Merge lp:~laurynas-biveinis/percona-server/bug1407941-5.5 into lp:percona-server/5.5

[Laurynas Biveinis]

Fix bug 1407941 (Thread id lost on srv/srv0start.cc).

The issue is a thread_ids array element index being reused on the call
to os_thread_create for the redo log tracker thread. Fix by allocating
a separate array item correctly. An alternative fix would be simply
passing NULL, but apparently accounting those thread ids is a good
idea on some of the platforms. Add a comment describing the thread_ids
array item allocation. Fix thread_ids and n array sizing.

    http://jenkins.percona.com/job/percona-server-5.5-param/1079/

[Laurynas Biveinis (laurynas-biveinis)]

review: Approve

=== modified file 'storage/innobase/srv/srv0start.c'
--- storage/innobase/srv/srv0start.c	2014-03-27 15:42:21 +0000
+++ storage/innobase/srv/srv0start.c	2015-01-07 07:55:37 +0000
@@ -122,9 +122,16 @@
 static os_file_t	files[1000];
 
 /** io_handler_thread parameters for thread identification */
-static ulint		n[SRV_MAX_N_IO_THREADS + 8];
-/** io_handler_thread identifiers */
-static os_thread_id_t	thread_ids[SRV_MAX_N_IO_THREADS + 8];
+static ulint		n[SRV_MAX_N_IO_THREADS];
+/** io_handler_thread identifiers. The extra elements at the end are allocated
+as follows:
+SRV_MAX_N_IO_THREADS + 1: srv_master_thread
+SRV_MAX_N_IO_THREADS + 2: srv_lock_timeout_thread
+SRV_MAX_N_IO_THREADS + 3: srv_error_monitor_thread
+SRV_MAX_N_IO_THREADS + 4: srv_monitor_thread
+SRV_MAX_N_IO_THREADS + 5: srv_LRU_dump_restore_thread
+SRV_MAX_N_IO_THREADS + 6: srv_redo_log_follow_thread */
+static os_thread_id_t	thread_ids[SRV_MAX_N_IO_THREADS + 7];
 
 /** We use this mutex to test the return value of pthread_mutex_trylock
    on successful locking. HP-UX does NOT return 0, though Linux et al do. */
@@ -1184,7 +1191,7 @@
 		/* Create the thread that follows the redo log to output the
 		   changed page bitmap */
 		os_thread_create(&srv_redo_log_follow_thread, NULL,
-				 thread_ids + 5 + SRV_MAX_N_IO_THREADS);
+				 thread_ids + 6 + SRV_MAX_N_IO_THREADS);
 	}
 }