MDEV-38180: Added XXH3 and XXH32 (64 bit) sql functions

include/m_ctype.h

@@ -495,6 +495,38 @@ typedef struct my_charset_loader_st
 
 extern int (*my_string_stack_guard)(int);
 
+typedef struct my_hasher_st
+{
+  ulong m_nr1;                  /* mysql5x hash value */
+  ulong m_nr2;                  /* mysql5x aux value */
+  uint32 m_nr;                  /* The hash value */
+  /* Whether a streaming algorithm has been in use */
+  my_bool m_streaming;
+  /* One-shot string hash function */
+  void (*m_hash_str)(struct my_hasher_st *hasher, const uchar* str,
+                     size_t len);
+  /*
+     Byte-streaming hash function, fallback to m_hash_str on one byte
+     if NULL
+  */
+  void (*m_hash_byte)(struct my_hasher_st *hasher, uchar value);
+  /* Function to clean up and return the hash value */
+  uint32 (*m_finalize)(struct my_hasher_st *hasher);
+  /* Custom pointer e.g. to a state */
+  void *m_specific;
+} my_hasher_st;
+
+/* Defined in hasher-xxx.c files */
+extern my_hasher_st my_hasher_mysql5x();
+extern my_hasher_st my_hasher_mysql5x_for_unique();
+extern my_hasher_st my_hasher_base31();
+extern my_hasher_st my_hasher_crc32c();
+extern my_hasher_st my_hasher_xxh32();
+extern my_hasher_st my_hasher_xxh3();
+
+#define MY_HASH_ADD_MARIADB(A, B, value) \
+  do { A^= (((A & 63)+B)*((value)))+ (A << 8); B+=3; } while(0)
+
 /* See strings/CHARSET_INFO.txt for information about this structure  */
 struct my_collation_handler_st
 {
@@ -580,8 +612,8 @@ struct my_collation_handler_st
                 my_match_t *match, uint nmatch);
 
   /* Hash calculation */
-  void (*hash_sort)(CHARSET_INFO *cs, const uchar *key, size_t len,
-		    ulong *nr1, ulong *nr2); 
+  void (*hash_sort)(my_hasher_st *hasher, CHARSET_INFO *cs, const uchar *key,
+                    size_t len);
   my_bool (*propagate)(CHARSET_INFO *cs, const uchar *str, size_t len);
   /*
     Make minimum and maximum strings for the collation.
@@ -1182,9 +1214,9 @@ struct charset_info_st
     return (coll->instr)(this, b, b_length, s, s_length, match, nmatch);
   }
 
-  void hash_sort(const uchar *key, size_t len, ulong *nr1, ulong *nr2) const
+  void hash_sort(my_hasher_st *hasher, const uchar *key, size_t len) const
   {
-    (coll->hash_sort)(this, key, len, nr1, nr2);
+    (coll->hash_sort)(hasher, this, key, len);
   }
 
   my_bool propagate(const uchar *str, size_t len) const
@@ -1462,11 +1494,10 @@ my_ci_instr(CHARSET_INFO *ci,
 
 
 static inline void
-my_ci_hash_sort(CHARSET_INFO *ci,
-                const uchar *key, size_t len,
-                ulong *nr1, ulong *nr2)
+my_ci_hash_sort(my_hasher_st *hasher, CHARSET_INFO *ci,
+                const uchar *key, size_t len)
 {
-  (ci->coll->hash_sort)(ci, key, len, nr1, nr2);
+  (ci->coll->hash_sort)(hasher, ci, key, len);
 }
 
 
@@ -1585,17 +1616,16 @@ extern int  my_strnncoll_simple(CHARSET_INFO *, const uchar *, size_t,
 extern int  my_strnncollsp_simple(CHARSET_INFO *, const uchar *, size_t,
                                   const uchar *, size_t);
 
-extern void my_hash_sort_simple(CHARSET_INFO *cs,
-				const uchar *key, size_t len,
-				ulong *nr1, ulong *nr2); 
+extern void my_hash_sort_simple(my_hasher_st *hasher,
+                                CHARSET_INFO *cs,
+                                const uchar *key, size_t len);
 
