ruby: fix deadlock bug in banked array resource checks

The Ruby banked array resource checks (initiated from SLICC) did a check and
allocate at the same time. If a transition needs more than one resource, then
it might check/allocate resource #1, then fail to get resource #2. Another
transition might then try to get the same resources, but in reverse order.
Deadlock.

This patch separates resource checking and resource reservation into two
steps to avoid deadlock.

1 files changed, 11 insertions, 1 deletions

diff --git a/src/mem/ruby/structures/CacheMemory.cc b/src/mem/ruby/structures/CacheMemory.cc
index c802d8fb6..d08724cff 100644
--- a/src/mem/ruby/structures/CacheMemory.cc
+++ b/src/mem/ruby/structures/CacheMemory.cc
@@ -528,22 +528,32 @@ CacheMemory::regStats()
         ;
 }
 
+// assumption: SLICC generated files will only call this function
+// once **all** resources are granted
 void
-CacheMemory::recordRequestType(CacheRequestType requestType)
+CacheMemory::recordRequestType(CacheRequestType requestType, Address addr)
 {
     DPRINTF(RubyStats, "Recorded statistic: %s\n",
             CacheRequestType_to_string(requestType));
     switch(requestType) {
     case CacheRequestType_DataArrayRead:
+        if (m_resource_stalls)
+            dataArray.reserve(addressToCacheSet(addr));
         numDataArrayReads++;
         return;
     case CacheRequestType_DataArrayWrite:
+        if (m_resource_stalls)
+            dataArray.reserve(addressToCacheSet(addr));
         numDataArrayWrites++;
         return;
     case CacheRequestType_TagArrayRead:
+        if (m_resource_stalls)
+            tagArray.reserve(addressToCacheSet(addr));
         numTagArrayReads++;
         return;
     case CacheRequestType_TagArrayWrite:
+        if (m_resource_stalls)
+            tagArray.reserve(addressToCacheSet(addr));
         numTagArrayWrites++;
         return;
     default: