[Skiboot] [PATCH V3 0/6] opal/errorlog : Generalize errorlog read/write routines and adds write to host support for BMC systems

[Mukesh Ojha]

Hi Stewart,

Thanks for your time.
Please find the response inline.

On Thursday 30 June 2016 12:14 PM, Stewart Smith wrote:
> Mukesh Ojha <mukesh02 at linux.vnet.ibm.com> writes:
>> Improve current errorlog infrastructure by splitting existing FSP specific
>> errorlog read/write routines into common and hardware(FSP) specific routines.
>> This would add errorlog writing to host support for BMC based
>> platforms.
> First of all, thanks for tackling this bit of code! It really hasn't
> seen much love since the Power8 GA1 timeframe where I managed to get
> things (both OPAL and kernel) into a state that would be okay to GA
> with.
>
> Unfortunately, the OPAL APIs haven't been documented clearly and
> probably too closely mirror the FSP interface rather than a good API
> that would make a good abstraction layer between kernel and whatever
> service processor (or lack thereof) we're running  on.
>
> So, I'd like to see a first step of properly documenting the existing
> OPAL APIs in doc/opal-api/ so that we at least have a good idea where
> we're starting from. Without this, it's harder to review the changes and
> if the platform abstraction is occuring at the right place.

I am sending out the patch on documentation of opal-api's  of errorlogs.

>> Patch 3 resolves the bug of not clearing the variable OPAL_EVENT_ERROR_LOG_AVAIL
>> properly.
> Is this an existing bug? Could this go in as a separate fix before the refactoring?
Vasant has sent the patches on the existing bug,
I will revise my patch on the basis of it.

Thanks,
Mukesh

>