On Wed 27-06-18 12:47:39, Vlastimil Babka wrote: > On 06/27/2018 09:54 AM, Michal Hocko wrote: > > On Wed 27-06-18 09:50:01, Vlastimil Babka wrote: > >> On 06/27/2018 09:34 AM, Michal Hocko wrote: > >>> On Tue 26-06-18 10:04:16, Andrew Morton wrote: > >>> > >>> And as I've argued before the code would be wrong regardless. We would > >>> leak the memory or worse touch somebody's else kmap without knowing > >>> that. So we have a choice between a mem leak, data corruption k or a > >>> silent fixup. I would prefer the last option. And blowing up on a BUG > >>> is not much better on something that is easily fixable. I am not really > >>> convinced that & ~__GFP_HIGHMEM is something to lose sleep over. > >> > >> Maybe put the fixup into a "#ifdef CONFIG_HIGHMEM" block and then modern > >> systems won't care? In that case it could even be if (WARN_ON_ONCE(...)) > >> so future cases with wrong expectations would become known. > > > > Yes that could be done as well. Or maybe we can make __GFP_HIGHMEM 0 for > > !HIGHMEM systems. Does something really rely on it being non-zero? > > I guess gfp_zone() would have to be checked, dunno about the rewrite of > GFP_ZONE_TABLE (CCing people). > In general checks like "if (flags & __GFP_HIGHMEM)" would become false, > which probably should not be a problem, unless something expect the flag > to be there and errors out if it isn't. Well, __GFP_HIGHMEM should be basically GFP_KERNEL for !highmem systems. But most checks I have seen try to mask it off. Having it 0 would help to reduce at least some code. -- Michal Hocko SUSE Labs
