Re: [PATCH 1/7] KVM: Write protect the updated slot only when we start dirty logging

[Gleb Natapov <gleb@redhat.com>]

On Tue, Dec 25, 2012 at 01:08:40PM +0900, Takuya Yoshikawa wrote:
> On Mon, 24 Dec 2012 15:27:17 +0200
> Gleb Natapov <gleb@redhat.com> wrote:
> 
> > > @@ -6897,7 +6897,14 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
> > >  	spin_lock(&kvm->mmu_lock);
> > >  	if (nr_mmu_pages)
> > >  		kvm_mmu_change_mmu_pages(kvm, nr_mmu_pages);
> > > -	kvm_mmu_slot_remove_write_access(kvm, mem->slot);
> > > +	/*
> > > +	 * Write protect all pages for dirty logging.
> > > +	 * Existing largepage mappings are destroyed here and new ones will
> > > +	 * not be created until the end of the logging.
> > > +	 */
> > > +	if ((mem->flags & KVM_MEM_LOG_DIRTY_PAGES) &&
> > > +	    !(old.flags & KVM_MEM_LOG_DIRTY_PAGES))
> > > +		kvm_mmu_slot_remove_write_access(kvm, mem->slot);
> > We should not check old slot flags here or at least check that
> > old.npages is not zero. Userspace may delete a slot using old flags,
> > then, if new memslot is created with dirty log enabled, it will not be
> > protected.
> 
> The flag, KVM_MEM_LOG_DIRTY_PAGES, is explicitely cleared when we
> delete a slot to free dirty_bitmap:
> 
> 	if (!npages)
> 		mem->flags &= ~KVM_MEM_LOG_DIRTY_PAGES;
> 
> So when old.npages is not zero, the second condition must be true.
> 
Right. I was looking for where it is cleared and missed that.

> Other parts are doing "if (!slot->dirty_bitmap)" to see if the slot
> is in dirty logging mode.  If you prefer, we can do the same here.
> 
Does not matter to me. If you think it will be more consistent do it.

--
			Gleb.