https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123172
--- Comment #2 from Zdenek Sojka <zsojka at seznam dot cz> --- As a user, I don't see what I did wrong; I only see gcc automatically generating call to __atomic_*_16() which needs 16B aligned argument, even though the argument is known not to be aligned, when I use the generic __atomic_*() call (I don't explicitly use __atomic_*_16() ). Could GCC generate call to __atomic_*() instead, using the locks internally?
