6e11407
Talk about silly byte-shaving ... the "stock" value of STACK_MIN_SIZE
6e11407
does not work on s390x, and probably not on some other platforms either.
6e11407
Knock it up to twice the pre-5.0.33 value to have some margin for future
6e11407
changes in compiler, glibc, etc.  Suspect this number if
6e11407
"execution_constants" regression test fails.
81f2dd6
0f4cc99
Note: STACK_MIN_SIZE has to be enough less than DEFAULT_THREAD_STACK / 8
0f4cc99
to provide some headroom, which means that its value can't be raised too much
0f4cc99
further without increasing the latter as well.
aec7501
c625e75
As of 5.1.33 the stock value is 16000, which might be enough ... or maybe
c625e75
not.  For the moment, we keep this patch around for reference, but it's
c625e75
not being applied.
c625e75
aec7501
f5da580
diff -Naur mysql-5.1.30.orig/sql/mysql_priv.h mysql-5.1.30/sql/mysql_priv.h
f5da580
--- mysql-5.1.30.orig/sql/mysql_priv.h	2008-11-14 11:37:13.000000000 -0500
f5da580
+++ mysql-5.1.30/sql/mysql_priv.h	2009-01-13 11:07:35.000000000 -0500
f5da580
@@ -286,7 +286,7 @@
aec7501
  Feel free to raise this by the smallest amount you can to get the
aec7501
  "execution_constants" test to pass.
aec7501
  */
f5da580
-#define STACK_MIN_SIZE          12000   ///< Abort if less stack during eval.
f5da580
+#define STACK_MIN_SIZE          16384   ///< Abort if less stack during eval.
aec7501
 
aec7501
 #define STACK_MIN_SIZE_FOR_OPEN 1024*80
f5da580
 #define STACK_BUFF_ALLOC        352     ///< For stack overrun checks