Variable length types

Ghidra/Features/Decompiler/src/decompile/cpp/coreaction.cc

@@ -4244,7 +4244,7 @@ Datatype *ActionInferTypes::propagateAddIn2Out(TypeFactory *typegrp,PcodeOp *op,
   int4 offset = propagateAddPointer(op,inslot);
   if (offset==-1) return op->getOut()->getTempType(); // Doesn't look like a good pointer add
   uintb uoffset = AddrSpace::addressToByte(offset,((TypePointer *)rettype)->getWordSize());
-  if (tstruct->getSize() > 0)
+  if (tstruct->getSize() > 0 && !tstruct->isVariableLength())
     uoffset = uoffset % tstruct->getSize();
   if (uoffset==0) {
     if (op->code() == CPUI_PTRSUB) // Go down at least one level
@@ -4419,7 +4419,7 @@ bool ActionInferTypes::propagateTypeEdge(TypeFactory *typegrp,PcodeOp *op,int4 i
     }
     else if (alttype->getMetatype()==TYPE_PTR) {
       newtype = ((TypePointer *)alttype)->getPtrTo();
-      if (newtype->getSize() != outvn->getTempType()->getSize()) // Size must be appropriate
+      if (newtype->getSize() != outvn->getTempType()->getSize() || newtype->isVariableLength()) // Size must be appropriate
 	newtype = outvn->getTempType();
     }
     else
@@ -4432,7 +4432,7 @@ bool ActionInferTypes::propagateTypeEdge(TypeFactory *typegrp,PcodeOp *op,int4 i
     }
     else if (alttype->getMetatype()==TYPE_PTR) {
       newtype = ((TypePointer *)alttype)->getPtrTo();
-      if (newtype->getSize() != outvn->getTempType()->getSize())
+      if (newtype->getSize() != outvn->getTempType()->getSize() || newtype->isVariableLength())
 	newtype = outvn->getTempType();
     }
     else

Ghidra/Features/Decompiler/src/decompile/cpp/database_ghidra.cc

@@ -38,7 +38,7 @@ ScopeGhidra::~ScopeGhidra(void)
 }
 
 /// The Ghidra client reports a \e namespace id associated with
-/// Symbol. Determine if a matching \e namespac Scope already exists in the cache and build
+/// Symbol. Determine if a matching \e namespace Scope already exists in the cache and build
 /// it if it isn't. This may mean creating a new \e namespace Scope.
 /// \param id is the ID associated with the Ghidra namespace
 /// \return the Scope matching the id.

Ghidra/Features/Decompiler/src/decompile/cpp/printc.cc

@@ -1670,21 +1670,7 @@ void PrintC::pushAnnotation(const Varnode *vn,const PcodeOp *op)
 void PrintC::pushSymbol(const Symbol *sym,const Varnode *vn,const PcodeOp *op)
 
 {
-  Datatype *ct = sym->getType();
   EmitXml::syntax_highlight tokenColor;
-  if (((sym->getFlags()&Varnode::readonly)!=0)&&(ct->getMetatype()==TYPE_ARRAY)) {
-    Datatype *subct = ((TypeArray *)ct)->getBase();
-    if (subct->isCharPrint()) {
-      SymbolEntry *entry = sym->getFirstWholeMap();
-      if (entry != (SymbolEntry *)0) {
-	ostringstream s;
-	if (printCharacterConstant(s,entry->getAddr(),subct)) {
-	  pushAtom(Atom(s.str(),vartoken,EmitXml::const_color,op,vn));
-	  return;
-	}
-      }
-    }
-  }
   if (sym->getScope()->isGlobal())
     tokenColor = EmitXml::global_color;
   else if (sym->getCategory() == 0)

Ghidra/Features/Decompiler/src/decompile/cpp/ruleaction.cc

@@ -5627,6 +5627,9 @@ AddTreeState::AddTreeState(Funcdata &d,PcodeOp *op,int4 slot)
   ptrsize = ptr->getSize();
   ptrmask = calc_mask(ptrsize);
   baseType = ct->getPtrTo();
+  if (baseType->isVariableLength())
+    size = 0;		// Open-ended size being pointed to, there will be no "multiples" component
+  else
     size = AddrSpace::byteToAddressInt(baseType->getSize(),ct->getWordSize());
   multsum = 0;		// Sums start out as zero
   nonmultsum = 0;

