bzr push into mysql-trunk branch (andrei.elkin:3478 to 3479) Bug#12960284

[Andrei Elkin]

 3479 Andrei Elkin	2011-10-07
      Bug#12960284 - RPL.RPL_PARALLEL_SWITCH_SEQUENTIAL TEST FAILED IN VALGRIND RUN ON
      
      There was a timeout at the very end of the test.
      Details of printout upon the failure does not reveal any issue with the text nor
      the server code. It's just slow and is used to pass on the same valgrind but the run
      time is apparently big (over 1 000 000 msec) -
        rpl.rpl_parallel_switch_sequential 'mix' w2 [ pass ]  1068101
      
      So it's fixed with decreasing load size when the test runs on valgrind.
     @ mysql-test/collections/default.experimental
        correcting ref:s to bugs to accord them with the format.
     @ mysql-test/suite/rpl/t/rpl_parallel_switch_sequential.test
        decreasing load size when the test runs on valgrind.

    modified:
      mysql-test/collections/default.experimental
      mysql-test/suite/rpl/t/rpl_parallel_switch_sequential.test
 3478 Mattias Jonsson	2011-10-07
      Bug#13007154: Crash in keys_to_use_for_scanning with ORDER BY
                                 and PARTITIONING
      
      The server could crash on a query that called
      ha_partition::keys_to_use_for_scanning when there was no partitions
      used.Due to the first used partition would then be out of bounds.
      
      The solution was to always use the first partitions handler since the
      function will return the same key_map for every partition, since they
      have the same engine and the same structure.
      
      This bug was introduced as a regression in WL#5217.

    modified:
      mysql-test/r/partition_innodb.result
      mysql-test/t/partition_innodb.test
      sql/ha_partition.cc
=== modified file 'mysql-test/collections/default.experimental'
--- a/mysql-test/collections/default.experimental	2011-10-04 10:55:09 +0000
+++ b/mysql-test/collections/default.experimental	2011-10-07 08:36:04 +0000
@@ -41,6 +41,6 @@ sys_vars.have_ndbcluster_basic
 sys_vars.ndb_log_updated_only_basic
 sys_vars.rpl_init_slave_func		 # Bug#12535301 2011-05-09 andrei sys_vars.rpl_init_slave_func mismatches in daily-5.5
 
-rpl.rpl_mixed_mts_rec_crash_safe @solaris          # Bug 12902514 2011-08-19 andrei mts recovery tests are slow
-rpl.rpl_mixed_mts_rec_crash_safe_checksum @solaris # same as rpl_mixed_mts_rec_crash_safe
-rpl.rpl_mixed_mts_crash_safe @solaris              # same as rpl_mixed_mts_rec_crash_safe
+rpl.rpl_mixed_mts_rec_crash_safe @solaris          # Bug#12902514 2011-08-19 andrei mts recovery tests are slow
+rpl.rpl_mixed_mts_rec_crash_safe_checksum @solaris # Bug#12902514 2011-08-19 andrei same as rpl_mixed_mts_rec_crash_safe
+rpl.rpl_mixed_mts_crash_safe @solaris              # Bug#12902514 2011-08-19 andrei same as rpl_mixed_mts_rec_crash_safe

=== modified file 'mysql-test/suite/rpl/t/rpl_parallel_switch_sequential.test'
--- a/mysql-test/suite/rpl/t/rpl_parallel_switch_sequential.test	2011-08-19 13:04:28 +0000
+++ b/mysql-test/suite/rpl/t/rpl_parallel_switch_sequential.test	2011-10-07 08:36:04 +0000
@@ -40,6 +40,11 @@ let $slave_status= 0;
 let $trx= 0;
 let $alter= 0;
 let $i= 200; # with bigger value test times out on some PB hosts 
+if ($VALGRIND_TEST)
+{
+  # test may be slow on Valgrind
+  let $i=50;
+}
 if (`select @@binlog_format like "STATEMENT"`) {
    # relax mtr to scan unsafe warnings
    let $i=50;

No bundle (reason: useless for push emails).