Isolate property on Symbols

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

@@ -232,6 +232,21 @@ bool Symbol::isNameUndefined(void) const
   return ((name.size()==15)&&(0==name.compare(0,7,"$$undef")));
 }
 
+/// If the given value is \b true, any Varnodes that map directly to \b this Symbol,
+/// will not be speculatively merged with other Varnodes.  (Required merges will still happen).
+/// \param val is the given boolean value
+void Symbol::setIsolated(bool val)
+
+{
+  if (val) {
+    dispflags |= isolate;
+    flags |= Varnode::typelock;		// Isolated Symbol must be typelocked
+    checkSizeTypeLock();
+  }
+  else
+    dispflags &= ~((uint4)isolate);
+}
+
 /// \return the first SymbolEntry
 SymbolEntry *Symbol::getFirstWholeMap(void) const
 
@@ -262,6 +277,24 @@ SymbolEntry *Symbol::getMapEntry(const Address &addr) const
   return (SymbolEntry *)0;
 }
 
+/// Among all the SymbolEntrys that map \b this entire Symbol, calculate
+/// the position of the given SymbolEntry within the list.
+/// \param entry is the given SymbolEntry
+/// \return its position within the list or -1 if it is not in the list
+int4 Symbol::getMapEntryPosition(const SymbolEntry *entry) const
+
+{
+  int4 pos = 0;
+  for(int4 i=0;i<mapentry.size();++i) {
+    const SymbolEntry *tmp = &(*mapentry[i]);
+    if (tmp == entry)
+      return pos;
+    if (entry->getSize() == type->getSize())
+      pos += 1;
+  }
+  return -1;
+}
+
 /// A value of 0 means the base Symbol name is visible and not overridden in the given use scope.
 /// A value of 1 means the base name may be overridden, but the parent scope name is not.
 /// The minimual number of names that distinguishes \b this Symbol uniquely within the
@@ -313,18 +346,20 @@ void Symbol::saveXmlHeader(ostream &s) const
     a_v_b(s,"indirectstorage",true);
   if ((flags&Varnode::hiddenretparm)!=0)
     a_v_b(s,"hiddenretparm",true);
+  if ((dispflags&isolate)!=0)
+    a_v_b(s,"merge",false);
   int4 format = getDisplayFormat();
   if (format != 0) {
     s << " format=\"";
-    if (format == Symbol::force_hex)
+    if (format == force_hex)
       s << "hex\"";
-    else if (format == Symbol::force_dec)
+    else if (format == force_dec)
       s << "dec\"";
-    else if (format == Symbol::force_char)
+    else if (format == force_char)
       s << "char\"";
-    else if (format == Symbol::force_oct)
+    else if (format == force_oct)
       s << "oct\"";
-    else if (format == Symbol::force_bin)
+    else if (format == force_bin)
       s << "bin\"";
     else
       s << "hex\"";
@@ -355,15 +390,15 @@ void Symbol::restoreXmlHeader(const Element *el)
 	if (attName == "format") {
 	  const string &formString(el->getAttributeValue(i));
 	  if (formString == "hex")
-	    dispflags |= Symbol::force_hex;
+	    dispflags |= force_hex;
 	  else if (formString == "dec")
-	    dispflags |= Symbol::force_dec;
+	    dispflags |= force_dec;
 	  else if (formString == "char")
-	    dispflags |= Symbol::force_char;
+	    dispflags |= force_char;
 	  else if (formString == "oct")
-	    dispflags |= Symbol::force_oct;
+	    dispflags |= force_oct;
 	  else if (formString == "bin")
-	    dispflags |= Symbol::force_bin;
+	    dispflags |= force_bin;
 	}
 	break;
       case 'h':
@@ -385,6 +420,14 @@ void Symbol::restoreXmlHeader(const Element *el)
 	    flags |= Varnode::indirectstorage;
 	}
 	break;
+      case 'm':
+	if (attName == "merge") {
+	  if (!xml_readbool(el->getAttributeValue(i))) {
+	    dispflags |= isolate;
+	    flags |= Varnode::typelock;
+	  }
+	}
+	break;
       case 'n':
 	if (attName == "name")
 	  name = el->getAttributeValue(i);

