On Fri, 15 May 2015, Ralf Baechle wrote: > After solving the include ordering issue, that is ... I think the issue just means these generic BIT* macros (and GENMASK* too) belong to a header that does not rely on machine dependencies. And quite possibly one that has no C code at all as they would be useful, in a variant appropriately defined, for assembly code too (e.g. to replace stuff in <asm/mipsregs.h> where we have `(_ULCAST_(1) << 0)', etc.). Putting them back to <linux/types.h> where some of them originally came from with d05be13b might be one option. Maciej
