On Tue, Aug 18, 2015 at 05:43:00PM +0200, Michal Hocko wrote: > On Mon 17-08-15 18:09:04, Kirill A. Shutemov wrote: > > The patch halves space occupied by compound_dtor and compound_order in > > struct page. > > > > For compound_order, it's trivial long -> int/short conversion. > > > > For get_compound_page_dtor(), we now use hardcoded table for destructor > > lookup and store its index in the struct page instead of direct pointer > > to destructor. It shouldn't be a big trouble to maintain the table: we > > have only two destructor and NULL currently. > > > > This patch free up one word in tail pages for reuse. This is preparation > > for the next patch. > > > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx> > > Reviewed-by: Michal Hocko <mhocko@xxxxxxxx> > > [...] > > @@ -145,8 +143,13 @@ struct page { > > */ > > /* First tail page of compound page */ > > struct { > > - compound_page_dtor *compound_dtor; > > - unsigned long compound_order; > > +#ifdef CONFIG_64BIT > > + unsigned int compound_dtor; > > + unsigned int compound_order; > > +#else > > + unsigned short int compound_dtor; > > + unsigned short int compound_order; > > +#endif > > }; > > Why do we need this ifdef? We can go with short for both 32b and 64b > AFAICS. My assumption was that operations on ints can be faster on some [micro]arhictectures. I'm not sure if it's ever true. > We do not use compound_order for anything else than the order, right? Right. > While I am looking at this, it seems we are jugling with type for order > quite a lot - int, unsing int and even unsigned long. Yeah. It's mess. I'll check if I can fix anything of it in v3. -- Kirill A. Shutemov -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx"> email@xxxxxxxxx </a>