Too complex/risky to try to fix
Instead, detect the problem and communicate that this is not supported

write()/read()

Sometimes stop/restart master or stop/restart salve can cause
network error, which can cause the 'invalid file descriptor
-1 in syscall write()/read()' warnings. All involved test
cases except rpl_slave_load_remove_tmpfile belong to the
kind of network error. So they are expected.
The 'rpl_slave_load_remove_tmpfile' belongs to file error,
but it is testing the file error as following code:
DBUG_EXECUTE_IF("remove_slave_load_file_before_write",
my_close(fd,MYF(0)); fd= -1; my_delete(fname, MYF(0)););
So it's expected too.

To fix the problem, add the valgrind warnings to the global
suppression list to suppress it.

 As suggested, convert internally to value of build_thread

delimiter not executed so does not recognize end of block
Always execute delimiter command, revert after false if() block.

valgrind pointed to a buffer allocated by my_realloc which looked fishy
Replaced size with what was probably intended, added test case.
Now also fixed line after review comment

Since the exec command line is passed on externally, it cannot take newlines
Simply replace \n with space
Now also added test case

- The arguments are properly quoted when mtr.pl calls my_safe_process but
  unfortunately the all off when running with active state perl and stays
  in cygwin perl.
- Extend the patch to only quote args that are not already quoted
This a redo of previous commit, will be included in next push

Undid amendment allowing pending reap after switching connections
Moved check for pending reap earlier; failed if running with ps-protocol

Small amendment: ignore pending reap when switching connection, add test

Added them
NB the 6th case is adapted to Bug no. 49269, gives wrong output without it

Set a flag after send to trap the case

Implemented --lowercase_result which lower cases next result

Was available in v1
Porting to v2 required some rewriting
Updated after review comments

Combinations beginning with -- not allowed
Allow them...

NULLable BIGINT and INT columns in comparison

Problem: a consequence of the fix for 43668.
Some Arg_comparator inner initialization missed,
that may lead to unpredictable (wrong) comparison
results.

Fix: always properly initialize Arg_comparator
before its usage.

Extracts last 20 lines if no output after failure

  was killed

Merge the fix from 5.1-bugteam to 5.1-main

Merge the fix from 5.1-bugteam to 5.1-main

Merge the fix from 5.1-bugteam to 5.1-main

      
Add the test to disabled list and remove from the experimental list.

the last IF ELSE part which decides the plugin name is not relevant as we still have
to substitute the occurences of INNOBASE with INNODB_PLUGIN.
Remove the last IF ELSE part in CMakeLists.txt as it doesn't make sense in 5.1.

manual merge 5.0-->5.1, updating InnoDB plugin.

The bug 38816 changed the lock that protects THD::query from
LOCK_thread_count to LOCK_thd_data, but didn't update the associated
InnoDB functions.

1. The innobase_mysql_prepare_print_arbitrary_thd and the
innobase_mysql_end_print_arbitrary_thd InnoDB functions have been
removed, since now we have a per-thread mutex: now we don't need to wrap
several inter-thread access tries to THD::query with a single global
LOCK_thread_count lock, so we can simplify the code.

2. The innobase_mysql_print_thd function has been modified to lock
LOCK_thd_data in direct way.