[PATCH] fix a leak in fcntl_dirnotify()

[PATCH] fix a leak in fcntl_dirnotify()

[Al Viro]

[into #fixes, unless somebody objects]

Lifetime of new_dn_mark is controlled by that of its ->fsn_mark,
pointed to by new_fsn_mark.  Unfortunately, a failure exit had
been inserted between the allocation of new_dn_mark and the
call of fsnotify_init_mark(), ending up with a leak.
    
Fixes: 1934b212615d "file: reclaim 24 bytes from f_owner"
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index c4cdaf5fa7ed..9fb73bafd41d 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -308,6 +308,10 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned int arg)
 		goto out_err;
 	}
 
+	error = file_f_owner_allocate(filp);
+	if (error)
+		goto out_err;
+
 	/* new fsnotify mark, we expect most fcntl calls to add a new mark */
 	new_dn_mark = kmem_cache_alloc(dnotify_mark_cache, GFP_KERNEL);
 	if (!new_dn_mark) {
@@ -315,10 +319,6 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned int arg)
 		goto out_err;
 	}
 
-	error = file_f_owner_allocate(filp);
-	if (error)
-		goto out_err;
-
 	/* set up the new_fsn_mark and new_dn_mark */
 	new_fsn_mark = &new_dn_mark->fsn_mark;
 	fsnotify_init_mark(new_fsn_mark, dnotify_group);

Re: [PATCH] fix a leak in fcntl_dirnotify()

[Christian Brauner]

On Sat, Jul 12, 2025 at 06:18:43PM +0100, Al Viro wrote:
> [into #fixes, unless somebody objects]
> 
> Lifetime of new_dn_mark is controlled by that of its ->fsn_mark,
> pointed to by new_fsn_mark.  Unfortunately, a failure exit had
> been inserted between the allocation of new_dn_mark and the
> call of fsnotify_init_mark(), ending up with a leak.
>     
> Fixes: 1934b212615d "file: reclaim 24 bytes from f_owner"
> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
> ---

I'll grab this into vfs.fixes and send it with the batch I have in
there this week.

Re: [PATCH] fix a leak in fcntl_dirnotify()

[Al Viro]

On Mon, Jul 14, 2025 at 10:14:38AM +0200, Christian Brauner wrote:
> On Sat, Jul 12, 2025 at 06:18:43PM +0100, Al Viro wrote:
> > [into #fixes, unless somebody objects]
> > 
> > Lifetime of new_dn_mark is controlled by that of its ->fsn_mark,
> > pointed to by new_fsn_mark.  Unfortunately, a failure exit had
> > been inserted between the allocation of new_dn_mark and the
> > call of fsnotify_init_mark(), ending up with a leak.
> >     
> > Fixes: 1934b212615d "file: reclaim 24 bytes from f_owner"
> > Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
> > ---
> 
> I'll grab this into vfs.fixes and send it with the batch I have in
> there this week.

OK...  Again, the only things I've got in the vicinity are post-rc1
fodder, so I've no preferences re which branch does that go through.