[<prev] [next>] [day] [month] [year] [list]
Message-Id: <20230710181959.2750269-1-longman@redhat.com>
Date: Mon, 10 Jul 2023 14:19:59 -0400
From: Waiman Long <longman@...hat.com>
To: Will Deacon <will@...nel.org>, Mark Rutland <mark.rutland@....com>
Cc: linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
Robin Murphy <robin.murphy@....com>,
Waiman Long <longman@...hat.com>
Subject: [PATCH v2] perf/arm-dmc620: Reverse locking order in dmc620_pmu_get_irq()
The following circular locking dependency was reported when running
cpus online/offline test on an arm64 system.
[ 84.195923] Chain exists of:
dmc620_pmu_irqs_lock --> cpu_hotplug_lock --> cpuhp_state-down
[ 84.207305] Possible unsafe locking scenario:
[ 84.213212] CPU0 CPU1
[ 84.217729] ---- ----
[ 84.222247] lock(cpuhp_state-down);
[ 84.225899] lock(cpu_hotplug_lock);
[ 84.232068] lock(cpuhp_state-down);
[ 84.238237] lock(dmc620_pmu_irqs_lock);
[ 84.242236]
*** DEADLOCK ***
The problematic locking order seems to be
lock(dmc620_pmu_irqs_lock) --> lock(cpu_hotplug_lock)
This locking order happens when dmc620_pmu_get_irq() is called from
dmc620_pmu_device_probe(). Since dmc620_pmu_irqs_lock is used for
protecting the dmc620_pmu_irqs structure only, we don't actually need
to hold the lock when adding a new instance to the CPU hotplug subsystem.
Fix this possible deadlock scenario by releasing the lock when a new
dmc620_pmu_irq needs to be created and reacquring it again when the
new irq is inserted into dmc620_pmu_irqs.
Suggested-by: Robin Murphy <robin.murphy@....com>
Signed-off-by: Waiman Long <longman@...hat.com>
---
drivers/perf/arm_dmc620_pmu.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
diff --git a/drivers/perf/arm_dmc620_pmu.c b/drivers/perf/arm_dmc620_pmu.c
index 9d0f01c4455a..dbf67c122420 100644
--- a/drivers/perf/arm_dmc620_pmu.c
+++ b/drivers/perf/arm_dmc620_pmu.c
@@ -419,13 +419,16 @@ static irqreturn_t dmc620_pmu_handle_irq(int irq_num, void *data)
}
static struct dmc620_pmu_irq *__dmc620_pmu_get_irq(int irq_num)
+ __releases(&dmc620_pmu_irqs_lock)
{
struct dmc620_pmu_irq *irq;
int ret;
list_for_each_entry(irq, &dmc620_pmu_irqs, irqs_node)
if (irq->irq_num == irq_num && refcount_inc_not_zero(&irq->refcount))
- return irq;
+ goto out_unlock;
+
+ mutex_unlock(&dmc620_pmu_irqs_lock);
irq = kzalloc(sizeof(*irq), GFP_KERNEL);
if (!irq)
@@ -452,8 +455,12 @@ static struct dmc620_pmu_irq *__dmc620_pmu_get_irq(int irq_num)
goto out_free_irq;
irq->irq_num = irq_num;
+
+ mutex_lock(&dmc620_pmu_irqs_lock);
list_add(&irq->irqs_node, &dmc620_pmu_irqs);
+out_unlock:
+ mutex_unlock(&dmc620_pmu_irqs_lock);
return irq;
out_free_irq:
@@ -469,7 +476,7 @@ static int dmc620_pmu_get_irq(struct dmc620_pmu *dmc620_pmu, int irq_num)
mutex_lock(&dmc620_pmu_irqs_lock);
irq = __dmc620_pmu_get_irq(irq_num);
- mutex_unlock(&dmc620_pmu_irqs_lock);
+ /* mutex_unlock() called inside __dmc620_pmu_get_irq() */
if (IS_ERR(irq))
return PTR_ERR(irq);
--
2.31.1
Powered by blists - more mailing lists