[OpenBMC] rflash -d command enhancements

[gurevichmark]

Opening this issue on Victor's suggestion.

He proposed the following rflash -d enhancements. Perhaps this issue can be turned into an Epic if work estimate is high.

1. Allow only one firmware update file to be in -d directory. Today both BMC and Host files are required to be in the directory. This would work similar to Boston nodes where only one file is required to be in the directory.

2. Do not reboot BMC or Host if firmware levels did not change. When firmware files are uploaded and happened to be the same level as was already active on the node there is no need to do a reboot. Perhaps after upload and activation, if there are no entries with (+) for BMC, we can skip the BMC reboot, if no entries with (+) for Host, we can skip Host reboot. This could speed up updating nodes in the lab. For example OP910 1742Fx Revisions 7 through 12 have the same BMC firmware, only Host firmware has been changing.

[zet809]

From my point of view, the logic shall be like this:

1. If the version of PNOR/BMC firmware file not exist on BMC, we shall upload+activate
2. If exist, but not active, just activate
3. If exist and active, do nothing.

Hi, @gurevichmark and @whowutwut , and as the state machine in Perl rflash is very very complex, and we are moving to python version, do you think we can enhance python code rather than Perl code if no issue found?

[whowutwut]

@zet809 I think one issue is that we currently do not default to Python, as there are dependencies issues... so if the user does not satisfy the dependencies, there's no option but Perl.

When discussing today... with @gurevichmark

I think we would want to let the firmware decide, so we just upload the file... the FW will extract and activate, (if new)... or extract and do nothing (if already there..) We are able to tell by the + that a reboot is needed for the BMC or PNOR to take effect.

Am I missing something, shouldn't we be able to do that without too much complexity?