| Bug #74844 | killed_state save in ndbcluster_binlog retry at shutdown doesn't need to be vola | ||
|---|---|---|---|
| Submitted: | 13 Nov 2014 22:39 | Modified: | 20 Mar 2015 15:26 |
| Reporter: | Stewart Smith | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S3 (Non-critical) |
| Version: | 5.7.5 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
[13 Nov 2014 22:40]
Stewart Smith
remove volatile in ndbcluster_binlog codepath (*) I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.
Contribution: volatile-ndb-killed.patch (text/x-patch), 746 bytes.
[14 Nov 2014 12:55]
MySQL Verification Team
Hello Stewart, Thank you for the report and contribution. Thanks, Umesh
[20 Mar 2015 15:26]
Jon Stephens
Fixed in trunk-cluster, currently known as NDB 7.5.0. This is an internal change only and not visible to users, so no changelog entry is required. Thanks for your contribution! Closed.
Description: --- mysql-5.7.5-m15.orig/sql/ha_ndbcluster_binlog.cc +++ mysql-5.7.5-m15/sql/ha_ndbcluster_binlog.cc @@ -7585,7 +7585,7 @@ restart_cluster_failure: { DBUG_PRINT("error", ("Failed to write to ndb_binlog_index at shu mysql_mutex_lock(&thd->LOCK_thd_data); - volatile THD::killed_state killed= thd->killed; + THD::killed_state killed= thd->killed; /* We are cleaning up, allow for flushing last epoch */ thd->killed= THD::NOT_KILLED; /* also clear error from last failing write */ The volatile keyword doesn't win us anything here, it will just prevent any optimizations and force loads/stores to local variable. At least in this case volatile isn't actively harmful, just not needed How to repeat: code analysis Suggested fix: apply my patch (attached)