Add POPCOUNT function (#2166)

include/helpers.hpp

@@ -42,12 +42,13 @@ static inline void unreachable_() {
 
 // Ideally we'd use `std::bit_width`, but it has insufficient compiler support
 #ifdef __GNUC__ // GCC or compatible
-	#define ctz __builtin_ctz
-	#define clz __builtin_clz
+	#define ctz      __builtin_ctz
+	#define clz      __builtin_clz
+	#define popcount __builtin_popcount
 
 #elif defined(_MSC_VER)
 	#include <intrin.h>
-	#pragma intrinsic(_BitScanReverse, _BitScanForward)
+	#pragma intrinsic(_BitScanReverse, _BitScanForward, __popcnt)
 
 static inline int ctz(unsigned int x) {
 	assume(x != 0);
@@ -63,6 +64,10 @@ static inline int clz(unsigned int x) {
 	return 31 - count;
 }
 
+static inline int popcount(unsigned int x) {
+	return __popcnt(x);
+}
+
 #else
 	#include <limits.h>
 
@@ -85,6 +90,15 @@ static inline int clz(unsigned int x) {
 	}
 	return count;
 }
+
+static inline int popcount(unsigned int x) {
+	int count = 0;
+	while (x) {
+		x &= x - 1;
+		++count;
+	}
+	return count;
+}
 #endif
 
 // Macros for stringification

include/linkdefs.hpp

@@ -57,6 +57,7 @@ enum RPNCommand {
 	RPN_LOW = 0x71,
 	RPN_BITWIDTH = 0x72,
 	RPN_TZCOUNT = 0x73,
+	RPN_POPCOUNT = 0x74,
 
 	RPN_CONST = 0x80,
 	RPN_SYM = 0x81

include/opmath.hpp

@@ -23,5 +23,6 @@ int32_t op_low(int32_t value);
 
 int32_t op_bitwidth(int32_t value);
 int32_t op_tzcount(int32_t value);
+int32_t op_popcount(int32_t value);
 
 #endif // RGBDS_OP_MATH_HPP