Backport of the fix for BUG#40368 "mysqld_safe not honouring underscore same as dash on server options" from 6.0
(revision-id:guilhem@mysql.com-20090505113602-l12kxupeatve18dh).
Such bug led "mysqld_safe --core_file_size=#" to not work because mysqld_safe wouldn't recognize
that "ulimit -c" is needed; only --core-file-size=# worked. Same for --open_files_limit and other
options with _ where mysqld_safe needs to do something more than passing to mysqld.
Original fix by Erik Ljungstrom erik at ibiblio dot org ; slightly modified here.
Tested on all internally accessible Unix.

Backport to 5.5: adjusted the test outputs in the funcs_1 test suite

   Although the MY_SYNC_DIR flag supported in my_create(), my_delete(),
   my_rename() and my_symlink(), this feature is not used in the
   mysql code now. So technically we can declare the MY_SYNC_DIR as 0,
   but I decided to assign a new value for it as it's probably safer and
   worths nothing.

per-file comments:
  include/my_sys.h
Bug#47126      equal flag values causing unexpected behaviour
     assign unique value for the MY_SYNC_DIR

  In fact this crashes in normal (not embedded) run also.
  The problem is in the memory mapping. Handling the ha_myisammrg::extra(MMAP)
  the MERGE engine tries to mmap all the tables it unites.
  Though some can be empty and then in the mi_dynmap_file()
  we call the my_mmap(0). Normally this call returns MAP_FAILED,
  but not on FreeBSD. There it returns like a 'normal' value,
  and after the consequitive munmap systems gets unstable and
  crashes on some system call later.

per-file comments:
  storage/myisam/mi_dynrec.c
Bug #47139      Test "merge" crashes in "embedded" run
    don't try to mmap zero-length area, just return at once.

   The additional patch. That 'loadxml.test' failure was actually about our testing system,
   not the code.
   Firstly we need a new mysqltest command, wich i called 'send_eval'. So the expression
   can be evaluated, then started in a parallel thread. We only have separane 'send' and
   'eval' commands at the moment.
   Then we need to add the waiting code after the 'KILL' to our test, so the thread will be killed
   before the test goes further. The present 'reap' command doesn't handle the killed threads
   well.
      
per-file comments:
  client/mysqltest.cc
Bug#42520      killing load .. infile Assertion failed: ! is_set(), file .\sql_error.cc, line 8
    The 'send_eval' command implemented.

  mysql-test/r/loadxml.result
Bug#42520      killing load .. infile Assertion failed: ! is_set(), file .\sql_error.cc, line 8
   test result updated.

  mysql-test/t/loadxml.test
Bug#42520      killing load .. infile Assertion failed: ! is_set(), file .\sql_error.cc, line 8
   test case added.

Backporting from 6.0.

# Bug#24690 Stored functions: RETURNing UTF8 strings
# do not return UTF8_UNICODE_CI collation
#
# Bug#17903: cast to char results in binary
# Regression. The character set was not being properly initialized
# for CAST() with a type like CHAR(2) BINARY, which resulted in
# incorrect results or even a server crash.
#

Backporting from mysql-6.0-codebase.

mysql-test/r/sp-ucs2.result:
mysql-test/t/sp-ucs2.test:

  Adding tests

sql/mysql_priv.h:
  Adding prototype

sql/sp.cc
  Remember COLLATE clause for non-default collations

sql/sql_parse.cc
  Adding a new helper function

sql/sql_yacc.yy
  - Allow "CHARACTER SET cs COLLATE cl" in
    SP parameters, RETURNS, DECLARE
  - Minor reorganization for "ASCII" and "UNICODE"
    related rules, to make the code more readable,
    also to allow these aliases:
    * "VARCHAR(10) ASCII BINARY"   -> CHARACTER SET latin1 COLLATE latin1_bin
    * "VARCHAR(10) BINARY ASCII"   -> CHARACTER SET latin1 COLLATE latin1_bin
    * "VARCHAR(10) UNICODE BINARY" -> CHARACTER SET ucs2 COLLATE ucs2_bin
    * "VARCHAR(10) BINARY UNICODE" -> CHARACTER SET ucs2 COLLATE ucs2_bin
    Previously these four aliases returned the error
    "This version of MySQL does not yet support return value collation".

Note:

   This patch allows  "VARCHAR(10) CHARACTER SET cs COLLATE cl"
   and the above four aliases.

   "VARCHAR(10) COLLATE cl" is still not allowed
   i.e. when COLLATE is given without CHARACTER SET.
   If we want to support this, we need an architecture decision
   which character set to use by default.