On 6/16/22 10:52, Christophe Leroy wrote: > > > Le 16/06/2022 à 06:09, Anshuman Khandual a écrit : >> __SXXX/__PXXX macros is an unnecessary abstraction layer in creating the >> generic protection_map[] array which is used for vm_get_page_prot(). This >> abstraction layer can be avoided, if the platforms just define the array >> protection_map[] for all possible vm_flags access permission combinations. >> >> This series drops __SXXX/__PXXX macros from across platforms in the tree. >> First it makes protection_map[] array private (static) on platforms which >> enable ARCH_HAS_VM_GET_PAGE_PROT, later moves protection_map[] array into >> arch for all remaining platforms (!ARCH_HAS_VM_GET_PAGE_PROT), dropping >> the generic one. In the process __SXXX/__PXXX macros become redundant and >> thus get dropped off completely. I understand that the diff stat is large >> here, but please do suggest if there is a better way. This series applies >> on v5.19-rc1 and has been build tested for multiple platforms. > > Maybe this patch could be split with one patch per architecture. All you > have to do for that is to guard the generic protection_map declaration > with #ifdef __S000 , then the architectures can be migrated one by one. > >> >> The CC list for this series has been reduced to just minimum, until there >> is some initial agreement. > > Agreement with who if people don't know this series exists ? > > I think you should keep the architecture lists in copy allthough you > don't include individual maintainers/reviewers for now. Sure, will do. > >> >> - Anshuman >> >> Changes in V3: >> >> - Fix build issues on powerpc and riscv >> >> Changes in V2: > > I guess V2 was only sent to linux-mm as well ? Too bad. I was in a dilemma, whether to first arrive at something more acceptable or just engage all stake holders from the beginning. I understand your concern and hence will copy architecture mailing lists from next time onward.
