nfsd4: check for negative dentry before use in nfsv4 readdir
[safe/jmp/linux-2.6] / fs / namespace.c
index 6e283c9..4119620 100644 (file)
@@ -27,6 +27,7 @@
 #include <linux/ramfs.h>
 #include <linux/log2.h>
 #include <linux/idr.h>
+#include <linux/fs_struct.h>
 #include <asm/uaccess.h>
 #include <asm/unistd.h>
 #include "pnode.h"
@@ -397,11 +398,10 @@ static void __mnt_unmake_readonly(struct vfsmount *mnt)
        spin_unlock(&vfsmount_lock);
 }
 
-int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
+void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
 {
        mnt->mnt_sb = sb;
        mnt->mnt_root = dget(sb->s_root);
-       return 0;
 }
 
 EXPORT_SYMBOL(simple_set_mnt);
@@ -614,9 +614,11 @@ static inline void __mntput(struct vfsmount *mnt)
         */
        for_each_possible_cpu(cpu) {
                struct mnt_writer *cpu_writer = &per_cpu(mnt_writers, cpu);
-               if (cpu_writer->mnt != mnt)
-                       continue;
                spin_lock(&cpu_writer->lock);
+               if (cpu_writer->mnt != mnt) {
+                       spin_unlock(&cpu_writer->lock);
+                       continue;
+               }
                atomic_add(cpu_writer->count, &mnt->__mnt_writers);
                cpu_writer->count = 0;
                /*
@@ -778,6 +780,7 @@ static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)
                { MNT_NOATIME, ",noatime" },
                { MNT_NODIRATIME, ",nodiratime" },
                { MNT_RELATIME, ",relatime" },
+               { MNT_STRICTATIME, ",strictatime" },
                { 0, NULL }
        };
        const struct proc_fs_info *fs_infop;
@@ -1128,7 +1131,7 @@ static int do_umount(struct vfsmount *mnt, int flags)
  * unixes. Our API is identical to OSF/1 to avoid making a mess of AMD
  */
 
-asmlinkage long sys_umount(char __user * name, int flags)
+SYSCALL_DEFINE2(umount, char __user *, name, int, flags)
 {
        struct path path;
        int retval;
@@ -1160,26 +1163,26 @@ out:
 /*
  *     The 2.0 compatible umount. No flags.
  */
-asmlinkage long sys_oldumount(char __user * name)
+SYSCALL_DEFINE1(oldumount, char __user *, name)
 {
        return sys_umount(name, 0);
 }
 
 #endif
 
-static int mount_is_safe(struct nameidata *nd)
+static int mount_is_safe(struct path *path)
 {
        if (capable(CAP_SYS_ADMIN))
                return 0;
        return -EPERM;
 #ifdef notyet
-       if (S_ISLNK(nd->path.dentry->d_inode->i_mode))
+       if (S_ISLNK(path->dentry->d_inode->i_mode))
                return -EPERM;
-       if (nd->path.dentry->d_inode->i_mode & S_ISVTX) {
-               if (current->uid != nd->path.dentry->d_inode->i_uid)
+       if (path->dentry->d_inode->i_mode & S_ISVTX) {
+               if (current_uid() != path->dentry->d_inode->i_uid)
                        return -EPERM;
        }
-       if (vfs_permission(nd, MAY_WRITE))
+       if (inode_permission(path->dentry->d_inode, MAY_WRITE))
                return -EPERM;
        return 0;
 #endif
@@ -1374,7 +1377,7 @@ static int attach_recursive_mnt(struct vfsmount *source_mnt,
        if (parent_path) {
                detach_mnt(source_mnt, parent_path);
                attach_mnt(source_mnt, path);
-               touch_mnt_namespace(current->nsproxy->mnt_ns);
+               touch_mnt_namespace(parent_path->mnt->mnt_ns);
        } else {
                mnt_set_mountpoint(dest_mnt, dest_dentry, source_mnt);
                commit_tree(source_mnt);
@@ -1425,11 +1428,10 @@ out_unlock:
 
