List:Commits« Previous MessageNext Message »
From:Christopher Powers Date:September 22 2011 9:31pm
Subject:bzr push into mysql-trunk branch (chris.powers:3427 to 3428) Bug#12981100
View as plain text  
 3428 Christopher Powers	2011-09-22
      Bug#12981100 - PERFORMANCE_SCHEMA NET IO IDLE AND RECV EVENTS NEST WITHIN WRONG STATEMENT
      
      - For idle events, set idle nesting_event_id to 0 when locker object created
      
      - For recv events, set nesting_event_id to parent nesting_event_id. This had
        been fixed but was regressed prior to merge for Fuschia. 
      
      - Fixed warning for 64-bit builds in create_socket()

    modified:
      storage/perfschema/pfs.cc
      storage/perfschema/pfs_instr.cc
 3427 Raghav Kapoor	2011-09-22
      BUG#12735829 - SPACE() FUNCTION WARNING REFERS TO REPEAT() 
      IN ER_WARN_ALLOWED_PACKET_OVERFLOWED 
      
      BACKGROUND:
      Previously in the bug,the warning text was incorrect.Instead 
      of printing space it was printing repeat in the warning message.
      
      FIX:
      
      The bug has been fixed by creating a new class Item_func_space
      in item_strfunc.h file and implementing fix_length_and_dec()
      and val_str(String *str) functions for it in item_strfunc.cc.
      Also Create_func_space now returns Item_func_space in 
      item_create.cc file.This patch also partially fixes the 
      bug#11746123 which arises because of wrong implementaion of
      space() function which "caches" session character set and 
      collation at creation time.Test cases have been written to 
      demonstrate the fix for the both the bugs in func_str.test 
      file and the corrsponding result files have also been updated.  
     @ mysql-test/t/func_str.test
        Test case for bug#12735829 and bug#11746123 have been written
     @ sql/item_create.cc
        Item_func_space is called instead of Item_func_repeat
     @ sql/item_strfunc.cc
        fix_length_and_dec() and val_str(String *str) functions 
        for item_func_space have been implemented.
     @ sql/item_strfunc.h
        new class item_func_space has been created

    modified:
      mysql-test/r/ctype_utf16.result
      mysql-test/r/func_str.result
      mysql-test/t/func_str.test
      sql/item_create.cc
      sql/item_strfunc.cc
      sql/item_strfunc.h
=== modified file 'storage/perfschema/pfs.cc'
--- a/storage/perfschema/pfs.cc	2011-09-21 15:41:50 +0000
+++ b/storage/perfschema/pfs.cc	2011-09-22 21:30:32 +0000
@@ -2878,8 +2878,8 @@ get_thread_socket_locker_v1(PSI_socket_l
 
       PFS_events_waits *parent_event= wait - 1;
       wait->m_event_type= EVENT_TYPE_WAIT;
-      wait->m_nesting_event_id=   parent_event->m_event_id;
-      wait->m_nesting_event_type= parent_event->m_event_type;
+      wait->m_nesting_event_id=   parent_event->m_nesting_event_id;
+      wait->m_nesting_event_type= parent_event->m_nesting_event_type;
       wait->m_thread=       pfs_thread;
       wait->m_class=        pfs_socket->m_class;
       wait->m_timer_start=  0;
@@ -3141,8 +3141,8 @@ start_idle_wait_v1(PSI_idle_locker_state
 
       PFS_events_waits *parent_event= wait - 1;
       wait->m_event_type= EVENT_TYPE_WAIT;
-      wait->m_nesting_event_id= parent_event->m_event_id;
-      wait->m_nesting_event_type= parent_event->m_event_type;
+      wait->m_nesting_event_id= 0;   // idle timer not linked to statements or stages
+      wait->m_nesting_event_type= EVENT_TYPE_WAIT;
 
       wait->m_thread= pfs_thread;
       wait->m_class= &global_idle_class;

=== modified file 'storage/perfschema/pfs_instr.cc'
--- a/storage/perfschema/pfs_instr.cc	2011-09-21 15:41:50 +0000
+++ b/storage/perfschema/pfs_instr.cc	2011-09-22 21:30:32 +0000
@@ -1473,7 +1473,8 @@ PFS_socket* create_socket(PFS_socket_cla
     */
   my_socket fd= likely(identity != NULL) ?
                 *(reinterpret_cast<const my_socket*>(identity)) : 0;
-  uint random= randomized_index((const void *)fd, socket_max);
+
+  uint random= randomized_index(reinterpret_cast<const void *>(fd), socket_max);
 
   for (scan.init(random, socket_max);
        scan.has_pass();

No bundle (reason: useless for push emails).
Thread
bzr push into mysql-trunk branch (chris.powers:3427 to 3428) Bug#12981100Christopher Powers23 Sep