[Skiboot] [PATCH] Document changes to how firmware versions are passed to host on P9

[Vasant Hegde]

On 08/05/2016 09:27 AM, Stewart Smith wrote:
> Vasant Hegde <hegdevasant at linux.vnet.ibm.com> writes:
>> On 07/19/2016 12:58 PM, Stewart Smith wrote:
>>> We won't do mi-version/ml-version, but we will always have version.
>>
>> Stewart,
>>
>> Sorry for late response.
>>
>>>
>>> Signed-off-by: Stewart Smith <stewart at linux.vnet.ibm.com>
>>> ---
>>>    doc/device-tree/ibm,opal/firmware.txt | 4 +++-
>>>    doc/opal-api/power9-changes.txt       | 1 +
>>>    2 files changed, 4 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/doc/device-tree/ibm,opal/firmware.txt b/doc/device-tree/ibm,opal/firmware.txt
>>> index eab497058be4..3376fba86f37 100644
>>> --- a/doc/device-tree/ibm,opal/firmware.txt
>>> +++ b/doc/device-tree/ibm,opal/firmware.txt
>>> @@ -18,11 +18,13 @@ firmware {
>>>    'symbol-map' property describes OPAL symbol start address and size.
>>>
>>>    'version' property describes OPAL version. Replaces 'git-id', so may
>>> -not be present.
>>> +not be present. On POWER9 and above, it is always present.
>>>
>>>    'mi-version' property describes Microcode Image. Only on IBM FSP systems.
>>> +Will (likely) not be present on POWER9 systems.
>>>
>>>    'ml-version' property describes Microcode Level. Only on IBM FSP systems.
>>> +Will (likely) not be present on POWER9 systems.
>>
>> I failed to understand why it will not be present on FSP based P9 system? Any
>> reason?
>>
>> update_flash (user space tool) makes use of these properties.

Stewart,

>
> I think we should put it in a better place, a node that describes the

mi/ml-versions related to FW . Hence I added under firmware node.
Do you think we have any other better place?

> state of the host firmware at boot time, and some API to get the
> *current* state of host firmware.

On FSP based system user space tools rely in DT information to display FW 
versions. In some corner cases where user did commit/reject operation and not 
rebooted machine then we endup displaying stale information.. I think we should 
have OPAL API to get current FW state.. I have been thinking about this for 
sometime.. But not yet implemented ..

We don't have this issue on BMC system as we use ipmitool to get FW information.


 > Same for BMC firmware.
>

Presently we are relying on ipmitool to get FW information... which seems to be 
fine. Do you think we should add it in DT (or via new OPAL API)?

-Vasant