Some questions about the handler of Add SEL Entry Command
patrick at stwcx.xyz
Fri Mar 27 00:53:33 AEDT 2020
On Thu, Mar 26, 2020 at 02:40:14AM +0000, Vijay Khemka wrote:
> I don’t think this command is even supported completely by IPMID, I think it was started with only one type and never got to completion.
> From: openbmc <openbmc-bounces+vijaykhemka=fb.com at lists.ozlabs.org> on behalf of Max Lai/WYHQ/Wiwynn <Max_Lai at wiwynn.com>
> Date: Wednesday, March 25, 2020 at 4:03 PM
> To: "tomjose at linux.vnet.ibm.com" <tomjose at linux.vnet.ibm.com>
> Cc: "openbmc at lists.ozlabs.org" <openbmc at lists.ozlabs.org>, "LF_OpenBMC.WYHQ.Wiwynn" <LF_OpenBMC.WYHQ.Wiwynn at Wiwynn.com>
> Subject: Some questions about the handler of Add SEL Entry Command
> Hi Tom,
> Our validation team met a problem in phosphor-host-ipmid (phosphor-ipmi-host) recently. We want to use the Add SEL Entry Command handled by ipmiStorageAddSEL. We look forward to getting correct sel log with parameter we offer. But We found that the handler only use sensortype to create log entry when recordType is equal to procedureType(0xDE). Why the other parameter is ignored to create log entry ? It seemed that it did’t follow the ipmi spec. There are more questions. What is the eSEL ? In function createProcedureLogEntry(), what information was stored in the eSELFile (/tmp/esel) ? And how to create the eSELFile ?
The IPMI code was originally written to support IBM's Power systems.
They had existing firmware that supported this "eSEL" format, which I
think stood for "extended SEL". The eSEL contains everything in a SEL
plus more and is sent down as multiple IPMI transactions. The handler
for the multiple IPMI transactions are what create the /tmp/esel file.
More information about the openbmc