| Message ID | 20200920110924.19741-4-Sergey.Semin@baikalelectronics.ru (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
| Series | hwmon: bt1-pvt: Fix PVT sensor being unpowered | expand |
On Sun, Sep 20, 2020 at 02:09:23PM +0300, Serge Semin wrote: > If the PVT sensor is suddenly powered down while a caller is waiting for > the conversion completion, the request won't be finished and the task will > hang up on this procedure until the power is back up again. Let's call the > wait_for_completion_timeout() method instead to prevent that. The cached > timeout is exactly what we need to predict for how long conversion could > normally last. > > Fixes: 87976ce2825d ("hwmon: Add Baikal-T1 PVT sensor driver") > Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> Applied. Thanks, Guenter > --- > drivers/hwmon/bt1-pvt.c | 13 ++++++++++++- > 1 file changed, 12 insertions(+), 1 deletion(-) > > diff --git a/drivers/hwmon/bt1-pvt.c b/drivers/hwmon/bt1-pvt.c > index 2600426a3b21..3e1d56585b91 100644 > --- a/drivers/hwmon/bt1-pvt.c > +++ b/drivers/hwmon/bt1-pvt.c > @@ -477,6 +477,7 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, > long *val) > { > struct pvt_cache *cache = &pvt->cache[type]; > + unsigned long timeout; > u32 data; > int ret; > > @@ -500,7 +501,14 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, > pvt_update(pvt->regs + PVT_INTR_MASK, PVT_INTR_DVALID, 0); > pvt_update(pvt->regs + PVT_CTRL, PVT_CTRL_EN, PVT_CTRL_EN); > > - wait_for_completion(&cache->conversion); > + /* > + * Wait with timeout since in case if the sensor is suddenly powered > + * down the request won't be completed and the caller will hang up on > + * this procedure until the power is back up again. Multiply the > + * timeout by the factor of two to prevent a false timeout. > + */ > + timeout = 2 * usecs_to_jiffies(ktime_to_us(pvt->timeout)); > + ret = wait_for_completion_timeout(&cache->conversion, timeout); > > pvt_update(pvt->regs + PVT_CTRL, PVT_CTRL_EN, 0); > pvt_update(pvt->regs + PVT_INTR_MASK, PVT_INTR_DVALID, > @@ -510,6 +518,9 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, > > mutex_unlock(&pvt->iface_mtx); > > + if (!ret) > + return -ETIMEDOUT; > + > if (type == PVT_TEMP) > *val = pvt_calc_poly(&poly_N_to_temp, data); > else
diff --git a/drivers/hwmon/bt1-pvt.c b/drivers/hwmon/bt1-pvt.c index 2600426a3b21..3e1d56585b91 100644 --- a/drivers/hwmon/bt1-pvt.c +++ b/drivers/hwmon/bt1-pvt.c @@ -477,6 +477,7 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, long *val) { struct pvt_cache *cache = &pvt->cache[type]; + unsigned long timeout; u32 data; int ret; @@ -500,7 +501,14 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, pvt_update(pvt->regs + PVT_INTR_MASK, PVT_INTR_DVALID, 0); pvt_update(pvt->regs + PVT_CTRL, PVT_CTRL_EN, PVT_CTRL_EN); - wait_for_completion(&cache->conversion); + /* + * Wait with timeout since in case if the sensor is suddenly powered + * down the request won't be completed and the caller will hang up on + * this procedure until the power is back up again. Multiply the + * timeout by the factor of two to prevent a false timeout. + */ + timeout = 2 * usecs_to_jiffies(ktime_to_us(pvt->timeout)); + ret = wait_for_completion_timeout(&cache->conversion, timeout); pvt_update(pvt->regs + PVT_CTRL, PVT_CTRL_EN, 0); pvt_update(pvt->regs + PVT_INTR_MASK, PVT_INTR_DVALID, @@ -510,6 +518,9 @@ static int pvt_read_data(struct pvt_hwmon *pvt, enum pvt_sensor_type type, mutex_unlock(&pvt->iface_mtx); + if (!ret) + return -ETIMEDOUT; + if (type == PVT_TEMP) *val = pvt_calc_poly(&poly_N_to_temp, data); else
If the PVT sensor is suddenly powered down while a caller is waiting for the conversion completion, the request won't be finished and the task will hang up on this procedure until the power is back up again. Let's call the wait_for_completion_timeout() method instead to prevent that. The cached timeout is exactly what we need to predict for how long conversion could normally last. Fixes: 87976ce2825d ("hwmon: Add Baikal-T1 PVT sensor driver") Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> --- drivers/hwmon/bt1-pvt.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-)