Comment 1 for bug 1379582

Revision history for this message
Roel Van de Paar (roel11) wrote : Re: tmp->start_utime > now_utime impossibility leading to huge TIME_MS values in processlist

===================== Testcase
1. Patch code

$ bzr diff
=== modified file 'sql/sql_show.cc'
--- sql/sql_show.cc 2014-09-26 04:43:24 +0000
+++ sql/sql_show.cc 2014-10-10 10:15:06 +0000
@@ -2301,6 +2301,8 @@
       table->field[8]->store(((tmp->start_utime ?
                                now_utime - tmp->start_utime : 0)/ 1000));

+DBUG_ASSERT( (tmp->start_utime != 0) && (((now_utime - tmp->start_utime) / 1000) > 100000 ) );
+
       mysql_mutex_lock(&tmp->LOCK_thd_data);
       /* ROWS_SENT */
       table->field[9]->store((ulonglong) tmp->get_sent_row_count());

2. Recompile

3. Modify uploaded pquery-run-hack.sh to set correct BASEDIR (point to modded build) and INFILE;
BASEDIR=/sda/Percona-Server-5.6.21-rel69.0-676-debug.Linux.x86_64
INFILE=/sda/814189/high_time0.sql # See upload

===================== core/mysqld/ldd
See upload