On Wed, 22 Jul 2015, Paolo Bonzini wrote: > > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c > > index 2d73807..a8723a8 100644 > > --- a/arch/x86/kvm/vmx.c > > +++ b/arch/x86/kvm/vmx.c > > @@ -3158,7 +3158,7 @@ static struct vmcs *alloc_vmcs_cpu(int cpu) > > struct page *pages; > > struct vmcs *vmcs; > > > > - pages = alloc_pages_exact_node(node, GFP_KERNEL, vmcs_config.order); > > + pages = alloc_pages_prefer_node(node, GFP_KERNEL, vmcs_config.order); > > if (!pages) > > return NULL; > > vmcs = page_address(pages); > > Even though there's a pretty strong preference for the "right" node, > things can work if the node is the wrong one. The order is always zero > in practice, so the allocation should succeed. >
You're code is fine both before and after the patch since __GFP_THISNODE isn't set. The allocation will eventually succeed but, as you said, may be from remote memory (and the success of allocating on node may be influenced by the global setting of zone_reclaim_mode). -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html