Merge adventure.(none):/home/thek/Development/cpp/bug28249/my50-bug28249

into  adventure.(none):/home/thek/Development/cpp/mysql-5.0-runtime

mysql-test/r/query_cache.result

@@ -1467,3 +1467,42 @@ insert into t1 values ('c');
 a
 drop table t1;
 set GLOBAL query_cache_size= default;
+Bug#28249 Query Cache returns wrong result with concurrent insert/ certain lock
+set GLOBAL query_cache_type=1;
+set GLOBAL query_cache_limit=10000;
+set GLOBAL query_cache_min_res_unit=0;
+set GLOBAL query_cache_size= 100000;
+flush tables;
+drop table if exists t1, t2;
+create table t1 (a int);
+create table t2 (a int);
+insert into t1 values (1),(2),(3);
+Locking table T2 with a write lock.
+lock table t2 write;
+Select blocked by write lock.
+select *, (select count(*) from t2) from t1;;
+Sleeing is ok, because selecting should be done very fast.
+Inserting into table T1.
+insert into t1 values (4);
+Unlocking the tables.
+unlock tables;
+Collecting result from previously blocked select.
+Next select should contain 4 rows, as the insert is long finished.
+select *, (select count(*) from t2) from t1;
+a	(select count(*) from t2)
+1	0
+2	0
+3	0
+4	0
+reset query cache;
+select *, (select count(*) from t2) from t1;
+a	(select count(*) from t2)
+1	0
+2	0
+3	0
+4	0
+drop table t1,t2;
+set GLOBAL query_cache_type=default;
+set GLOBAL query_cache_limit=default;
+set GLOBAL query_cache_min_res_unit=default;
+set GLOBAL query_cache_size=default;

mysql-test/t/query_cache.test

@@ -1028,4 +1028,77 @@ drop table t1;
 
 set GLOBAL query_cache_size= default;
 
+#
+# Bug #28249 Query Cache returns wrong result with concurrent insert / certain lock
+#
+--echo Bug#28249 Query Cache returns wrong result with concurrent insert/ certain lock
+connect (user1,localhost,root,,test,,);
+connect (user2,localhost,root,,test,,);
+connect (user3,localhost,root,,test,,);
+
+connection user1;
+
+set GLOBAL query_cache_type=1;
+set GLOBAL query_cache_limit=10000;
+set GLOBAL query_cache_min_res_unit=0;
+set GLOBAL query_cache_size= 100000;
+
+flush tables;
+--disable_warnings
+drop table if exists t1, t2;
+--enable_warnings
+create table t1 (a int);
+create table t2 (a int);
+insert into t1 values (1),(2),(3);
+connection user2;
+--echo Locking table T2 with a write lock.
+lock table t2 write;
+
+connection user1;
+--echo Select blocked by write lock.
+--send select *, (select count(*) from t2) from t1;
+--echo Sleeing is ok, because selecting should be done very fast.
+sleep 5;
+
+connection user3;
+--echo Inserting into table T1.
+insert into t1 values (4);
+
+connection user2;
+--echo Unlocking the tables.
+unlock tables;
+
+connection user1;
+--echo Collecting result from previously blocked select.
+# 
+# Since the lock ordering rule in thr_multi_lock depends on
+# pointer values, from execution to execution we might have
+# different lock order, and therefore, sometimes lock t1 and block
+# on t2, and sometimes block on t2 right away. In the second case,
+# the following insert succeeds, and only then this select can
+# proceed, and we actually test nothing, as the very first select
+# returns 4 rows right away.
+# It's fine to have a test case that covers the problematic area
+# at least once in a while.
+# We, however, need to disable the result log here to make the
+# test repeatable.
+--disable_result_log
+--reap
+--enable_result_log
+--echo Next select should contain 4 rows, as the insert is long finished.
+select *, (select count(*) from t2) from t1;
+reset query cache;
+select *, (select count(*) from t2) from t1;
+
+drop table t1,t2;
+
+connection default;
+disconnect user1;
+disconnect user2;
+disconnect user3;
+set GLOBAL query_cache_type=default;
+set GLOBAL query_cache_limit=default;
+set GLOBAL query_cache_min_res_unit=default;
+set GLOBAL query_cache_size=default;
 # End of 5.0 tests
+

sql/ha_myisam.cc

@@ -1921,3 +1921,77 @@ uint ha_myisam::checksum() const
   return (uint)file->state->checksum;
 }
 