 /*
  * recursively change the type of the mountpoint.
- * noinline this do_mount helper to save do_mount stack space.
  */
-static noinline int do_change_type(struct nameidata *nd, int flag)
+static int do_change_type(struct path *path, int flag)
 {
-       struct vfsmount *m, *mnt = nd->path.mnt;
+       struct vfsmount *m, *mnt = path->mnt;
        int recurse = flag & MS_REC;
        int type = flag & ~MS_REC;
        int err = 0;
@@ -1437,7 +1439,7 @@ static noinline int do_change_type(struct nameidata *nd, int flag)
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
 
-       if (nd->path.dentry != nd->path.mnt->mnt_root)
+       if (path->dentry != path->mnt->mnt_root)
                return -EINVAL;
 
        down_write(&namespace_sem);
@@ -1459,40 +1461,39 @@ static noinline int do_change_type(struct nameidata *nd, int flag)
 
 /*
  * do loopback mount.
- * noinline this do_mount helper to save do_mount stack space.
  */
-static noinline int do_loopback(struct nameidata *nd, char *old_name,
+static int do_loopback(struct path *path, char *old_name,
                                int recurse)
 {
-       struct nameidata old_nd;
+       struct path old_path;
        struct vfsmount *mnt = NULL;
-       int err = mount_is_safe(nd);
+       int err = mount_is_safe(path);
        if (err)
                return err;
        if (!old_name || !*old_name)
                return -EINVAL;
-       err = path_lookup(old_name, LOOKUP_FOLLOW, &old_nd);
+       err = kern_path(old_name, LOOKUP_FOLLOW, &old_path);
        if (err)
                return err;
 
        down_write(&namespace_sem);
        err = -EINVAL;
-       if (IS_MNT_UNBINDABLE(old_nd.path.mnt))
+       if (IS_MNT_UNBINDABLE(old_path.mnt))
                goto out;
 
-       if (!check_mnt(nd->path.mnt) || !check_mnt(old_nd.path.mnt))
+       if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
                goto out;
 
        err = -ENOMEM;
        if (recurse)
-               mnt = copy_tree(old_nd.path.mnt, old_nd.path.dentry, 0);
+               mnt = copy_tree(old_path.mnt, old_path.dentry, 0);
        else
-               mnt = clone_mnt(old_nd.path.mnt, old_nd.path.dentry, 0);
+               mnt = clone_mnt(old_path.mnt, old_path.dentry, 0);
 
        if (!mnt)
                goto out;
 
-       err = graft_tree(mnt, &nd->path);
+       err = graft_tree(mnt, path);
        if (err) {
                LIST_HEAD(umount_list);
                spin_lock(&vfsmount_lock);
@@ -1503,7 +1504,7 @@ static noinline int do_loopback(struct nameidata *nd, char *old_name,
 
 out:
        up_write(&namespace_sem);
-       path_put(&old_nd.path);
+       path_put(&old_path);
        return err;
 }
 
@@ -1528,33 +1529,37 @@ static int change_mount_flags(struct vfsmount *mnt, int ms_flags)
  * change filesystem flags. dir should be a physical root of filesystem.
  * If you've mounted a non-root directory somewhere and want to do remount
  * on it - tough luck.
- * noinline this do_mount helper to save do_mount stack space.
  */
-static noinline int do_remount(struct nameidata *nd, int flags, int mnt_flags,
+static int do_remount(struct path *path, int flags, int mnt_flags,
                      void *data)
 {
        int err;
-       struct super_block *sb = nd->path.mnt->mnt_sb;
+       struct super_block *sb = path->mnt->mnt_sb;
 
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
 
-       if (!check_mnt(nd->path.mnt))
+       if (!check_mnt(path->mnt))
                return -EINVAL;
 
