> <feature creep> > If we decide to proceed with the RTE_PTR_ADD/SUB macros not discarding > qualifiers, similar RTE_PTR_ macros should be updated similarly. > But let's cross that bridge if we go there. > </feature creep>
Is RTE_PTR_DIFF the only remaining? There is no difference in assembly by casting to uintptr_t https://godbolt.org/z/z77nEr6Ts (I assume bcz integer return means pointer provenance isn't broken). Native pointer types will throw an error if the pointers are not of the same type which can strengthen type checking, but similar call-site analysis will be needed as there is mixed usage (different ptr types, int + ptr types).
