Re: [PATCH v19 011/130] KVM: Add new members to struct kvm_gfn_range to operate on

From: Isaku Yamahata
Date: Tue Mar 19 2024 - 17:50:33 EST


On Tue, Mar 19, 2024 at 02:47:47PM +0000,
"Edgecombe, Rick P" <rick.p.edgecombe@xxxxxxxxx> wrote:

> On Mon, 2024-03-18 at 19:50 -0700, Rick Edgecombe wrote:
> > On Wed, 2024-03-13 at 10:14 -0700, Isaku Yamahata wrote:
> > > > IMO, an enum will be clearer than the two flags.
> > > >
> > > >     enum {
> > > >         PROCESS_PRIVATE_AND_SHARED,
> > > >         PROCESS_ONLY_PRIVATE,
> > > >         PROCESS_ONLY_SHARED,
> > > >     };
> > >
> > > The code will be ugly like
> > > "if (== PRIVATE || == PRIVATE_AND_SHARED)" or
> > > "if (== SHARED || == PRIVATE_AND_SHARED)"
> > >
> > > two boolean (or two flags) is less error-prone.
> >
> > Yes the enum would be awkward to handle. But I also thought the way
> > this is specified in struct kvm_gfn_range is a little strange.
> >
> > It is ambiguous what it should mean if you set:
> >  .only_private=true;
> >  .only_shared=true;
> > ...as happens later in the series (although it may be a mistake).
> >
> > Reading the original conversation, it seems Sean suggested this
> > specifically. But it wasn't clear to me from the discussion what the
> > intention of the "only" semantics was. Like why not?
> >  bool private;
> >  bool shared;
>
> I see Binbin brought up this point on v18 as well:
> https://lore.kernel.org/kvm/6220164a-aa1d-43d2-b918-6a6eaad769fb@xxxxxxxxxxxxxxx/#t
>
> and helpfully dug up some other discussion with Sean where he agreed
> the "_only" is confusing and proposed the the enum:
> https://lore.kernel.org/kvm/ZUO1Giju0GkUdF0o@xxxxxxxxxx/
>
> He wanted the default value (in the case the caller forgets to set
> them), to be to include both private and shared. I think the enum has
> the issues that Isaku mentioned. What about?
>
> bool exclude_private;
> bool exclude_shared;
>
> It will become onerous if more types of aliases grow, but it clearer
> semantically and has the safe default behavior.

I'm fine with those names. Anyway, I'm fine with wither way, two bools or enum.
--
Isaku Yamahata <isaku.yamahata@xxxxxxxxx>