data->version); | 1729 | data->version); |
1730 | goto out_err; | 1730 | goto out_err_noserver; |
1731 | } | 1731 | } |
1732 | case 5: | 1732 | case 5: |
1733 | memset(data->context, 0, sizeof(data->context)); | 1733 | memset(data->context, 0, sizeof(data->context)); |
1734 | } | 1734 | } |
1735 | #ifndef CONFIG_NFS_V3 | 1735 | #ifndef CONFIG_NFS_V3 |
1736 | /* If NFSv3 is not compiled in, return -EPROTONOSUPPORT */ | 1736 | /* If NFSv3 is not compiled in, return -EPROTONOSUPPORT */ |
1737 | s = ERR_PTR(-EPROTONOSUPPORT); | 1737 | error = -EPROTONOSUPPORT; |
1738 | if (data->flags & NFS_MOUNT_VER3) { | 1738 | if (data->flags & NFS_MOUNT_VER3) { |
1739 | dprintk("%s: NFSv3 not compiled into kernel\n", __FUNCTION__); | 1739 | dprintk("%s: NFSv3 not compiled into kernel\n", __FUNCTION__); |
1740 | goto out_err; | 1740 | goto out_err_noserver; |
1741 | } | 1741 | } |
1742 | #endif /* CONFIG_NFS_V3 */ | 1742 | #endif /* CONFIG_NFS_V3 */ |
1743 | 1743 | ||
1744 | s = ERR_PTR(-ENOMEM); | 1744 | error = -ENOMEM; |
1745 | server = kzalloc(sizeof(struct nfs_server), GFP_KERNEL); | 1745 | server = kzalloc(sizeof(struct nfs_server), GFP_KERNEL); |
1746 | if (!server) | 1746 | if (!server) |
1747 | goto out_err; | 1747 | goto out_err_noserver; |
1748 | /* Zero out the NFS state stuff */ | 1748 | /* Zero out the NFS state stuff */ |
1749 | init_nfsv4_state(server); | 1749 | init_nfsv4_state(server); |
1750 | server->client = server->client_sys = server->client_acl = ERR_PTR(-EINVAL); | 1750 | server->client = server->client_sys = server->client_acl = ERR_PTR(-EINVAL); |
@@ -1754,7 +1754,7 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1754 | root->size = data->root.size; | 1754 | root->size = data->root.size; |
1755 | else | 1755 | else |
1756 | root->size = NFS2_FHSIZE; | 1756 | root->size = NFS2_FHSIZE; |
1757 | s = ERR_PTR(-EINVAL); | 1757 | error = -EINVAL; |
1758 | if (root->size > sizeof(root->data)) { | 1758 | if (root->size > sizeof(root->data)) { |
1759 | dprintk("%s: invalid root filehandle\n", __FUNCTION__); | 1759 | dprintk("%s: invalid root filehandle\n", __FUNCTION__); |
1760 | goto out_err; | 1760 | goto out_err; |
@@ -1770,15 +1770,20 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1770 | } | 1770 | } |
1771 | 1771 | ||
1772 | /* Fire up rpciod if not yet running */ | 1772 | /* Fire up rpciod if not yet running */ |
1773 | s = ERR_PTR(rpciod_up()); | 1773 | error = rpciod_up(); |
1774 | if (IS_ERR(s)) { | 1774 | if (error < 0) { |
1775 | dprintk("%s: couldn't start rpciod! Error = %ld\n", | 1775 | dprintk("%s: couldn't start rpciod! Error = %d\n", |
1776 | __FUNCTION__, PTR_ERR(s)); | 1776 | __FUNCTION__, error); |
1777 | goto out_err; | 1777 | goto out_err; |
1778 | } | 1778 | } |
1779 | 1779 | ||
1780 | s = sget(fs_type, nfs_compare_super, nfs_set_super, server); | 1780 | s = sget(fs_type, nfs_compare_super, nfs_set_super, server); |
1781 | if (IS_ERR(s) || s->s_root) | 1781 | if (IS_ERR(s)) { |
1782 | error = PTR_ERR(s); | ||
1783 | goto out_err_rpciod; | ||
1784 | } | ||
1785 | |||
1786 | if (s->s_root) | ||
1782 | goto out_rpciod_down; | 1787 | goto out_rpciod_down; |
1783 | 1788 | ||
1784 | s->s_flags = flags; | 1789 | s->s_flags = flags; |
