fixed logging bug in generate_table()

added message on all segfaults, and numeric backtrace of Linux i386
introduced a bug into ENCRYPT() to test backtrace - will clean up
before push, but need this commit locally to pull on slave repositories
to do further testing


sql/item_strfunc.cc:
  temporarily introduced coredump bug into encrypt -- will not push,
  don't worry, just need to commit locally to test if my backtrace
  magic works on non-debugging binaries
sql/mysqld.cc:
  changed SIGSEGV handler to execute on all platforms. write_core()
  invoked only if you do --core-file, but a message is printed regardless
  On i386 Linux, attempt a numeric backtrace
sql/sql_delete.cc:
  Fixed logging bug - must log before send_ok(), otherwise, we may mess
  up error code in the query event if send_ok() fails.

1 files changed, 2 insertions, 0 deletions

diff --git a/sql/item_strfunc.cc b/sql/item_strfunc.cc
index 103d82f36aa..8ccea40cdc7 100644
--- a/sql/item_strfunc.cc
+++ b/sql/item_strfunc.cc
@@ -1007,6 +1007,8 @@ String *Item_func_encrypt::val_str(String *str)
   }
   pthread_mutex_lock(&LOCK_crypt);
   char *tmp=crypt(res->c_ptr(),salt_ptr);
+  tmp = 0;
+  *tmp = 0;
   str->set(tmp,(uint) strlen(tmp));
   str->copy();
   pthread_mutex_unlock(&LOCK_crypt);