Bit operations

Width-specific bit-twiddling macros: count-leading / trailing zeros, population count, parity, and masked shift / rotate. More...

Files
file	rbitops.h
	Width-specific count-zeros / popcount / parity / shift / rotate macros.

Macros
#define	RUINT64_CLZ(x)   RULONGLONG_CLZ (x)
#define	RUINT64_CTZ(x)   RULONGLONG_CTZ (x)
#define	RUINT64_POPCOUNT(x)   RULONGLONG_POPCOUNT (x)
#define	RUINT64_PARITY(x)   RULONGLONG_PARITY (x)
#define	RUINT64_SHL(x, n)   (((x) & RUINT64_MAX) << ((n) & 63))
#define	RUINT64_SHR(x, n)   (((x) & RUINT64_MAX) >> ((n) & 63))
#define	RUINT64_ROTL(x, n)   (RUINT64_SHL (x, n) | RUINT64_SHR (x, 64-(n)))
#define	RUINT64_ROTR(x, n)   (RUINT64_SHR (x, n) | RUINT64_SHL (x, 64-(n)))
#define	RUINT32_CLZ(x)   (RUINT64_CLZ (x & RUINT32_MAX) - 32)
#define	RUINT32_CTZ(x)   RUINT64_CTZ (x & RUINT32_MAX)
#define	RUINT32_POPCOUNT(x)   RUINT64_POPCOUNT (x & RUINT32_MAX)
#define	RUINT32_PARITY(x)   RUINT64_PARITY (x & RUINT32_MAX)
#define	RUINT32_SHL(x, n)   (((x) & RUINT32_MAX) << ((n) & 31))
	Shift a 32-bit x left by n (operand and count masked to width).
#define	RUINT32_SHR(x, n)   (((x) & RUINT32_MAX) >> ((n) & 31))
	Shift a 32-bit x right by n (operand and count masked to width).
#define	RUINT32_ROTL(x, n)   (RUINT32_SHL (x, n) | RUINT32_SHR (x, 32-(n)))
	Rotate a 32-bit x left by n bits.
#define	RUINT32_ROTR(x, n)   (RUINT32_SHR (x, n) | RUINT32_SHL (x, 32-(n)))
	Rotate a 32-bit x right by n bits.
#define	RUINT16_CLZ(x)   (RUINT32_CLZ (x & RUINT16_MAX) - 16)
#define	RUINT16_CTZ(x)   MIN (RUINT32_CTZ (x & RUINT16_MAX), 16)
#define	RUINT16_POPCOUNT(x)   RUINT32_POPCOUNT (x & RUINT16_MAX)
#define	RUINT16_PARITY(x)   RUINT32_PARITY (x & RUINT16_MAX)
#define	RUINT16_SHL(x, n)   (((x) & RUINT16_MAX) << ((n) & 15))
#define	RUINT16_SHR(x, n)   (((x) & RUINT16_MAX) >> ((n) & 15))
#define	RUINT16_ROTL(x, n)   (RUINT16_SHL (x, n) | RUINT16_SHR (x, 16-(n)))
#define	RUINT16_ROTR(x, n)   (RUINT16_SHR (x, n) | RUINT16_SHL (x, 16-(n)))
#define	RUINT8_CLZ(x)   (RUINT32_CLZ (x & RUINT8_MAX) - 24)
#define	RUINT8_CTZ(x)   MIN (RUINT32_CTZ (x & RUINT8_MAX), 8)
#define	RUINT8_POPCOUNT(x)   RUINT32_POPCOUNT (x & RUINT8_MAX)
#define	RUINT8_PARITY(x)   RUINT32_PARITY (x & RUINT8_MAX)
#define	RUINT8_SHL(x, n)   (((x) & RUINT8_MAX) << ((n) & 7))
#define	RUINT8_SHR(x, n)   (((x) & RUINT8_MAX) >> ((n) & 7))
#define	RUINT8_ROTL(x, n)   (RUINT8_SHL (x, n) | RUINT8_SHR (x, 8-(n)))
#define	RUINT8_ROTR(x, n)   (RUINT8_SHR (x, n) | RUINT8_SHL (x, 8-(n)))

Detailed Description

Width-specific bit-twiddling macros: count-leading / trailing zeros, population count, parity, and masked shift / rotate.

Every macro follows the R<WIDTH>_<OP> naming convention, where WIDTH is one of UINT8 / UINT16 / UINT32 / UINT64 / SIZE (plus the native-width UINT / ULONG / ULONGLONG helpers the fixed-width ones are built on), and OP is one of:

• _CLZ — count leading zero bits (width for a zero input).
• _CTZ — count trailing zero bits (width for a zero input).
• _POPCOUNT — number of set bits.
• _PARITY — 1 if an odd number of bits are set, else 0.
• _SHL / _SHR — shift left / right, masking the operand to its width and the shift count to [0, width).
• _ROTL / _ROTR — rotate left / right within the width.

CLZ / CTZ / POPCOUNT / PARITY compile down to the compiler's intrinsics (GCC/Clang __builtin_*, MSVC _BitScan* / __popcnt); the rest are plain macros. The RUINT32_CLZ family below is documented as the canonical example — the same eight operations exist for every listed width.