since when does linkat() on deleted /proc/$PID/fd/$num return ENOENT ?

[David Madore]

It used to be the case (last time I checked was around late 2008 or
early 2009) that deleted entries from /proc/$PID/fd/ could be linked
back to the filesystem by using linkat(,,,,AT_SYMLINK_FOLLOW).

Now this just returns ENOENT.

I'd like to understand when, how and why this change took place. What
commit introduced it and was it a deliberate move (e.g., because the
feature was a security issue of itself, or came into conflict with
something else) or was it accidental? Does it depend on the /proc
filesystem itself or on the target filesystem where the deleted file
used to reside?

(There's a Reddit thread, <URL:
http://www.reddit.com/r/programming/comments/7yx6f/how_to_undelete_any_open_deleted_file_in_linux/
>, where some people are reporting ENOENT on 2.6.27 or perhaps even
2.6.26, which helps but a bound on the change.)

(See also this thread: <URL:
http://comments.gmane.org/gmane.linux.kernel/1224071
>, where the question is not answered, however.)

--
David A. Madore
( http://www.madore.org/~david/ )
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

[Alexey Dobriyan]

On Fri, Mar 30, 2012 at 12:21:21PM +0200, David Madore wrote:
> It used to be the case (last time I checked was around late 2008 or
> early 2009) that deleted entries from /proc/$PID/fd/ could be linked
> back to the filesystem by using linkat(,,,,AT_SYMLINK_FOLLOW).
>
> Now this just returns ENOENT.
>
> I'd like to understand when, how and why this change took place. What
> commit introduced it and was it a deliberate move (e.g., because the
> feature was a security issue of itself, or came into conflict with
> something else) or was it accidental?

It was explicitly prohibited since 2.6.39:

commit aae8a97d3ec30788790d1720b71d76fd8eb44b73
Author: Aneesh Kumar K.V <aneesh...@linux.vnet.ibm.com>
Date: Sat Jan 29 18:43:27 2011 +0530

fs: Don't allow to create hardlink for deleted file

Add inode->i_nlink == 0 check in VFS. Some of the file systems
do this internally. A followup patch will remove those instance.
This is needed to ensure that with link by handle we don't allow
to create hardlink of an unlinked file. The check also prevent a race
between unlink and link

Signed-off-by: Aneesh Kumar K.V <aneesh...@linux.vnet.ibm.com>
Signed-off-by: Al Viro <vi...@zeniv.linux.org.uk>

diff --git a/fs/namei.c b/fs/namei.c
index 83e92ba..33be51a 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2906,7 +2906,11 @@ int vfs_link(struct dentry *old_dentry, struct inode *dir, struct dentry *new_de
return error;

mutex_lock(&inode->i_mutex);
- error = dir->i_op->link(old_dentry, dir, new_dentry);
+ /* Make sure we don't allow creating hardlink to an unlinked file */
+ if (inode->i_nlink == 0)
+ error = -ENOENT;
+ else
+ error = dir->i_op->link(old_dentry, dir, new_dentry);
mutex_unlock(&inode->i_mutex);
if (!error)
fsnotify_link(dir, inode, new_dentry);

[David Madore]

On Sat, Mar 31, 2012 at 09:53:05AM +0300, Alexey Dobriyan wrote:
> It was explicitly prohibited since 2.6.39:
>
> commit aae8a97d3ec30788790d1720b71d76fd8eb44b73
> Author: Aneesh Kumar K.V <aneesh...@linux.vnet.ibm.com>
> Date: Sat Jan 29 18:43:27 2011 +0530
>
> fs: Don't allow to create hardlink for deleted file

<snip>

Thanks for digging this for me!

--
David A. Madore
( http://www.madore.org/~david/ )

[Alexey Dobriyan]

On Sat, Mar 31, 2012 at 01:06:40PM +0200, David Madore wrote:
> On Sat, Mar 31, 2012 at 09:53:05AM +0300, Alexey Dobriyan wrote:
> > It was explicitly prohibited since 2.6.39:
> >
> > commit aae8a97d3ec30788790d1720b71d76fd8eb44b73
> > Author: Aneesh Kumar K.V <aneesh...@linux.vnet.ibm.com>
> > Date: Sat Jan 29 18:43:27 2011 +0530
> >
> > fs: Don't allow to create hardlink for deleted file
> <snip>
>
> Thanks for digging this for me!

Well, the question if this counts as userspace-visible change.