>>> On 17.08.17 at 17:04, <boris.ostrov...@oracle.com> wrote: > On 08/17/2017 10:59 AM, Jan Beulich wrote: >>>>> On 17.08.17 at 16:19, <boris.ostrov...@oracle.com> wrote: >>> --- a/xen/arch/x86/hvm/vmx/vmcs.c >>> +++ b/xen/arch/x86/hvm/vmx/vmcs.c >>> @@ -652,7 +652,7 @@ int vmx_cpu_up(void) >>> >>> INIT_LIST_HEAD(&this_cpu(active_vmcs_list)); >>> >>> - if ( (rc = vmx_cpu_up_prepare(cpu)) != 0 ) >>> + if ( (cpu == 0) && (rc = vmx_cpu_up_prepare(cpu)) != 0 ) >> Let's please not introduce any further "CPU0 is always to BSP" >> assumptions - we ought to get to the point where CPU0 can be >> hot-unplugged and then hot-plugged again. > > Pass bool bsp to (added) _vmx_cpu_up(), just like we do for SVM?
That's the best you can do, I suppose. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org https://lists.xen.org/xen-devel
