REORG: include: move integer manipulation functions from standard.h to intops.h

There are quite a number of integer manipulation functions defined in
standard.h, which is one of the reasons why standard.h is included from
many places and participates to the dependencies loop.

Let's just have a new file, intops.h to place all these operations.
These are a few bitops, 32/64 bit mul/div/rotate, integer parsing and
encoding (including varints), the full avalanche hash function, and
the my_htonll/my_ntohll functions.

For now no new C file was created for these yet.
diff --git a/include/common/standard.h b/include/common/standard.h
index 2f1dffe..7a2ba2e 100644
--- a/include/common/standard.h
+++ b/include/common/standard.h
@@ -38,6 +38,7 @@
 #include <netinet/in.h>
 #include <arpa/inet.h>
 #include <haproxy/api.h>
+#include <haproxy/intops.h>
 #include <common/chunk.h>
 #include <common/namespace.h>
 #include <import/eb32tree.h>
@@ -75,26 +76,6 @@
 /* return the largest possible integer of type <ret>, with all bits set */
 #define MAX_RANGE(ret) (~(typeof(ret))0)
 
-/* rotate left a 64-bit integer by <bits:[0-5]> bits */
-static inline uint64_t rotl64(uint64_t v, uint8_t bits)
-{
-#if !defined(__ARM_ARCH_8A) && !defined(__x86_64__)
-	bits &= 63;
-#endif
-	v = (v << bits) | (v >> (-bits & 63));
-	return v;
-}
-
-/* rotate right a 64-bit integer by <bits:[0-5]> bits */
-static inline uint64_t rotr64(uint64_t v, uint8_t bits)
-{
-#if !defined(__ARM_ARCH_8A) && !defined(__x86_64__)
-	bits &= 63;
-#endif
-	v = (v >> bits) | (v << (-bits & 63));
-	return v;
-}
-
 /* DEFNULL() returns either the argument as-is, or NULL if absent. This is for
  * use in macros arguments.
  */
@@ -241,104 +222,6 @@
 	return ret;
 }
 
-/* returns the number of bytes needed to encode <v> as a varint. Be careful, use
- * it only with constants as it generates a large code (typ. 180 bytes). Use the
- * varint_bytes() version instead in case of doubt.
- */
-int varint_bytes(uint64_t v);
-static inline int __varint_bytes(uint64_t v)
-{
-	switch (v) {
-	case 0x0000000000000000 ... 0x00000000000000ef: return 1;
-	case 0x00000000000000f0 ... 0x00000000000008ef: return 2;
-	case 0x00000000000008f0 ... 0x00000000000408ef: return 3;
-	case 0x00000000000408f0 ... 0x00000000020408ef: return 4;
-	case 0x00000000020408f0 ... 0x00000001020408ef: return 5;
-	case 0x00000001020408f0 ... 0x00000081020408ef: return 6;
-	case 0x00000081020408f0 ... 0x00004081020408ef: return 7;
-	case 0x00004081020408f0 ... 0x00204081020408ef: return 8;
-	case 0x00204081020408f0 ... 0x10204081020408ef: return 9;
-	default: return 10;
-	}
-}
-
-/* Encode the integer <i> into a varint (variable-length integer). The encoded
- * value is copied in <*buf>. Here is the encoding format:
- *
- *        0 <= X < 240        : 1 byte  (7.875 bits)  [ XXXX XXXX ]
- *      240 <= X < 2288       : 2 bytes (11 bits)     [ 1111 XXXX ] [ 0XXX XXXX ]
- *     2288 <= X < 264432     : 3 bytes (18 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]   [ 0XXX XXXX ]
- *   264432 <= X < 33818864   : 4 bytes (25 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]*2 [ 0XXX XXXX ]
- * 33818864 <= X < 4328786160 : 5 bytes (32 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]*3 [ 0XXX XXXX ]
- * ...
- *
- * On success, it returns the number of written bytes and <*buf> is moved after
- * the encoded value. Otherwise, it returns -1. */
-static inline int
-encode_varint(uint64_t i, char **buf, char *end)
-{
-	unsigned char *p = (unsigned char *)*buf;
-	int r;
-
-	if (p >= (unsigned char *)end)
-		return -1;
-
-	if (i < 240) {
-		*p++ = i;
-		*buf = (char *)p;
-		return 1;
-	}
-
-	*p++ = (unsigned char)i | 240;
-	i = (i - 240) >> 4;
-	while (i >= 128) {
-		if (p >= (unsigned char *)end)
-			return -1;
-		*p++ = (unsigned char)i | 128;
-		i = (i - 128) >> 7;
-	}
-
-	if (p >= (unsigned char *)end)
-		return -1;
-	*p++ = (unsigned char)i;
-
-	r    = ((char *)p - *buf);
-	*buf = (char *)p;
-	return r;
-}
-
-/* Decode a varint from <*buf> and save the decoded value in <*i>. See
- * 'spoe_encode_varint' for details about varint.
