Re: [PATCH v3 02/70] RAMBlock: Add support of KVM private guest memfd

[David Hildenbrand]

On 30.11.23 08:37, Xiaoyao Li wrote:
> On 11/20/2023 5:24 PM, David Hildenbrand wrote:
> > >    uint8_t memory_region_get_dirty_log_mask(MemoryRegion *mr)
> > >    {
> > >        uint8_t mask = mr->dirty_log_mask;
> > > diff --git a/system/physmem.c b/system/physmem.c
> > > index fc2b0fee0188..0af2213cbd9c 100644
> > > --- a/system/physmem.c
> > > +++ b/system/physmem.c
> > > @@ -1841,6 +1841,20 @@ static void ram_block_add(RAMBlock *new_block,
> > > Error **errp)
> > >            }
> > >        }
> > > +#ifdef CONFIG_KVM
> > > +    if (kvm_enabled() && new_block->flags & RAM_GUEST_MEMFD &&
> >
> >
> > I recall that we prefer to write this as
> >
> >       if (kvm_enabled() && (new_block->flags & RAM_GUEST_MEMFD) &&
>
> get it.
>
> Thanks!
>
> > > +        new_block->guest_memfd < 0) {
> > > +        /* TODO: to decide if KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is
> > > supported */
> > > +        uint64_t flags = 0;
> > > +        new_block->guest_memfd =
> > > kvm_create_guest_memfd(new_block->max_length,
> > > +                                                        flags, errp);
> >
> > Get rid of "flags" and just pass 0". Whatever code wants to pass flags
> > later can decide how to do that.
>
>
> How to handle it please see the reply to patch 3.

If patch #3 cannot go in now and has to be deferred, then please clean 
this here up. Otherwise, as suggested, squash with #3.

Depending on KVM_GUEST_MEMFD_ALLOW_HUGEPAGE support :)

--
Cheers,

David / dhildenb