[
https://issues.apache.org/jira/browse/CASSANDRA-4111?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vijay updated CASSANDRA-4111:
-----------------------------
Attachment: 0001-CASSANDRA-4111-v2.patch
ahaaa missed that.... V2 fixes it. Thanks!
> Serializing cache can cause Segfault in 1.1
> -------------------------------------------
>
> Key: CASSANDRA-4111
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4111
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.1.0
> Reporter: Vijay
> Assignee: Vijay
> Fix For: 1.1.0
>
> Attachments: 0001-CASSANDRA-4111-v2.patch, 0001-CASSANDRA-4111.patch
>
>
> Rare but this can happen per sure, looks like this issue is after
> CASSANDRA-3862 hence affectes only 1.1
> FreeableMemory old = map.get(key);
> if (old == null)
> return false;
> // see if the old value matches the one we want to replace
> FreeableMemory mem = serialize(value);
> if (mem == null)
> return false; // out of memory. never mind.
> V oldValue = deserialize(old);
> boolean success = oldValue.equals(oldToReplace) && map.replace(key,
> old, mem);
> if (success)
> old.unreference();
> else
> mem.unreference();
> return success;
> in the above code block we deserialize(old) without taking reference to the
> old memory, this can case seg faults when the old is reclaimed (free is
> called)
> Fix is to get the reference just for deserialization
> V oldValue;
> // reference old guy before de-serializing
> old.reference();
> try
> {
> oldValue = deserialize(old);
> }
> finally
> {
> old.unreference();
> }
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
