> On Jul 30, 2019, at 1:26 PM, Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote: > > On Mon, Jul 29, 2019 at 10:03:40PM +0000, Song Liu wrote: >>> +/* If you add more flags, increment FGP_ORDER_SHIFT */ >>> +#define FGP_ORDER_SHIFT 7 >>> +#define FGP_PMD ((PMD_SHIFT - PAGE_SHIFT) << FGP_ORDER_SHIFT) >>> +#define FGP_PUD ((PUD_SHIFT - PAGE_SHIFT) << FGP_ORDER_SHIFT) >>> +#define fgp_get_order(fgp) ((fgp) >> FGP_ORDER_SHIFT) >> >> This looks like we want support order up to 25 (32 - 7). I guess we don't >> need that many. How about we specify the highest order to support here? > > We can support all the way up to order 64 with just 6 bits, leaving 32 - > 6 - 7 = 19 bits free. We haven't been adding FGP flags very quickly, > so I doubt we'll need anything larger. lol. I misread the bit usage. > >> Also, fgp_flags is signed int, so we need to make sure fgp_flags is not >> negative. > > If we ever get there, I expect people to convert the parameter from signed > int to unsigned long. Agreed. Thanks, Song