+#ifdef HAVE_QUERY_CACHE
+/**
+  @brief Register a named table with a call back function to the query cache.
+
+  @param thd The thread handle
+  @param table_key A pointer to the table name in the table cache
+  @param key_length The length of the table name
+  @param[out] engine_callback The pointer to the storage engine call back
+    function, currently 0
+  @param[out] engine_data Engine data will be set to 0.
+
+  @note Despite the name of this function, it is used to check each statement
+    before it is cached and not to register a table or callback function.
+
+  @see handler::register_query_cache_table
+
+  @return The error code. The engine_data and engine_callback will be set to 0.
+    @retval TRUE Success
+    @retval FALSE An error occured
+*/
+
+my_bool ha_myisam::register_query_cache_table(THD *thd, char *table_name,
+                                              uint table_name_len,
+                                              qc_engine_callback
+                                              *engine_callback,
+                                              ulonglong *engine_data)
+{
+  /*
+    No call back function is needed to determine if a cached statement
+    is valid or not.
+  */
+  *engine_callback= 0;
+
+  /*
+    No engine data is needed.
+  */
+  *engine_data= 0;
+
+  /*
+    If a concurrent INSERT has happened just before the currently processed
+    SELECT statement, the total size of the table is unknown.
+
+    To determine if the table size is known, the current thread's snap shot of
+    the table size with the actual table size are compared.
+
+    If the table size is unknown the SELECT statement can't be cached.
+  */
+  ulonglong actual_data_file_length;
+  ulonglong current_data_file_length;
+
+  /*
+    POSIX visibility rules specify that "2. Whatever memory values a
+    thread can see when it unlocks a mutex <...> can also be seen by any
+    thread that later locks the same mutex". In this particular case,
+    concurrent insert thread had modified the data_file_length in
+    MYISAM_SHARE before it has unlocked (or even locked)
+    structure_guard_mutex. So, here we're guaranteed to see at least that
+    value after we've locked the same mutex. We can see a later value
+    (modified by some other thread) though, but it's ok, as we only want
+    to know if the variable was changed, the actual new value doesn't matter
+  */
+  actual_data_file_length= file->s->state.state.data_file_length;
+  current_data_file_length= file->save_state.data_file_length;
+
+  if (current_data_file_length != actual_data_file_length)
+  {
+    /* Don't cache current statement. */
+    return FALSE;
+  }
+
+  /* It is ok to try to cache current statement. */
+  return TRUE;
+}
+#endif

sql/ha_myisam.h

@@ -127,4 +127,11 @@ class ha_myisam: public handler
   int dump(THD* thd, int fd);
   int net_read_dump(NET* net);
 #endif
+#ifdef HAVE_QUERY_CACHE
+  my_bool register_query_cache_table(THD *thd, char *table_key,
+                                     uint key_length,
+                                     qc_engine_callback
+                                     *engine_callback,
+                                     ulonglong *engine_data);
+#endif
 };

sql/handler.h

@@ -843,16 +843,49 @@ class handler :public Sql_alloc
 
   /* Type of table for caching query */
   virtual uint8 table_cache_type() { return HA_CACHE_TBL_NONTRANSACT; }
-  /* ask handler about permission to cache table when query is to be cached */
+
+
+  /**
+    @brief Register a named table with a call back function to the query cache.
+
+    @param thd The thread handle
+    @param table_key A pointer to the table name in the table cache
+    @param key_length The length of the table name
+    @param[out] engine_callback The pointer to the storage engine call back
+      function
+    @param[out] engine_data Storage engine specific data which could be
+      anything
+
+    This method offers the storage engine, the possibility to store a reference
+    to a table name which is going to be used with query cache. 
+    The method is called each time a statement is written to the cache and can
+    be used to verify if a specific statement is cachable. It also offers
+    the possibility to register a generic (but static) call back function which
+    is called each time a statement is matched against the query cache.
+
+    @note If engine_data supplied with this function is different from
+      engine_data supplied with the callback function, and the callback returns
+      FALSE, a table invalidation on the current table will occur.
+
+    @return Upon success the engine_callback will point to the storage engine
+      call back function, if any, and engine_data will point to any storage
+      engine data used in the specific implementation.
+      @retval TRUE Success
+      @retval FALSE The specified table or current statement should not be
+        cached
+  */
+
   virtual my_bool register_query_cache_table(THD *thd, char *table_key,
-					     uint key_length,
-					     qc_engine_callback
-					     *engine_callback,
-					     ulonglong *engine_data)
+                                             uint key_length,
+                                             qc_engine_callback
+                                             *engine_callback,
+                                             ulonglong *engine_data)
   {
     *engine_callback= 0;
-    return 1;
+    return TRUE;
   }
+
+
  /*
   RETURN
     true  Primary key (if there is one) is clustered key covering all fields