Crash happens in dlopen() code when trying to load the library. Crash does not happen when library is 
not DTrace instrumented . Additionally, crash does not happen with default Solaris 10 GCC 3.4.3 and
it does not happen if main executable is instrumented.

So , just check for this specific situation (32 bit, GCC3.4.6 , Solaris) and disable Dtrace in shared libraries.
We have only single plugin so far that is instrumented (ha_example)

-lthread works fine in most cases, but at least with gcc 3.4.6 on x86, dlopen() crashes when libpthread is not used. 
Note : the workaround existed prior and did not work  since CMAKE_THREADS_LIBS_INIT was already
 in cache.  Now, use SET(.. CACHE FORCE)  to overwrite the cached value.

Fixing cmake files.

 - Remove INSTALL-BINARY from installed docs directory, we provide a copy
   in the root directory (but perhaps this should be revisited later).

 - Disable audit_null and daemon_example plugins.

 - Fix the docs directory.

 - Remove mysql-test/Makefile.in

 - Build and install mysql_tzinfo_to_sql

 - Remove share/charsets/languages.html

In the worst case possible scenario (no bzr, in-source build),
make dist produced a package that compiled ok with autotools
but failed to package because extra make_binary_distribution was
found in source package and was not built. make_binary_distribution
contained paths of the build machine.

Fix:  exclude some scripts that are produced in cmake build. 
Note that there is no good general fix for it in this specific scenario.
it is advisable to build source packages out of source or in bzr repo.

actually "system"), --with-ssl should be "bundled".

Fixes error on sol-gcc-x86, where build machine had  openssl but not
the test box.

HAVE_IBGCC_ATOMIC_BUILTINS=>HAVE_IB_GCC_ATOMIC_BUILTINS.
Due to the typo, detection of atomics was broken. 
It also lead to valgrind error during shutdown 
(access to freed memory),which is likely present 
in all builds where atomics are not used.

Backport to 5.5.99

Reason: implementation of send/reap in mysqltest uses the same "embedded" connection in a thread different from current, so thread stack has to change when connection is used in different  OS thread..

Backport to 5.5.99

It appears that stack overflow checks for recusrive stored procedure
calls, that run in the normal server, did not work in embedded and were
 dummified with preprocessor magic( #ifndef EMBEDDED_SERVER ).
      
 The fix is to remove ifdefs, there is no reason not to run overflow checks
 and crash in deeply recursive calls. 
      
 Note: Start of the stack (thd->thread_stack variable) in embedded is not
necessarily exact but stil provides the best guess. Unless the caller of 
mysql_read_connect()   is already deep in the stack, thd->thread_stack 
variable should approximate stack  start address well.

The bug was that ./configure was  passing paramers to subscripts as $@, and to handle embedded spaces it needs
to be quoted as "$@".
This resulting into a bug when ./configure was called e.g with CFLAGS='-m64 -Xstrconst'..

Additionally, fixed cmake/configure.pl did not handle environment variables passed on the command line.
this is fixed in this push

Fix contains of :
-  splitting CC/CXX if there are more that 2 space-delimited tokens, add the rest to CFLAGS
(in this case CC was set to "ccache gcc --pipe", and this broke recognition of gcc compiler
as CMake understands CC consisting of 2 space delimited tokens but not more)

-  add my_new.cc to mysys fle list if C++ operator new is not found. Always, not only
for gcc  (the original problem was that missing operator new when compiling with CXX=gcc)

Conflicts:
  - mysql-test/r/mysqld--help-notwin.result
  - mysql-test/r/mysqld--help-win.result

The patch is ready and will be pushed via -bugfixing.

There was an erroneous parameter when calling flush_master_info
from write_ignored_events_info_to_relay_log which could lead to a
server crash. This happens because the I/O thread releases the
log_lock before calling the flush_master_info.

Set the function to call flush_master_info with third parameter
to true, so that the mutex is properly taken.