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

@@ -181,7 +181,8 @@ public:
     force_bin = 4,		///< Force binary printing of constant symbol
     force_char = 5,		///< Force integer to be printed as a character constant
     size_typelock = 8,	        ///< Only the size of the symbol is typelocked
-    isolate = 16		///< Symbol should not speculatively merge automatically
+    isolate = 16,		///< Symbol should not speculatively merge automatically
+    merge_problems = 32		///< Set if some SymbolEntrys did not get merged
   };
   /// \brief Construct given a name and data-type
   Symbol(Scope *sc,const string &nm,Datatype *ct)
@@ -203,11 +204,17 @@ public:
   bool isIndirectStorage(void) const { return ((flags&Varnode::indirectstorage)!=0); }	///< Is storage really a pointer to the true Symbol
   bool isHiddenReturn(void) const { return ((flags&Varnode::hiddenretparm)!=0); }	///< Is this a reference to the function return value
   bool isNameUndefined(void) const;				///< Does \b this have an undefined name
+  bool isMultiEntry(void) const { return (wholeCount > 1); }	///< Does \b this have more than one \e entire mapping
+  bool hasMergeProblems(void) const { return ((dispflags & merge_problems)!=0); } ///< Were some SymbolEntrys not merged
+  void setMergeProblems(void) { dispflags |= merge_problems; }	///< Mark that some SymbolEntrys could not be merged
+  bool isIsolated(void) const { return ((dispflags & isolate)!=0); }	///< Return \b true if \b this is isolated from speculative merging
+  void setIsolated(bool val);					///< Set whether \b this Symbol should be speculatively merged
   Scope *getScope(void) const { return scope; }			///< Get the scope owning \b this Symbol
   SymbolEntry *getFirstWholeMap(void) const;	 		///< Get the first entire mapping of the symbol
   SymbolEntry *getMapEntry(const Address &addr) const;	 	///< Get first mapping of the symbol that contains the given Address
   int4 numEntries(void) const { return mapentry.size(); }	///< Return the number of SymbolEntrys
   SymbolEntry *getMapEntry(int4 i) const { return &(*mapentry[i]); }	///< Return the i-th SymbolEntry for \b this Symbol
+  int4 getMapEntryPosition(const SymbolEntry *entry) const;	///< Position of given SymbolEntry within \b this multi-entry Symbol
   int4 getResolutionDepth(const Scope *useScope) const;		///< Get the number of scope names to print to resolve symbol in given context
   void saveXmlHeader(ostream &s) const;				///< Save basic Symbol properties as XML attributes
   void restoreXmlHeader(const Element *el);			///< Restore basic Symbol properties from XML

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

@@ -115,6 +115,7 @@ void IfaceDecompCapability::registerCommands(IfaceStatus *status)
   status->registerCom(new IfcRename(),"rename");
   status->registerCom(new IfcRetype(),"retype");
   status->registerCom(new IfcRemove(),"remove");
+  status->registerCom(new IfcIsolate(),"isolate");
   status->registerCom(new IfcLockPrototype(),"prototype","lock");
   status->registerCom(new IfcUnlockPrototype(),"prototype","unlock");
   status->registerCom(new IfcCommentInstr(),"comment","instruction");
@@ -1057,6 +1058,32 @@ void IfcRetype::execute(istream &s)
   }
 }
   
+void IfcIsolate::execute(istream &s)
+
+{
+  string name;
+
+  s >> ws >> name;
+  if (name.size()==0)
+    throw IfaceParseError("Must specify name of symbol");
+
+  Symbol *sym;
+  vector<Symbol *> symList;
+  if (dcp->fd != (Funcdata *)0)
+    dcp->fd->getScopeLocal()->queryByName(name,symList);
+  else
+    dcp->conf->symboltab->getGlobalScope()->queryByName(name,symList);
+
+  if (symList.empty())
+    throw IfaceExecutionError("No symbol named: "+name);
+  if (symList.size() > 1)
+    throw IfaceExecutionError("More than one symbol named : "+name);
+  else
+    sym = symList[0];
+
+  sym->setIsolated(true);
+}
+
 static Varnode *iface_read_varnode(IfaceDecompData *dcp,istream &s)
 
