Thanks for the other replies, Srihari. A followup on the
strand enum code below.
-scott
On Fri, 2009-09-04 at 16:40 -0700, Srihari Venkatesan wrote:
> > 271 if (FM_AWARE_SMBIOS(mod)) {
> > 272 (void) topo_node_label_set(strand, NULL,
> &perr);
> > 273
> > 274 if (topo_node_resource(strand, &fmri, &perr) !=
> 0)
> > 275 whinge(mod, &nerr, "create_strand: "
> > 276 "topo_node_resource failed\n");
> > 277
> > 278 perr += nvlist_lookup_string(fmri,
> > 279 FM_FMRI_HC_PART, &part);
> > 280 perr += nvlist_lookup_string(fmri,
> > 281 FM_FMRI_HC_REVISION, &rev);
> > 282
> > 283 if (perr != 0)
> > 284 whinge(mod, NULL,
> > 285 "create_strand: nvlist_add_string
> failed\n");
> > 286
> > 287 perr += topo_prop_set_string(strand,
> PGNAME(STRAND),
> > 288 FM_FMRI_HC_SERIAL_ID, TOPO_PROP_IMMUTABLE,
> serial, &perr);
> > 289 perr += topo_prop_set_string(strand,
> PGNAME(STRAND),
> > 290 FM_FMRI_HC_PART, TOPO_PROP_IMMUTABLE, part,
> &perr);
> > 291 perr += topo_prop_set_string(strand,
> PGNAME(STRAND),
> > 292 FM_FMRI_HC_REVISION, TOPO_PROP_IMMUTABLE,
> rev, &perr);
> > 293
> > 294 if (perr != 0)
> > 295 whinge(mod, NULL, "create_strand:
> topo_prop_set_string"
> > 296 "failed\n");
> > 297
> > 298 nvlist_free(fmri);
> > 299 topo_mod_strfree(mod, serial);
> > 300 }
> >
> > Does any of this code belong? This is enumerating a strand,
> > which from a label perspective will be the same as its parent
> > core (which will be the same as its parent chip). Why isn't
> > a single topo_node_label_set(strand, NULL, &perr) sufficient?
> >
>
> After setting the Label for the Strand node, we also add Serial,
> Part & Revision in the "strand-properties" ( - this was based upon
> a comment from the portfolio-review that - chip/core/strand should
> all carry the serial, part, revision properties separately.. )
Alright....I don't recall that comment. Or there's a variance on
the interpretation of "separately". If Serviceability has a
need for this (perhaps ASR?), then OK.
My understanding is that the FRU FMRI certainly needs to have the
part/serial/rev of the FRU in question. For a strand, this will be
the same as the parent chip. I'd think that topo_node_fru_set() with
the 2nd parameter as NULL would cover this. I'm less clear that
strand-properties itself needs the same info.
Thx,
-scott
_______________________________________________
fm-discuss mailing list
fm-discuss@opensolaris.org