-       if (nd->path.dentry != nd->path.mnt->mnt_root)
+       if (path->dentry != path->mnt->mnt_root)
                return -EINVAL;
 
        down_write(&sb->s_umount);
        if (flags & MS_BIND)
-               err = change_mount_flags(nd->path.mnt, flags);
+               err = change_mount_flags(path->mnt, flags);
        else
                err = do_remount_sb(sb, flags, data, 0);
        if (!err)
-               nd->path.mnt->mnt_flags = mnt_flags;
+               path->mnt->mnt_flags = mnt_flags;
        up_write(&sb->s_umount);
-       if (!err)
-               security_sb_post_remount(nd->path.mnt, flags, data);
+       if (!err) {
+               security_sb_post_remount(path->mnt, flags, data);
+
+               spin_lock(&vfsmount_lock);
+               touch_mnt_namespace(path->mnt->mnt_ns);
+               spin_unlock(&vfsmount_lock);
+       }
        return err;
 }
 
@@ -1568,90 +1573,85 @@ static inline int tree_contains_unbindable(struct vfsmount *mnt)
        return 0;
 }
 
-/*
- * noinline this do_mount helper to save do_mount stack space.
- */
-static noinline int do_move_mount(struct nameidata *nd, char *old_name)
+static int do_move_mount(struct path *path, char *old_name)
 {
-       struct nameidata old_nd;
-       struct path parent_path;
+       struct path old_path, parent_path;
        struct vfsmount *p;
        int err = 0;
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
        if (!old_name || !*old_name)
                return -EINVAL;
-       err = path_lookup(old_name, LOOKUP_FOLLOW, &old_nd);
+       err = kern_path(old_name, LOOKUP_FOLLOW, &old_path);
        if (err)
                return err;
 
        down_write(&namespace_sem);
-       while (d_mountpoint(nd->path.dentry) &&
-              follow_down(&nd->path.mnt, &nd->path.dentry))
+       while (d_mountpoint(path->dentry) &&
+              follow_down(&path->mnt, &path->dentry))
                ;
        err = -EINVAL;
-       if (!check_mnt(nd->path.mnt) || !check_mnt(old_nd.path.mnt))
+       if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
                goto out;
 
        err = -ENOENT;
-       mutex_lock(&nd->path.dentry->d_inode->i_mutex);
-       if (IS_DEADDIR(nd->path.dentry->d_inode))
+       mutex_lock(&path->dentry->d_inode->i_mutex);
+       if (IS_DEADDIR(path->dentry->d_inode))
                goto out1;
 
-       if (!IS_ROOT(nd->path.dentry) && d_unhashed(nd->path.dentry))
+       if (!IS_ROOT(path->dentry) && d_unhashed(path->dentry))
                goto out1;
 
        err = -EINVAL;
-       if (old_nd.path.dentry != old_nd.path.mnt->mnt_root)
+       if (old_path.dentry != old_path.mnt->mnt_root)
                goto out1;
 
-       if (old_nd.path.mnt == old_nd.path.mnt->mnt_parent)
+       if (old_path.mnt == old_path.mnt->mnt_parent)
                goto out1;
 
-       if (S_ISDIR(nd->path.dentry->d_inode->i_mode) !=
-             S_ISDIR(old_nd.path.dentry->d_inode->i_mode))
+       if (S_ISDIR(path->dentry->d_inode->i_mode) !=
+             S_ISDIR(old_path.dentry->d_inode->i_mode))
                goto out1;
        /*
         * Don't move a mount residing in a shared parent.
         */
-       if (old_nd.path.mnt->mnt_parent &&
-           IS_MNT_SHARED(old_nd.path.mnt->mnt_parent))
+       if (old_path.mnt->mnt_parent &&
+           IS_MNT_SHARED(old_path.mnt->mnt_parent))
                goto out1;
        /*
         * Don't move a mount tree containing unbindable mounts to a destination
         * mount which is shared.
         */
-       if (IS_MNT_SHARED(nd->path.mnt) &&
-           tree_contains_unbindable(old_nd.path.mnt))
+       if (IS_MNT_SHARED(path->mnt) &&
+           tree_contains_unbindable(old_path.mnt))
                goto out1;
        err = -ELOOP;
-       for (p = nd->path.mnt; p->mnt_parent != p; p = p->mnt_parent)
-               if (p == old_nd.path.mnt)
+       for (p = path->mnt; p->mnt_parent != p; p = p->mnt_parent)
+               if (p == old_path.mnt)
                        goto out1;
 