 {				// Return varnode identified by input stream

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

@@ -304,6 +304,11 @@ public:
   virtual void execute(istream &s);
 };
 
+class IfcIsolate : public IfaceDecompCommand {
+public:
+  virtual void execute(istream &s);
+};
+
 class IfcRemove : public IfaceDecompCommand {
 public:
   virtual void execute(istream &s);

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

@@ -102,14 +102,14 @@ bool Merge::mergeTestRequired(HighVariable *high_out,HighVariable *high_in)
   }
   else if (high_out->isExtraOut())
     return false;
-  if (high_in->isMapped() && high_out->isMapped()) {
+
-    Symbol *symbolIn = high_in->getSymbol();
+  Symbol *symbolIn = high_in->getSymbol();
-    Symbol *symbolOut = high_out->getSymbol();
+  Symbol *symbolOut = high_out->getSymbol();
-    if (symbolIn != (Symbol *)0 && symbolOut != (Symbol *)0) {
+  if (symbolIn != (Symbol *) 0 && symbolOut != (Symbol *) 0) {
-      if (symbolIn != symbolOut) return false;		// Map to different symbols
+    if (symbolIn != symbolOut)
-      if (high_in->getSymbolOffset() != high_out->getSymbolOffset())
+      return false;		// Map to different symbols
-	return false;					// Map to different parts of same symbol
+    if (high_in->getSymbolOffset() != high_out->getSymbolOffset())
-    }
+      return false;			// Map to different parts of same symbol
   }
 
   return true;
@@ -145,6 +145,14 @@ bool Merge::mergeTestAdjacent(HighVariable *high_out,HighVariable *high_in)
     Varnode *vn = high_in->getInputVarnode();
     if (vn->isIllegalInput()&&(!vn->isIndirectOnly())) return false;
   }
+  Symbol *symbol = high_in->getSymbol();
+  if (symbol != (Symbol *)0)
+    if (symbol->isIsolated())
+      return false;
+  symbol = high_out->getSymbol();
+  if (symbol != (Symbol *)0)
+    if (symbol->isIsolated())
+      return false;
   return true;
 }
 
@@ -837,10 +845,14 @@ void Merge::mergeMultiEntry(void)
       if (newHigh == high) continue;		// Varnodes already merged
       updateHigh(newHigh);
       if (!mergeTestRequired(high, newHigh)) {
+	symbol->setMergeProblems();
+	newHigh->setUnmerged();
 	conflictCount += 1;
 	continue;
       }
       if (!merge(high,newHigh,false)) {		// Attempt the merge
+	symbol->setMergeProblems();
+	newHigh->setUnmerged();
 	conflictCount += 1;
 	continue;
       }

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

@@ -1639,6 +1639,21 @@ void PrintC::pushSymbol(const Symbol *sym,const Varnode *vn,const PcodeOp *op)
   else
     tokenColor = EmitXml::var_color;
   // FIXME: resolve scopes
+  if (sym->hasMergeProblems() && vn != (Varnode *)0) {
+    HighVariable *high = vn->getHigh();
+    if (high->isUnmerged()) {
+      ostringstream s;
+      s << sym->getName();
+      SymbolEntry *entry = high->getSymbolEntry();
+      if (entry != (SymbolEntry *)0) {
+	s << '$' << dec << entry->getSymbol()->getMapEntryPosition(entry);
+      }
+      else
+	s << "$$";
+      pushAtom(Atom(s.str(),vartoken,tokenColor,op,vn));
+      return;
+    }
+  }
   pushAtom(Atom(sym->getName(),vartoken,tokenColor,op,vn));
 }
 
