Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Sender: "elver via sendgmr" <elver@elver.muc.corp.google.com>
Date:   Mon,  4 Jan 2021 14:07:49 +0100
Message-Id: <20210104130749.1768991-1-elver@google.com>
Mime-Version: 1.0
Subject: [PATCH mm] kfence: fix potential deadlock due to wake_up()
From:   Marco Elver <elver@google.com>
To:     elver@google.com, akpm@linux-foundation.org
Cc:     glider@google.com, dvyukov@google.com, jannh@google.com,
        mark.rutland@arm.com, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, kasan-dev@googlegroups.com,
        syzbot+8983d6d4f7df556be565@syzkaller.appspotmail.com,
        Hillf Danton <hdanton@sina.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Lockdep reports that we may deadlock when calling wake_up() in
__kfence_alloc(), because we may already hold base->lock. This can
happen if debug objects are enabled:

    ...
    __kfence_alloc+0xa0/0xbc0 mm/kfence/core.c:710
    kfence_alloc include/linux/kfence.h:108 [inline]
    ...
    kmem_cache_zalloc include/linux/slab.h:672 [inline]
    fill_pool+0x264/0x5c0 lib/debugobjects.c:171
    __debug_object_init+0x7a/0xd10 lib/debugobjects.c:560
    debug_object_init lib/debugobjects.c:615 [inline]
    debug_object_activate+0x32c/0x3e0 lib/debugobjects.c:701
    debug_timer_activate kernel/time/timer.c:727 [inline]
    __mod_timer+0x77d/0xe30 kernel/time/timer.c:1048
    ...

Therefore, switch to an open-coded wait loop. The difference to before
is that the waiter wakes up and rechecks the condition after 1 jiffy;
however, given the infrequency of kfence allocations, the difference is
insignificant.

Link: https://lkml.kernel.org/r/000000000000c0645805b7f982e4@google.com
Reported-by: syzbot+8983d6d4f7df556be565@syzkaller.appspotmail.com
Suggested-by: Hillf Danton <hdanton@sina.com>
Signed-off-by: Marco Elver <elver@google.com>
---
 mm/kfence/core.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/mm/kfence/core.c b/mm/kfence/core.c
index 933b197b8634..f0816d5f5913 100644
--- a/mm/kfence/core.c
+++ b/mm/kfence/core.c
@@ -94,9 +94,6 @@ DEFINE_STATIC_KEY_FALSE(kfence_allocation_key);
 /* Gates the allocation, ensuring only one succeeds in a given period. */
 static atomic_t allocation_gate = ATOMIC_INIT(1);
 
-/* Wait queue to wake up allocation-gate timer task. */
-static DECLARE_WAIT_QUEUE_HEAD(allocation_wait);
-
 /* Statistics counters for debugfs. */
 enum kfence_counter_id {
 	KFENCE_COUNTER_ALLOCATED,
@@ -586,6 +583,8 @@ late_initcall(kfence_debugfs_init);
 static struct delayed_work kfence_timer;
 static void toggle_allocation_gate(struct work_struct *work)
 {
+	unsigned long end_wait;
+
 	if (!READ_ONCE(kfence_enabled))
 		return;
 
@@ -596,7 +595,14 @@ static void toggle_allocation_gate(struct work_struct *work)
 	 * Await an allocation. Timeout after 1 second, in case the kernel stops
 	 * doing allocations, to avoid stalling this worker task for too long.
 	 */
-	wait_event_timeout(allocation_wait, atomic_read(&allocation_gate) != 0, HZ);
+	end_wait = jiffies + HZ;
+	do {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (atomic_read(&allocation_gate) != 0)
+			break;
+		schedule_timeout(1);
+	} while (time_before(jiffies, end_wait));
+	__set_current_state(TASK_RUNNING);
 
 	/* Disable static key and reset timer. */
 	static_branch_disable(&kfence_allocation_key);
@@ -707,7 +713,6 @@ void *__kfence_alloc(struct kmem_cache *s, size_t size, gfp_t flags)
 	 */
 	if (atomic_read(&allocation_gate) || atomic_inc_return(&allocation_gate) > 1)
 		return NULL;
-	wake_up(&allocation_wait);
 
 	if (!READ_ONCE(kfence_enabled))
 		return NULL;
-- 
2.29.2.729.g45daf8777d-goog