@@ -1787,15 +1792,22 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type, | |||
1787 | if (error) { | 1792 | if (error) { |
1788 | up_write(&s->s_umount); | 1793 | up_write(&s->s_umount); |
1789 | deactivate_super(s); | 1794 | deactivate_super(s); |
1790 | return ERR_PTR(error); | 1795 | return error; |
1791 | } | 1796 | } |
1792 | s->s_flags |= MS_ACTIVE; | 1797 | s->s_flags |= MS_ACTIVE; |
1793 | return s; | 1798 | return simple_set_mnt(mnt, s); |
1799 | |||
1794 | out_rpciod_down: | 1800 | out_rpciod_down: |
1795 | rpciod_down(); | 1801 | rpciod_down(); |
1802 | kfree(server); | ||
1803 | return simple_set_mnt(mnt, s); | ||
1804 | |||
1805 | out_err_rpciod: | ||
1806 | rpciod_down(); | ||
1796 | out_err: | 1807 | out_err: |
1797 | kfree(server); | 1808 | kfree(server); |
1798 | return s; | 1809 | out_err_noserver: |
1810 | return error; | ||
1799 | } | 1811 | } |
1800 | 1812 | ||
1801 | static void nfs_kill_super(struct super_block *s) | 1813 | static void nfs_kill_super(struct super_block *s) |
@@ -2032,8 +2044,8 @@ nfs_copy_user_string(char *dst, struct nfs_string *src, int maxlen) | |||
2032 | return dst; | 2044 | return dst; |
2033 | } | 2045 | } |
2034 | 2046 | ||
2035 | static struct super_block *nfs4_get_sb(struct file_system_type *fs_type, | 2047 | static int nfs4_get_sb(struct file_system_type *fs_type, |
2036 | int flags, const char *dev_name, void *raw_data) | 2048 | int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt) |
2037 | { | 2049 | { |
2038 | int error; | 2050 | int error; |
2039 | struct nfs_server *server; | 2051 | struct nfs_server *server; |
@@ -2043,16 +2055,16 @@ static struct super_block *nfs4_get_sb(struct file_system_type *fs_type, | |||
2043 | 2055 | ||
2044 | if (data == NULL) { | 2056 | if (data == NULL) { |
2045 | dprintk("%s: missing data argument\n", __FUNCTION__); | 2057 | dprintk("%s: missing data argument\n", __FUNCTION__); |
2046 | return ERR_PTR(-EINVAL); | 2058 | return -EINVAL; |
2047 | } | 2059 | } |
2048 | if (data->version <= 0 || data->version > NFS4_MOUNT_VERSION) { | 2060 | if (data->version <= 0 || data->version > NFS4_MOUNT_VERSION) { |
2049 | dprintk("%s: bad mount version\n", __FUNCTION__); | 2061 | dprintk("%s: bad mount version\n", __FUNCTION__); |
2050 | return ERR_PTR(-EINVAL); | 2062 | return -EINVAL; |
2051 | } | 2063 | } |
2052 | 2064 | ||
2053 | server = kzalloc(sizeof(struct nfs_server), GFP_KERNEL); | 2065 | server = kzalloc(sizeof(struct nfs_server), GFP_KERNEL); |
2054 | if (!server) | 2066 | if (!server) |
2055 | return ERR_PTR(-ENOMEM); | 2067 | return -ENOMEM; |
2056 | /* Zero out the NFS state stuff */ | 2068 | /* Zero out the NFS state stuff */ |
2057 | init_nfsv4_state(server); | 2069 | init_nfsv4_state(server); |
2058 | server->client = server->client_sys = server->client_acl = ERR_PTR(-EINVAL); | 2070 | server->client = server->client_sys = server->client_acl = ERR_PTR(-EINVAL); |
@@ -2074,33 +2086,41 @@ static struct super_block *nfs4_get_sb(struct file_system_type *fs_type, | |||
2074 | 2086 | ||
2075 | /* We now require that the mount process passes the remote address */ | 2087 | /* We now require that the mount process passes the remote address */ |
2076 | if (data->host_addrlen != sizeof(server->addr)) { | 2088 | if (data->host_addrlen != sizeof(server->addr)) { |
2077 | s = ERR_PTR(-EINVAL); | 2089 | error = -EINVAL; |
2078 | goto out_free; | 2090 | goto out_free; |
2079 | } | 2091 | } |
2080 | if (copy_from_user(&server->addr, data->host_addr, sizeof(server->addr))) { | 2092 | if (copy_from_user(&server->addr, data->host_addr, sizeof(server->addr))) { |
2081 | s = ERR_PTR(-EFAULT); | 2093 | error = -EFAULT; |
2082 | goto out_free; | 2094 | goto out_free; |
2083 | } | 2095 | } |
2084 | if (server->addr.sin_family != AF_INET || | 2096 | if (server->addr.sin_family != AF_INET || |
2085 | server->addr.sin_addr.s_addr == INADDR_ANY) { | 2097 | server->addr.sin_addr.s_addr == INADDR_ANY) { |
2086 | dprintk("%s: mount program didn't pass remote IP address!\n", | 2098 | dprintk("%s: mount program didn't pass remote IP address!\n", |
2087 | __FUNCTION__); | 2099 | __FUNCTION__); |
2088 | s = ERR_PTR(-EINVAL); | 2100 | error = -EINVAL; |
2089 | goto out_free; | 2101 | goto out_free; |
2090 | } | 2102 | } |
2091 | 2103 | ||
2092 | /* Fire up rpciod if not yet running */ | 2104 | /* Fire up rpciod if not yet running */ |
2093 | s = ERR_PTR(rpciod_up()); | 2105 | error = rpciod_up(); |
2094 | if (IS_ERR(s)) { | 2106 | if (error < 0) { |
2095 | dprintk("%s: couldn't start rpciod! Error = %ld\n", | 2107 | dprintk("%s: couldn't start rpciod! Error = %d\n", |
2096 | __FUNCTION__, PTR_ERR(s)); | 2108 | __FUNCTION__, error); |
2097 | goto out_free; | 2109 | goto out_free; |
2098 | } | 2110 | } |
2099 | 2111 | ||
2100 | s = sget(fs_type, nfs4_compare_super, nfs_set_super, server); | 2112 | s = sget(fs_type, nfs4_compare_super, nfs_set_super, server); |
2101 | 2113 | if (IS_ERR(s)) { | |
2102 | if (IS_ERR(s) || s->s_root) | 2114 | error = PTR_ERR(s); |
2103 | goto out_free; | 2115 | goto out_free; |
2116 | } | ||
2117 | |||
2118 | if (s->s_root) { | ||
2119 | kfree(server->mnt_path); | ||
2120 | kfree(server->hostname); | ||
2121 | kfree(server); | ||
2122 | return simple_set_mnt(mnt, s); | ||
2123 | } | ||
2104 | 2124 | ||
2105 | s->s_flags = flags; | 2125 | s->s_flags = flags; |
2106 | 2126 | ||
@@ -2108,17 +2128,17 @@ static struct super_block *nfs4_get_sb(struct file_system_type *fs_type, | |||
2108 | if (error) { | 2128 | if (error) { |
2109 | up_write(&s->s_umount); | 2129 | up_write(&s->s_umount); |
2110 | deactivate_super(s); | 2130 | deactivate_super(s); |
2111 | return ERR_PTR(error); | 2131 | return error; |
2112 | } | 2132 | } |
2113 | s->s_flags |= MS_ACTIVE; | 2133 | s->s_flags |= MS_ACTIVE; |
2114 | return s; | 2134 | return simple_set_mnt(mnt, s); |
2115 | out_err: | 2135 | out_err: |
2116 | s = (struct super_block *)p; | 2136 | error = PTR_ERR(p); |
2117 | out_free: | 2137 | out_free: |
2118 | kfree(server->mnt_path); | 2138 | kfree(server->mnt_path); |
2119 | kfree(server->hostname); | 2139 | kfree(server->hostname); |
2120 | kfree(server); | 2140 | kfree(server); |
2121 | return s; | 2141 | return error; |
2122 | } | 2142 | } |
2123 | 2143 | ||
2124 | static void nfs4_kill_super(struct super_block *sb) | 2144 | static void nfs4_kill_super(struct super_block *sb) |