- * On success, it returns the number of read bytes and <*buf> is moved after the
- * varint. Otherwise, it returns -1. */
-static inline int
-decode_varint(char **buf, char *end, uint64_t *i)
-{
-	unsigned char *p = (unsigned char *)*buf;
-	int r;
-
-	if (p >= (unsigned char *)end)
-		return -1;
-
-	*i = *p++;
-	if (*i < 240) {
-		*buf = (char *)p;
-		return 1;
-	}
-
-	r = 4;
-	do {
-		if (p >= (unsigned char *)end)
-			return -1;
-		*i += (uint64_t)*p << r;
-		r  += 7;
-	} while (*p++ >= 128);
-
-	r    = ((char *)p - *buf);
-	*buf = (char *)p;
-	return r;
-}
-
 /* returns a locally allocated string containing the quoted encoding of the
  * input string. The output may be truncated to QSTR_SIZE chars, but it is
  * guaranteed that the string will always be properly terminated. Quotes are
@@ -368,24 +251,6 @@
 extern int ishex(char s);
 
 /*
- * Return integer equivalent of character <c> for a hex digit (0-9, a-f, A-F),
- * otherwise -1. This compact form helps gcc produce efficient code.
- */
-static inline int hex2i(int c)
-{
-	if ((unsigned char)(c -= '0') > 9) {
-		if ((unsigned char)(c -= 'A' - '0') > 5 &&
-			      (unsigned char)(c -= 'a' - 'A') > 5)
-			c = -11;
-		c += 10;
-	}
-	return c;
-}
-
-/* rounds <i> down to the closest value having max 2 digits */
-unsigned int round_2dig(unsigned int i);
-
-/*
  * Checks <name> for invalid characters. Valid chars are [A-Za-z0-9_:.-]. If an
  * invalid character is found, a pointer to it is returned. If everything is
  * fine, NULL is returned.
@@ -625,101 +490,6 @@
  */
 int url_decode(char *string, int in_form);
 
-/* This one is 6 times faster than strtoul() on athlon, but does
- * no check at all.
- */
-static inline unsigned int __str2ui(const char *s)
-{
-	unsigned int i = 0;
-	while (*s) {
-		i = i * 10 - '0';
-		i += (unsigned char)*s++;
-	}
-	return i;
-}
-
-/* This one is 5 times faster than strtoul() on athlon with checks.
- * It returns the value of the number composed of all valid digits read.
- */
-static inline unsigned int __str2uic(const char *s)
-{
-	unsigned int i = 0;
-	unsigned int j;
-	while (1) {
-		j = (*s++) - '0';
-		if (j > 9)
-			break;
-		i *= 10;
-		i += j;
-	}
-	return i;
-}
-
-/* This one is 28 times faster than strtoul() on athlon, but does
- * no check at all!
- */
-static inline unsigned int __strl2ui(const char *s, int len)
-{
-	unsigned int i = 0;
-	while (len-- > 0) {
-		i = i * 10 - '0';
-		i += (unsigned char)*s++;
-	}
-	return i;
-}
-
-/* This one is 7 times faster than strtoul() on athlon with checks.
- * It returns the value of the number composed of all valid digits read.
- */
-static inline unsigned int __strl2uic(const char *s, int len)
-{
-	unsigned int i = 0;
-	unsigned int j, k;
-
-	while (len-- > 0) {
-		j = (*s++) - '0';
-		k = i * 10;
-		if (j > 9)
-			break;
-		i = k + j;
-	}
-	return i;
-}
-
-/* This function reads an unsigned integer from the string pointed to by <s>
- * and returns it. The <s> pointer is adjusted to point to the first unread
- * char. The function automatically stops at <end>.
