Messages in this thread | | | Date | Tue, 13 May 2014 00:18:10 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: [patch 3/3] futex: Prevent attaching to kernel threads |
| |
On Mon, 12 May 2014, Davidlohr Bueso wrote: > On Mon, 2014-05-12 at 20:45 +0000, Thomas Gleixner wrote: > > We happily allow userspace to declare a random kernel thread to be the > > owner of a user space PI futex. > > > > Found while analysing the fallout of Dave Jones syscall fuzzer. > > > > Signed-off-by: Thomas Gleixner <tglx@linutronix.de> > > Cc: stable@vger.kernel.org > > --- > > kernel/futex.c | 7 +++++++ > > 1 file changed, 7 insertions(+) > > > > Index: linux-2.6/kernel/futex.c > > =================================================================== > > --- linux-2.6.orig/kernel/futex.c > > +++ linux-2.6/kernel/futex.c > > @@ -682,6 +682,8 @@ static struct task_struct * futex_find_g > > p = find_task_by_vpid(pid); > > if (p) > > get_task_struct(p); > > + else > > + p = NULL;
Duh, that hunk is actually left over from the attempt to do what you suggest below :(
> > > > rcu_read_unlock(); > > > > @@ -814,6 +816,11 @@ lookup_pi_state(u32 uval, struct futex_h > > if (!p) > > return -ESRCH; > > > > + if (!p->mm) { > > + put_task_struct(p); > > + return -EPERM; > > + } > > It might make sense to check the mm inside futex_find_get_task(), and > simply never take the refcount if NULL.
Right, but for some reason it was a more ugly patch than the one I posted (w/o the stupid hunk above) and I tried to minimize the impact for stable backporting.
Thanks,
tglx
| |