2015-05-03 12:32 GMT+02:00 Felix Schumacher <
[email protected]>:
> The problem apears to be caused by this part of the patch:
>
> diff --git a/java/org/apache/tomcat/util/net/SecureNio2Channel.java
> b/java/org/apache/tomcat/util/net/SecureNio2Channel.java
> index ba16492..bbc18b0 100644
> --- a/java/org/apache/tomcat/util/net/SecureNio2Channel.java
> +++ b/java/org/apache/tomcat/util/net/SecureNio2Channel.java
> @@ -784,21 +753,15 @@ public class SecureNio2Channel extends Nio2Channel {
> handler.completed(Integer.valueOf(-1), attachment);
> return;
> }
> if (!handshakeComplete) {
> throw new
> IllegalStateException(sm.getString("channel.nio.ssl.incompleteHandshake"));
> }
>
> if (readPending) {
> throw new ReadPendingException();
> } else {
> readPending = true;
> }
> .
> - ReadCompletionHandler<A> readCompletionHandler = new
> ReadCompletionHandler<>(dst, handler);
> - if (netInBuffer.position() > 0 ) {
> - readCompletionHandler.completed(Integer.valueOf(netInBuffer.position()),
> attachment);
> } else {
> - sc.read(netInBuffer, timeout, unit, attachment,
> readCompletionHandler);
> + sc.read(netInBuffer, timeout, unit, attachment, new
> ReadCompletionHandler<>(dst, handler));
> }
> .
> @Override
>
> The rest of the patch will not influence the jmeter test.
>
> I suppose then netInBuffer.position() > 0 really does happen, and the
change was a bad idea. OTOH, I guess I added it for a reason originally,
but I thought it was no longer needed. I'll add back the code and test
again the testsuite.
Rémy