- */
-static inline unsigned int __read_uint(const char **s, const char *end)
-{
-	const char *ptr = *s;
-	unsigned int i = 0;
-	unsigned int j, k;
-
-	while (ptr < end) {
-		j = *ptr - '0';
-		k = i * 10;
-		if (j > 9)
-			break;
-		i = k + j;
-		ptr++;
-	}
-	*s = ptr;
-	return i;
-}
-
-unsigned long long int read_uint64(const char **s, const char *end);
-long long int read_int64(const char **s, const char *end);
-
-extern unsigned int str2ui(const char *s);
-extern unsigned int str2uic(const char *s);
-extern unsigned int strl2ui(const char *s, int len);
-extern unsigned int strl2uic(const char *s, int len);
-extern int strl2ic(const char *s, int len);
-extern int strl2irc(const char *s, int len, int *ret);
-extern int strl2llrc(const char *s, int len, long long *ret);
-extern int strl2llrc_dotted(const char *text, int len, long long *ret);
-extern unsigned int read_uint(const char **s, const char *end);
 unsigned int inetaddr_host(const char *text);
 unsigned int inetaddr_host_lim(const char *text, const char *stop);
 unsigned int inetaddr_host_lim_ret(char *text, char *stop, char **ret);
@@ -836,168 +606,6 @@
 #define HOUR (60 * MINUTE)
 #define DAY (24 * HOUR)
 
-/* Multiply the two 32-bit operands and shift the 64-bit result right 32 bits.
- * This is used to compute fixed ratios by setting one of the operands to
- * (2^32*ratio).
- */
-static inline unsigned int mul32hi(unsigned int a, unsigned int b)
-{
-	return ((unsigned long long)a * b) >> 32;
-}
-
-/* gcc does not know when it can safely divide 64 bits by 32 bits. Use this
- * function when you know for sure that the result fits in 32 bits, because
- * it is optimal on x86 and on 64bit processors.
- */
-static inline unsigned int div64_32(unsigned long long o1, unsigned int o2)
-{
-	unsigned long long result;
-#ifdef __i386__
-	asm("divl %2"
-	    : "=A" (result)
-	    : "A"(o1), "rm"(o2));
-#else
-	result = o1 / o2;
-#endif
-	return result;
-}
-
-/* Simple popcountl implementation. It returns the number of ones in a word.
- * Described here : https://graphics.stanford.edu/~seander/bithacks.html
- */
-static inline unsigned int my_popcountl(unsigned long a)
-{
-	a = a - ((a >> 1) & ~0UL/3);
-	a = (a & ~0UL/15*3) + ((a >> 2) & ~0UL/15*3);
-	a = (a + (a >> 4)) & ~0UL/255*15;
-	return (unsigned long)(a * (~0UL/255)) >> (sizeof(unsigned long) - 1) * 8;
-}
-
-/* returns non-zero if <a> has at least 2 bits set */
-static inline unsigned long atleast2(unsigned long a)
-{
-	return a & (a - 1);
-}
-
-/* Simple ffs implementation. It returns the position of the lowest bit set to
- * one, starting at 1. It is illegal to call it with a==0 (undefined result).
- */
-static inline unsigned int my_ffsl(unsigned long a)
-{
-	unsigned long cnt;
-
-#if defined(__x86_64__)
-	__asm__("bsf %1,%0\n" : "=r" (cnt) : "rm" (a));
-	cnt++;
-#else
-
-	cnt = 1;
-#if LONG_MAX > 0x7FFFFFFFL /* 64bits */
-	if (!(a & 0xFFFFFFFFUL)) {
-		a >>= 32;
-		cnt += 32;
-	}
-#endif
-	if (!(a & 0XFFFFU)) {
-		a >>= 16;
-		cnt += 16;
-	}
-	if (!(a & 0XFF)) {
-		a >>= 8;
-		cnt += 8;
-	}
-	if (!(a & 0xf)) {
-		a >>= 4;
-		cnt += 4;
-	}
-	if (!(a & 0x3)) {
-		a >>= 2;
-		cnt += 2;
-	}
-	if (!(a & 0x1)) {
-		cnt += 1;
-	}
-#endif /* x86_64 */
-
-	return cnt;
-}
-
-/* Simple fls implementation. It returns the position of the highest bit set to
- * one, starting at 1. It is illegal to call it with a==0 (undefined result).
