mem: Make returning snoop responses occupy response layer

This patch introduces a mirrored internal snoop port to facilitate
easy addition of flow control for the snoop responses that are turned
into normal responses on their return. To perform this, the slave
ports of the coherent bus are wrapped in internal master ports that
are passed as the source ports to the response layer in question.

As a result of this patch, there is more contention for the response
resources, and as such system performance will decrease slightly.

A consequence of the mirrored internal port is that the port the bus
tells to retry (the internal one) and the port actually retrying (the
mirrored) one are not the same. Thus, the existing check in tryTiming
is not longer correct. In fact, the test is redundant as the layer is
only in the retry state while calling sendRetry on the waiting port,
and if the latter does not immediately call the bus then the retry
state is left. Consequently the check is removed.

1 files changed, 14 insertions, 6 deletions

diff --git a/src/mem/bus.cc b/src/mem/bus.cc
index 855f82db4..d94bd6a28 100644
--- a/src/mem/bus.cc
+++ b/src/mem/bus.cc
@@ -189,12 +189,20 @@ template <typename SrcType, typename DstType>
 bool
 BaseBus::Layer<SrcType,DstType>::tryTiming(SrcType* src_port)
 {
-    // first we see if the layer is busy, next we check if we are in a
-    // retry with a port other than the current one, lastly we check
-    // if the destination port is already engaged in a transaction
-    // waiting for a retry from the peer
-    if (state == BUSY || (state == RETRY && src_port != retryingPort) ||
-        waitingForPeer != NULL) {
+    // if we are in the retry state, we will not see anything but the
+    // retrying port (or in the case of the snoop ports the snoop
+    // response port that mirrors the actual slave port) as we leave
+    // this state again in zero time if the peer does not immediately
+    // call the bus when receiving the retry
+
+    // first we see if the layer is busy, next we check if the
+    // destination port is already engaged in a transaction waiting
+    // for a retry from the peer
+    if (state == BUSY || waitingForPeer != NULL) {
+        // the port should not be waiting already
+        assert(std::find(waitingForLayer.begin(), waitingForLayer.end(),
+                         src_port) == waitingForLayer.end());
+
         // put the port at the end of the retry list waiting for the
         // layer to be freed up (and in the case of a busy peer, for
         // that transaction to go through, and then the bus to free