From: Jiri Slaby > Sent: Wednesday, August 16, 2023 9:59 AM ... > > 'Vote up' my patches to minmax.h that make this all work. > > Then it won't care provided both values have the same signedness. > > (or, with patch 5, are non-negative 31bit compile time constants.) > > Oh yeah, that [1] looks great. Why should one care in min(4096, > sizeof()) after all… > > So what's the current status of those? Waiting... :-( The only comment is from Linus who really doesn't like the idea that min(signed_var, 4u) should be the same as min(signed_var, 4). I think he is ok with min(unsigned_var, 4) though. The min_t(u16,...) I quoted from the console buffer code is a real bug that was identified by someone else last week. Really min_t() is just an accident waiting to happen. David > > [1] > https://lore.kernel.org/all/b4ce9dad748e489f9314a2dc95615033@xxxxxxxxxxxxxxxx/ > > thanks, > -- > js > suse labs - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
