Re: [PATCH 0/7] KVM: Alleviate mmu_lock hold time when we start dirty logging

[Alex Williamson <alex.williamson@redhat.com>]

On Thu, 2012-12-20 at 23:35 +0900, Takuya Yoshikawa wrote:
> On Thu, 20 Dec 2012 06:41:27 -0700
> Alex Williamson <alex.williamson@redhat.com> wrote:
> 
> > Hmm, isn't the fix as simple as:
> > 
> > --- a/virt/kvm/kvm_main.c
> > +++ b/virt/kvm/kvm_main.c
> > @@ -847,7 +847,8 @@ int __kvm_set_memory_region(struct kvm *kvm,
> >                                 GFP_KERNEL);
> >                 if (!slots)
> >                         goto out_free;
> > -       }
> > +       } else
> > +               slots->generation = kvm->memslots->generation;
> >  
> >         /* map new memory slot into the iommu */
> >         if (npages) {
> > 
> > Or even just slots->generation++ since we're holding the lock across all
> > of this.
> 
> Yes, the fix should work, but I do not want to update the
> generation from outside of update_memslots().

Ok, then:

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 87089dd..c7b5061 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -413,7 +413,8 @@ void kvm_exit(void);
 
 void kvm_get_kvm(struct kvm *kvm);
 void kvm_put_kvm(struct kvm *kvm);
-void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new);
+void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new,
+                     u64 last_generation);
 
 static inline struct kvm_memslots *kvm_memslots(struct kvm *kvm)
 {
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index c4c8ec1..06961ea 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -667,7 +667,8 @@ static void sort_memslots(struct kvm_memslots *slots)
 		slots->id_to_index[slots->memslots[i].id] = i;
 }
 
-void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new)
+void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new,
+                     u64 last_generation)
 {
 	if (new) {
 		int id = new->id;
@@ -679,7 +680,7 @@ void update_memslots(struct kvm_memslots *slots, struct kvm_memory_slot *new)
 			sort_memslots(slots);
 	}
 
-	slots->generation++;
+	slots->generation = last_generation + 1;
 }
 
 static int check_memory_region_flags(struct kvm_userspace_memory_region *mem)
@@ -814,7 +815,7 @@ int __kvm_set_memory_region(struct kvm *kvm,
 		slot = id_to_memslot(slots, mem->slot);
 		slot->flags |= KVM_MEMSLOT_INVALID;
 
-		update_memslots(slots, NULL);
+		update_memslots(slots, NULL, kvm->memslots->generation);
 
 		old_memslots = kvm->memslots;
 		rcu_assign_pointer(kvm->memslots, slots);
@@ -862,7 +863,7 @@ int __kvm_set_memory_region(struct kvm *kvm,
 		memset(&new.arch, 0, sizeof(new.arch));
 	}
 
-	update_memslots(slots, &new);
+	update_memslots(slots, &new, kvm->memslots->generation);
 	old_memslots = kvm->memslots;
 	rcu_assign_pointer(kvm->memslots, slots);
 	synchronize_srcu_expedited(&kvm->srcu);

> > The original patch can be reverted, there are no following dependencies,
> > but the idea was that we're making the memslot array larger, so there
> > could be more pressure in allocating it, so let's not trivially do extra
> > frees and allocs.  Thanks,
> 
> I agree that the current set_memory_region() is not good for frequent updates.
> But the alloc/free is not a major factor at the moment: flushing shadows should
> be more problematic.

I don't understand why we'd throw away even a minor optimization that's
so easy to fix.  We're not only removing a free/alloc, but we're being
more friendly to the cache by avoiding the extra memcpy.  The above also
makes the generation management a bit more explicit.  Thanks,

Alex