MDEV-22851: Engine independent index statistics are incorrect for large tables on Windows

An oveflow was happening on windows because on Windows sizeof(ulong) is 4 bytes
while it is 8 bytes on Linux.
Switched avg_frequency and avg length for column statistics to ulonglong.
Switched avg_frequency for index statistics to ulonglong.

4 files changed, 67 insertions, 15 deletions

diff --git a/mysql-test/r/stat_tables_innodb.result b/mysql-test/r/stat_tables_innodb.result
index b8bed681465..5f393dde05e 100644
--- a/mysql-test/r/stat_tables_innodb.result
+++ b/mysql-test/r/stat_tables_innodb.result
@@ -770,4 +770,33 @@ set @@optimizer_use_condition_selectivity= @save_optimizer_use_condition_selecti
 set @save_optimizer_switch=@@optimizer_switch;
 set use_stat_tables=@save_use_stat_tables;
 set optimizer_switch=@save_optimizer_switch_for_stat_tables_test;
+#
+# MDEV-22851: Engine independent index statistics are incorrect for large tables on Windows.
+#
+CREATE TABLE t1 AS SELECT TABLE_CATALOG FROM INFORMATION_SCHEMA.COLUMNS;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+SET @save_use_stat_tables= @@use_stat_tables;
+SET use_stat_tables= preferably;
+SELECT count(*) FROM t1;
+count(*)
+54336
+CREATE INDEX idx ON t1(TABLE_CATALOG);
+Warnings:
+Warning	1071	Specified key was too long; max key length is 767 bytes
+ANALYZE TABLE t1 PERSISTENT FOR COLUMNS (TABLE_CATALOG) INDEXES (idx);
+Table	Op	Msg_type	Msg_text
+test.t1	analyze	status	Engine-independent statistics collected
+test.t1	analyze	status	OK
+SELECT * FROM mysql.index_stats where table_name='t1';
+db_name	table_name	index_name	prefix_arity	avg_frequency
+test	t1	idx	1	NULL
+SELECT * FROM mysql.column_stats where table_name='t1';
+db_name	table_name	column_name	min_value	max_value	nulls_ratio	avg_length	avg_frequency	hist_size	hist_type	histogram
+test	t1	TABLE_CATALOG	def	def	0.0000	3.0000	54336.0000	0	NULL	NULL
+SET use_stat_tables= @save_use_stat_tables;
+DROP TABLE t1;
 SET SESSION STORAGE_ENGINE=DEFAULT;
diff --git a/mysql-test/t/stat_tables_innodb.test b/mysql-test/t/stat_tables_innodb.test
index 04e81de8f9d..e4a11955aba 100644
--- a/mysql-test/t/stat_tables_innodb.test
+++ b/mysql-test/t/stat_tables_innodb.test
@@ -9,4 +9,25 @@ set optimizer_switch='extended_keys=on';
 
 set optimizer_switch=@save_optimizer_switch_for_stat_tables_test;
 
+--echo #
+--echo # MDEV-22851: Engine independent index statistics are incorrect for large tables on Windows.
+--echo #
+
+CREATE TABLE t1 AS SELECT TABLE_CATALOG FROM INFORMATION_SCHEMA.COLUMNS;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+INSERT INTO t1 SELECT * FROM t1;
+
+SET @save_use_stat_tables= @@use_stat_tables;
+SET use_stat_tables= preferably;
+SELECT count(*) FROM t1;
+CREATE INDEX idx ON t1(TABLE_CATALOG);
+ANALYZE TABLE t1 PERSISTENT FOR COLUMNS (TABLE_CATALOG) INDEXES (idx);
+SELECT * FROM mysql.index_stats where table_name='t1';
+SELECT * FROM mysql.column_stats where table_name='t1';
+SET use_stat_tables= @save_use_stat_tables;
+DROP TABLE t1;
+
 SET SESSION STORAGE_ENGINE=DEFAULT;
diff --git a/sql/sql_statistics.cc b/sql/sql_statistics.cc
index ae02254c745..c2581fd6105 100644
--- a/sql/sql_statistics.cc
+++ b/sql/sql_statistics.cc
@@ -2096,8 +2096,8 @@ int alloc_statistics_for_table(THD* thd, TABLE *table)
                                     sizeof(Index_statistics) * keys);
 
   uint key_parts= table->s->ext_key_parts;
-  ulong *idx_avg_frequency= (ulong*) alloc_root(&table->mem_root,
-                                                sizeof(ulong) * key_parts);
+  ulonglong *idx_avg_frequency= (ulonglong*) alloc_root(&table->mem_root,
+                                               sizeof(ulonglong) * key_parts);
 
   uint columns= 0;
   for (field_ptr= table->field; *field_ptr; field_ptr++)
@@ -2142,7 +2142,7 @@ int alloc_statistics_for_table(THD* thd, TABLE *table)
     }
   }
 
