From: Al Viro <viro@ZenIV.linux.org.uk>
To: David Howells <dhowells@redhat.com>
Cc: linux-fsdevel@vger.kernel.org, linux-afs@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 20/32] vfs: Make close() unmount the attached mount if so flagged [ver #8]
Date: Fri, 1 Jun 2018 02:52:55 +0100 [thread overview]
Message-ID: <20180601015255.GP30522@ZenIV.linux.org.uk> (raw)
In-Reply-To: <20180531191955.GG30522@ZenIV.linux.org.uk>
On Thu, May 31, 2018 at 08:19:55PM +0100, Al Viro wrote:
> On Fri, May 25, 2018 at 01:07:34AM +0100, David Howells wrote:
> > + if (unlikely(file->f_mode & FMODE_NEED_UNMOUNT))
> > + __detach_mounts(dentry);
> > +
>
> This is completely wrong. First of all, you want to dissolve the mount tree
> on file->f_path.mount, not every tree rooted at dentry equal to file->f_path.dentry.
> This is easily done - it would be a simple call of drop_collected_mounts(mnt)
> if not for one detail. You want it to happen only if the sucker isn't attached
> anywhere by that point. IOW,
> namespace_lock();
> lock_mount_hash();
> if (!real_mount(mnt)->mnt_ns)
> umount_tree(real_mount(mnt), UMOUNT_SYNC);
> unlock_mount_hash();
> namespace_unlock();
> and that's it. You don't need that magical mystery turd in move_mount() later
> in the series and all the infrastructure you grow for it.
>
> FWIW, I would've suggested this
> void drop_collected_mounts(struct vfsmount *mnt)
> {
> namespace_lock();
> lock_mount_hash();
> + if (!real_mount(mnt)->mnt_ns)
> + umount_tree(real_mount(mnt), UMOUNT_SYNC);
> - umount_tree(real_mount(mnt), UMOUNT_SYNC);
> unlock_mount_hash();
> namespace_unlock();
> }
>
> and in __fput()
> if (unlikely(file->f_mode & FMODE_NEED_UNMOUNT))
> drop_collected_mounts(mnt);
>
> All there is to it, AFAICS...
... except that it should be a separate primitive - drop_collected_mounts() is
used put_mnt_ns(), where the root definitely has non-NULL ->mnt_ns.
Another thing: the same issue (misuse of __detach_mounts()) exists in cleanup
path of do_o_path(). What's more, doing it there is pointless - if
do_dentry_open() has set FMODE_NEED_UNMOUNT, it either succeeds or calls fput()
itself. Either way, the caller should *not* do the cleanups done by fput().
Another thing: copy_mount_for_o_path() is bogus. Horrible calling conventions
aside, what the hell is that lock_mount() for? In do_loopback() we lock the
*mountpoint*; here the source gets locked, for no visible reason. What we
should do is something like this:
1) common helper -
static struct mount *__do_loopback(struct path *from, bool recurse)
{
struct mount *mnt = ERR_PTR(-EINVAL), *f = real_mount(from->mnt);
if (IS_MNT_UNBINDABLE(f))
return mnt;
if (!check_mnt(f) && from->dentry->d_op != &ns_dentry_operations)
return mnt;
if (!recurse && has_locked_children(f, from->dentry))
return mnt;
if (recurse)
mnt = copy_tree(f, from->dentry, CL_COPY_MNT_NS_FILE);
else
mnt = clone_mnt(f, from->dentry, 0);
if (!IS_ERR(mnt))
mnt->mnt.mnt_flags &= ~MNT_LOCKED;
return mnt;
}
2) in do_loopback() we are left with
static int do_loopback(struct path *path, const char *old_name,
int recurse)
{
struct path old_path;
struct mount *mnt, *parent;
struct mountpoint *mp;
int err;
if (!old_name || !*old_name)
return -EINVAL;
err = kern_path(old_name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &old_path);
if (err)
return err;
err = -EINVAL;
if (mnt_ns_loop(old_path.dentry))
goto out;
mp = lock_mount(path);
if (IS_ERR(mp)) {
err = PTR_ERR(mp);
goto out;
}
parent = real_mount(path->mnt);
if (!check_mnt(parent))
goto out2;
mnt = __do_loopback(&old_path, recurse);
if (IS_ERR(mnt)) {
err = PTR_ERR(mnt);
goto out2;
}
err = graft_tree(mnt, parent, mp);
if (err) {
lock_mount_hash();
umount_tree(mnt, UMOUNT_SYNC);
unlock_mount_hash();
}
out2:
unlock_mount(mp);
out:
path_put(&old_path);
return err;
}
3) copy_mount_for_o_path() with saner calling conventions:
int copy_mount_for_o_path(struct path *path, bool recurse)
{
struct mount *mnt = __do_loopback(path, recurse);
if (IS_ERR(mnt)) {
path_put(path);
return PTR_ERR(mnt);
}
mntput(path->mnt);
path->mnt = &mnt->mnt;
return 0;
}
4) in do_o_path():
static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
{
struct path path;
int error = path_lookupat(nd, flags, &path);
if (error)
return error;
if (file->f_flags & O_CLONE_MOUNT) {
error = copy_mount_for_o_path(&path,
!(file->f_flags & O_NON_RECURSIVE));
if (error < 0)
return error;
}
audit_inode(nd->name, path.dentry, 0);
error = vfs_open(&path, file, current_cred());
path_put(&path);
return error;
}
next prev parent reply other threads:[~2018-06-01 1:52 UTC|newest]
Thread overview: 87+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-25 0:05 [PATCH 00/32] VFS: Introduce filesystem context [ver #8] David Howells
2018-05-25 0:05 ` [PATCH 01/32] VFS: Suppress MS_* flag defs within the kernel unless explicitly enabled " David Howells
2018-05-25 0:05 ` [PATCH 02/32] vfs: Provide documentation for new mount API " David Howells
2018-05-25 0:05 ` [PATCH 03/32] VFS: Introduce the basic header for the new mount API's filesystem context " David Howells
2018-05-31 23:11 ` Al Viro
2018-05-31 23:13 ` Al Viro
2018-05-25 0:05 ` [PATCH 04/32] VFS: Add LSM hooks for the new mount API " David Howells
2018-05-25 0:05 ` [PATCH 05/32] selinux: Implement the new mount API LSM hooks " David Howells
2018-05-25 0:06 ` [PATCH 06/32] smack: Implement filesystem context security " David Howells
2018-05-25 0:06 ` [PATCH 07/32] apparmor: Implement security hooks for the new mount API " David Howells
2018-05-25 0:06 ` [PATCH 08/32] tomoyo: " David Howells
2018-05-25 0:06 ` [PATCH 09/32] VFS: Require specification of size of mount data for internal mounts " David Howells
2018-05-25 0:06 ` [PATCH 10/32] VFS: Implement a filesystem superblock creation/configuration context " David Howells
2018-06-07 19:50 ` Miklos Szeredi
2018-07-03 18:33 ` Eric Biggers
2018-07-03 21:53 ` David Howells
2018-07-03 21:58 ` Al Viro
2018-07-03 22:06 ` David Howells
2018-05-25 0:06 ` [PATCH 11/32] VFS: Remove unused code after filesystem context changes " David Howells
2018-05-25 0:06 ` [PATCH 12/32] procfs: Move proc_fill_super() to fs/proc/root.c " David Howells
2018-05-25 0:06 ` [PATCH 13/32] proc: Add fs_context support to procfs " David Howells
2018-05-25 0:06 ` [PATCH 14/32] ipc: Convert mqueue fs to fs_context " David Howells
2018-05-25 0:07 ` [PATCH 15/32] cpuset: Use " David Howells
2018-05-25 0:07 ` [PATCH 16/32] kernfs, sysfs, cgroup, intel_rdt: Support " David Howells
2018-06-21 18:47 ` [16/32] " Andrei Vagin
2018-06-22 12:52 ` David Howells
2018-06-22 15:30 ` Andrei Vagin
2018-06-22 16:57 ` Andrei Vagin
2018-06-23 23:34 ` David Howells
2018-05-25 0:07 ` [PATCH 17/32] hugetlbfs: Convert to " David Howells
