Re: [PATCH v3] powernv/sensor: Handle OPAL_WRONG_STATE error return

On Monday 06 March 2017 03:03 PM, Michael Ellerman wrote:

Vipin K Parashar [off-list ref] writes:

quoted

OPAL returns OPAL_WRONG_STATE upon failing to provide
sensor data due to core sleeping/offline. Added check
for OPAL_WRONG_STATE rerurn code with sensor read failure.
Also added a log message indicating sensor data being
queried for sleeping/offline core.

Signed-off-by: Vipin K Parashar <redacted>
---
Changes in v3:
  - Added a new case for OPAL_WRONG_STATE in sensor read
    along with a log message indicating sleeping/offline core
    causing read fail.

  arch/powerpc/platforms/powernv/opal-sensor.c | 6 ++++++
  1 file changed, 6 insertions(+)

diff --git a/arch/powerpc/platforms/powernv/opal-sensor.c b/arch/powerpc/platforms/powernv/opal-sensor.c
index 308efd1..fb6d6bb 100644
--- a/arch/powerpc/platforms/powernv/opal-sensor.c
+++ b/arch/powerpc/platforms/powernv/opal-sensor.c

@@ -64,6 +64,12 @@ int opal_get_sensor_data(u32 sensor_hndl, u32 *sensor_data)
  		*sensor_data = be32_to_cpu(data);
  		break;
  
+	case OPAL_WRONG_STATE:
+		pr_notice("%s: Sensor data read failure due to "
+				"core sleeping/offline\n", __func__);

I don't think it should print.

It's not the users fault, or anything they can prevent. It's a
mis-feature (aka. bug) in the driver that it queries sensors for offline
CPUs. At least it should be ratelimited.

I thought the entire motivation for the patch in the first place was
that we were spamming the console with messages?

Yes. Correct.
To avoid flooding console, logged message using pr_notice.
I think, it was suggested in previous reviews to log a message before
returning -EIO.
Shall i directly return -EIO without any log message ?

cheers