Currently such attempts lead to assertions like this:

mysqld.cc:3757: void my_malloc_size_cb_func(long long int, my_bool): Assertion `!is_thread_specific || (mysqld_server_initialized && thd)' failed.

The reason is that memory accounting is done in THD constructor (which is called after member/base class constructor). Thus memory allocated by member/base constructor is not accounted. Same problem (but turn it upside down) with destructor.

This causes ugly constructor/destructor simulation like:

THD::~THD()

{

  ...

  /* trick to make happy memory accounting system */

#ifndef EMBEDDED_LIBRARY

  session_tracker.sysvars.deinit();

#endif //EMBEDDED_LIBRARY

  if (status_var.local_memory_used != 0)

  {

    DBUG_PRINT("error", ("memory_used: %lld", status_var.local_memory_used));

    SAFEMALLOC_REPORT_MEMORY(thread_id);

    DBUG_ASSERT(status_var.local_memory_used == 0 ||

                !debug_assert_on_not_freed_memory);

  }

  update_global_memory_status(status_var.global_memory_used);

  set_current_thd(orig_thd == this ? 0 : orig_thd);

  DBUG_VOID_RETURN;

}

Original implementation was declined due to convoluted current_thd juggling. This juggling is needed, because in some cases THD objects are created by thread creator, not thread itself.

Patches implementing this task are in bb-10.5-svoj-MDEV-19439, but they are incomplete and fail on assertions. Move THD creation from creator to creature and do whatever is needed to make them work.