[Skiboot] [RFC PATCH] IPMI: Enable synchronous eSEL logging option

Vasant Hegde hegdevasant at linux.vnet.ibm.com
Fri Jan 22 21:21:05 AEDT 2016

On 01/22/2016 10:33 AM, Stewart Smith wrote:
> Vasant Hegde <hegdevasant at linux.vnet.ibm.com> writes:
>> On 08/10/2015 03:52 PM, Vasant Hegde wrote:
>>> Presently we use queue method (ipmi_queue_msg) to send eSEL logs
>>> to BMC.
>>> There are cases like assert() where we want to commit messages
>>> synchronously. This patch checks for log severity and logs PANIC
>>> messages synchronously to BMC (Similar to what we do in FSP based
>>> system).
>>> Note:
>>>   Presently we rely on event severity to decided the logging path.
>>>   With this we should create log with severity PANIC to push events
>>>   via synchrous path.
>>>   Alternatively we can pass boolean variable to log_commit(). (like
>>>   below). This approach allows users to decide on the logging path.
>>>   int ipmi_elog_commit(struct errorlog *elog_buf, bool sync);
>>>   struct platform {
>>> 	  ....
>>> 	  int             (*elog_commit)(struct errorlog *buf, bool sync);
>>> 	  ....
>>>   };
>>>  Let me know which is the best approach here.
>>> -Vasant
>>> Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>
>>> ---
>>> Stewart,
>>>   I've observed "reservation getting lost" issue [1] with synchronous path.
>>>   Hence please *don't* merge this patch.
>>> [1] https://github.com/open-power/tyan-openpower/issues/284
>> Stewart,
>> This issue is fixed in latest skiboot.. I guess Alistair's bt timeout patch
>> helped here..
>> Can you please take a look of this patch?
> I'm certainly not opposed to it. Should this be going to stable? Want to
> resend it and maybe get Alistair to ack it?


Current code works fine as we have infinite while loop after logging eSEL event
in assert path. Hence I think we don't need this patch in stable branch.

I will send next version of patch later next week .


More information about the Skiboot mailing list