| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1204641443.3842.46.camel@yangyi-dev.bj.intel.com>
Date: Tue, 04 Mar 2008 22:37:23 +0800
From: Yi Yang <yi.y.yang@...el.com>
To: Oleg Nesterov <oleg@...sign.ru>
Cc: Gautham R Shenoy <ego@...ibm.com>, Ingo Molnar <mingo@...e.hu>,
akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
"Rafael J. Wysocki" <rjw@...k.pl>,
Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [BUG 2.6.25-rc3] scheduler/hotplug: some processes are
dealocked when cpu is set to offline
On Tue, 2008-03-04 at 18:01 +0300, Oleg Nesterov wrote:
> On 03/04, Gautham R Shenoy wrote:
> >
> > So at times, the callback thread is blocked on kthread_stop(k) in
> > softlockup.c, while other time, it was blocked in
> > cleanup_workqueue_threads() in workqueue.c.
>
> From another message:
> >
> > However, it remains in R< state
>
> What about cwq->thread? Was it TASK_RUNNING too?
>
> Perhaps, for some reason the task can't get CPU after migrating from
> the now dead CPU.
>
> I can't reproduce this problem on my one cpu P4-ht, perhaps you can
> try something like the untested/uncompiled patch below?
You need one multiprocessor system to regenerate, 4 CPUs or more are
best.
At the beginning, i thought Detect Soft Lockups
(CONFIG_DETECT_SOFTLOCKUP) is culprit because it added much code to
detect lock dependencies and deadlocks, but the issue is still there
after i disabled soft lock detection, it can be regenerated but will
take much long time.
At this time, group_balance and kblockd are in "R<" state.
[root@...wich-rh ~]# ps aux | grep kstop
root 15466 0.0 0.0 0 0 ? S< 01:03 0:00 [kstopmachine]
root 15467 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15468 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15469 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15470 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15471 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15472 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15473 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15474 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15475 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 15476 0.0 0.0 0 0 ? Z< 01:03 0:00 [kstopmachine] <defunct>
root 17155 0.0 0.0 61148 708 pts/3 R+ 14:47 0:00 grep kstop
[root@...wich-rh ~]# ps aux | grep "R<"
root 35 0.0 0.0 0 0 ? R< Mar13 0:00 [group_balance]
root 182 0.1 0.0 0 0 ? R< Mar13 1:09 [kblockd/0]
root 17157 0.0 0.0 61148 704 pts/3 R+ 14:47 0:00 grep R<
[root@...wich-rh ~]#
> Oleg.
>
> --- include/linux/sched.h 2008-02-15 16:59:17.000000000 +0300
> +++ include/linux/sched.h 2008-03-04 17:44:53.136738605 +0300
> @@ -1121,6 +1121,7 @@ struct task_struct {
> /* hung task detection */
> unsigned long last_switch_timestamp;
> unsigned long last_switch_count;
> + unsigned long xxx;
> #endif
> /* CPU-specific state of this task */
> struct thread_struct thread;
> --- kernel/fork.c 2008-02-15 16:59:17.000000000 +0300
> +++ kernel/fork.c 2008-03-04 17:45:14.773033839 +0300
> @@ -1097,6 +1097,7 @@ static struct task_struct *copy_process(
> #ifdef CONFIG_DETECT_SOFTLOCKUP
> p->last_switch_count = 0;
> p->last_switch_timestamp = 0;
> + p->xxx = 0;
> #endif
>
> #ifdef CONFIG_TASK_XACCT
> --- kernel/sched.c 2008-02-15 16:59:17.000000000 +0300
> +++ kernel/sched.c 2008-03-04 17:48:42.308798646 +0300
> @@ -1291,6 +1291,7 @@ static void enqueue_task(struct rq *rq,
> sched_info_queued(p);
> p->sched_class->enqueue_task(rq, p, wakeup);
> p->se.on_rq = 1;
> + p->xxx = jiffies | 1;
> }
>
> static void dequeue_task(struct rq *rq, struct task_struct *p, int sleep)
> @@ -3944,6 +3945,8 @@ need_resched_nonpreemptible:
> preempt_enable_no_resched();
> if (unlikely(test_thread_flag(TIF_NEED_RESCHED)))
> goto need_resched;
> +
> + current->xxx = 0;
> }
> EXPORT_SYMBOL(schedule);
>
> --- kernel/softlockup.c 2008-02-15 16:59:17.000000000 +0300
> +++ kernel/softlockup.c 2008-03-04 17:49:05.584414763 +0300
> @@ -174,6 +174,27 @@ static void check_hung_task(struct task_
> touch_nmi_watchdog();
> }
>
> +static void check_running_task(struct task_struct *t, unsigned long now)
> +{
> + if (!sysctl_hung_task_timeout_secs)
> + return;
> +
> + if (time_before(now, t->xxx + HZ * sysctl_hung_task_timeout_secs)
> + return;
> +
> + printk(KERN_ERR "INFO: task %s:%d can't get CPU for more than "
> + "%ld seconds.\n", t->comm, t->pid,
> + sysctl_hung_task_timeout_secs);
> +
> + if (!cpus_intersects(t->cpus_allowed, cpu_online_map))
> + printk(KERN_ERR "bad ->cpus_allowed\n");
> + if (!cpu_online(task_cpu(t)))
> + printk(KERN_ERR "bad ->cpu\n");
> +
> + sched_show_task(t);
> + touch_nmi_watchdog();
> +}
> +
> /*
> * Check whether a TASK_UNINTERRUPTIBLE does not get woken up for
> * a really long time (120 seconds). If that happens, print out
> @@ -183,6 +204,7 @@ static void check_hung_uninterruptible_t
> {
> int max_count = sysctl_hung_task_check_count;
> unsigned long now = get_timestamp(this_cpu);
> + unsigned long jiff = jiffies;
> struct task_struct *g, *t;
>
> /*
> @@ -192,15 +214,17 @@ static void check_hung_uninterruptible_t
> if ((tainted & TAINT_DIE) || did_panic)
> return;
>
> - read_lock(&tasklist_lock);
> + rcu_read_lock();
> do_each_thread(g, t) {
> if (!--max_count)
> goto unlock;
> if (t->state & TASK_UNINTERRUPTIBLE)
> check_hung_task(t, now);
> + if (!t->xxx)
> + check_running_task(t, jiff);
> } while_each_thread(g, t);
> unlock:
> - read_unlock(&tasklist_lock);
> + rcu_read_unlock();
> }
>
> /*
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists