<div class="socmaildefaultfont" dir="ltr" style="font-family:Arial, Helvetica, sans-serif;font-size:10.5pt" ><div dir="ltr" >Nice Patrick, As discussed in test work group meeting,</div>
<div dir="ltr" >Shall we have this feature with identified</div>
<div dir="ltr" >Use case Scenarios -> Requirement -> Design -> Test cases</div>
<div dir="ltr" > </div>
<div dir="ltr" >Because code update is the feature which all end users will do at least once.</div>
<div dir="ltr" >We can have a wiki as IPMI Inband code update and then populate design over there.</div>
<div dir="ltr" >Once design is baselined, we can move it under presentations / document of github.</div>
<div dir="ltr" > </div>
<div dir="ltr" >Please let know your thoughts.</div>
<div dir="ltr" > </div>
<div dir="ltr" > </div>
<blockquote data-history-content-modified="1" dir="ltr" style="border-left:solid #aaaaaa 2px; margin-left:5px; padding-left:5px; direction:ltr; margin-right:0px" >----- Original message -----<br>From: Patrick Venture <venture@google.com><br>Sent by: "openbmc" <openbmc-bounces+sivas.srr=in.ibm.com@lists.ozlabs.org><br>To: OpenBMC Maillist <openbmc@lists.ozlabs.org><br>Cc:<br>Subject: In-band IPMI Update over IPMI Blobs<br>Date: Fri, Oct 5, 2018 8:16 PM<br> <br><tt><font size="3" face="" >For those who have been watching, we have been upstreaming how we</font></tt><br><tt><font size="3" face="" >currently handle in-band updates over IPMI/LPC/PCI to</font></tt><br><tt><font size="3" face="" >phosphor-ipmi-flash. However, recently we also upstreamed a generic</font></tt><br><tt><font size="3" face="" >blob interface, phosphor-ipmi-blobs. Until</font></tt><br><tt><font size="3" face="" ><a href="https://gerrit.openbmc-project.xyz/13620" target="_blank">https://gerrit.openbmc-project.xyz/13620</a> it wasn't a suitable</font></tt><br><tt><font size="3" face="" >mechanism for the firmware update due to an inability to do a clean</font></tt><br><tt><font size="3" face="" >method mapping between the protocols.</font></tt><br><br><tt><font size="3" face="" >I've written up a proposal (downstream for now) to port the</font></tt><br><tt><font size="3" face="" >phosphor-ipmi-flash implementation into a blob handler that just goes</font></tt><br><tt><font size="3" face="" >through phosphor-ipmi-blobs. The thinking is to have more things</font></tt><br><tt><font size="3" face="" >leverage the generic blob handler instead of having a fully separate</font></tt><br><tt><font size="3" face="" >OEM IPMI command set for everything (although phosphor-ipmi-flash goes</font></tt><br><tt><font size="3" face="" >over the Firmware Netfn).</font></tt><br><br><tt><font size="3" face="" >It's now a pretty trivial transformation, and once the design has some</font></tt><br><tt><font size="3" face="" >age downstream (1-2 days :D), then I'm going to send it here for</font></tt><br><tt><font size="3" face="" >review.</font></tt><br><br><tt><font size="3" face="" >Unless there are obvious objections.</font></tt><br><br><tt><font size="3" face="" >And as food for thought, other Blob protocol use-cases in the works:</font></tt><br><tt><font size="3" face="" >- mechanism for downloading core dumps and crash logs over IPMI (no</font></tt><br><tt><font size="3" face="" >network access)</font></tt><br><tt><font size="3" face="" >- mechanism for sending commands down for special hardware</font></tt><br><tt><font size="3" face="" >- mechanism for reading back configuration files for whitelisted debug</font></tt><br><tt><font size="3" face="" >- mechanism for randomly creating large files such that the flash</font></tt><br><tt><font size="3" face="" >fills up and the system never boots again.</font></tt><br><br><tt><font size="3" face="" >Patrick</font></tt></blockquote>
<div dir="ltr" > </div></div><BR>