Florian Manschwetus
2009-Jun-27 09:58 UTC
diffrences between xvm-3.3/matrix-unstable and xvm-3.4/onnv-3.4
I have setup both in different BEs on my server. Using the xvm-3.4 BE I could access my fc san devices (used for guests). Using the xvm-3.3 BE the /devices/scsi_vhci/ is empty so san isn''t accessible could it be related to the msi problems??? Thx, Florian
Florian Manschwetus
2009-Jun-27 10:10 UTC
Re: diffrences between xvm-3.3/matrix-unstable and xvm-3.4/onnv-3.4
Am 27.06.2009 11:58, schrieb Florian Manschwetus:> I have setup both in different BEs on my server. > Using the xvm-3.4 BE I could access my fc san devices (used for guests). > Using the xvm-3.3 BE the /devices/scsi_vhci/ is empty so san isn''t > accessible could it be related to the msi problems???Ok, I checked, yes it is this problem. Adding: set xpv_psm:xen_support_msi = -1 to /etc/system and reboot gives me back my san luns. Commenting this line and reboot takes them away again... Definitely this has to be fixed before the putback is done ;) Btw: Does b118 still stands for the putback? Thx, Florian
Mark Johnson
2009-Jun-27 12:44 UTC
Re: diffrences between xvm-3.3/matrix-unstable and xvm-3.4/onnv-3.4
Florian Manschwetus wrote:> Am 27.06.2009 11:58, schrieb Florian Manschwetus: >> I have setup both in different BEs on my server. >> Using the xvm-3.4 BE I could access my fc san devices (used for guests). >> Using the xvm-3.3 BE the /devices/scsi_vhci/ is empty so san isn''t >> accessible could it be related to the msi problems??? > Ok, I checked, yes it is this problem. > Adding: > set xpv_psm:xen_support_msi = -1 > to /etc/system and reboot gives me back my san luns. > Commenting this line and reboot takes them away again... > > Definitely this has to be fixed before the putback is done ;) > > Btw: Does b118 still stands for the putback?No, we missed the window.. We ran into 4 bugs in final PIT testing. A domU console hang which has since been fixed A dom0 console panic (due to BIOS bug) which has since been fixed MSI problems :-) A Xen shadow page table bug when running a Solaris HVM guest with more than 2 CPUs and hap disabled. We''re talking about disabling MSIs for the putback and fixing it with a follow up putback. No final decision has been made for this. We are looking into the Xen shadow page table bug currently. MRJ