Hi,
Most likely you would need to set the “optimize” attribute on some of the inline functions in the kernel headers. Please check the build log for the actual function that calls __smp_store_release(), e.g.:
1><...>\include\asm-generic\qrwlock.h(165,2): note : in expansion of macro `smp_store_release'
E.g. with kernel 4.10 you would need to set the optimization for the queued_write_unlock() function:
static inline void queued_write_unlock(struct qrwlock *lock)
{
smp_store_release(__qrwlock_write_byte(lock), 0);
}