Just a bit of a side question here....
What about IO?
Where does it fall in this plan? Throughout this thread there has been
a number of mentions
about PCI and IPMI. No one has gone back and cleared up the issues
mentioned there and
also there has been no mention of VPD as a source for enumeration.
Please don't push this off to a 'phase 2' project.
-Leslie
On 03/04/09 05:54, Steve Hanson wrote:
The implementation approach aside, is there a general disagreement
with the project concept - enumerate from an industry standard source?
I'm certainly in agreement with the concept, though there are a few
caveats.
- the data quality of the SMBIOS records is certainly a concern, and
perhaps we need to
take some precautions. Presumably FMA still needs to do something
sensible on
some 3rd party box with no useful data in its SMBIOS records. Also I
think there
needs to continue to be a simple mechanism for Solaris to override
the labels taken
from SMBIOS for when we find that SMBIOS has got them wrong (which I
fear will
still occasionally happen).
- As Gavin said earlier, you need to scope exactly what information
is to be gleaned
from SMBIOS and what is to be gleaned from elsewhere. Also it is
important that the
hc scheme is a physical representation not a logical one (so it is
invariant if FRUs are
added or disabled) and should have translation properties to find the
logical cpuid,
the IPMI FRU information etc.
Steve
_______________________________________________
fm-discuss mailing list
fm-discuss@opensolaris.org
begin:vcard
fn:Leslie Keller
n:Keller;Leslie
adr;dom:Room 1108;;5300 Riata Park Ct.;Austin;Tx;78727
email;internet:leslie.kel...@sun.com
title:Serviceability Engineering
tel;work:512-401-1787 (x64787)
x-mozilla-html:TRUE
version:2.1
end:vcard
_______________________________________________
fm-discuss mailing list
fm-discuss@opensolaris.org
