[PATCH 1/3] audit: don't free_chunk() after fsnotify_add_mark()

[PATCH 1/3] audit: don't free_chunk() after fsnotify_add_mark()

[Miklos Szeredi]

From: Miklos Szeredi <mszeredi@suse.cz>

Don't do free_chunk() after fsnotify_add_mark().  That one does a delayed unref
via the destroy list and this results in use-after-free.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
CC: stable@vger.kernel.org
---
 kernel/audit_tree.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
index 5bf0790..d52d247 100644
--- a/kernel/audit_tree.c
+++ b/kernel/audit_tree.c
@@ -259,7 +259,7 @@ static void untag_chunk(struct node *p)
 
 	fsnotify_duplicate_mark(&new->mark, entry);
 	if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, NULL, 1)) {
-		free_chunk(new);
+		fsnotify_put_mark(&new->mark);
 		goto Fallback;
 	}
 
@@ -322,7 +322,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
 
 	entry = &chunk->mark;
 	if (fsnotify_add_mark(entry, audit_tree_group, inode, NULL, 0)) {
-		free_chunk(chunk);
+		fsnotify_put_mark(entry);
 		return -ENOSPC;
 	}
 
@@ -396,7 +396,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 	fsnotify_duplicate_mark(chunk_entry, old_entry);
 	if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, NULL, 1)) {
 		spin_unlock(&old_entry->lock);
-		free_chunk(chunk);
+		fsnotify_put_mark(chunk_entry);
 		fsnotify_put_mark(old_entry);
 		return -ENOSPC;
 	}
-- 
1.7.7

[PATCH 2/3] audit: fix refcounting in audit-tree

[Miklos Szeredi]

From: Miklos Szeredi <mszeredi@suse.cz>

Refcounting of fsnotify_mark in audit tree is broken.  E.g:

                              refcount
create_chunk
  alloc_chunk                 1
  fsnotify_add_mark           2

untag_chunk
  fsnotify_get_mark           3
  fsnotify_destroy_mark
    audit_tree_freeing_mark   2
  fsnotify_put_mark           1
  fsnotify_put_mark           0
  via destroy_list
    fsnotify_mark_destroy    -1

This was reported by various people as triggering Oops when stopping auditd.

We could just remove the put_mark from audit_tree_freeing_mark() but that would
break freeing via inode destruction.  So this patch simply omits a put_mark
after calling destroy_mark (or adds a get_mark before).

Next patch will clean up the remaining mess.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
CC: stable@vger.kernel.org
---
 kernel/audit_tree.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
index d52d247..31fdc48 100644
--- a/kernel/audit_tree.c
+++ b/kernel/audit_tree.c
@@ -250,7 +250,6 @@ static void untag_chunk(struct node *p)
 		spin_unlock(&hash_lock);
 		spin_unlock(&entry->lock);
 		fsnotify_destroy_mark(entry);
-		fsnotify_put_mark(entry);
 		goto out;
 	}
 
@@ -293,7 +292,6 @@ static void untag_chunk(struct node *p)
 	spin_unlock(&hash_lock);
 	spin_unlock(&entry->lock);
 	fsnotify_destroy_mark(entry);
-	fsnotify_put_mark(entry);
 	goto out;
 
 Fallback:
@@ -332,6 +330,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
 		spin_unlock(&hash_lock);
 		chunk->dead = 1;
 		spin_unlock(&entry->lock);
+		fsnotify_get_mark(entry);
 		fsnotify_destroy_mark(entry);
 		fsnotify_put_mark(entry);
 		return 0;
@@ -412,6 +411,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 		spin_unlock(&chunk_entry->lock);
 		spin_unlock(&old_entry->lock);
 
+		fsnotify_get_mark(chunk_entry);
 		fsnotify_destroy_mark(chunk_entry);
 
 		fsnotify_put_mark(chunk_entry);
@@ -445,7 +445,6 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 	spin_unlock(&old_entry->lock);
 	fsnotify_destroy_mark(old_entry);
 	fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */
-	fsnotify_put_mark(old_entry); /* and kill it */
 	return 0;
 }
 
-- 
1.7.7

Re: [PATCH 1/3] audit: don't free_chunk() after fsnotify_add_mark()

[Eric Paris]

On Mon, 2012-06-25 at 19:46 +0200, Miklos Szeredi wrote:
> From: Miklos Szeredi <mszeredi@suse.cz>
> 
> Don't do free_chunk() after fsnotify_add_mark().  That one does a delayed unref
> via the destroy list and this results in use-after-free.
> 
> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
> CC: stable@vger.kernel.org

Al, can you send these along?  I am not going to see a computer again
for a month.  Hurray!

