On 06.10.2011 [18:25:51 -0400], Don Dutile wrote: > On 10/06/2011 06:18 PM, Nishanth Aravamudan wrote: > >On 06.10.2011 [18:05:37 -0400], Prarit Bhargava wrote: > >> > >> > >>On 10/06/2011 05:03 PM, Nishanth Aravamudan wrote: > >>>We have observed the following on Power systems with SR-IOV capable > >>>adapters: > >>> > >>>lpfc 0002:01:00.0: device not available because of BAR 7 [0x000000-0x00ffff] collisions > >>> > >>>The issue is that on Power systems, PCI BARs cannot be remapped and VF > >>>BARs might have values that collide. As far as I can tell, the current > >>>SR-IOV code cannot be supported on Power and so it seems like we could > >>>provide a hook for an architecture that might set CONFIG_PCI_IOV to > >>>disable SR-IOV support (potentially at run-time). > >>> > >> > >>I'd rather see an off/on switch because in a short while when powerpc > >>does support SRIOV there maybe no users of the code you've suggested. > > > >Well, I think this is orthogonal to my patch and seems useful > >independent of it. > > > >There are two use cases being discussed here, I think. > > > >1) Mis-behaving implementations. > > Your patch does resolve this issue by allowing the *user* (not > > the kernel itself at run-time) to determine if SR-IOV is the > > root cause by trying to disable it. > > > >2) Platforms that are known not to support SR-IOV as currently > >implemented. > > My patch resolves this issue by not attemping SR-IOV support on > > those platforms. > > > >I suppose your hammer is bigger than mine :) But I don't think it's good > >to have a flag to workaround a code issue that the user must pass to > >ensure functioning adapters. > > > It's a BIOS (or PPC BIOS-equiv) issue, and the kernel has a number of > boot-time parameters like the one proposed to resolve such nuances > relative to known, working, architected technology that works for the > majority of cases it is employed. I guess it's all a matter of perspective. Yes, it is true that system firmware on Power does not allow reallocation of resources. That is the "issue" I think you are referring to above. However, that is by design. So it's not something that is going to change, at least in my understanding. What *will* need to change, though, to support Power is the in-kernel SR-IOV implementation, which assumes that it can do various things it won't be able to do on Power, because it appears to be written with only x86 in mind. In any case, I think I'm talking in circles, so I will go and test your patch. Thanks, Nish -- Nishanth Aravamudan <nacc@xxxxxxxxxx> IBM Linux Technology Center -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
