sahvx655-wq commented on PR #310: URL: https://github.com/apache/calcite-avatica/pull/310#issuecomment-4830063188
Sure, I'll raise it on the Apache JIRA and put the CALCITE key in the PR title so it's tracked properly. For the writeup I'll describe what I found while tracing the server parse path: CursorFactory.fromProto resolves the wire-supplied className with the single-arg Class.forName, which loads and initialises the class in one go. That path is reachable on the server straight from an untrusted ExecuteRequest, through StatementHandle.fromProto then Signature.fromProto then CursorFactory.fromProto, so a client can name any class on the server's classpath and have its static initialiser run at request-parse time. Left unfixed, the concern is initialisation side effects firing before the class is ever legitimately used, which is the usual CWE-470 unsafe class loading case. The change keeps the three-arg Class.forName with initialize=false so the class is still linked and resolved, and genuine initialisation still happens lazily when a cursor actually gets built. I'll post the JIRA link back here once it's raised. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
