Clarify some questions about BIOS update (phosphor-ipmi-flash)

[Patrick Venture]

On Mon, Aug 5, 2019 at 9:32 AM Patrick Venture <venture at google.com> wrote:
>
> On Mon, Aug 5, 2019 at 9:17 AM Patrick Venture <venture at google.com> wrote:
> >
> > On Mon, Aug 5, 2019 at 4:56 AM Andrew MS1 Peng <pengms1 at lenovo.com> wrote:
> > >
> > > Hi Patrick,
> > >
> > >
> > >
> > > I have some questions about BIOS update via phosphor-ipmi-flash with L2A, please help to clarify. Thanks.
> > >
> > >
> > >
> >
> > Thanks for giving it a shot!  I don't have any systems using this for
> > BIOS update, so I'm curious where there may be issues lurking :)
> >
> >
> > > 1. How can I update the status (success or failure) after upgrade BIOS firmware? Currently, the update status only appear "running".
> > >
> >
> > I see, the update mechanism doesn't have a verification option.  A bit
> > of an oversight.  I'll have something up for review later today.  Then
> > it'll be something you configure similarly to the verifySystemd json
> > entry.
> >
> > > 2. It was failed when triggered "phosphor-ipmi-flash-bios-update.target" since the job-mode is null string. It can work well when I add string "replace" to job-mode. How can I config (json file or configure parameter) to support it?
> >
> > I'll take a look!  The parameter for job-mode, I thought, was
> > optional?  Since we don't specify it for other jobs (verification,
> > etc).  The string should be empty by default and then not added as a
> > null-string:
> >
> > https://github.com/openbmc/phosphor-ipmi-flash/blob/f0c71df2514927e933f6fc33d9669ab0663a0706/bmc/update_systemd.cpp#L49
>
> I see it is required!  Since I always configure with reboot, I never
> saw this failure. I'll default it to "replace"

First patch in the series generated from your email:

https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-ipmi-flash/+/24165

Next: json configuration option to control the job-mode.
Then merge verification and update objects, s.t. you can specify the
status path.

*** A later patchset can have the status be not a path, but a dbus
property or dbus path -- if someone wants to implement a new object
for this, it's somewhat trivial.

>
> >
> > I don't think I have a unit-test that verifies the parameters are sent
> > down properly, so I'll also add that today AND I'll add an optional
> > mode parameter to the json.
> >
> > Sending for review today.
> >
> > ...
> >
> > >
> > >
> > >
> > > Regards,
> > > Andrew