-       err = attach_recursive_mnt(old_nd.path.mnt, &nd->path, &parent_path);
+       err = attach_recursive_mnt(old_path.mnt, path, &parent_path);
        if (err)
                goto out1;
 
        /* if the mount is moved, it should no longer be expire
         * automatically */
-       list_del_init(&old_nd.path.mnt->mnt_expire);
+       list_del_init(&old_path.mnt->mnt_expire);
 out1:
-       mutex_unlock(&nd->path.dentry->d_inode->i_mutex);
+       mutex_unlock(&path->dentry->d_inode->i_mutex);
 out:
        up_write(&namespace_sem);
        if (!err)
                path_put(&parent_path);
-       path_put(&old_nd.path);
+       path_put(&old_path);
        return err;
 }
 
 /*
  * create a new mount for userspace and request it to be added into the
  * namespace's tree
- * noinline this do_mount helper to save do_mount stack space.
  */
-static noinline int do_new_mount(struct nameidata *nd, char *type, int flags,
+static int do_new_mount(struct path *path, char *type, int flags,
                        int mnt_flags, char *name, void *data)
 {
        struct vfsmount *mnt;
@@ -1667,7 +1667,7 @@ static noinline int do_new_mount(struct nameidata *nd, char *type, int flags,
        if (IS_ERR(mnt))
                return PTR_ERR(mnt);
 
-       return do_add_mount(mnt, &nd->path, mnt_flags, NULL);
+       return do_add_mount(mnt, path, mnt_flags, NULL);
 }
 
 /*
@@ -1818,8 +1818,8 @@ static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts)
                while (!list_empty(&graveyard)) {
                        m = list_first_entry(&graveyard, struct vfsmount,
                                                mnt_expire);
-                       touch_mnt_namespace(mnt->mnt_ns);
-                       umount_tree(mnt, 1, umounts);
+                       touch_mnt_namespace(m->mnt_ns);
+                       umount_tree(m, 1, umounts);
                }
        }
 }
@@ -1902,7 +1902,7 @@ int copy_mount_options(const void __user * data, unsigned long *where)
 long do_mount(char *dev_name, char *dir_name, char *type_page,
                  unsigned long flags, void *data_page)
 {
-       struct nameidata nd;
+       struct path path;
        int retval = 0;
        int mnt_flags = 0;
 
@@ -1920,6 +1920,10 @@ long do_mount(char *dev_name, char *dir_name, char *type_page,
        if (data_page)
                ((char *)data_page)[PAGE_SIZE - 1] = 0;
 
+       /* Default to relatime unless overriden */
+       if (!(flags & MS_NOATIME))
+               mnt_flags |= MNT_RELATIME;
+
        /* Separate the per-mountpoint flags */
        if (flags & MS_NOSUID)
                mnt_flags |= MNT_NOSUID;
@@ -1931,38 +1935,39 @@ long do_mount(char *dev_name, char *dir_name, char *type_page,
                mnt_flags |= MNT_NOATIME;
        if (flags & MS_NODIRATIME)
                mnt_flags |= MNT_NODIRATIME;
-       if (flags & MS_RELATIME)
-               mnt_flags |= MNT_RELATIME;
+       if (flags & MS_STRICTATIME)
+               mnt_flags &= ~(MNT_RELATIME | MNT_NOATIME);
        if (flags & MS_RDONLY)
                mnt_flags |= MNT_READONLY;
 
        flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE |
-                  MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT);
+                  MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT |
+                  MS_STRICTATIME);
 
        /* ... and get the mountpoint */
-       retval = path_lookup(dir_name, LOOKUP_FOLLOW, &nd);
+       retval = kern_path(dir_name, LOOKUP_FOLLOW, &path);
        if (retval)
                return retval;
 
