mads1980 commented on issue #246: OpenSSLEngine improvements to guard against multiple shutdown() calls triggered by construction exception and finalize() later URL: https://github.com/apache/tomcat/pull/246#issuecomment-588286325 > Well, personally I don't approve until there's an explanation. > > * I still don't see what the atomic boolean brings OpenSSLContext has a very similar protection (this is were I got the idea for this solution), but it uses AtomicInteger instead of AtomicBoolean (not sure why, since AtomicBoolean is more semantically "readable", and both implementations internally use a volatile int for storage, so memory usage is the same). > all relevant code is already synchronized and the check on the boolean flag should be identical. The fact that shutdown() is synchronized does not prevent it from executing concurrently with a failed constructor. There is no synchronization within the constructor, while shutdown() is being concurrently execute as a result from finalize() being invoked by the JVM GC concurrently with the failed constructor. > I suppose the actual "fix" is the check if (networkBIO != 0). Actually the HotStop error log shows the stack trace on freeSSL() but I guess that it does not hurt protecting both networkBIO and ssl from null pointer frees. > * "This can happen if there are uncaught exceptions within the OpenSSLEngine constructor, as finalization can execute concurrently with object construction": ok, so what is the uncaught exception ? This is not being captured by the HotSpot error logs, since a SIGSEGV crash occurs before any useful logging. However, it must be some kind of exception arising from the JNI calls within the OpenSSLEngine constructor.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
