Re: [PATCH v4] sched/numa: fix unsafe get_task_struct() in task_numa_assign()

[Kirill Tkhai <ktkhai@parallels.com>]

В Пн, 10/11/2014 в 23:01 +0300, Kirill Tkhai пишет:
> 
> 10.11.2014, 19:45, "Sasha Levin" <sasha.levin@oracle.com>:
> > On 11/10/2014 11:36 AM, Kirill Tkhai wrote:
> >>  I mean task_numa_find_cpu(). If a garbage is in cpumask_of_node(env->dst_nid)
> >>  and cpu is bigger than mask, the check
> >>
> >>  cpumask_test_cpu(cpu, tsk_cpus_allowed(env->p)
> >>
> >>  may be true.
> >>
> >>  So, we dereference wrong rq in task_numa_compare(). It's not rq at all.
> >>  Strange cpu may be from here. It's just a int number in a wrong memory.
> >
> > But the odds of the spinlock magic and owner pointer matching up are slim
> > to none in that case. The memory is also likely to be valid since KASAN didn't
> > complain about the access, so I don't believe it to be an access to freed memory.
> 
> I'm not good with lockdep checks, so I can't judge right now...
> Just a hypothesis.
> 
> >>  A hypothesis that below may help:
> >>
> >>  diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> >>  index 826fdf3..a2b4a8a 100644
> >>  --- a/kernel/sched/fair.c
> >>  +++ b/kernel/sched/fair.c
> >>  @@ -1376,6 +1376,9 @@ static void task_numa_find_cpu(struct task_numa_env *env,
> >>   {
> >>           int cpu;
> >>
> >>  + if (!node_online(env->dst_nid))
> >>  + return;
> >
> > I've changed that to BUG_ON(!node_online(env->dst_nid)) and will run it for a
> > bit.
> 
> I've looked one more time, and it looks like it's better to check for
> BUG_ON(env->dst_nid > MAX_NUMNODES). node_online() may do not work for
> insane nids.
> 
> Anyway, even if it's not connected, we need to initialize numa_preferred_nid
> of init_task, because it's inherited by kernel_init() (and /sbin/init too).
> I'll send the patch tomorrow.

Also we can check for cpu. A problem may be in how nodes are built etc..

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 826fdf3..8f5c316 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1381,6 +1381,14 @@ static void task_numa_find_cpu(struct task_numa_env *env,
 		if (!cpumask_test_cpu(cpu, tsk_cpus_allowed(env->p)))
 			continue;
 
+		/*
+		 * Is cpumask_of_node() broken? In sched_init() we
+		 * initialize only possible RQs (their rq->lock etc).
+		 */
+		BUG_ON(cpu >= NR_CPUS || !cpu_possible(cpu));
+		/* Insane node id? */
+		BUG_ON(env->dst_nid >= MAX_NUMNODES || !node_online(env->dst_nid));
+
 		env->dst_cpu = cpu;
 		task_numa_compare(env, taskimp, groupimp);
 	}