Acked-by: Eric Paris <eparis@redhat.com>

> ---
>  kernel/audit_tree.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
> index 5bf0790..d52d247 100644
> --- a/kernel/audit_tree.c
> +++ b/kernel/audit_tree.c
> @@ -259,7 +259,7 @@ static void untag_chunk(struct node *p)
>  
>  	fsnotify_duplicate_mark(&new->mark, entry);
>  	if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, NULL, 1)) {
> -		free_chunk(new);
> +		fsnotify_put_mark(&new->mark);
>  		goto Fallback;
>  	}
>  
> @@ -322,7 +322,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
>  
>  	entry = &chunk->mark;
>  	if (fsnotify_add_mark(entry, audit_tree_group, inode, NULL, 0)) {
> -		free_chunk(chunk);
> +		fsnotify_put_mark(entry);
>  		return -ENOSPC;
>  	}
>  
> @@ -396,7 +396,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
>  	fsnotify_duplicate_mark(chunk_entry, old_entry);
>  	if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, NULL, 1)) {
>  		spin_unlock(&old_entry->lock);
> -		free_chunk(chunk);
> +		fsnotify_put_mark(chunk_entry);
>  		fsnotify_put_mark(old_entry);
>  		return -ENOSPC;
>  	}

Re: [PATCH 2/3] audit: fix refcounting in audit-tree

[Eric Paris]

On Mon, 2012-06-25 at 19:46 +0200, Miklos Szeredi wrote:
> From: Miklos Szeredi <mszeredi@suse.cz>
> 
> Refcounting of fsnotify_mark in audit tree is broken.  E.g:
> 
>                               refcount
> create_chunk
>   alloc_chunk                 1
>   fsnotify_add_mark           2
> 
> untag_chunk
>   fsnotify_get_mark           3
>   fsnotify_destroy_mark
>     audit_tree_freeing_mark   2
>   fsnotify_put_mark           1
>   fsnotify_put_mark           0
>   via destroy_list
>     fsnotify_mark_destroy    -1
> 
> This was reported by various people as triggering Oops when stopping auditd.
> 
> We could just remove the put_mark from audit_tree_freeing_mark() but that would
> break freeing via inode destruction.  So this patch simply omits a put_mark
> after calling destroy_mark (or adds a get_mark before).
> 
> Next patch will clean up the remaining mess.
> 
> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
> CC: stable@vger.kernel.org

Agreed this is needed.  My changelog was:

    audit: fix ref count problems in audit trees
    
    Before the switch from in kernel inotify to fsnotify for audit trees the
    code regularly did:
        inotify_evict_watch(&chunk->watch);
        put_inotify_watch(&chunk->watch);
    
    I translated this in fsnotify_speak into:
        fsnotify_destroy_mark_by_entry(chunk_entry);
        fsnotify_put_mark(chunk_entry);
    
    The problem is that the inotify_evict_watch function actually took a
    reference on chunk->watch, which is what was being dropped by
    put_inotify_watch().  The fsnotify code does not take such a reference
    during fsnotify_destroy_mark_by_entry().  Thus we are dropping reference
    counts prematurely and eventually we hit a use after free!  Whoops!
    
    Fix these call sites to not drop the extra reference.
    
    Reported-by: Valentin Avram <aval13@gmail.com>
    Reported-by: Peter Moody <pmoody@google.com>
    Partial-patch-by: Marcelo Cerri <mhcerri@linux.vnet.ibm.com>
    Signed-off-by: Eric Paris <eparis@redhat.com>

Maybe you can use some of that changelog in your next post?  (If you do
one?)  The only reason you would repost is because I don't understand
why you took a ref in some places instead of just not dropping it
everywhere...

> ---
>  kernel/audit_tree.c |    5 ++---
>  1 files changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
> index d52d247..31fdc48 100644
> --- a/kernel/audit_tree.c
> +++ b/kernel/audit_tree.c
> @@ -250,7 +250,6 @@ static void untag_chunk(struct node *p)
>  		spin_unlock(&hash_lock);
>  		spin_unlock(&entry->lock);
>  		fsnotify_destroy_mark(entry);
> -		fsnotify_put_mark(entry);
>  		goto out;
>  	}
>  
> @@ -293,7 +292,6 @@ static void untag_chunk(struct node *p)
>  	spin_unlock(&hash_lock);
>  	spin_unlock(&entry->lock);
>  	fsnotify_destroy_mark(entry);
> -	fsnotify_put_mark(entry);
>  	goto out;
>  
>  Fallback:
> @@ -332,6 +330,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
>  		spin_unlock(&hash_lock);
>  		chunk->dead = 1;
>  		spin_unlock(&entry->lock);
> +		fsnotify_get_mark(entry);
>  		fsnotify_destroy_mark(entry);
>  		fsnotify_put_mark(entry);
>  		return 0;

Like here?  Why not just avoid the atomic op altogether?

