MDEV-6274 Collation usage statistics

Adding collation usage statistics into the feedback plugin I_S table.

include/my_sys.h

@@ -242,6 +242,11 @@ extern MYSQL_PLUGIN_IMPORT CHARSET_INFO *default_charset_info;
 extern MYSQL_PLUGIN_IMPORT CHARSET_INFO *all_charsets[MY_ALL_CHARSETS_SIZE];
 extern struct charset_info_st compiled_charsets[];
 
+/* Collation properties and use statistics */
+extern my_bool my_collation_is_known_id(uint id);
+extern ulonglong my_collation_statistics_get_use_count(uint id);
+extern const char *my_collation_get_tailoring(uint id);
+
 /* statistics */
 extern ulong	my_file_opened,my_stream_opened, my_tmp_file_created;
 extern ulong    my_file_total_opened;

mysql-test/suite/plugins/r/feedback_plugin_load.result

@@ -10,3 +10,12 @@ FEEDBACK_SEND_RETRY_WAIT	60
 FEEDBACK_SEND_TIMEOUT	60
 FEEDBACK_URL	http://mariadb.org/feedback_plugin/post
 FEEDBACK_USER_INFO	mysql-test
+SELECT VARIABLE_VALUE>0, VARIABLE_NAME FROM INFORMATION_SCHEMA.FEEDBACK
+WHERE VARIABLE_NAME LIKE 'Collation used %'
+ORDER BY VARIABLE_NAME;
+VARIABLE_VALUE>0	VARIABLE_NAME
+1	Collation used binary
+1	Collation used latin1_bin
+1	Collation used latin1_swedish_ci
+1	Collation used utf8_bin
+1	Collation used utf8_general_ci

mysql-test/suite/plugins/t/feedback_plugin_load.test

@@ -8,3 +8,16 @@ select plugin_status from information_schema.plugins where plugin_name='feedback
 --sorted_result
 select * from information_schema.feedback where variable_name like 'feed%'
        and variable_name not like  '%_uid';
+
+
+# Embedded server does not use the table mysqld.user and thus
+# does not automatically use latin1_bin on startup. Use it manually.
+--disable_query_log
+if (`SELECT VERSION() LIKE '%embedded%'`)
+{
+  DO _latin1'test' COLLATE latin1_bin;
+}
+--enable_query_log
+SELECT VARIABLE_VALUE>0, VARIABLE_NAME FROM INFORMATION_SCHEMA.FEEDBACK
+WHERE VARIABLE_NAME LIKE 'Collation used %'
+ORDER BY VARIABLE_NAME;

mysys/charset.c

@@ -483,6 +483,50 @@ void add_compiled_collation(struct charset_info_st *cs)
 static my_pthread_once_t charsets_initialized= MY_PTHREAD_ONCE_INIT;
 static my_pthread_once_t charsets_template= MY_PTHREAD_ONCE_INIT;
 
+typedef struct
+{
+  ulonglong use_count;
+} MY_COLLATION_STATISTICS;
+
+
+static MY_COLLATION_STATISTICS my_collation_statistics[MY_ALL_CHARSETS_SIZE];
+
+
+my_bool my_collation_is_known_id(uint id)
+{
+  return id > 0 && id < array_elements(all_charsets) && all_charsets[id] ?
+         TRUE : FALSE;
+}
+
+
+/*
+  Collation use statistics functions do not lock
+  counters to avoid mutex contention. This can lose
+  some counter increments with high thread concurrency.
+  But this should be Ok, as we don't need exact numbers.
+*/
+static inline void my_collation_statistics_inc_use_count(uint id)
+{
+  DBUG_ASSERT(my_collation_is_known_id(id));
+  my_collation_statistics[id].use_count++;
+}
+
+
+ulonglong my_collation_statistics_get_use_count(uint id)
+{
+  DBUG_ASSERT(my_collation_is_known_id(id));
+  return my_collation_statistics[id].use_count;
+}
+
+
+const char *my_collation_get_tailoring(uint id)
+{
+  /* all_charsets[id]->tailoring is never changed after server startup. */
+  DBUG_ASSERT(my_collation_is_known_id(id));
+  return all_charsets[id]->tailoring;
+}
+
+
 static void init_available_charsets(void)
 {
   char fname[FN_REFLEN + sizeof(MY_CHARSET_INDEX)];
@@ -490,6 +534,7 @@ static void init_available_charsets(void)
   MY_CHARSET_LOADER loader;
 
   bzero((char*) &all_charsets,sizeof(all_charsets));
+  bzero((char*) &my_collation_statistics, sizeof(my_collation_statistics));
   init_compiled_charsets(MYF(0));
 
   /* Copy compiled charsets */
@@ -608,7 +653,10 @@ get_internal_charset(MY_CHARSET_LOADER *loader, uint cs_number, myf flags)
   if ((cs= (struct charset_info_st*) all_charsets[cs_number]))
   {
     if (cs->state & MY_CS_READY)  /* if CS is already initialized */
-        return cs;
+    {
+      my_collation_statistics_inc_use_count(cs_number);
+      return cs;
+    }
 
     /*
       To make things thread safe we are not allowing other threads to interfere
@@ -636,6 +684,7 @@ get_internal_charset(MY_CHARSET_LOADER *loader, uint cs_number, myf flags)
         else
           cs->state|= MY_CS_READY;
       }
+      my_collation_statistics_inc_use_count(cs_number);
     }
     else
       cs= NULL;

plugin/feedback/feedback.cc

@@ -217,7 +217,8 @@ int fill_feedback(THD *thd, TABLE_LIST *tables, COND *unused)
   tables->schema_table= i_s_feedback;
   res= res || fill_plugin_version(thd, tables)
            || fill_misc_data(thd, tables)
-           || fill_linux_info(thd, tables);
+           || fill_linux_info(thd, tables)
+           || fill_collation_statistics(thd, tables);
 
   return res;
 }

plugin/feedback/feedback.h

@@ -22,6 +22,7 @@ int fill_feedback(THD *thd, TABLE_LIST *tables, COND *cond);
 int fill_plugin_version(THD *thd, TABLE_LIST *tables);
 int fill_misc_data(THD *thd, TABLE_LIST *tables);
 int fill_linux_info(THD *thd, TABLE_LIST *tables);
+int fill_collation_statistics(THD *thd, TABLE_LIST *tables);
 
 static const int SERVER_UID_SIZE= 29;
 extern char server_uid_buf[SERVER_UID_SIZE+1], *user_info;

plugin/feedback/utils.cc

@@ -383,6 +383,25 @@ int fill_misc_data(THD *thd, TABLE_LIST *tables)
   return 0;
 }
 
+int fill_collation_statistics(THD *thd, TABLE_LIST *tables)
+{
+  TABLE *table= tables->table;
+  for (uint id= 1; id < MY_ALL_CHARSETS_SIZE; id++)
+  {
+    ulonglong count;
+    if (my_collation_is_known_id(id) &&
+        (count= my_collation_statistics_get_use_count(id)))
+    {
+      char name[MY_CS_NAME_SIZE + 32];
+      size_t namelen= my_snprintf(name, sizeof(name),
+                                 "Collation used %s",
+                                 get_charset_name(id));
+      INSERT2(name, namelen, (count, UNSIGNED));
+    }
+  }
+  return 0;
+};
+
 /**
   calculates the server unique identifier