-  memset(idx_avg_frequency, 0, sizeof(ulong) * key_parts);
+  memset(idx_avg_frequency, 0, sizeof(ulonglong) * key_parts);
 
   KEY *key_info, *end;
   for (key_info= table->key_info, end= key_info + table->s->keys;
@@ -2258,14 +2258,14 @@ static int alloc_statistics_for_table_share(THD* thd, TABLE_SHARE *table_share)
   }
 
   uint key_parts= table_share->ext_key_parts;
-  ulong *idx_avg_frequency=  table_stats->idx_avg_frequency;
+  ulonglong *idx_avg_frequency=  table_stats->idx_avg_frequency;
   if (!idx_avg_frequency)
   {
-    idx_avg_frequency= (ulong*) alloc_root(&stats_cb->mem_root,
-                                           sizeof(ulong) * key_parts);
+    idx_avg_frequency= (ulonglong*) alloc_root(&stats_cb->mem_root,
+                                               sizeof(ulonglong) * key_parts);
     if (idx_avg_frequency)
     {
-      memset(idx_avg_frequency, 0, sizeof(ulong) * key_parts);
+      memset(idx_avg_frequency, 0, sizeof(ulonglong) * key_parts);
       table_stats->idx_avg_frequency= idx_avg_frequency;
       for (key_info= table_share->key_info, end= key_info + keys;
            key_info < end; 
diff --git a/sql/sql_statistics.h b/sql/sql_statistics.h
index 151618ca365..903d164ca99 100644
--- a/sql/sql_statistics.h
+++ b/sql/sql_statistics.h
@@ -279,7 +279,9 @@ public:
   uchar *min_max_record_buffers;    /* Record buffers for min/max values  */
   Column_statistics *column_stats;  /* Array of statistical data for columns */
   Index_statistics *index_stats;    /* Array of statistical data for indexes */
-  ulong *idx_avg_frequency;   /* Array of records per key for index prefixes */
+
+  /* Array of records per key for index prefixes */
+  ulonglong *idx_avg_frequency;
   uchar *histograms;                /* Sequence of histograms       */                    
 };
 
@@ -331,7 +333,7 @@ private:
     CHAR values are stripped of trailing spaces.
     Flexible values are stripped of their length prefixes.
   */
-  ulong avg_length;
+  ulonglong avg_length;
 
   /*
     The ratio N/D multiplied by the scale factor Scale_factor_avg_frequency,
@@ -339,7 +341,7 @@ private:
        N is the number of rows with not null value in the column,
        D the number of distinct values among them
   */
-  ulong avg_frequency;
+  ulonglong avg_frequency;
 
 public:
 
@@ -389,12 +391,12 @@ public:
 
   void set_avg_length (double val)
   {
-    avg_length= (ulong) (val * Scale_factor_avg_length);
+    avg_length= (ulonglong) (val * Scale_factor_avg_length);
   }
 
   void set_avg_frequency (double val)
   {
-    avg_frequency= (ulong) (val * Scale_factor_avg_frequency);
+    avg_frequency= (ulonglong) (val * Scale_factor_avg_frequency);
   }
 
   bool min_max_values_are_provided()
@@ -433,11 +435,11 @@ private:
     in the first k components, and D is the number of distinct
     k-component prefixes among them 
   */
-  ulong *avg_frequency;
+  ulonglong *avg_frequency;
 
 public:
 
-  void init_avg_frequency(ulong *ptr) { avg_frequency= ptr; }
+  void init_avg_frequency(ulonglong *ptr) { avg_frequency= ptr; }
 
   bool avg_frequency_is_inited() { return avg_frequency != NULL; }
 
@@ -448,7 +450,7 @@ public:
 
   void set_avg_frequency(uint i, double val)
   {
-    avg_frequency[i]= (ulong) (val * Scale_factor_avg_frequency);
+    avg_frequency[i]= (ulonglong) (val * Scale_factor_avg_frequency);
   }
 
 };