@@ -2180,14 +2195,19 @@ bool PrintC::emitScopeVarDecls(const Scope *scope,int4 cat)
   MapIterator iter = scope->begin();
   MapIterator enditer = scope->end();
   for(;iter!=enditer;++iter) {
-    if ((*iter)->isPiece()) continue; // Don't do a partial entry
+    const SymbolEntry *entry = *iter;
-    Symbol *sym = (*iter)->getSymbol();
+    if (entry->isPiece()) continue; // Don't do a partial entry
+    Symbol *sym = entry->getSymbol();
     if (sym->getCategory() != cat) continue;
     if (sym->getName().size() == 0) continue;
     if (dynamic_cast<FunctionSymbol *>(sym) != (FunctionSymbol *)0)
       continue;
     if (dynamic_cast<LabSymbol *>(sym) != (LabSymbol *)0)
       continue;
+    if (sym->isMultiEntry()) {
+      if (sym->getFirstWholeMap() != entry)
+	continue;		// Only emit the first SymbolEntry for declaration of multi-entry Symbol
+    }
     notempty = true;
     emitVarDeclStatement(sym);
   }

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

@@ -262,6 +262,20 @@ void HighVariable::remove(Varnode *vn)
   }
 }
 
+/// Assuming there is a Symbol attached to \b this, run through the Varnode members
+/// until we find one with a SymbolEntry corresponding to the Symbol and return it.
+/// \return the SymbolEntry that mapped the Symbol to \b this or null if no Symbol is attached
+SymbolEntry *HighVariable::getSymbolEntry(void) const
+
+{
+  for(int4 i=0;i<inst.size();++i) {
+    SymbolEntry *entry = inst[i]->getSymbolEntry();
+    if (entry != (SymbolEntry *)0 && entry->getSymbol() == symbol)
+      return entry;
+  }
+  return (SymbolEntry *)0;
+}
+
 /// The data-type its dirtying mechanism is disabled.  The data-type will not change, unless
 /// this method is called again.
 /// \param tp is the data-type to set

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

@@ -49,7 +49,8 @@ public:
     symboldirty = 8,		///< The symbol attachment is dirty
     copy_in1 = 8,		///< There exists at least 1 COPY into \b this HighVariable from other HighVariables
     copy_in2 = 16,		///< There exists at least 2 COPYs into \b this HighVariable from other HighVariables
-    type_finalized = 32		///< Set if a final data-type is locked in and dirtying is disabled
+    type_finalized = 32,	///< Set if a final data-type is locked in and dirtying is disabled
+    unmerged = 64		///< Set if part of a multi-entry Symbol but did not get merged with other SymbolEntrys
   };
 private:
   friend class Varnode;
@@ -79,10 +80,12 @@ private:
   void flagsDirty(void) const { highflags |= HighVariable::flagsdirty; }	///< Mark the boolean properties as \e dirty
   void coverDirty(void) const { highflags |= HighVariable::coverdirty; }	///< Mark the cover as \e dirty
   void typeDirty(void) const { highflags |= HighVariable::typedirty; }		///< Mark the data-type as \e dirty
+  void setUnmerged(void) const { highflags |= unmerged; }	///< Mark \b this as having merge problems
 public:
   HighVariable(Varnode *vn);		///< Construct a HighVariable with a single member Varnode
   Datatype *getType(void) const { updateType(); return type; }	///< Get the data-type
-  Symbol *getSymbol(void) const { updateSymbol(); return symbol; }	///< Get the Symbol associated with \b this
+  Symbol *getSymbol(void) const { updateSymbol(); return symbol; }	///< Get the Symbol associated with \b this or null
+  SymbolEntry *getSymbolEntry(void) const;			/// Get the SymbolEntry mapping to \b this or null
   int4 getSymbolOffset(void) const { return symboloffset; }	///< Get the Symbol offset associated with \b this
   int4 numInstances(void) const { return inst.size(); }		///< Get the number of member Varnodes \b this has
   Varnode *getInstance(int4 i) const { return inst[i]; }	///< Get the i-th member Varnode
@@ -111,6 +114,7 @@ public:
   void setMark(void) const { flags |= Varnode::mark; }		///< Set the mark on this variable
   void clearMark(void) const { flags &= ~Varnode::mark; }	///< Clear the mark on this variable
   bool isMark(void) const { return ((flags&Varnode::mark)!=0); }	///< Return \b true if \b this is marked
+  bool isUnmerged(void) const { return ((highflags&unmerged)!=0); }	///< Return \b true if \b this has merge problems
 
   /// \brief Determine if \b this HighVariable has an associated cover.
   ///