[PATCH v6 21/22] KVM: Allow for different capacities in kvm_mmu_memory_cache structs

[David Matlack]

On Mon, May 23, 2022 at 10:37 AM Sean Christopherson <seanjc at google.com> wrote:
>
> On Fri, May 20, 2022, Mingwei Zhang wrote:
> > On Mon, May 16, 2022 at 4:24 PM David Matlack <dmatlack at google.com> wrote:
> > > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> > > index e089db822c12..5e2e75014256 100644
> > > --- a/virt/kvm/kvm_main.c
> > > +++ b/virt/kvm/kvm_main.c
> > > @@ -369,14 +369,31 @@ static inline void *mmu_memory_cache_alloc_obj(struct kvm_mmu_memory_cache *mc,
> > >                 return (void *)__get_free_page(gfp_flags);
> > >  }
> > >
> > > -int kvm_mmu_topup_memory_cache(struct kvm_mmu_memory_cache *mc, int min)
> > > +static int __kvm_mmu_topup_memory_cache(struct kvm_mmu_memory_cache *mc, int capacity, int min)
> > >  {
> > > +       gfp_t gfp = GFP_KERNEL_ACCOUNT;
> > >         void *obj;
> > >
> > >         if (mc->nobjs >= min)
> > >                 return 0;
> > > -       while (mc->nobjs < ARRAY_SIZE(mc->objects)) {
> > > -               obj = mmu_memory_cache_alloc_obj(mc, GFP_KERNEL_ACCOUNT);
> > > +
> > > +       if (unlikely(!mc->objects)) {
> > > +               if (WARN_ON_ONCE(!capacity))
> > > +                       return -EIO;
> > > +
> > > +               mc->objects = kvmalloc_array(sizeof(void *), capacity, gfp);
> > > +               if (!mc->objects)
> > > +                       return -ENOMEM;
> > > +
> > > +               mc->capacity = capacity;
> >
> > Do we want to ensure the minimum value of the capacity? I think
> > otherwise, we may more likely start using memory from GFP_ATOMIC if
> > the capacity is less than, say 5? But the minimum value seems related
> > to each cache type.
>
> Eh, if we specify a minimum, just make the arch default the minimum.  That way we
> avoid adding even more magic/arbitrary numbers.  E.g. for whatever reason, MIPS's
> default is '4'.

I'm not exactly sure what you had in mind Mingwei. But there is a bug
in this code if min > capacity. This function will happily return 0
after filling up the cache, even though it did not allocate min
objects. The same bug existed before this patch if min >
ARRAY_SIZE(mc->objects). I can include a separate patch to fix this
bug (e.g. WARN and return -ENOMEM if min > capacity).