Question for phosphor bmc code - BMC update

Adriana Kobylak anoo at linux.ibm.com
Wed Feb 19 08:16:33 AEDT 2020


On 2020-02-17 23:24, Lei YU wrote:
> + Adriana
> 
> AFAIK the restriction is added by Adriana, she should know the details.
> I think it's reasonable in production that if the revision is the same
> there is no need to do the code update if the running code is already
> the same version.
> But it does cause trouble in lab debug that one does want to update
> the code with the same revision.

Yeah, that's part of the reason, no need to rewrite the same version if 
it's already running. Also for systems that support dual versions, 
there's not currently a way to represent that both 'sides' have the same 
version via D-Bus objects, since the version id would be the same, so if 
side A is running version X and backup side B is running version Y, if 
side B is allowed to be updated to version X, there would only be one 
D-Bus object.

Is being able to overwrite the same version something that people would 
want? Maybe based on a lab vs production setting?
For configurations that support multiple versions, we'll looking to add 
the ability to have more than one 'side' have the same version.

> 
> On Mon, Feb 17, 2020 at 6:23 PM Eli Huang/WYHQ/Wiwynn
> <Eli_Huang at wiwynn.com> wrote:
>> 
>> Hi Lei Yu,
>> 
>> Confirm a question with you.
>> 
>> When we do the BMC update using phosphor-bmc-code-mgmt,
>> 
>> The code in the image_manager.cpp will get the version in the manifest 
>> file.
>> 
>> If the version is same as path on d-bus, the BMC will not update and 
>> print message “Software Object with the same version already 
>> exists”.
>> 
>> We want to know why upstream code can’t update the same version 
>> image if the BMC are valid.
>> 
>> (phosphor-bmc-code-mgmt commit number : 
>> a013560f96a9ee5c2db4e1778c7dcee199c3acf1)
>> 
>> Regards,
>> 
>> Eli
>> 
>> 


More information about the openbmc mailing list