HE> I saw that virDomainInfo contains a field nrVirtCpu. Maybe this is HE> an alternative. Okay, that was stupid of me :) HE> Wouldn't the case "virDomainGetVcpus = 32" cause an endless loop ? HE> On the other hand ... I can not quite follow the reason of this HE> for loop, which exits right after the first run in case of "count HE> < 32" ? No, max gets incremented by 32 each time. The only time we loop again is if count is equal to max. So, if there are, say 65 vcpus, then the first time through, we set max=32, and we get back our maximum. That means count == max == 32. The next time, max is 64, and we get back 64 as well. The next time, max is 96, but we only get back 65, so we exit the loop. However, as you pointed out, there is a *much* better way. I just didn't see it grepping through the API, but I was looking for a function instead of virDomainInfo, so I missed it :) HE> This function returns -1 in case of failure. Shouldn't this error HE> case be reported to the caller ? With the current setup the caller HE> expects that everything worked. The caller actually shouldn't care if this fails, because this is only the dynamic part. If this fails, we still re-write the XML as expected. However, for the sake of correctness, you're absolutely right. Thanks a lot! -- Dan Smith IBM Linux Technology Center Open Hypervisor Team email: danms@us.ibm.com