On Thu, Apr 11, 2024 at 11:11:59AM -0700, John Stultz wrote: > On Thu, Apr 11, 2024 at 8:39 AM Nathan Chancellor <nathan@xxxxxxxxxx> wrote: > > On Wed, Apr 10, 2024 at 04:26:29PM -0700, John Stultz wrote: > > > Building with clang, I see the following warning: > > > > > > In file included from posix_timers.c:17: > > > ./../kselftest.h:398:6: warning: variable 'major' is used uninitialized whenever '||' condition is true [-Wsometimes-uninitialized] > > > if (uname(&info) || sscanf(info.release, "%u.%u.", &major, &minor) != 2) > > > ^~~~~~~~~~~~ > > > ./../kselftest.h:401:9: note: uninitialized use occurs here > > > return major > min_major || (major == min_major && minor >= min_minor); > > > ^~~~~ > > > > > > This is a bit of a red-herring as if the uname() call did fail, > > > we would hit ksft_exit_fail_msg() which should exit. > > > > Correct, although we have not really conveyed that to the compiler, > > right? exit() is noreturn, which means all functions that call exit() > > unconditionally are also noreturn, such as ksft_exit_fail_msg(). LLVM > > will figure this out once it performs inlining and such but that happens > > after clang's static analysis phase that this warning occurs in. I think > > a better solution would be to add __noreturn to the functions in > > tools/testing/selftests/kselftest.h that call exit(), so that the > > compiler is aware of this through all pipeline phases, maybe something > > like this? It resolves the wawrning for me. > > No objection from me if this is the better approach. > > Would you send that patch out? Done: https://lore.kernel.org/20240411-mark-kselftest-exit-funcs-noreturn-v1-1-b027c948f586@xxxxxxxxxx/ If you have to respin this series for some reason, feel free to include that change so that they go together, up to you though. Cheers, nathan
