Stanley Pilton
2013-Nov-09 10:37 UTC
disk controller not working with xen: Marvell Technology Group Ltd. 88SE9172
Dear Xen folk, I''m trying to use xen as shipped with debian 7.0. The system has 13 disks over 3 controllers. When I boot normal linux, I get this (condensed output from lshw): product: MV64460/64461/64462 System Controller, Revision B disk:0 disk:1 disk:2 disk:3 disk:4 disk:5 product: 88SE9172 SATA 6Gb/s Controller disk:0 disk:1 product: 7 Series/C210 Series Chipset Family 6-port SATA Controller disk:0 disk:1 disk:2 disk:3 disk:4 When I boot the xen system, I get this: product: MV64460/64461/64462 System Controller, Revision B disk:0 disk:1 disk:2 disk:3 disk:4 disk:5 product: 88SE9172 SATA 6Gb/s Controller product: 7 Series/C210 Series Chipset Family 6-port SATA Controller disk:0 disk:1 disk:2 disk:3 disk:4 The relevant output from lspci is: 06:00.0 SATA controller: Marvell Technology Group Ltd. 88SE9172 SATA 6Gb/s Controller (rev 11) or for lspci -n : 06:00.0 0106: 1b4b:9172 (rev 11) From searching the web, I tried appending pci-phantom=06:00,1 as an argument to xen at boot time, but this did not result in the disks being visible. I based my attempted workaround on <http://lists.xen.org/archives/html/xen-users/2013-06/msg00460.html>. Can anyone help with this? regards, Stanley
Gordan Bobic
2013-Nov-11 12:05 UTC
Re: disk controller not working with xen: Marvell Technology Group Ltd. 88SE9172
On Sat, 9 Nov 2013 10:37:12 +0000, Stanley Pilton <stanley.pilton@gmail.com> wrote:> The relevant output from lspci is: > > 06:00.0 SATA controller: Marvell Technology Group Ltd. 88SE9172 SATA > 6Gb/s Controller (rev 11) > > or for lspci -n : > > 06:00.0 0106: 1b4b:9172 (rev 11) > > From searching the web, I tried appending pci-phantom=06:00,1 as an > argument to xen at boot time, but this did not result in the disks > being visible. > > I based my attempted workaround on > <http://lists.xen.org/archives/html/xen-users/2013-06/msg00460.html>.Not that IS interesting - what version of Xen is pci-phantom implemented from? Does it work for phantom devices as opposed to functions? e.g. some LSI and Adaptec SAS cards are based on PCI/PCI-X SAS controllers with PCIe bridges, but the bridges don''t show up in lspci, so when transfers actually happen there are AER PCIe errors all over the place as IOMMU forbids the DMA transfers from seemingly non-existant devices. Last time I mentioned such a thing here, the workaround was to write a kernel module for dom0 that informs the hypervisor of the devices in question, but using something like pci-phantom would be much more convenient. Gordan