-extern void my_hash_sort_simple_nopad(CHARSET_INFO *cs,
-				      const uchar *key, size_t len,
-				      ulong *nr1, ulong *nr2);
+extern void my_hash_sort_simple_nopad(my_hasher_st *hasher,
+                                      CHARSET_INFO *cs,
+                                      const uchar *key, size_t len);
 
-extern void my_hash_sort_bin(CHARSET_INFO *cs,
-                             const uchar *key, size_t len, ulong *nr1,
-                             ulong *nr2);
+extern void my_hash_sort_bin(my_hasher_st *hasher, CHARSET_INFO *cs,
+                             const uchar *key, size_t len);
 
 /**
   Compare a string to an array of spaces, for PAD SPACE comparison.
@@ -1761,12 +1791,13 @@ int my_wildcmp_mb_bin(CHARSET_INFO *cs,
                       const char *wildstr,const char *wildend,
                       int escape, int w_one, int w_many);
 
-void my_hash_sort_mb_bin(CHARSET_INFO *cs __attribute__((unused)),
-                         const uchar *key, size_t len,ulong *nr1, ulong *nr2);
+void my_hash_sort_mb_bin(my_hasher_st *hasher,
+                         CHARSET_INFO *cs __attribute__((unused)),
+                         const uchar *key, size_t len);
 
-void my_hash_sort_mb_nopad_bin(CHARSET_INFO *cs __attribute__((unused)),
-                               const uchar *key, size_t len,
-                               ulong *nr1, ulong *nr2);
+void my_hash_sort_mb_nopad_bin(my_hasher_st *hasher,
+                               CHARSET_INFO *cs __attribute__((unused)),
+                               const uchar *key, size_t len);
 
 
 extern my_bool my_parse_charset_xml(MY_CHARSET_LOADER *loader,

mysql-test/main/func_xxh.result

@@ -0,0 +1,31 @@
+# Testing XXH SQL functions
+SELECT XXH32('abc') = 852579327 AS xxh32_expected;
+xxh32_expected
+1
+SELECT XXH32(11223344) = 408040606 AS xxh32_expected;
+xxh32_expected
+1
+SELECT XXH32('abc') = XXH32('abc') AS xxh32_eq;
+xxh32_eq
+1
+SELECT XXH32(NULL) IS NULL AS xxh32_null;
+xxh32_null
+1
+SELECT XXH32('') IS NULL AS xxh32_null;
+xxh32_null
+1
+SELECT XXH3('abc') = 8696274497037089104 AS xxh3_expected;
+xxh3_expected
+1
+SELECT XXH3(11223344) = 9329478082249925753 AS xxh3_expected;
+xxh3_expected
+1
+SELECT XXH3('abc') = XXH32('abc') AS xxh3_eq;
+xxh3_eq
+0
+SELECT XXH3(NULL) IS NULL AS xxh3_null;
+xxh3_null
+1
+SELECT XXH3('') IS NULL AS xxh3_null;
+xxh3_null
+1

mysql-test/main/func_xxh.test

@@ -0,0 +1,13 @@
+--echo # Testing XXH SQL functions
+
+SELECT XXH32('abc') = 852579327 AS xxh32_expected;
+SELECT XXH32(11223344) = 408040606 AS xxh32_expected;
+SELECT XXH32('abc') = XXH32('abc') AS xxh32_eq;
+SELECT XXH32(NULL) IS NULL AS xxh32_null;
+SELECT XXH32('') IS NULL AS xxh32_null;
+
+SELECT XXH3('abc') = 8696274497037089104 AS xxh3_expected;
+SELECT XXH3(11223344) = 9329478082249925753 AS xxh3_expected;
+SELECT XXH3('abc') = XXH32('abc') AS xxh3_eq;
+SELECT XXH3(NULL) IS NULL AS xxh3_null;
+SELECT XXH3('') IS NULL AS xxh3_null;