-       retval = security_sb_mount(dev_name, &nd.path,
+       retval = security_sb_mount(dev_name, &path,
                                   type_page, flags, data_page);
        if (retval)
                goto dput_out;
 
        if (flags & MS_REMOUNT)
-               retval = do_remount(&nd, flags & ~MS_REMOUNT, mnt_flags,
+               retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
                                    data_page);
        else if (flags & MS_BIND)
-               retval = do_loopback(&nd, dev_name, flags & MS_REC);
+               retval = do_loopback(&path, dev_name, flags & MS_REC);
        else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
-               retval = do_change_type(&nd, flags);
+               retval = do_change_type(&path, flags);
        else if (flags & MS_MOVE)
-               retval = do_move_mount(&nd, dev_name);
+               retval = do_move_mount(&path, dev_name);
        else
-               retval = do_new_mount(&nd, type_page, flags, mnt_flags,
+               retval = do_new_mount(&path, type_page, flags, mnt_flags,
                                      dev_name, data_page);
 dput_out:
-       path_put(&nd.path);
+       path_put(&path);
        return retval;
 }
 
@@ -1993,7 +1998,7 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
        if (!new_ns->root) {
                up_write(&namespace_sem);
                kfree(new_ns);
-               return ERR_PTR(-ENOMEM);;
+               return ERR_PTR(-ENOMEM);
        }
        spin_lock(&vfsmount_lock);
        list_add_tail(&new_ns->list, &new_ns->root->mnt_list);
@@ -2048,9 +2053,8 @@ struct mnt_namespace *copy_mnt_ns(unsigned long flags, struct mnt_namespace *ns,
        return new_ns;
 }
 
-asmlinkage long sys_mount(char __user * dev_name, char __user * dir_name,
-                         char __user * type, unsigned long flags,
-                         void __user * data)
+SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name,
+               char __user *, type, unsigned long, flags, void __user *, data)
 {
        int retval;
        unsigned long data_page;
@@ -2091,66 +2095,6 @@ out1:
 }
 
 /*
- * Replace the fs->{rootmnt,root} with {mnt,dentry}. Put the old values.
- * It can block. Requires the big lock held.
- */
-void set_fs_root(struct fs_struct *fs, struct path *path)
-{
-       struct path old_root;
-
-       write_lock(&fs->lock);
-       old_root = fs->root;
-       fs->root = *path;
-       path_get(path);
-       write_unlock(&fs->lock);
-       if (old_root.dentry)
-               path_put(&old_root);
-}
-
-/*
- * Replace the fs->{pwdmnt,pwd} with {mnt,dentry}. Put the old values.
- * It can block. Requires the big lock held.
- */
-void set_fs_pwd(struct fs_struct *fs, struct path *path)
-{
-       struct path old_pwd;
-
-       write_lock(&fs->lock);
-       old_pwd = fs->pwd;
-       fs->pwd = *path;
-       path_get(path);
-       write_unlock(&fs->lock);
-
-       if (old_pwd.dentry)
-               path_put(&old_pwd);
-}
-
-static void chroot_fs_refs(struct path *old_root, struct path *new_root)
-{
-       struct task_struct *g, *p;
-       struct fs_struct *fs;
-
-       read_lock(&tasklist_lock);
-       do_each_thread(g, p) {
-               task_lock(p);
-               fs = p->fs;
-               if (fs) {
-                       atomic_inc(&fs->count);
-                       task_unlock(p);
-                       if (fs->root.dentry == old_root->dentry
-                           && fs->root.mnt == old_root->mnt)
-                               set_fs_root(fs, new_root);
-                       if (fs->pwd.dentry == old_root->dentry
-                           && fs->pwd.mnt == old_root->mnt)
-                               set_fs_pwd(fs, new_root);
-                       put_fs_struct(fs);
-               } else
-                       task_unlock(p);
-       } while_each_thread(g, p);
-       read_unlock(&tasklist_lock);
-}
-
-/*
  * pivot_root Semantics:
  * Moves the root file system of the current process to the directory put_old,
  * makes new_root as the new root file system of the current process, and sets
@@ -2175,8 +2119,8 @@ static void chroot_fs_refs(struct path *old_root, struct path *new_root)
  *    though, so you may need to say mount --bind /nfs/my_root /nfs/my_root
  *    first.
  */
-asmlinkage long sys_pivot_root(const char __user * new_root,
-                              const char __user * put_old)
+SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
+               const char __user *, put_old)
 {
        struct vfsmount *tmp;
        struct path new, old, parent_path, root_parent, root;