- */
-static inline unsigned int my_flsl(unsigned long a)
-{
-	unsigned long cnt;
-
-#if defined(__x86_64__)
-	__asm__("bsr %1,%0\n" : "=r" (cnt) : "rm" (a));
-	cnt++;
-#else
-
-	cnt = 1;
-#if LONG_MAX > 0x7FFFFFFFUL /* 64bits */
-	if (a & 0xFFFFFFFF00000000UL) {
-		a >>= 32;
-		cnt += 32;
-	}
-#endif
-	if (a & 0XFFFF0000U) {
-		a >>= 16;
-		cnt += 16;
-	}
-	if (a & 0XFF00) {
-		a >>= 8;
-		cnt += 8;
-	}
-	if (a & 0xf0) {
-		a >>= 4;
-		cnt += 4;
-	}
-	if (a & 0xc) {
-		a >>= 2;
-		cnt += 2;
-	}
-	if (a & 0x2) {
-		cnt += 1;
-	}
-#endif /* x86_64 */
-
-	return cnt;
-}
-
-/* Build a word with the <bits> lower bits set (reverse of my_popcountl) */
-static inline unsigned long nbits(int bits)
-{
-	if (--bits < 0)
-		return 0;
-	else
-		return (2UL << bits) - 1;
-}
-
-/* sets bit <bit> into map <map>, which must be long-aligned */
-static inline void ha_bit_set(unsigned long bit, long *map)
-{
-	map[bit / (8 * sizeof(*map))] |= 1UL << (bit & (8 * sizeof(*map) - 1));
-}
-
-/* clears bit <bit> from map <map>, which must be long-aligned */
-static inline void ha_bit_clr(unsigned long bit, long *map)
-{
-	map[bit / (8 * sizeof(*map))] &= ~(1UL << (bit & (8 * sizeof(*map) - 1)));
-}
-
-/* flips bit <bit> from map <map>, which must be long-aligned */
-static inline void ha_bit_flip(unsigned long bit, long *map)
-{
-	map[bit / (8 * sizeof(*map))] ^= 1UL << (bit & (8 * sizeof(*map) - 1));
-}
-
-/* returns non-zero if bit <bit> from map <map> is set, otherwise 0 */
-static inline int ha_bit_test(unsigned long bit, const long *map)
-{
-	return !!(map[bit / (8 * sizeof(*map))] & 1UL << (bit & (8 * sizeof(*map) - 1)));
-}
-
 /*
  * Parse binary string written in hexadecimal (source) and store the decoded
  * result into binstr and set binstrlen to the length of binstr. Memory for
@@ -1062,44 +670,6 @@
 /* returns an operator among STD_OP_* for string <str> or < 0 if unknown */
 int get_std_op(const char *str);
 
-/* hash a 32-bit integer to another 32-bit integer */
-extern unsigned int full_hash(unsigned int a);
-static inline unsigned int __full_hash(unsigned int a)
-{
-	/* This function is one of Bob Jenkins' full avalanche hashing
-	 * functions, which when provides quite a good distribution for little
-	 * input variations. The result is quite suited to fit over a 32-bit
-	 * space with enough variations so that a randomly picked number falls
-	 * equally before any server position.
-	 * Check http://burtleburtle.net/bob/hash/integer.html for more info.
-	 */
-	a = (a+0x7ed55d16) + (a<<12);
-	a = (a^0xc761c23c) ^ (a>>19);
-	a = (a+0x165667b1) + (a<<5);
-	a = (a+0xd3a2646c) ^ (a<<9);
-	a = (a+0xfd7046c5) + (a<<3);
-	a = (a^0xb55a4f09) ^ (a>>16);
-
-	/* ensure values are better spread all around the tree by multiplying
-	 * by a large prime close to 3/4 of the tree.
-	 */
-	return a * 3221225473U;
-}
-
-/* Return the bit position in mask <m> of the nth bit set of rank <r>, between
- * 0 and LONGBITS-1 included, starting from the left. For example ranks 0,1,2,3
- * for mask 0x55 will be 6, 4, 2 and 0 respectively. This algorithm is based on
- * a popcount variant and is described here :
- *   https://graphics.stanford.edu/~seander/bithacks.html
- */
-unsigned int mask_find_rank_bit(unsigned int r, unsigned long m);
-unsigned int mask_find_rank_bit_fast(unsigned int r, unsigned long m,
-                                     unsigned long a, unsigned long b,
-                                     unsigned long c, unsigned long d);
-void mask_prep_rank_map(unsigned long m,
-                        unsigned long *a, unsigned long *b,
-                        unsigned long *c, unsigned long *d);
-
 /* sets the address family to AF_UNSPEC so that is_addr() does not match */
 static inline void clear_addr(struct sockaddr_storage *addr)
 {
@@ -1451,39 +1021,6 @@
 	return code & 0x0f;
 }
 
-/* Turns 64-bit value <a> from host byte order to network byte order.