mysql-test/main/partition_alter.result

@@ -391,3 +391,49 @@ partitio values in (2, 3, 4),
 pn values in (52, 53, 54));
 drop table t1;
 # End of 10.7 tests
+# MDEV-9826 change the hash algorithm for PARTITION BY KEY
+create table t1(c1 int, c2 date) partition by key (c2) partitions 8;
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `c1` int(11) DEFAULT NULL,
+  `c2` date DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+ PARTITION BY KEY (`c2`)
+PARTITIONS 8
+insert into t1 values (1, NULL);
+insert into t1 select 1, from_days(seq) from seq_735000_to_736000;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+partition_name	table_rows
+p0	297
+p1	0
+p2	1
+p3	0
+p4	428
+p5	0
+p6	276
+p7	0
+alter table t1 partition by key algorithm=crc32c (c2) partitions 8;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+partition_name	table_rows
+p0	126
+p1	127
+p2	126
+p3	126
+p4	126
+p5	125
+p6	122
+p7	124
+alter table t1 partition by key algorithm=xxh3 (c2) partitions 8;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+partition_name	table_rows
+p0	122
+p1	132
+p2	151
+p3	121
+p4	113
+p5	125
+p6	128
+p7	110
+drop table t1;
+# End of 12.3 tests

mysql-test/main/partition_alter.test

@@ -323,3 +323,18 @@ partition by list(x) (
 drop table t1;
 
 --echo # End of 10.7 tests
+
+--echo # MDEV-9826 change the hash algorithm for PARTITION BY KEY
+--source include/have_sequence.inc
+create table t1(c1 int, c2 date) partition by key (c2) partitions 8;
+show create table t1;
+insert into t1 values (1, NULL);
+insert into t1 select 1, from_days(seq) from seq_735000_to_736000;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+alter table t1 partition by key algorithm=crc32c (c2) partitions 8;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+alter table t1 partition by key algorithm=xxh3 (c2) partitions 8;
+select partition_name, table_rows from information_schema.partitions where table_name='t1';
+drop table t1;
+
+--echo # End of 12.3 tests

