Add support for function pointers.

Both uniform and varying function pointers are supported; when a function
is called through a varying function pointer, each unique function pointer
value across the running program instances is called once for the set of
active program instances that want to call it.

builtins.m4

@@ -1080,6 +1080,14 @@ define internal <$1 x i32> @__sext_varying_bool(<$1 x i32>) nounwind readnone al
   ret <$1 x i32> %0
 }
 
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; count trailing zeros
+
+define internal i32 @__count_trailing_zeros(i32) nounwind readnone alwaysinline {
+  %c = call i32 @llvm.cttz.i32(i32 %0)
+  ret i32 %c
+}
+
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;; AOS/SOA conversion primitives
 
@@ -2457,7 +2465,7 @@ done:
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;; reduce_equal
 
-; count leading zeros
+; count trailing zeros
 declare i32 @llvm.cttz.i32(i32)
 
 define(`reduce_equal_aux', `