BugID: 447

Two faults in the source code of the GCFCommon are the reason of the segmentation fault in the KeyValueLoggerMaster.
As expected the seg. fault occours at unpacking of the GCFPVDynArr. This
is solved now.

MAC/GCF/GCFCommon/src/GCF_PVDynArr.cc

@@ -69,7 +69,7 @@ unsigned int GCFPVDynArr::unpackConcrete(const char* valBuf)
 
   for (unsigned int i = 0; i < arraySize; i++)
   {
-    pNewValue = GCFPValue::createMACTypeObject((TMACValueType) (getType() | LPT_DYNARR));
+    pNewValue = GCFPValue::createMACTypeObject((TMACValueType) (getType() & ~LPT_DYNARR));
     
     pNewValue->setDataFormat(_dataFormat);
     

MAC/GCF/GCFCommon/src/GCF_PValue.cc

@@ -96,7 +96,7 @@ GCFPValue* GCFPValue::createMACTypeObject(TMACValueType type)
 
 GCFPValue* GCFPValue::unpackValue(const char* valBuf)
 {
-  GCFPValue* pValue = createMACTypeObject((TMACValueType) *valBuf);
+  GCFPValue* pValue = createMACTypeObject((TMACValueType) (uchar) *valBuf);
   if (pValue)
   {
     unsigned int readLength = pValue->unpack(valBuf);
@@ -111,7 +111,7 @@ GCFPValue* GCFPValue::unpackValue(const char* valBuf)
 
 unsigned int GCFPValue::unpack(const char* valBuf)
 {
-  ASSERT(_type == (TMACValueType) *valBuf);
+  ASSERT(_type == (TMACValueType) (uchar) *valBuf);
   _dataFormat = (LOFAR::DataFormat) valBuf[1];
   // the type was already set, because it was set on construction of this class
   // later maybe also a timestamp can be assigned to a value.