2018-05-25 0:07 ` [PATCH 18/32] VFS: Remove kern_mount_data() " David Howells
2018-05-25 0:07 ` [PATCH 19/32] VFS: Implement fsopen() to prepare for a mount " David Howells
2018-05-31 21:25 ` Al Viro
2018-05-25 0:07 ` [PATCH 20/32] vfs: Make close() unmount the attached mount if so flagged " David Howells
2018-05-31 19:19 ` Al Viro
2018-05-31 19:26 ` Al Viro
2018-06-01 1:52 ` Al Viro [this message]
2018-06-01 3:18 ` Al Viro
2018-06-01 5:16 ` Al Viro
2018-05-25 0:07 ` [PATCH 21/32] VFS: Implement fsmount() to effect a pre-configured mount " David Howells
2018-06-04 15:05 ` Arnd Bergmann
2018-06-04 15:24 ` David Howells
2018-05-25 0:07 ` [PATCH 22/32] vfs: Provide an fspick() system call " David Howells
2018-05-25 0:07 ` [PATCH 23/32] VFS: Implement logging through fs_context " David Howells
2018-05-25 1:48 ` Joe Perches
2018-05-25 0:07 ` [PATCH 24/32] vfs: Add some logging to the core users of the fs_context log " David Howells
2018-05-25 0:08 ` [PATCH 25/32] afs: Add fs_context support " David Howells
2018-05-25 0:08 ` [PATCH 26/32] afs: Use fs_context to pass parameters over automount " David Howells
2018-06-07 1:58 ` Goldwyn Rodrigues
2018-06-07 20:45 ` David Howells
2018-05-25 0:08 ` [PATCH 27/32] vfs: Use a 'struct fd_cookie *' type for light fd handling " David Howells
2018-05-25 0:08 ` [PATCH 28/32] vfs: Store the fd_cookie in nameidata, not the dfd int " David Howells
2018-05-25 0:08 ` [PATCH 29/32] vfs: Don't mix FMODE_* flags with O_* flags " David Howells
2018-05-25 0:08 ` [PATCH 30/32] vfs: Allow cloning of a mount tree with open(O_PATH|O_CLONE_MOUNT) " David Howells
2018-06-01 6:26 ` Christoph Hellwig
2018-06-01 6:39 ` Al Viro
2018-06-01 8:27 ` David Howells
2018-06-02 3:09 ` Al Viro
2018-06-02 3:42 ` Al Viro
2018-06-02 4:04 ` Al Viro
2018-06-02 15:45 ` David Howells
2018-06-02 17:49 ` Al Viro
2018-06-03 0:55 ` [PATCH][RFC] open_tree(2) (was Re: [PATCH 30/32] vfs: Allow cloning of a mount tree with open(O_PATH|O_CLONE_MOUNT) [ver #8]) Al Viro
2018-06-04 10:34 ` Miklos Szeredi
2018-06-04 15:52 ` Al Viro
2018-06-04 15:59 ` Al Viro
2018-06-04 19:27 ` Miklos Szeredi
2018-06-04 15:27 ` David Howells
2018-06-04 17:16 ` Matthew Wilcox
2018-06-04 17:35 ` Al Viro
2018-06-04 19:38 ` Miklos Szeredi
2018-06-01 8:02 ` [PATCH 30/32] vfs: Allow cloning of a mount tree with open(O_PATH|O_CLONE_MOUNT) [ver #8] Amir Goldstein
2018-06-01 8:42 ` David Howells
2018-05-25 0:08 ` [PATCH 31/32] [RFC] fs: Add a move_mount() system call " David Howells
2018-05-31 21:20 ` Al Viro
2018-05-25 0:08 ` [PATCH 32/32] [RFC] fsinfo: Add a system call to allow querying of filesystem information " David Howells
2018-06-04 13:10 ` Arnd Bergmann
2018-06-04 15:01 ` David Howells
2018-06-04 16:00 ` Arnd Bergmann
2018-06-04 19:03 ` David Howells
2018-06-04 20:45 ` Arnd Bergmann
2018-05-31 20:56 ` Test program for move_mount() David Howells
2018-05-31 20:57 ` fsinfo test program David Howells
2018-06-15 4:18 ` [PATCH 00/32] VFS: Introduce filesystem context [ver #8] Eric W. Biederman
2018-06-18 20:30 ` David Howells
2018-06-18 21:33 ` Eric W. Biederman
2018-06-18 23:33 ` Theodore Y. Ts'o
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180601015255.GP30522@ZenIV.linux.org.uk \
--to=viro@zeniv.linux.org.uk \
--cc=dhowells@redhat.com \
--cc=linux-afs@lists.infradead.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).