- * The principle consists in letting the compiler detect we're playing
- * with a union and simplify most or all operations. The asm-optimized
- * htonl() version involving bswap (x86) / rev (arm) / other is a single
- * operation on little endian, or a NOP on big-endian. In both cases,
- * this lets the compiler "see" that we're rebuilding a 64-bit word from
- * two 32-bit quantities that fit into a 32-bit register. In big endian,
- * the whole code is optimized out. In little endian, with a decent compiler,
- * a few bswap and 2 shifts are left, which is the minimum acceptable.
- */
-static inline unsigned long long my_htonll(unsigned long long a)
-{
-#if defined(__x86_64__)
-	__asm__ volatile("bswap %0" : "=r"(a) : "0"(a));
-	return a;
-#else
-	union {
-		struct {
-			unsigned int w1;
-			unsigned int w2;
-		} by32;
-		unsigned long long by64;
-	} w = { .by64 = a };
-	return ((unsigned long long)htonl(w.by32.w1) << 32) | htonl(w.by32.w2);
-#endif
-}
-
-/* Turns 64-bit value <a> from network byte order to host byte order. */
-static inline unsigned long long my_ntohll(unsigned long long a)
-{
-	return my_htonll(a);
-}
-
 /* returns a 64-bit a timestamp with the finest resolution available. The
  * unit is intentionally not specified. It's mostly used to compare dates.
  */
diff --git a/include/haproxy/intops.h b/include/haproxy/intops.h
new file mode 100644
index 0000000..960c6cb
--- /dev/null
+++ b/include/haproxy/intops.h
@@ -0,0 +1,494 @@
+/*
+ * include/haproxy/intops.h
+ * Functions for integer operations.
+ *
+ * Copyright (C) 2020 Willy Tarreau - w@1wt.eu
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation, version 2.1
+ * exclusively.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
+*/
+
+#ifndef _HAPROXY_INTOPS_H
+#define _HAPROXY_INTOPS_H
+
+#include <haproxy/api.h>
+
+/* exported functions, mostly integer parsing */
+/* rounds <i> down to the closest value having max 2 digits */
+unsigned int round_2dig(unsigned int i);
+unsigned int full_hash(unsigned int a);
+int varint_bytes(uint64_t v);
+unsigned int read_uint(const char **s, const char *end);
+long long read_int64(const char **s, const char *end);
+unsigned long long read_uint64(const char **s, const char *end);
+unsigned int str2ui(const char *s);
+unsigned int str2uic(const char *s);
+unsigned int strl2ui(const char *s, int len);
+unsigned int strl2uic(const char *s, int len);
+int strl2ic(const char *s, int len);
+int strl2irc(const char *s, int len, int *ret);
+int strl2llrc(const char *s, int len, long long *ret);
+int strl2llrc_dotted(const char *text, int len, long long *ret);
+unsigned int mask_find_rank_bit(unsigned int r, unsigned long m);
+unsigned int mask_find_rank_bit_fast(unsigned int r, unsigned long m,
+                                     unsigned long a, unsigned long b,
+                                     unsigned long c, unsigned long d);
+void mask_prep_rank_map(unsigned long m,
+                        unsigned long *a, unsigned long *b,
+                        unsigned long *c, unsigned long *d);
+
+
+/* Multiply the two 32-bit operands and shift the 64-bit result right 32 bits.
+ * This is used to compute fixed ratios by setting one of the operands to
+ * (2^32*ratio).
+ */
+static inline unsigned int mul32hi(unsigned int a, unsigned int b)
+{
+	return ((unsigned long long)a * b) >> 32;
+}
+
+/* gcc does not know when it can safely divide 64 bits by 32 bits. Use this
+ * function when you know for sure that the result fits in 32 bits, because
+ * it is optimal on x86 and on 64bit processors.
+ */
+static inline unsigned int div64_32(unsigned long long o1, unsigned int o2)
+{
+	unsigned long long result;
+#ifdef __i386__
+	asm("divl %2"
+	    : "=A" (result)
+	    : "A"(o1), "rm"(o2));
+#else
+	result = o1 / o2;
+#endif
+	return result;
+}
+
+/* rotate left a 64-bit integer by <bits:[0-5]> bits */
+static inline uint64_t rotl64(uint64_t v, uint8_t bits)
+{
+#if !defined(__ARM_ARCH_8A) && !defined(__x86_64__)
+	bits &= 63;
+#endif
+	v = (v << bits) | (v >> (-bits & 63));
+	return v;
+}
+
+/* rotate right a 64-bit integer by <bits:[0-5]> bits */
+static inline uint64_t rotr64(uint64_t v, uint8_t bits)
+{
+#if !defined(__ARM_ARCH_8A) && !defined(__x86_64__)
+	bits &= 63;
+#endif
+	v = (v >> bits) | (v << (-bits & 63));
+	return v;
+}
+
+/* Simple popcountl implementation. It returns the number of ones in a word.
