MySQL Bugs: #10351: Max_Heap_Table_Size cannot be set to 4 gb

Bug #10351	Max_Heap_Table_Size cannot be set to 4 gb
Submitted:	4 May 2005 5:23	Modified:	6 Aug 2005 22:44
Reporter:	Disha	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Server	Severity:	S3 (Non-critical)
Version:	5.0.4 Beta	OS:	Any (All)
Assigned to:	Jim Winstead	CPU Architecture:	Any

Description:
On trying to set Max_Heap_table_size to exactly 4 gb, that is 4294967296 in bytes, it does not get correctly set and is set to 0.

How to repeat:
1) set @@max_heap_table_size= 4294967296;
2) select @@max_heap_table_size;

ACTUAL RESULT:
Returns that @max_heap_table_size is 0.

EXPECTED RESULT:
Returns that @max_heap_table_size should be set to 4gb.

The machine are you tried this are 64 bit machine ?, on Windows for
example a 32 bit machine can't allocates more than 2GB of memory.

Hi Miguel,
The same behaviour is observed on both 64 as well as 32 bit machines.

On a windows machine, we can allocate close to 4 gb.
That is...
1) When we try to set it as follows: set @@max_heap_table_size= 4294967295;
    ... it actually sets the parameter to 4294966272, which is ok.
2) BUT, When we try to set it as: set @@max_heap_table_size= 4294967296;
    ... it actually sets the parameter to 0!. This is NOT acceptable.

thanks, disha

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

I think only "set @@max_heap_table_size" does't work right . 
Tested on Windows 2000 prof SP4

mysql> set @@max_heap_table_size= 4294967296;
Query OK, 0 rows affected (0.06 sec)

mysql> select @@max_heap_table_size;
+-----------------------+
| @@max_heap_table_size |
+-----------------------+
|                     0 |
+-----------------------+
1 row in set (0.00 sec)

mysql> set global max_heap_table_size= 4294967296;
Query OK, 0 rows affected (0.00 sec)

mysql> select @@max_heap_table_size;
+-----------------------+
| @@max_heap_table_size |
+-----------------------+
|            4294966272 |
+-----------------------+
1 row in set (0.00 sec)

mysql> set @@max_heap_table_size= 4294967295;
Query OK, 0 rows affected (0.00 sec)

mysql> select @@max_heap_table_size;
+-----------------------+
| @@max_heap_table_size |
+-----------------------+
|            4294966272 |
+-----------------------+
1 row in set (0.00 sec)

mysql>

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/internals/26268

+    return new Item_int((ulonglong)value, 10);

10 seems a little short, why dont you use the default value 21?

The length is set to 10 because that is the maximum length of an unsigned 32-bit int as a string.

And this is how it is done in Item_uint::Item_uint(uint32 i), which is what was being used before.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/internals/27507

Fixed in 5.0.14.

Still happens on x86, 5.0 tree compiled by gcc 3.3.5 with compile-pentium-debug-max. 
variables.test fails:
*** r/variables.result  2005-07-29 01:12:02.000000000 +0300
--- r/variables.reject  2005-07-29 01:29:36.000000000 +0300
***************
*** 528,539 ****
  set @@max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 4294967296
  set global max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 4294967296
  set @@max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 4294967296
--- 528,539 ----
  set @@max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 0
  set global max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 0
  set @@max_heap_table_size= 4294967296;
  select @@max_heap_table_size;
  @@max_heap_table_size
! 0

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/internals/27705

Fixed in 5.0.14.

Documented in 5.0.14 changelog:

<listitem><para>
 Unsigned <literal>LONG</literal> system variables may return incorrect value when retreived with a <literal>SELECT</literal> for certain values. (Bug #10351)
</para></listitem>