implemented without an HP oem specific switch by modifying each interface
open routine to read and cache the OEM manufacturer (via the Get Device ID
command) immediately upon open and to reset the cached OEM manufacturer
id to unknown upon interface close. The cached manufacturer value in the
interface is then be used at run time to implement the manufacturer
specific code.
A non threshold sensor has an analog value if it is defined as having
analog units and it has either valid percentage or modifier units
defined and the manufacturer is IPMI_OEM_HP. The determination of
whether or not a given sensor has an analog value is contained in a
single function (sdr_sensor_has_analog_reading).
This patch also encapsulates the reading of all sensor values in
ipmitool into a a single ipmitool function, ipmi_sdr_read_sensor_value.
The prior independent sensor reading code paths have been collapsed into
one sensor reading code path. This encapsulation enabled consistent
validation of all data returned by the IPMI get sensor reading command.
In implementing a single sensor value read function, I resolved the
inconsistencies between what the "sensor" and "sdr" arguments displayed
and I also resolved numerous cases where uninitialized information in
the IPMI reply from get sensor reading was being interpreted and
incorrectly displayed.
Running an "ipmitool sensor list" command on a system where remote sensors are
not on channel 0 results in unexpected behavior. The SDR functions were fixed
in January of 2009 (lib/ipmi_sdr.c, rev 1.86) to use target I2C addresses and
LUNs for non-BMC-owned sensors. Sensors owned by satellite controllers on other
channels were read as if they were on channel 0.
I've fitted patch, posted alongside bug report, to CVS version.
