Re: [PATCH] fsnotify: don't put user context if it was never assigned

[Sasha Levin <sasha.levin@oracle.com>]

On 09/11/2014 04:43 PM, Andrew Morton wrote:
> On Tue, 29 Jul 2014 09:25:14 -0400 Sasha Levin <sasha.levin@oracle.com> wrote:
> 
>> > On some failure paths we may attempt to free user context even
>> > if it wasn't assigned yet. This will cause a NULL ptr deref
>> > and a kernel BUG.
> Are you able to identify "some failure paths"?  I spent some time
> grepping, but it's a pain.
> 
> Please try to include such info in changelogs because reviewers (ie,
> me) might want to review those callers to decide whether the bug lies
> elsewhere.
> 

Sorry about that.

The path I was looking at is in inotify_new_group():

        oevent = kmalloc(sizeof(struct inotify_event_info), GFP_KERNEL);
        if (unlikely(!oevent)) {
                fsnotify_destroy_group(group);
                return ERR_PTR(-ENOMEM);
        }

fsnotify_destroy_group() would get called here, but group->inotify_data.user
is only getting assigned later:

	group->inotify_data.user = get_current_user();


Thanks,
Sasha