> @@ -412,6 +411,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
>  		spin_unlock(&chunk_entry->lock);
>  		spin_unlock(&old_entry->lock);
>  
> +		fsnotify_get_mark(chunk_entry);
>  		fsnotify_destroy_mark(chunk_entry);
>  
>  		fsnotify_put_mark(chunk_entry);
> @@ -445,7 +445,6 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
>  	spin_unlock(&old_entry->lock);
>  	fsnotify_destroy_mark(old_entry);
>  	fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */
> -	fsnotify_put_mark(old_entry); /* and kill it */
>  	return 0;
>  }
>  

Re: [PATCH 2/3] audit: fix refcounting in audit-tree

[Miklos Szeredi]

On Fri, 2012-07-06 at 13:52 -0400, Eric Paris wrote:
> On Mon, 2012-06-25 at 19:46 +0200, Miklos Szeredi wrote:
> > From: Miklos Szeredi <mszeredi@suse.cz>
> > 
> > Refcounting of fsnotify_mark in audit tree is broken.  E.g:
> > 
> >                               refcount
> > create_chunk
> >   alloc_chunk                 1
> >   fsnotify_add_mark           2
> > 
> > untag_chunk
> >   fsnotify_get_mark           3
> >   fsnotify_destroy_mark
> >     audit_tree_freeing_mark   2
> >   fsnotify_put_mark           1
> >   fsnotify_put_mark           0
> >   via destroy_list
> >     fsnotify_mark_destroy    -1
> > 
> > This was reported by various people as triggering Oops when stopping auditd.
> > 
> > We could just remove the put_mark from audit_tree_freeing_mark() but that would
> > break freeing via inode destruction.  So this patch simply omits a put_mark
> > after calling destroy_mark (or adds a get_mark before).
> > 
> > Next patch will clean up the remaining mess.
> > 
> > Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
> > CC: stable@vger.kernel.org
> 
> Agreed this is needed.  My changelog was:
> 
>     audit: fix ref count problems in audit trees
>     
>     Before the switch from in kernel inotify to fsnotify for audit trees the
>     code regularly did:
>         inotify_evict_watch(&chunk->watch);
>         put_inotify_watch(&chunk->watch);
>     
>     I translated this in fsnotify_speak into:
>         fsnotify_destroy_mark_by_entry(chunk_entry);
>         fsnotify_put_mark(chunk_entry);
>     
>     The problem is that the inotify_evict_watch function actually took a
>     reference on chunk->watch, which is what was being dropped by
>     put_inotify_watch().  The fsnotify code does not take such a reference
>     during fsnotify_destroy_mark_by_entry().  Thus we are dropping reference
>     counts prematurely and eventually we hit a use after free!  Whoops!
>     
>     Fix these call sites to not drop the extra reference.
>     
>     Reported-by: Valentin Avram <aval13@gmail.com>
>     Reported-by: Peter Moody <pmoody@google.com>
>     Partial-patch-by: Marcelo Cerri <mhcerri@linux.vnet.ibm.com>
>     Signed-off-by: Eric Paris <eparis@redhat.com>
> 
> Maybe you can use some of that changelog in your next post?  (If you do
> one?)  The only reason you would repost is because I don't understand
> why you took a ref in some places instead of just not dropping it
> everywhere...
> 
> > ---
> >  kernel/audit_tree.c |    5 ++---
> >  1 files changed, 2 insertions(+), 3 deletions(-)
> > 
> > diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
> > index d52d247..31fdc48 100644
> > --- a/kernel/audit_tree.c
> > +++ b/kernel/audit_tree.c
> > @@ -250,7 +250,6 @@ static void untag_chunk(struct node *p)
> >  		spin_unlock(&hash_lock);
> >  		spin_unlock(&entry->lock);
> >  		fsnotify_destroy_mark(entry);
> > -		fsnotify_put_mark(entry);
> >  		goto out;
> >  	}
> >  
> > @@ -293,7 +292,6 @@ static void untag_chunk(struct node *p)
> >  	spin_unlock(&hash_lock);
> >  	spin_unlock(&entry->lock);
> >  	fsnotify_destroy_mark(entry);
> > -	fsnotify_put_mark(entry);
> >  	goto out;
> >  
> >  Fallback:
> > @@ -332,6 +330,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
> >  		spin_unlock(&hash_lock);
> >  		chunk->dead = 1;
> >  		spin_unlock(&entry->lock);
> > +		fsnotify_get_mark(entry);
> >  		fsnotify_destroy_mark(entry);
> >  		fsnotify_put_mark(entry);
> >  		return 0;
> 
> Like here?  Why not just avoid the atomic op altogether?

Simply because fsnotify_destroy_mark() assumes that the caller is
holding a reference to the mark (or the inode is keeping the mark
pinned, not the case here AFAICS).  Without that ref it can Oops.

> 
> > @@ -412,6 +411,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
> >  		spin_unlock(&chunk_entry->lock);
> >  		spin_unlock(&old_entry->lock);
> >  
> > +		fsnotify_get_mark(chunk_entry);
> >  		fsnotify_destroy_mark(chunk_entry);
> >  
> >  		fsnotify_put_mark(chunk_entry);
> > @@ -445,7 +445,6 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
> >  	spin_unlock(&old_entry->lock);
> >  	fsnotify_destroy_mark(old_entry);
> >  	fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */
> > -	fsnotify_put_mark(old_entry); /* and kill it */
> >  	return 0;
> >  }
> >  
> 
>