Ghidra/Features/Decompiler/src/decompile/cpp/ruleaction.hh

@@ -46,7 +46,7 @@ class AddTreeState {
   const TypePointer *ct;	///< The pointer data-type
   const Datatype *baseType;	///< The base data-type being pointed at
   int4 ptrsize;			///< Size of the pointer
-  int4 size;			///< Size of data-type being pointed to (in address units)
+  int4 size;			///< Size of data-type being pointed to (in address units) or 0 for open ended pointer
   uintb ptrmask;		///< Mask for modulo calculations in ptr space
   uintb offset;			///< Number of bytes we dig into the base data-type
   uintb correct;		///< Number of bytes being double counted

Ghidra/Features/Decompiler/src/decompile/cpp/type.cc

@@ -1092,15 +1092,18 @@ void TypeStruct::restoreXml(const Element *el,TypeFactory &typegrp)
 }
 
 /// Turn on the data-type's function prototype
+/// \param tfact is the factory that owns \b this
 /// \param model is the prototype model
 /// \param outtype is the return type of the prototype
 /// \param intypes is the list of input parameters
 /// \param dotdotdot is true if the prototype takes variable arguments
 /// \param voidtype is the reference "void" data-type
-void TypeCode::set(ProtoModel *model,
+void TypeCode::set(TypeFactory *tfact,ProtoModel *model,
 		    Datatype *outtype,const vector<Datatype *> &intypes,
 		    bool dotdotdot,Datatype *voidtype)
 {
+  factory = tfact;
+  flags |= variable_length;
   if (proto != (FuncProto *)0)
     delete proto;
   proto = new FuncProto();
@@ -1123,6 +1126,7 @@ TypeCode::TypeCode(const TypeCode &op) : Datatype(op)
 
 {
   proto = (FuncProto *)0;
+  factory = op.factory;
   if (op.proto != (FuncProto *)0) {
     proto = new FuncProto();
     proto->copy(*op.proto);
@@ -1133,6 +1137,7 @@ TypeCode::TypeCode(const string &nm) : Datatype(1,TYPE_CODE,nm)
 
 {
   proto = (FuncProto *)0;
+  factory = (TypeFactory *)0;
 }
 
 TypeCode::~TypeCode(void)
@@ -1204,6 +1209,14 @@ int4 TypeCode::compareBasic(const TypeCode *op) const
   return 2;			// Carry on with comparison of parameters
 }
 
+Datatype *TypeCode::getSubType(uintb off,uintb *newoff) const
+
+{
+  if (factory == (TypeFactory *)0) return (Datatype *)0;
+  *newoff = 0;
+  return factory->getBase(1, TYPE_CODE);	// Return code byte unattached to function prototype
+}
+
 int4 TypeCode::compare(const Datatype &op,int4 level) const
 
 {
@@ -1284,6 +1297,8 @@ void TypeCode::restoreXml(const Element *el,TypeFactory &typegrp)
   iter = list.begin();
   if (iter == list.end()) return; // No underlying prototype
   Architecture *glb = typegrp.getArch();
+  factory = &typegrp;
+  flags |= variable_length;
   proto = new FuncProto();
   proto->setInternal( glb->defaultfp, typegrp.getTypeVoid() );
   proto->restoreXml(*iter,glb);
@@ -2107,7 +2122,7 @@ TypeCode *TypeFactory::getTypeCode(ProtoModel *model,Datatype *outtype,
 				   bool dotdotdot)
 {
   TypeCode tc("");		// getFuncdata type with no name
-  tc.set(model,outtype,intypes,dotdotdot,getTypeVoid());
+  tc.set(this,model,outtype,intypes,dotdotdot,getTypeVoid());
   return (TypeCode *) findAdd(tc);
 }
 

Ghidra/Features/Decompiler/src/decompile/cpp/type.hh

@@ -342,7 +342,8 @@ class TypeCode : public Datatype {
 protected:
   friend class TypeFactory;
   FuncProto *proto;		///< If non-null, this describes the prototype of the underlying function
-  void set(ProtoModel *model,
+  TypeFactory *factory;		///< Factory owning \b this
+  void set(TypeFactory *tfact,ProtoModel *model,
 	   Datatype *outtype,const vector<Datatype *> &intypes,
 	   bool dotdotdot,Datatype *voidtype);	///< Establish a function pointer
   virtual void restoreXml(const Element *el,TypeFactory &typegrp);
@@ -354,6 +355,7 @@ public:
   void setProperties(bool isConstructor,bool isDestructor);	///< Set additional function properties
   virtual ~TypeCode(void);
   virtual void printRaw(ostream &s) const;
+  virtual Datatype *getSubType(uintb off,uintb *newoff) const;
   virtual int4 compare(const Datatype &op,int4 level) const;
   virtual int4 compareDependency(const Datatype &op) const;
   virtual Datatype *clone(void) const { return new TypeCode(*this); }