Re: [PATCH 00/17] VFS: Filesystem information and notifications [ver #17]

From: Al Viro <viro@zeniv.linux.org.uk>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: Ian Kent <raven@themaw.net>, David Howells <dhowells@redhat.com>,
	Christian Brauner <christian.brauner@ubuntu.com>,
	James Bottomley <James.Bottomley@hansenpartnership.com>,
	Steven Whitehouse <swhiteho@redhat.com>,
	Miklos Szeredi <mszeredi@redhat.com>,
	Christian Brauner <christian@brauner.io>,
	Jann Horn <jannh@google.com>,
	"Darrick J. Wong" <darrick.wong@oracle.com>,
	Linux API <linux-api@vger.kernel.org>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	lkml <linux-kernel@vger.kernel.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [PATCH 00/17] VFS: Filesystem information and notifications [ver #17]
Date: Fri, 6 Mar 2020 20:56:18 +0000	[thread overview]
Message-ID: <20200306205618.GF23230@ZenIV.linux.org.uk> (raw)
In-Reply-To: <20200306204926.GE23230@ZenIV.linux.org.uk>

On Fri, Mar 06, 2020 at 08:49:26PM +0000, Al Viro wrote:
> On Fri, Mar 06, 2020 at 08:45:23PM +0000, Al Viro wrote:
> > On Fri, Mar 06, 2020 at 08:38:44PM +0000, Al Viro wrote:
> > > On Fri, Mar 06, 2020 at 08:37:05PM +0000, Al Viro wrote:
> > > 
> > > > You are misreading mntput_no_expire(), BTW - your get_mount() can
> > > > bloody well race with umount(2), hitting the moment when we are done
> > > > figuring out whether it's busy but hadn't cleaned ->mnt_ns (let alone
> > > > set MNT_DOOMED) yet.  If somebody calls umount(2) on a filesystem that
> > > > is not mounted anywhere else, they are not supposed to see the sucker
> > > > return 0 until the filesystem is shut down.  You break that.
> > > 
> > > While we are at it, d_alloc_parallel() requires i_rwsem on parent held
> > > at least shared.
> > 
> > Egads...  Let me see if I got it right - you are providing procfs symlinks
> > to objects on the internal mount of that thing.  And those objects happen
> > to be directories, so one can get to their parent that way.  Or am I misreading
> > that thing?
> 
> IDGI.  You have (in your lookup) kstrtoul, followed by snprintf, followed
> by strcmp and WARN_ON() in case of mismatch?  Is there any point in having
> stat(2) on "00" spew into syslog?  Confused...

AFAICS, refcounting in there cannot be right:
+static struct dentry *mountfs_lookup(struct inode *dir, struct dentry *dentry,
+                                    unsigned int flags)
+{
+       struct mountfs_entry *entry = dir->i_private;
+       int i = 0;
+               
+       if (entry) {
+               for (i = 0; i < ARRAY_SIZE(mountfs_attrs); i++)
+                       if (strcmp(mountfs_attrs[i], dentry->d_name.name) == 0)
+                               break;
+               if (i == ARRAY_SIZE(mountfs_attrs))
+                       return ERR_PTR(-ENOMEM);
+               i++;
+       } else {
+               entry = mountfs_get_entry(dentry->d_name.name);
+               if (!entry)
+                       return ERR_PTR(-ENOENT);
+       }
+                          
+       return mountfs_lookup_entry(dentry, entry, i);
+}
ends up consuming a reference in mountfs_lookup_entry() (at the very least,
drops it in case of inode allocation hitting OOM) and nothing in the
that loop in mountfs_lookup() appears to do a matching reference grab.