<br><br>
<div class="gmail_quote">On Sun, Aug 30, 2009 at 8:00 AM, Sasha Khapyorsky <span dir="ltr"><<a href="mailto:sashak@voltaire.com" target="_blank">sashak@voltaire.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">
<div>On 07:32 Sun 30 Aug , Hal Rosenstock wrote:<br>> > ><br>> > > osm_link_mgr.c:link_mgr_get_smsl has the following:<br>> > ><br>> > > /* Find osm_port of the source = p_physp */<br>
> > > slid = osm_physp_get_base_lid(p_physp);<br>> > > p_src_port =<br>> > > cl_ptr_vector_get(&sm->p_subn->port_lid_tbl,<br>> > cl_ntoh16(slid));<br>
> > ><br>> > > /* Call lash to find proper SL */<br>> > > sl = osm_get_lash_sl(p_osm, p_src_port, p_sm_port);<br>> > ><br>> > > It may be that this code is invoked prior to the LID being assigned<br>
> ><br>> > How is it possible? In the code I can see that link_mgr_process() is<br>> > always executed after lid_mgr run.<br>><br>> When nodes use gPXE, the LID is not passed from the gPXE to the Linux<br>
> environment.<br><br></div>How is it related to gPXE?<br><br>OpenSM's lid manager runs and assigns lids to all available endports,<br>only after this link manager runs and try with SMSL - at this point all<br>lids should be in place and p_subn->port_lid_tbl should be fine.<br>
</blockquote>
<div> </div>
<div>Is that (lids in place) always the case ? What about if the sets of PortInfo for LID fail.</div>
<div> </div>
<div>-- Hal</div>
<div> </div>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid"><br>Am I missing something?<br><font color="#888888"><br>Sasha<br></font></blockquote></div><br>