Automerged bundle from committed and approved cset.

Replace part of the patch that Kevin apparently forgot to push.
Fix the bug also in the built-in InnoDB of MySQL 5.1.

I cannot explain why the test case was not failing without the
full patch.

This was rb:762, approved by me.

Extra fix: 'if (p5 < p5_a + P5A_MAX)' is not portable.
p5 starts out pointing to a static array, then may point
to a buffer on the stack, then may point to malloc()ed memory.

The problem occurred when indexes are added between the time that an
UNDO record is created and the time that the purge thread comes around
and deletes the old secondary index entries.  The purge thread would
hit an assert when trying to build a secondary index entry for
searching.  The problem was that the old value of those fields were not
in the UNDO record since they were not part of an index when the UPDATE
occured. 
A test case was added to innodb-index.test.

The problem occurred when indexes are added between the time that an
UNDO record is created and the time that the purge thread comes around
and deletes the old secondary index entries.  The purge thread would
hit an assert when trying to build a secondary index entry for
searching.  The problem was that the old value of those fields were not
in the UNDO record since they were not part of an index when the UPDATE
occured. 
A test case was added to innodb-index.test.

WITH MYSQL_REFRESH()

reset_slave_info.all was not initialized.

We fix this by setting lex->reset_slave_info.all= false in
the lex_start routine, which is called before every statement.

  Added options --boot-gdb etc.
  Extended gdb_arguments() with optional input argument
  Cannot use set args in gdb init file, as run < <input> kills them (?)

  Added simple cut-off w/warning if > one million lines

For some reason the test authentication plugin accepted connection with arbitrary password. But the intention of the plugin is that password should equal to the authentication string and in the later versions of the server connection fails if password is wrong. So I have updated auth_rpl test to specify the correct password.

               FULLTEXT INDEXES

myisamchk may create incorrect fulltext index for compressed
tables. Incorrect data pointer size was used while creating
fulltext index.

  Quick fix: run mysql-stress-test.pl via a wrapper test
  Amend mtr to run just that test when using --stress
  Updated mysql-stress-test.pl to exit(1) if wrong options

  Followup fixes for --ps-protocol:
  1) Incorrectly set ps_protocol variable instead of ps_protocol_enabled
  2) disable_result_log was tested after calling handle_no_error()
        which would revert a temporary setting.

Connection of slave to master using a replication account which authenticates
with an external plugin was not possible.

Fixed by making sure that the CLIENT_PLUGIN_AUTH capability is set when client connects using mysql_real_connect(). Also, a plugin-dir path used by client library to locate authentication plugins is set based on the analogous server setting. This is done in connect_to_master() function before a call to mysql_real_connect().

  Just check that core file name does not end in .gz
  Not a problem for Windows, as we check for files *ending* in .dmp

  Export $MYSQLD as full path to executable
  $MYSQLD_CMD is executable with minimum required args

  Call handle_error() instead of die() when evaluating these
  Must remember "current command" with link to errors to ignore
  Added test cases to mysqltest.test 

  Added a keyword ONCE to add to those commands
  Some internal tables to keep track of which properties are
    temporarily overriden
  Added tests in mysqltest.test
  Updates to other tests will be done later

when buffered changes are to be discarded

sync_thread_add_level(level = SYNC_IBUF_INDEX_TREE): Relax a too strict
condition that the thread must not be holding locks below SYNC_FSP_PAGE.
It is perfectly valid to hold any latch above SYNC_IBUF_INDEX_TREE when
acquiring the insert buffer tree latch.