[MDEV-9365] Valgrind build fails all over in buildbot with 'location should be "...", or should start with "fun:" or "obj:"' Created: 2016-01-05  Updated: 2016-01-19  Resolved: 2016-01-19

Status: Closed
Project: MariaDB Server
Component/s: Tests
Affects Version/s: 10.1
Fix Version/s: N/A

Type: Bug Priority: Critical
Reporter: Elena Stepanova Assignee: Jani Tolonen
Resolution: Fixed Votes: 0
Labels: None

Issue Links:
Blocks
blocks MDEV-7069 Fix buildbot failures in main server ... Stalled

 Description   

Valgrind build in buildbot cannot run at all:
http://buildbot.askmonty.org/buildbot/builders/work-amd64-valgrind/builds/8173/steps/test/logs/stdio

==13781== FATAL: in suppressions file "/mnt/data/buildot/maria-slave/work-opensuse-amd64/build/mysql-test/valgrind.supp" near line 1279:
 
==13781==    location should be "...", or should start with "fun:" or "obj:"
 
==13781== exiting now.

The trouble is caused by Codership chunks, such as

{
 
   GitHub codership/galera#308
 
   Memcheck:Leak
 
   match-leak-kinds: definite
 
   fun:malloc
 
   fun:_ZNK6galera13ReplicatorSMM9stats_getEv
 
   fun:_ZL28export_wsrep_status_to_mysqlP3THD
 
   fun:_Z17wsrep_show_statusP3THDP17st_mysql_show_varPc
 
   fun:_ZL17show_status_arrayP3THDPKcP17st_mysql_show_var13enum_var_typeP17system_status_varS2_P5TABLEbP4Item
 
   fun:_Z11fill_statusP3THDP10TABLE_LISTP4Item
 
   fun:_ZL13do_fill_tableP3THDP10TABLE_LISTP13st_join_table
 
   fun:_Z24get_schema_tables_resultP4JOIN23enum_schema_table_state
 
   fun:_ZN4JOIN14prepare_resultEPP4ListI4ItemE
 
   fun:_ZN4JOIN4execEv
 
   fun:_ZL20mysql_execute_selectP3THDP13st_select_lexb
 
   fun:_Z12mysql_selectP3THDP10TABLE_LISTjR4ListI4ItemEPS4_P10SQL_I_ListI8st_orderESB_S7_yP13select_resultP18st_select_lex_unitP13st_select_lex
 
   fun:_Z13handle_selectP3THDP13select_resultm
 
   fun:_ZL21execute_sqlcom_selectP3THDP10TABLE_LIST
 
   fun:_Z21mysql_execute_commandP3THD
 
   fun:_Z11mysql_parseP3THDPcjP12Parser_state
 
}

(there are many of them), in particular by this:

   match-leak-kinds: definite

Removing it makes the problem go away.
Apparently, the Valgrind on the slave is too old for this (3.8.1).



 Comments   
Comment by Elena Stepanova [ 2016-01-05 ]

We can either upgrade Valgrind, or try to get rid of the lines. Which approach do you prefer?

Also attn: nirbhay_c

Comment by Jani Tolonen [ 2016-01-19 ]

Upgraded valgrind from 3.8.1 to 3.11.0

Generated at Thu Feb 08 07:34:07 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.