Messages in this thread | | | Date | Thu, 15 Feb 2024 11:21:58 +0100 | From | "Pankaj Raghav (Samsung)" <> | Subject | Re: [RFC v2 01/14] fs: Allow fine-grained control of folio sizes |
| |
On Wed, Feb 14, 2024 at 06:49:49PM +0000, Matthew Wilcox wrote: > On Tue, Feb 13, 2024 at 10:37:00AM +0100, Pankaj Raghav (Samsung) wrote: > > +static inline void mapping_set_folio_orders(struct address_space *mapping, > > + unsigned int min, unsigned int max) > > +{ > > + if (min == 1) > > + min = 2; > > If you order the "support order-1 folios" patch first, you can drop > these two lines. > Thanks for pointing this out. I actually forgot to update this later in my series.
The only failure I was noticing for LBS in 8k block sizes (generic/630) gets fixed by this change as well :) .
> > +static inline unsigned int mapping_min_folio_nrpages(struct address_space *mapping) > > I'm not sure if you need this, but it should return unsigned long, not > unsigned int. With 64KiB pages on Arm, a PMD page is 512MiB (order 13) > and a PUD page will be order 26, which is far too close to 2^32 for > my comfort.
There were some suggestions from Chinner which might make this function go away. But in case I need it, I will update it to unsigned long to be on the safe side.
| |