| Method from org.apache.jk.common.ChannelJni Detail: |
public int flush(Msg msg,
MsgContext ep) throws IOException {
ep.setNote( receivedNote, null );
return OK;
}
|
public String getChannelName() {
return getName();
}
|
public void init() throws IOException {
super.initNative("channel.jni:jni");
if( apr==null ) return;
// We'll be called from C. This deals with that.
apr.addJkHandler( "channelJni", this );
log.info("JK: listening on channel.jni:jni" );
if( next==null ) {
if( nextName!=null )
setNext( wEnv.getHandler( nextName ) );
if( next==null )
next=wEnv.getHandler( "dispatch" );
if( next==null )
next=wEnv.getHandler( "request" );
if( log.isDebugEnabled() )
log.debug("Setting default next " + next.getClass().getName());
}
}
|
public int invoke(Msg msg,
MsgContext ep) throws IOException {
if( apr==null ) return -1;
long xEnv=ep.getJniEnv();
long cEndpointP=ep.getJniContext();
int type=ep.getType();
if( log.isDebugEnabled() ) log.debug("ChannelJni.invoke: " + ep + " " + type);
switch( type ) {
case JkHandler.HANDLE_RECEIVE_PACKET:
return receive( msg, ep );
case JkHandler.HANDLE_SEND_PACKET:
return send( msg, ep );
case JkHandler.HANDLE_FLUSH:
return flush(msg, ep);
}
// Reset receivedNote. It'll be visible only after a SEND and before a receive.
ep.setNote( receivedNote, null );
// Default is FORWARD - called from C
try {
// first, we need to get an endpoint. It should be
// per/thread - and probably stored by the C side.
if( log.isDebugEnabled() ) log.debug("Received request " + xEnv);
// The endpoint will store the message pt.
msg.processHeader();
if( log.isTraceEnabled() ) msg.dump("Incoming msg ");
int status= next.invoke( msg, ep );
if( log.isDebugEnabled() ) log.debug("after processCallbacks " + status);
return status;
} catch( Exception ex ) {
ex.printStackTrace();
}
return 0;
}
Receive a packet from the C side. This is called from the C
code using invocation, but only for the first packet - to avoid
recursivity and thread problems.
This may look strange, but seems the best solution for the
problem ( the problem is that we don't have 'continuation' ).
sendPacket will move the thread execution on the C side, and
return when another packet is available. For packets that
are one way it'll return after it is processed too ( having
2 threads is far more expensive ).
Again, the goal is to be efficient and behave like all other
Channels ( so the rest of the code can be shared ). Playing with
java objects on C is extremely difficult to optimize and do
right ( IMHO ), so we'll try to keep it simple - byte[] passing,
the conversion done in java ( after we know the encoding and
if anyone asks for it - same lazy behavior as in 3.3 ). |
public boolean isSameAddress(MsgContext ep) {
return true;
}
|
public int receive(Msg msg,
MsgContext ep) throws IOException {
Msg sentResponse=(Msg)ep.getNote( receivedNote );
ep.setNote( receivedNote, null );
if( sentResponse == null ) {
if( log.isDebugEnabled() )
log.debug("No send() prior to receive(), no data buffer");
// No sent() was done prior to receive.
msg.reset();
msg.end();
sentResponse = msg;
}
sentResponse.processHeader();
if( log.isTraceEnabled() )
sentResponse.dump("received response ");
if( msg != sentResponse ) {
log.error( "Error, in JNI mode the msg used for receive() must be identical with the one used for send()");
}
return 0;
}
Receives does nothing - send will put the response
in the same buffer |
public void registerRequest(Request req,
MsgContext ep,
int count) {
// Not supported.
}
|
public int send(Msg msg,
MsgContext ep) throws IOException {
ep.setNote( receivedNote, null );
if( log.isDebugEnabled() ) log.debug("ChannelJni.send: " + msg );
int rc=super.nativeDispatch( msg, ep, JK_HANDLE_JNI_DISPATCH, 0);
// nativeDispatch will put the response in the same buffer.
// Next receive() will just get it from there. Very tricky to do
// things in one thread instead of 2.
ep.setNote( receivedNote, msg );
return rc;
}
Send the packet. XXX This will modify msg !!!
We could use 2 packets, or sendAndReceive(). |