From: Annamalai Gurusami
Date: December 20 2012 6:28am
Subject: bzr push into mysql-5.6 branch (annamalai.gurusami:4807 to 4808) Bug#14556349
List-Archive: http://lists.mysql.com/commits/145541
X-Bug: 14556349
Message-Id: <20121220062858.29042.71802.4808@rainbow>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

 4808 Annamalai Gurusami	2012-12-20
      Bug #14556349 RENAME OF COMPRESSED TABLE AND INSERT BUFFER MERGE CAUSE
      HANG
      
      Problem Statement:
      
      When the operation RENAME TABLE is about rename the tablespace of the
      table, it will stop all i/o operations on the tablespace temporarily.
      For this the fil_space_t::stop_ios member is used.
      
      Once the fil_space_t::stop_ios member is set to TRUE in the RENAME
      TABLE operation, it is expected that no new i/o operation will be done
      on the tablespace and all pending i/o operation can be completed on
      the tablespace.
      
      If the pending i/o operations initiate any new i/o operations then
      there will be deadlock.  The RENAME TABLE operation will be waiting
      for pending i/o on the tablespace to be completed, and the pending i/o
      operations will be waiting on the RENAME TABLE operation to set the
      file_space_t::stop_ios flag to be set to FALSE.
      
      But in the given scenario the pending i/o operations did not initiate
      new i/o.  But they where still unnecessarily checking the
      fil_space_t::stop_ios flag.  This resulted in deadlock.
      
      Solution:
      
      I noticed that this deadlock happens in fil_space_get_size() and
      fil_space_get_zip_size() in the i/o threads.  These functions check
      the stop_ios flag even when no i/o will be initiated.  I modified
      these functions to ensure that they check the stop_ios flag only when
      they will be initiating an i/o operation.  This solves the problem.
      
      rb://1635 (mysql-5.5)
      rb://1660 (mysql-trunk) approved by Inaam, Jimmy, and ima.

    modified:
      storage/innobase/fil/fil0fil.cc
 4807 Tor Didriksen	2012-12-19 [merge]
      empty merge 5.5 => 5.6

=== modified file 'storage/innobase/fil/fil0fil.cc'
--- a/storage/innobase/fil/fil0fil.cc	revid:tor.didriksen@stripped
+++ b/storage/innobase/fil/fil0fil.cc	revid:annamalai.gurusami@stripped
@@ -1481,6 +1481,21 @@ fil_space_get_space(
 	if (space->size == 0 && space->purpose == FIL_TABLESPACE) {
 		ut_a(id != 0);
 
+		mutex_exit(&fil_system->mutex);
+
+		/* It is possible that the space gets evicted at this point
+		before the fil_mutex_enter_and_prepare_for_io() acquires
+		the fil_system->mutex. Check for this after completing the
+		call to fil_mutex_enter_and_prepare_for_io(). */
+		fil_mutex_enter_and_prepare_for_io(id);
+
+		/* We are still holding the fil_system->mutex. Check if
+		the space is still in memory cache. */
+		space = fil_space_get_by_id(id);
+		if (space == NULL) {
+			return(NULL);
+		}
+
 		/* The following code must change when InnoDB supports
 		multiple datafiles per tablespace. */
 		ut_a(1 == UT_LIST_GET_LEN(space->chain));
@@ -1552,8 +1567,7 @@ fil_space_get_size(
 	ulint		size;
 
 	ut_ad(fil_system);
-
-	fil_mutex_enter_and_prepare_for_io(id);
+	mutex_enter(&fil_system->mutex);
 
 	space = fil_space_get_space(id);
 
@@ -1583,7 +1597,7 @@ fil_space_get_flags(
 		return(0);
 	}
 
-	fil_mutex_enter_and_prepare_for_io(id);
+	mutex_enter(&fil_system->mutex);
 
 	space = fil_space_get_space(id);
 

No bundle (reason: useless for push emails).