On Thu, 22 Nov 2018 12:14:20 +0900 Masahiro Yamada <yamada.masah...@socionext.com> wrote:
> When I tried to enable BUILD_BUG_ON for Sparse, the kbuild test robot > reported lots of "unknown expression" warnings from container_of(), > which seemed false positive. > > I addressed this in [1], but fixing Sparse is the right thing to do. > > The issue was fixed by Sparse commit 0eb8175d3e9c ("fix expansion of > function designator"), but it will take time until the fixed version > of Sparse is widely available. > > Disable the container_of() type checks for Sparse for now. > > [1] > https://lore.kernel.org/lkml/1542623503-3755-1-git-send-email-yamada.masah...@socionext.com/ > > ... > > --- a/include/linux/kernel.h > +++ b/include/linux/kernel.h > @@ -985,6 +985,21 @@ static inline void ftrace_dump(enum ftrace_dump_mode > oops_dump_mode) { } > #define __CONCAT(a, b) a ## b > #define CONCATENATE(a, b) __CONCAT(a, b) > > +/* > + * TODO: > + * Sparse emits "unknown expression" warnings. > + * It was fixed by commit 0eb8175d3e9c0d20354763d07ce3d4c0e543d988 in Sparse. > + * Remove the following workaround when the fixed Sparse is widely available. > + */ > +#ifdef __CHECKER__ > +#define TYPE_CHECK_CONTAINER_OF(ptr, type, member) do {} while (0) > +#else > +#define TYPE_CHECK_CONTAINER_OF(ptr, type, member) \ > + BUILD_BUG_ON_MSG(!__same_type(*(ptr), ((type *)0)->member) && \ > + !__same_type(*(ptr), void), \ > + "pointer type mismatch in container_of()") > +#endif I think that's OK. A few years hence, someone will happen upon this comment and will perform the obvious cleanup.