Reporting alarms using hwmon_notify_event() may result in a callback
from the thermal subsystem. This means that such notifications must
not hold the update lock to avoid a deadlock. To avoid this situation,
use a worker to handle notifications.
Reported-by: Dmitry Osipenko <[email protected]>
Fixes: e182cf579f0f ("hwmon: (lm90) Rework alarm/status handling")
Signed-off-by: Guenter Roeck <[email protected]>
---
Dmitry,
can you give this patch a try in your system ?
Thanks,
Guenter
drivers/hwmon/lm90.c | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)
diff --git a/drivers/hwmon/lm90.c b/drivers/hwmon/lm90.c
index 9d878163a1f2..03d07da8c2dc 100644
--- a/drivers/hwmon/lm90.c
+++ b/drivers/hwmon/lm90.c
@@ -693,6 +693,7 @@ struct lm90_data {
struct hwmon_chip_info chip;
struct mutex update_lock;
struct delayed_work alert_work;
+ struct work_struct report_work;
bool valid; /* true if register values are valid */
bool alarms_valid; /* true if status register values are valid */
unsigned long last_updated; /* in jiffies */
@@ -1043,13 +1044,18 @@ static int lm90_update_limits(struct device *dev)
return 0;
}
-static void lm90_report_alarms(struct device *dev, struct lm90_data *data)
+static void lm90_report_alarms(struct work_struct *work)
{
- u16 cleared_alarms = data->reported_alarms & ~data->current_alarms;
- u16 new_alarms = data->current_alarms & ~data->reported_alarms;
+ struct lm90_data *data = container_of(work, struct lm90_data, report_work);
+ u16 cleared_alarms, new_alarms, current_alarms;
struct device *hwmon_dev = data->hwmon_dev;
+ struct device *dev = &data->client->dev;
int st, st2;
+ current_alarms = data->current_alarms;
+ cleared_alarms = data->reported_alarms & ~current_alarms;
+ new_alarms = current_alarms & ~data->reported_alarms;
+
if (!cleared_alarms && !new_alarms)
return;
@@ -1101,7 +1107,7 @@ static void lm90_report_alarms(struct device *dev, struct lm90_data *data)
if (st2 & MAX6696_STATUS2_R2OT2)
hwmon_notify_event(hwmon_dev, hwmon_temp, hwmon_temp_emergency_alarm, 2);
- data->reported_alarms = data->current_alarms;
+ data->reported_alarms = current_alarms;
}
static int lm90_update_alarms_locked(struct lm90_data *data, bool force)
@@ -1143,7 +1149,7 @@ static int lm90_update_alarms_locked(struct lm90_data *data, bool force)
(data->config & 0x80);
if (force || check_enable)
- lm90_report_alarms(&client->dev, data);
+ schedule_work(&data->report_work);
/*
* Re-enable ALERT# output if it was originally enabled, relevant
@@ -2552,6 +2558,7 @@ static void lm90_restore_conf(void *_data)
struct i2c_client *client = data->client;
cancel_delayed_work_sync(&data->alert_work);
+ cancel_work_sync(&data->report_work);
/* Restore initial configuration */
if (data->flags & LM90_HAVE_CONVRATE)
@@ -2769,6 +2776,7 @@ static int lm90_probe(struct i2c_client *client)
i2c_set_clientdata(client, data);
mutex_init(&data->update_lock);
INIT_DELAYED_WORK(&data->alert_work, lm90_alert_work);
+ INIT_WORK(&data->report_work, lm90_report_alarms);
/* Set the device type */
if (client->dev.of_node)
--
2.35.1
On 6/29/22 18:42, Guenter Roeck wrote:
> Reporting alarms using hwmon_notify_event() may result in a callback
> from the thermal subsystem. This means that such notifications must
> not hold the update lock to avoid a deadlock. To avoid this situation,
> use a worker to handle notifications.
>
> Reported-by: Dmitry Osipenko <[email protected]>
> Fixes: e182cf579f0f ("hwmon: (lm90) Rework alarm/status handling")
> Signed-off-by: Guenter Roeck <[email protected]>
> ---
> Dmitry,
>
> can you give this patch a try in your system ?
It works, thanks!
Tested-by: Dmitry Osipenko <[email protected]>
--
Best regards,
Dmitry
On 6/29/22 13:02, Dmitry Osipenko wrote:
> On 6/29/22 18:42, Guenter Roeck wrote:
>> Reporting alarms using hwmon_notify_event() may result in a callback
>> from the thermal subsystem. This means that such notifications must
>> not hold the update lock to avoid a deadlock. To avoid this situation,
>> use a worker to handle notifications.
>>
>> Reported-by: Dmitry Osipenko <[email protected]>
>> Fixes: e182cf579f0f ("hwmon: (lm90) Rework alarm/status handling")
>> Signed-off-by: Guenter Roeck <[email protected]>
>> ---
>> Dmitry,
>>
>> can you give this patch a try in your system ?
>
> It works, thanks!
>
> Tested-by: Dmitry Osipenko <[email protected]>
>
Thanks a lot for testing!
Guenter