Hi,
(I would have created a github issue but the "create an issue" button
isn't there. Not sure if there was a recent permissions change or
something.)
Was helping a user with an issue in FreeIPMI
https://savannah.gnu.org/support/index.php?110963
where ipmitool was outputting some fan RPMs but FreeIPMI was not.
FreeIPMI reports errors from the system when sensors are read.
After looking into it, I believe there is a bug in ipmitool.
In `ipmi_sdr_read_sensor_value()` we have
```
if (rsp->ccode) {
if ( !((sr.full && rsp->ccode == 0xcb) ||
(sr.compact && rsp->ccode == 0xcd)) ) {
lprintf(LOG_DEBUG,
"Error reading sensor %s (#%02x): %s",
sr.s_id,
sensor->keys.sensor_num,
val2str(rsp->ccode, completion_code_vals));
}
return &sr;
}
```
It appears that if a response has a non-zero completion code, optionally
output some debugging and then return the sensor reading data structure
back to the caller.
In `ipmi_sdr_print_sensor_fc()` we have the call
```
sr = ipmi_sdr_read_sensor_value(intf, sensor, sdr_record_type, 2);
if (!sr)
return -1;
```
So if a response has a non-zero completion code when reading a sensor,
we know the sensor reading struct will be returned non-NULL.
So I think the code in `ipmi_sdr_print_sensor_fc()` will continue on its
way outputting things assuming the sensor reading was successful, even
if it was not.
Al
--
Al Chu
Livermore Computing
Lawrence Livermore National Laboratory
_______________________________________________
Ipmitool-devel mailing list
Ipmitool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ipmitool-devel
