Avi Kivity <avi@xxxxxxxxxx> wrote:
> > + * Iteration must be started by this function. This should also be used after
> > + * removing/dropping sptes from rmap because in such cases the information in
> > + * the itererator may not be valid.
>
> Note: this suggests rmap_remove(struct rmap_iterator *ri) to remove an
> rmap from the iterator while keeping it valid. Converts a potentially
> quadratic kvm_mmu_rmap_write_protect() to linear.
Yes, but a bit scary because we need to touch drop_spte() and so on.
Actually drop_spte() itself should be changed to take rmap as an argument to
avoid recalculating it from sp.
I personally do not like current rmap_* interface. We should think of
a better design for the future.
> > +static bool rmap_get_first(unsigned long rmap, struct rmap_iterator *iter)
> > +{
> > + if (!rmap) {
> > + iter->sptep = NULL;
> > + return false;
> > + }
> > +
> > + if (!(rmap & 1)) {
> > + iter->sptep = (u64 *)rmap;
> > + iter->desc = NULL;
> > + } else {
> > + iter->desc = (struct pte_list_desc *)(rmap & ~1ul);
> > + iter->pos = 0;
> > + iter->sptep = iter->desc->sptes[iter->pos];
> > + }
> > +
> > + return true;
> > +}
>
> Might be simplified to return the sptep or NULL; so you don't have to
> copy it to the iterator.
while ((spte = rmap_...)) or for (rmap_...; iter.sptep; ...)
Will think again!
Takuya
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
