Just thought I'd send some notes + updates.
The plan is to commit one more tool (ipmi-fru) into FreeIPMI before
releasing 0.4.0-beta. The tool has been complete for quite some time,
however I have not received GPL approval from my organization (it
normally doesn't take this long, not sure why it's hung up) to release
it into FreeIPMI publically.
I had planned to release the 0.4.0 beta in early July. Hopefully I get
GPL approval soon and we can still release around that time.
I looked into and actually began development of IPMI 2.0 protocol
support into UDM so it could be supported in most of the FreeIPMI tools
(i.e. ipmi-sensors, ipmi-sel, etc.). However, after a half a day of
effort, I have now abandoned it for the 0.4.0 release. The way that UDM
is architected makes it quite difficult to implement.
In essence, the issue lies in the implementation of
ipmi_lan_open_session() and how it calls other UDM functions (i.e.
ipmi_cmd_set_session_privilege()) before UDM is setup and has completed
for the user. Due to the nature of the IPMI 1.5 protocol, these issues
could be worked around relatively easily.
However, w/ IPMI 2.0, the above architecture becomes far more difficult
to workaround. Ipmi_lan_open_session() should be re-written to not use
any other UDM functions.
A.B., is someone on the Z-research staff (perhaps Ragha) interested in
re-architecting this? I could do the subsequent IPMI 2.0 support
afterwards? I still need to think a bit more about the code
architecture before we do it.
High Performance Systems Division
Lawrence Livermore National Laboratory
Freeipmi-devel mailing list