On Wed, Sep 10, 2025 at 1:16 PM NightStrike <nightstr...@gmail.com> wrote: > > > > On Mon, Sep 8, 2025, 03:47 Chris Packham via Gcc <gcc@gcc.gnu.org> wrote: >> >> Hi GCC, >> >> For various reasons I find myself working with a few code bases that define >> their own wrappers for memset(). Unfortunately these wrappers defeat gccs >> ability to detect when the size of the pointer is passed instead of the >> size of the pointed to object. >> >> Is there any attribute I can set on the memset like functions that will let >> gcc know to perform the same kinds of checks as the standard memset >> function? >> >> Thanks, >> Chris > > > Would nonnull_if_nonzero help?
It's in the ball-park but not quite. I'm trying to catch cases where a non-zero size is passed but it's wrong. GCC will warn if I do something like struct foo *pt = &some_foo; memset(pt, 0, sizeof(pt)); test.c:13:29: warning: argument to ‘sizeof’ in ‘memset’ call is the same expression as the destination; did you mean to dereference it? [-Wsizeof-pointer-memaccess] But if for some reason I have struct foo *pt = &some_foo; super_sdk_memset(pt, 0, sizeof(pt)); Then it doesn't know to apply the same heuristic to my vendors[*] memset wrapper. [*] quite why vendors insist on wrapping basic C functions with these terrible wrappers is beyond me but I don't think I can change 30 years of industry practice.
