consistently use _NSIG/8 idiom for kernel sigset size in sigaction

this code in sigaction was the only place where sizeof was being
applied to the kernel sigaction's mask member to get the size argument
to pass to the kernel. everywhere else, _NSIG/8 is used for this
purpose.

src/signal/sigaction.c

@@ -53,9 +53,9 @@ int __libc_sigaction(int sig, const struct sigaction *restrict sa, struct sigact
 		ksa.handler = sa->sa_handler;
 		ksa.flags = sa->sa_flags | SA_RESTORER;
 		ksa.restorer = (sa->sa_flags & SA_SIGINFO) ? __restore_rt : __restore;
-		memcpy(&ksa.mask, &sa->sa_mask, sizeof ksa.mask);
+		memcpy(&ksa.mask, &sa->sa_mask, _NSIG/8);
 	}
-	int r = __syscall(SYS_rt_sigaction, sig, sa?&ksa:0, old?&ksa_old:0, sizeof ksa.mask);
+	int r = __syscall(SYS_rt_sigaction, sig, sa?&ksa:0, old?&ksa_old:0, _NSIG/8);
 	if (sig == SIGABRT && sa && sa->sa_handler != SIG_DFL) {
 		UNLOCK(__abort_lock);
 		__restore_sigs(&set);
@@ -63,7 +63,7 @@ int __libc_sigaction(int sig, const struct sigaction *restrict sa, struct sigact
 	if (old && !r) {
 		old->sa_handler = ksa_old.handler;
 		old->sa_flags = ksa_old.flags;
-		memcpy(&old->sa_mask, &ksa_old.mask, sizeof ksa_old.mask);
+		memcpy(&old->sa_mask, &ksa_old.mask, _NSIG/8);
 	}
 	return __syscall_ret(r);
 }