Re: bk commit into 5.1 tree (mats:1.2559) BUG#36197

[Konstantin Osipov]

* Mats Kindahl <mats@stripped> [08/05/09 17:53]:

Good: open_tables() is removed from
Table_map_event::do_apply_event(), and now all tables are opened
and locked in Rows_log_event::do_apply_event(), by means of
simple_open_and_lock_tables(). simple_open_and_lock_tables() has
an assert that there are no other open/locked tables in the
thread, thus this code should be change-safe in case someone in
future breaks the open table cache invariant like the old code
did.

Now Table_map_event only collects the used tables in the
table list, and they all are opened at once.
It's nice that it also simplifies the implementation quite a bit.

The patch is OK to push.

> ChangeSet@stripped, 2008-05-09 15:30:54+02:00, mats@mats-laptop.(none) +8 -0
>   BUG#36197: flush tables (or little table cache) can cause crash on slave
>   
>   When flushing tables, there were a slight chance that the flush was occuring
>   between processing of two table map events. Since the tables are opened
>   one by one, it might result in that the tables were not valid and that sub-
>   sequent locking of tables would cause the slave to crash.
>   
>   The problem is solved by opening and locking all tables at once using
>   simple_open_n_lock_tables(). Also, the patch contain a change to open_tables()
>   so that pre-locking only takes place when the trg_event_map is zero, which

is not zero

>   was not the case before (this caused the lock to be placed in
> thd->locked_tables
>   instead of thd->lock since the assumption was that triggers would be called
>   later and therefore the tables should be pre-locked).


--