On 08/30/22 11:09, Sidhartha Kumar wrote:
>
>
> On 8/29/22 8:33 PM, Matthew Wilcox wrote:
> > On Mon, Aug 29, 2022 at 04:00:08PM -0700, Sidhartha Kumar wrote:
> > > #define TESTHPAGEFLAG(uname, flname) \
> > > +static __always_inline \
> > > +int folio_test_hugetlb_##flname(struct folio *folio) \
> > One change I made was to have folio_test_foo() return bool instead of
> > int. It helps the compiler really understand what's going on. Maybe
> > some humans too ;-)
> >
>
> I went with returning an int to stay consistent with the page version
> of the macros which return an int. I'm fine with changing it to return
> a bool.
I believe the page test macros returned an int when I added the hugetlb
specific versions. So, I just did the same. Since they are now bool,
it makes sense to have these be consistent.
--
Mike Kravetz
>
> > > + { void **private = &folio->private; \
> > > + return test_bit(HPG_##flname, (void *)((unsigned long)private)); \
> > I've made this tricky for you by making folio->private a void * instead
> > of the unsigned long in page. Would this look better as ...
> >
> > { \
> > void *private = &folio->private; \
> > return test_bit(HPG_##flname, private); \
> >
> > perhaps?
>
> Ya this looks much better and passes the tests, will add to v2.
>
>
>