mysql-test/main/partition_key_algorithm,1.rdiff

@@ -0,0 +1,123 @@
+--- ../src/mysql-test/main/partition_key_algorithm.result	2025-12-08 13:49:44.869944988 +1100
++++ ../src/mysql-test/main/partition_key_algorithm,1.reject	2025-12-08 13:51:13.221162943 +1100
+@@ -1,13 +1,13 @@
+ # MDEV-9826 change the hash algorithm for PARTITION BY KEY
+ ## MDEV-9826 case with NULL
+-create table t1(c1 int, c2 date) partition by key  (c2) partitions 8;
++create table t1(c1 int, c2 date) partition by key ALGORITHM=1 (c2) partitions 8;
+ show create table t1;
+ Table	Create Table
+ t1	CREATE TABLE `t1` (
+   `c1` int(11) DEFAULT NULL,
+   `c2` date DEFAULT NULL
+ ) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+- PARTITION BY KEY (`c2`)
++ PARTITION BY KEY ALGORITHM = MYSQL51 (`c2`)
+ PARTITIONS 8
+ insert into t1 values (1, NULL);
+ insert into t1 values (1, '2014-04-22');
+@@ -43,14 +43,14 @@
+ 1	2014-04-29
+ drop table t1;
+ ## MDEV-9826 case with more dates
+-create table t1(c1 int, c2 date) partition by key  (c2) partitions 8;
++create table t1(c1 int, c2 date) partition by key ALGORITHM=1 (c2) partitions 8;
+ show create table t1;
+ Table	Create Table
+ t1	CREATE TABLE `t1` (
+   `c1` int(11) DEFAULT NULL,
+   `c2` date DEFAULT NULL
+ ) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+- PARTITION BY KEY (`c2`)
++ PARTITION BY KEY ALGORITHM = MYSQL51 (`c2`)
+ PARTITIONS 8
+ insert into t1 values (1, NULL);
+ insert into t1 select 1, from_days(seq) from seq_735000_to_736000;
+@@ -66,14 +66,14 @@
+ p7	0
+ drop table t1;
+ ## MDEV-9826 case with more dates and linear key
+-create table t1(c1 int, c2 date) partition by linear key  (c2) partitions 8;
++create table t1(c1 int, c2 date) partition by linear key ALGORITHM=1 (c2) partitions 8;
+ show create table t1;
+ Table	Create Table
+ t1	CREATE TABLE `t1` (
+   `c1` int(11) DEFAULT NULL,
+   `c2` date DEFAULT NULL
+ ) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+- PARTITION BY LINEAR KEY (`c2`)
++ PARTITION BY LINEAR KEY ALGORITHM = MYSQL51 (`c2`)
+ PARTITIONS 8
+ insert into t1 values (1, NULL);
+ insert into t1 select 1, from_days(seq) from seq_735000_to_736000;
+@@ -89,14 +89,14 @@
+ p7	0
+ drop table t1;
+ ## MDEV-9826 case with more dates and multiple keys
+-create table t1(c1 int, c2 date) partition by key  (c1, c2) partitions 8;
++create table t1(c1 int, c2 date) partition by key ALGORITHM=1 (c1, c2) partitions 8;
+ show create table t1;
+ Table	Create Table
+ t1	CREATE TABLE `t1` (
+   `c1` int(11) DEFAULT NULL,
+   `c2` date DEFAULT NULL
+ ) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+- PARTITION BY KEY (`c1`,`c2`)
++ PARTITION BY KEY ALGORITHM = MYSQL51 (`c1`,`c2`)
+ PARTITIONS 8
+ insert into t1 values (1, NULL);
+ insert into t1 select 1, from_days(seq) from seq_735000_to_736000;
+@@ -112,7 +112,7 @@
+ p7	0
+ drop table t1;
+ ## MDEV-20791 case with INET6
+-create table t1 (a inet6) partition by key  (a) partitions 2;
++create table t1 (a inet6) partition by key ALGORITHM=1 (a) partitions 2;
+ insert into t1 values ('0:db8::ff00:42:8329'),('2001:db8::ff00:42:8329'),('::'),('::192.0.2.128'),('::ffff:192.0.2.128');
+ select partition_name, table_rows from information_schema.partitions where table_name='t1';
+ partition_name	table_rows
+@@ -120,7 +120,7 @@
+ p1	5
+ drop table t1;
+ ## MDEV-20791 case with BINARY
+-CREATE TABLE t1 (a BINARY(16)) PARTITION BY KEY  (a) PARTITIONS 2;
++CREATE TABLE t1 (a BINARY(16)) PARTITION BY KEY ALGORITHM=1 (a) PARTITIONS 2;
+ INSERT INTO t1 VALUES (X'00000DB8000000000000FF0000428329');
+ INSERT INTO t1 VALUES (X'20010DB8000000000000FF0000428329');
+ INSERT INTO t1 VALUES (X'00000000000000000000000000000000');
+@@ -132,7 +132,7 @@
+ p1	5
+ drop table t1;
+ ## MDEV-20791 case with more rows
+-create table t1 (a inet6) partition by key  (a) partitions 2;
++create table t1 (a inet6) partition by key ALGORITHM=1 (a) partitions 2;
+ insert into t1 select concat(hex(seq), ':db8::ff00:42:8329') from seq_0_to_65535;
+ select partition_name, table_rows from information_schema.partitions where table_name='t1';
+ partition_name	table_rows
+@@ -155,7 +155,7 @@
+ 60
+ 6060
+ 606060
+-ALTER TABLE t1  PARTITION BY KEY  (a) PARTITIONS 3;
++ALTER TABLE t1  PARTITION BY KEY ALGORITHM=1 (a) PARTITIONS 3;
+ SELECT HEX(a) FROM t1 WHERE a=0x60;
+ HEX(a)
+ 20
+@@ -172,7 +172,7 @@
+ 60
+ 6060
+ 606060
+-ALTER TABLE t1  PARTITION BY KEY  (a) PARTITIONS 3;
++ALTER TABLE t1  PARTITION BY KEY ALGORITHM=1 (a) PARTITIONS 3;
+ SELECT HEX(a) FROM t1 WHERE a=0x60;
+ HEX(a)
+ 20
+@@ -189,7 +189,7 @@
+ A0
+ A0A0
+ A0A0A0
+-ALTER TABLE t1  PARTITION BY KEY  (a) PARTITIONS 3;
++ALTER TABLE t1  PARTITION BY KEY ALGORITHM=1 (a) PARTITIONS 3;
+ SELECT HEX(a) FROM t1 WHERE a=0xA0;
+ HEX(a)
+ 20