[External] Re: Clarify some questions about host tool (burn_my_bmc)

Patrick Venture venture at google.com
Fri Aug 16 00:00:54 AEST 2019


On Thu, Aug 15, 2019 at 12:16 AM Andrew MS1 Peng <pengms1 at lenovo.com> wrote:
>
>
>
> -----邮件原件-----
> 发件人: Patrick Venture <venture at google.com>
> 发送时间: 2019年8月14日 4:55
> 收件人: Andrew MS1 Peng <pengms1 at lenovo.com>
> 抄送: openbmc at lists.ozlabs.org
> 主题: Re: [External] Re: Clarify some questions about host tool (burn_my_bmc)
>
> On Tue, Aug 13, 2019 at 8:49 AM Patrick Venture <venture at google.com> wrote:
> >
> > On Tue, Aug 13, 2019 at 8:15 AM Andrew MS1 Peng <pengms1 at lenovo.com> wrote:
> > >
> > >
> > >
> > > -----邮件原件-----
> > > 发件人: Patrick Venture <venture at google.com>
> > > 发送时间: 2019年8月13日 22:34
> > > 收件人: Andrew MS1 Peng <pengms1 at lenovo.com>
> > > 抄送: openbmc at lists.ozlabs.org
> > > 主题: [External] Re: Clarify some questions about host tool
> > > (burn_my_bmc)
> > >
> > > On Mon, Aug 12, 2019 at 2:43 AM Andrew MS1 Peng <pengms1 at lenovo.com> wrote:
> > > >
> > > > Hi Patrick,
> > > >
> > > >
> > > >
> > > > 1.      It took about 4 minutes and 30 seconds for BIOS update with 64MB rom image. Could we extension the pollstatus time from 100 sec to 300 sec to get the final status?
> > >
> > > I'm curious why it's harmful to check more frequently?
> > >
> > >                 I showed some source code of helper.cpp as below, the checking status total time was 100 sec (5*20) but it was not enough for bios update and therefore I would like to extension the time from 100 sec to 300 sec.
> > >
> > >                                         static constexpr auto verificationSleep = 5s;
> > >                                 ipmi_flash::ActionStatus result =
> > > ipmi_flash::ActionStatus::unknown;
> > >
> > >                                         try
> > >                                 {
> > >                                         static constexpr int commandAttempts = 20;
> > >                                          int attempts = 0;
> > >                                                 ............................................
> >
> > So you're not asking to check it less frequently, you're asking it to
> > allow for a longer period before timing out.
> >
> > >
> > > >
> > > > 2.      If user can choose to preserve BMC configuration (rw area) or BIOS configuration when upgrade firmware, do you have any suggestions with regards to preserve configuration implementation or could the host tool support to send a parameter to support it?
>
> Today I'm submitting https://gerrit.openbmc-project.xyz/24402 for review to just lengthen the wait.  I'll submit a patch parameterizing it later.
>
> > >
> > > Since one of the design goals was to keep the interface from the host simple, and mostly agonistic to what was taking place, consideration wasn't given for a mechanism for adding extra parameters.  We're in a similar boat where we want a mechanism for changing the level of upgrade, so I'll go over the design today at some point and see whether we can do it agnostically.
>
> For this, I think it could make sense to just have multiple targets for BMC updates.  You can define multiple in a single json (by the way), so you could have multiple /flash/image, /flash/factory or something and choosing one would have a different behavior because it would start a different update service.
>                 I added two object in single json file, the configuration could be loaded as below and it seems to work well. Do you have any plans to support flexible blob-id in host flash tool or we need to modify it by ourselves?

I'll have that patch up today.  And update the README to require that
the blob id starts with "/flash/."

>                         Aug 12 20:54:22 hr630 ipmid[778]: config loaded: /flash/bios
>                         Aug 12 20:54:22 hr630 ipmid[778]: config loaded: /flash/bios1
>                         Aug 12 20:54:22 hr630 ipmid[778]: config loaded: /flash/image
>
> > >
> > > >
> > > >
> > > >
> > > > Regards,
> > > >
> > > > Andrew
> > > >
> > > >


More information about the openbmc mailing list