+ * Described here : https://graphics.stanford.edu/~seander/bithacks.html
+ */
+static inline unsigned int my_popcountl(unsigned long a)
+{
+	a = a - ((a >> 1) & ~0UL/3);
+	a = (a & ~0UL/15*3) + ((a >> 2) & ~0UL/15*3);
+	a = (a + (a >> 4)) & ~0UL/255*15;
+	return (unsigned long)(a * (~0UL/255)) >> (sizeof(unsigned long) - 1) * 8;
+}
+
+/* returns non-zero if <a> has at least 2 bits set */
+static inline unsigned long atleast2(unsigned long a)
+{
+	return a & (a - 1);
+}
+
+/* Simple ffs implementation. It returns the position of the lowest bit set to
+ * one, starting at 1. It is illegal to call it with a==0 (undefined result).
+ */
+static inline unsigned int my_ffsl(unsigned long a)
+{
+	unsigned long cnt;
+
+#if defined(__x86_64__)
+	__asm__("bsf %1,%0\n" : "=r" (cnt) : "rm" (a));
+	cnt++;
+#else
+
+	cnt = 1;
+#if LONG_MAX > 0x7FFFFFFFL /* 64bits */
+	if (!(a & 0xFFFFFFFFUL)) {
+		a >>= 32;
+		cnt += 32;
+	}
+#endif
+	if (!(a & 0XFFFFU)) {
+		a >>= 16;
+		cnt += 16;
+	}
+	if (!(a & 0XFF)) {
+		a >>= 8;
+		cnt += 8;
+	}
+	if (!(a & 0xf)) {
+		a >>= 4;
+		cnt += 4;
+	}
+	if (!(a & 0x3)) {
+		a >>= 2;
+		cnt += 2;
+	}
+	if (!(a & 0x1)) {
+		cnt += 1;
+	}
+#endif /* x86_64 */
+
+	return cnt;
+}
+
+/* Simple fls implementation. It returns the position of the highest bit set to
+ * one, starting at 1. It is illegal to call it with a==0 (undefined result).
+ */
+static inline unsigned int my_flsl(unsigned long a)
+{
+	unsigned long cnt;
+
+#if defined(__x86_64__)
+	__asm__("bsr %1,%0\n" : "=r" (cnt) : "rm" (a));
+	cnt++;
+#else
+
+	cnt = 1;
+#if LONG_MAX > 0x7FFFFFFFUL /* 64bits */
+	if (a & 0xFFFFFFFF00000000UL) {
+		a >>= 32;
+		cnt += 32;
+	}
+#endif
+	if (a & 0XFFFF0000U) {
+		a >>= 16;
+		cnt += 16;
+	}
+	if (a & 0XFF00) {
+		a >>= 8;
+		cnt += 8;
+	}
+	if (a & 0xf0) {
+		a >>= 4;
+		cnt += 4;
+	}
+	if (a & 0xc) {
+		a >>= 2;
+		cnt += 2;
+	}
+	if (a & 0x2) {
+		cnt += 1;
+	}
+#endif /* x86_64 */
+
+	return cnt;
+}
+
+/* Build a word with the <bits> lower bits set (reverse of my_popcountl) */
+static inline unsigned long nbits(int bits)
+{
+	if (--bits < 0)
+		return 0;
+	else
+		return (2UL << bits) - 1;
+}
+
+/* Turns 64-bit value <a> from host byte order to network byte order.
+ * The principle consists in letting the compiler detect we're playing
+ * with a union and simplify most or all operations. The asm-optimized
+ * htonl() version involving bswap (x86) / rev (arm) / other is a single
+ * operation on little endian, or a NOP on big-endian. In both cases,
+ * this lets the compiler "see" that we're rebuilding a 64-bit word from
+ * two 32-bit quantities that fit into a 32-bit register. In big endian,
+ * the whole code is optimized out. In little endian, with a decent compiler,
+ * a few bswap and 2 shifts are left, which is the minimum acceptable.
+ */
+static inline unsigned long long my_htonll(unsigned long long a)
+{
+#if defined(__x86_64__)
+	__asm__ volatile("bswap %0" : "=r"(a) : "0"(a));
+	return a;
+#else
+	union {
+		struct {
+			unsigned int w1;
+			unsigned int w2;
+		} by32;
+		unsigned long long by64;
+	} w = { .by64 = a };
+	return ((unsigned long long)htonl(w.by32.w1) << 32) | htonl(w.by32.w2);
+#endif
+}
+
+/* Turns 64-bit value <a> from network byte order to host byte order. */
+static inline unsigned long long my_ntohll(unsigned long long a)
+{
+	return my_htonll(a);
+}
+
+/* sets bit <bit> into map <map>, which must be long-aligned */
+static inline void ha_bit_set(unsigned long bit, long *map)
+{
+	map[bit / (8 * sizeof(*map))] |= 1UL << (bit & (8 * sizeof(*map) - 1));
+}
+
+/* clears bit <bit> from map <map>, which must be long-aligned */
+static inline void ha_bit_clr(unsigned long bit, long *map)
+{
+	map[bit / (8 * sizeof(*map))] &= ~(1UL << (bit & (8 * sizeof(*map) - 1)));
+}
+
+/* flips bit <bit> from map <map>, which must be long-aligned */
+static inline void ha_bit_flip(unsigned long bit, long *map)
+{
+	map[bit / (8 * sizeof(*map))] ^= 1UL << (bit & (8 * sizeof(*map) - 1));
+}
+
+/* returns non-zero if bit <bit> from map <map> is set, otherwise 0 */
+static inline int ha_bit_test(unsigned long bit, const long *map)
+{
+	return !!(map[bit / (8 * sizeof(*map))] & 1UL << (bit & (8 * sizeof(*map) - 1)));
+}
+
+/* hash a 32-bit integer to another 32-bit integer. This code may be large when
+ * inlined, use full_hash() instead.
+ */
+static inline unsigned int __full_hash(unsigned int a)
+{
+	/* This function is one of Bob Jenkins' full avalanche hashing
+	 * functions, which when provides quite a good distribution for little
+	 * input variations. The result is quite suited to fit over a 32-bit
+	 * space with enough variations so that a randomly picked number falls
+	 * equally before any server position.
+	 * Check http://burtleburtle.net/bob/hash/integer.html for more info.
+	 */
+	a = (a+0x7ed55d16) + (a<<12);
+	a = (a^0xc761c23c) ^ (a>>19);
+	a = (a+0x165667b1) + (a<<5);
+	a = (a+0xd3a2646c) ^ (a<<9);
+	a = (a+0xfd7046c5) + (a<<3);
+	a = (a^0xb55a4f09) ^ (a>>16);
+
+	/* ensure values are better spread all around the tree by multiplying
+	 * by a large prime close to 3/4 of the tree.
+	 */
+	return a * 3221225473U;
+}
+
+/*
+ * Return integer equivalent of character <c> for a hex digit (0-9, a-f, A-F),
+ * otherwise -1. This compact form helps gcc produce efficient code.
+ */
+static inline int hex2i(int c)
+{
+	if ((unsigned char)(c -= '0') > 9) {
+		if ((unsigned char)(c -= 'A' - '0') > 5 &&
+			      (unsigned char)(c -= 'a' - 'A') > 5)
+			c = -11;
+		c += 10;
+	}
+	return c;
+}
+
+/* This one is 6 times faster than strtoul() on athlon, but does
+ * no check at all.
+ */
+static inline unsigned int __str2ui(const char *s)
+{
+	unsigned int i = 0;
+	while (*s) {
+		i = i * 10 - '0';
+		i += (unsigned char)*s++;
+	}
+	return i;
+}
+
+/* This one is 5 times faster than strtoul() on athlon with checks.
+ * It returns the value of the number composed of all valid digits read.
+ */
+static inline unsigned int __str2uic(const char *s)
+{
+	unsigned int i = 0;
+	unsigned int j;
+
+	while (1) {
+		j = (*s++) - '0';
+		if (j > 9)
+			break;
+		i *= 10;
+		i += j;
+	}
+	return i;
+}
+
+/* This one is 28 times faster than strtoul() on athlon, but does
+ * no check at all!
+ */
+static inline unsigned int __strl2ui(const char *s, int len)
+{
+	unsigned int i = 0;
+
+	while (len-- > 0) {
+		i = i * 10 - '0';
+		i += (unsigned char)*s++;
+	}
+	return i;
+}
+
+/* This one is 7 times faster than strtoul() on athlon with checks.
+ * It returns the value of the number composed of all valid digits read.
+ */
+static inline unsigned int __strl2uic(const char *s, int len)
+{
+	unsigned int i = 0;
+	unsigned int j, k;
+
+	while (len-- > 0) {
+		j = (*s++) - '0';
+		k = i * 10;
+		if (j > 9)
+			break;
+		i = k + j;
+	}
+	return i;
+}
+
+/* This function reads an unsigned integer from the string pointed to by <s>
+ * and returns it. The <s> pointer is adjusted to point to the first unread
+ * char. The function automatically stops at <end>.
+ */
+static inline unsigned int __read_uint(const char **s, const char *end)
+{
+	const char *ptr = *s;
+	unsigned int i = 0;
+	unsigned int j, k;
+
+	while (ptr < end) {
+		j = *ptr - '0';
+		k = i * 10;
+		if (j > 9)
+			break;
+		i = k + j;
+		ptr++;
+	}
+	*s = ptr;
+	return i;
+}
+
+/* returns the number of bytes needed to encode <v> as a varint. Be careful, use
+ * it only with constants as it generates a large code (typ. 180 bytes). Use the
+ * varint_bytes() version instead in case of doubt.
+ */
+static inline int __varint_bytes(uint64_t v)
+{
+	switch (v) {
+	case 0x0000000000000000 ... 0x00000000000000ef: return 1;
+	case 0x00000000000000f0 ... 0x00000000000008ef: return 2;
+	case 0x00000000000008f0 ... 0x00000000000408ef: return 3;
+	case 0x00000000000408f0 ... 0x00000000020408ef: return 4;
+	case 0x00000000020408f0 ... 0x00000001020408ef: return 5;
+	case 0x00000001020408f0 ... 0x00000081020408ef: return 6;
+	case 0x00000081020408f0 ... 0x00004081020408ef: return 7;
+	case 0x00004081020408f0 ... 0x00204081020408ef: return 8;
+	case 0x00204081020408f0 ... 0x10204081020408ef: return 9;
+	default: return 10;
+	}
+}
+
+/* Encode the integer <i> into a varint (variable-length integer). The encoded
+ * value is copied in <*buf>. Here is the encoding format:
+ *
+ *        0 <= X < 240        : 1 byte  (7.875 bits)  [ XXXX XXXX ]
+ *      240 <= X < 2288       : 2 bytes (11 bits)     [ 1111 XXXX ] [ 0XXX XXXX ]
+ *     2288 <= X < 264432     : 3 bytes (18 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]   [ 0XXX XXXX ]
+ *   264432 <= X < 33818864   : 4 bytes (25 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]*2 [ 0XXX XXXX ]
+ * 33818864 <= X < 4328786160 : 5 bytes (32 bits)     [ 1111 XXXX ] [ 1XXX XXXX ]*3 [ 0XXX XXXX ]
+ * ...
+ *
+ * On success, it returns the number of written bytes and <*buf> is moved after
+ * the encoded value. Otherwise, it returns -1. */
+static inline int encode_varint(uint64_t i, char **buf, char *end)
+{
+	unsigned char *p = (unsigned char *)*buf;
+	int r;
+
+	if (p >= (unsigned char *)end)
+		return -1;
+
+	if (i < 240) {
+		*p++ = i;
+		*buf = (char *)p;
+		return 1;
+	}
+
+	*p++ = (unsigned char)i | 240;
+	i = (i - 240) >> 4;
+	while (i >= 128) {
+		if (p >= (unsigned char *)end)
+			return -1;
+		*p++ = (unsigned char)i | 128;
+		i = (i - 128) >> 7;
+	}
+
+	if (p >= (unsigned char *)end)
+		return -1;
+	*p++ = (unsigned char)i;
+
+	r    = ((char *)p - *buf);
+	*buf = (char *)p;
+	return r;
+}
+
+/* Decode a varint from <*buf> and save the decoded value in <*i>. See
+ * 'spoe_encode_varint' for details about varint.
+ * On success, it returns the number of read bytes and <*buf> is moved after the
+ * varint. Otherwise, it returns -1. */
+static inline int decode_varint(char **buf, char *end, uint64_t *i)
+{
+	unsigned char *p = (unsigned char *)*buf;
+	int r;
+
+	if (p >= (unsigned char *)end)
+		return -1;
+
+	*i = *p++;
+	if (*i < 240) {
+		*buf = (char *)p;
+		return 1;
+	}
+
+	r = 4;
+	do {
+		if (p >= (unsigned char *)end)
+			return -1;
+		*i += (uint64_t)*p << r;
+		r  += 7;
+	} while (*p++ >= 128);
+
+	r    = ((char *)p - *buf);
+	*buf = (char *)p;
+	return r;
+}
+
+#endif /* _HAPROXY_INTOPS_H */
+
+/*
+ * Local variables:
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ * End:
+ */