From mboxrd@z Thu Jan 1 00:00:00 1970 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751500AbeACUhR (ORCPT + 1 other); Wed, 3 Jan 2018 15:37:17 -0500 Received: from forwardcorp1j.cmail.yandex.net ([5.255.227.105]:34517 "EHLO forwardcorp1j.cmail.yandex.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750945AbeACUhQ (ORCPT ); Wed, 3 Jan 2018 15:37:16 -0500 X-Greylist: delayed 437 seconds by postgrey-1.27 at vger.kernel.org; Wed, 03 Jan 2018 15:37:15 EST Authentication-Results: smtpcorp1p.mail.yandex.net; dkim=pass header.i=@yandex-team.ru Subject: [PATCH 2/2] sched/core: fix cpu_hotplug_lock recursion in tg_set_cfs_bandwidth() From: Konstantin Khlebnikov To: Peter Zijlstra , Ingo Molnar , linux-kernel@vger.kernel.org Cc: Thomas Gleixner Date: Wed, 03 Jan 2018 23:29:57 +0300 Message-ID: <151501139745.928921.3600916888493232637.stgit@buzz> In-Reply-To: <151501139053.928921.10437479983023205360.stgit@buzz> References: <151501139053.928921.10437479983023205360.stgit@buzz> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: After commit fc8dffd379ca ("cpu/hotplug: Convert hotplug locking to percpu rwsem") get_online_cpus() must be non-recursive. Cpu hotplug is already locked for read in tg_set_cfs_bandwidth() and static_key_slow_inc() in cfs_bandwidth_usage_inc() locks it again. Switch to cpus_read_[un]lock() and static_key_slow_inc/dec_cpuslocked(). This fixes lockdep warning: ============================================ WARNING: possible recursive locking detected 4.14.11-debug-test #5 Not tainted -------------------------------------------- portod-worker27/4773 is trying to acquire lock: (cpu_hotplug_lock.rw_sem){++++}, at: static_key_slow_inc+0xe/0x170 but task is already holding lock: (cpu_hotplug_lock.rw_sem){++++}, at: tg_set_cfs_bandwidth+0xc6/0x890 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(cpu_hotplug_lock.rw_sem); lock(cpu_hotplug_lock.rw_sem); *** DEADLOCK *** Signed-off-by: Konstantin Khlebnikov --- kernel/sched/core.c | 4 ++-- kernel/sched/fair.c | 4 ++-- kernel/sched/sched.h | 1 + 3 files changed, 5 insertions(+), 4 deletions(-) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 644fa2e3d993..584832b49fdc 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -6448,7 +6448,7 @@ static int tg_set_cfs_bandwidth(struct task_group *tg, u64 period, u64 quota) * Prevent race between setting of cfs_rq->runtime_enabled and * unthrottle_offline_cfs_rqs(). */ - get_online_cpus(); + cpus_read_lock(); mutex_lock(&cfs_constraints_mutex); ret = __cfs_schedulable(tg, period, quota); if (ret) @@ -6491,7 +6491,7 @@ static int tg_set_cfs_bandwidth(struct task_group *tg, u64 period, u64 quota) cfs_bandwidth_usage_dec(); out_unlock: mutex_unlock(&cfs_constraints_mutex); - put_online_cpus(); + cpus_read_unlock(); return ret; } diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 2fe3aa853e4d..26a71ebcd3c2 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4365,12 +4365,12 @@ static inline bool cfs_bandwidth_used(void) void cfs_bandwidth_usage_inc(void) { - static_key_slow_inc(&__cfs_bandwidth_used); + static_key_slow_inc_cpuslocked(&__cfs_bandwidth_used); } void cfs_bandwidth_usage_dec(void) { - static_key_slow_dec(&__cfs_bandwidth_used); + static_key_slow_dec_cpuslocked(&__cfs_bandwidth_used); } #else /* HAVE_JUMP_LABEL */ static bool cfs_bandwidth_used(void) diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index b19552a212de..7dddc531ba63 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -2007,6 +2007,7 @@ extern void init_cfs_rq(struct cfs_rq *cfs_rq); extern void init_rt_rq(struct rt_rq *rt_rq); extern void init_dl_rq(struct dl_rq *dl_rq); +/* Must be called under cpus_read_lock() */ extern void cfs_bandwidth_usage_inc(void); extern void cfs_bandwidth_usage_dec(void);