Yes, will remove the err=0 in line 268.
On 09/21/2010 09:53 AM, He Zhenxing wrote:
> Nice work, patch approved!
> One minor suggestion, I think you can remove the err=0 in line 268 then.
> Luis Soares wrote:
> file:///home/lsoares/Workspace/bzr/work/bugfixing/56882/mysql-next-mr-bugfixing/ based on
>> 3295 Luis Soares 2010-09-21
>> BUG#56882: binlog_base64_flag fails on valgrind and win debug
>> max platforms
>> In some cases, the variable err in mysql_client_binlog_statement
>> would not be initialized but still some conditional jumps would
>> rely on its value. This is for instance the case when a BINLOG
>> event is truncated and it returns an error. In this case the
>> 'err' variable is not initialized at that time, but we jump to
>> the label 'end' and then check the following condition:
>> if ((error || err) && ... )
>> This makes valgrind complain that there is a conditional jump or
>> move depending on an uninitialized value.
>> We fix this by initializing the variable err to 0.
>> === modified file 'sql/sql_binlog.cc'
>> --- a/sql/sql_binlog.cc 2010-09-01 02:51:08 +0000
>> +++ b/sql/sql_binlog.cc 2010-09-20 23:00:48 +0000
>> @@ -147,7 +147,7 @@ void mysql_client_binlog_statement(THD*
>> - int err;
>> + int err= 0;
>> Relay_log_info *rli;
>> rli= thd->rli_fake;
>> if (!rli && (rli= thd->rli_fake= new Relay_log_info(FALSE)))