Merge remote-tracking branch

'origin/GP-233_James_add_sleigh_filenames--SQUASHED'

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

@@ -15,7 +15,48 @@
  */
 #include "sleighbase.hh"
 
-const int4 SleighBase::SLA_FORMAT_VERSION = 2;
+const int4 SleighBase::SLA_FORMAT_VERSION = 3;
+
+int4 SourceFileIndexer::index(const string filename){
+	auto it = fileToIndex.find(filename);
+	if (fileToIndex.end() != it){
+		return it->second;
+	}
+	fileToIndex[filename] = leastUnusedIndex;
+	indexToFile[leastUnusedIndex] = filename;
+	return leastUnusedIndex++;
+}
+
+int4 SourceFileIndexer::getIndex(string filename){
+	return fileToIndex[filename];
+}
+
+string SourceFileIndexer::getFilename(int4 index){
+	return indexToFile[index];
+}
+
+void SourceFileIndexer::restoreXml(const Element *el){
+	const List &sourceFiles(el->getChildren());
+	List::const_iterator iter = sourceFiles.begin();
+	for (; iter != sourceFiles.end(); ++iter){
+		string filename = (*iter)->getAttributeValue("name");
+		int4 index = stoi((*iter)->getAttributeValue("index"),NULL,10);
+		fileToIndex[filename] = index;
+		indexToFile[index] = filename;
+	}
+}
+
+void SourceFileIndexer::saveXml(ostream& s) const {
+	s << "<sourcefiles>\n";
+	for (int4 i = 0; i < leastUnusedIndex; ++i){
+		s << ("<sourcefile name=\"");
+		const char *str = indexToFile.at(i).c_str();
+		xml_escape(s,str);
+		s << "\" index=\"" << dec << i << "\"/>\n";
+	}
+	s << "</sourcefiles>\n";
+}
+
 
 SleighBase::SleighBase(void)
 
@@ -155,7 +196,7 @@ void SleighBase::saveXml(ostream &s) const
   if (numSections != 0)
     a_v_u(s,"numsections",numSections);
   s << ">\n";
-
+  indexer.saveXml(s);
   s << "<spaces";
   a_v(s,"defaultspace",getDefaultCodeSpace()->getName());
   s << ">\n";
@@ -231,6 +272,8 @@ void SleighBase::restoreXml(const Element *el)
     floatformats.back().restoreXml(*iter);
     ++iter;
   }
+  indexer.restoreXml(*iter);
+  iter++;
   restoreXmlSpaces(*iter,this);
   iter++;
   symtab.restoreXml(*iter,this);

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

@@ -21,7 +21,35 @@
 #include "translate.hh"
 #include "slghsymbol.hh"
 
+/// \brief class for recording source file information for SLEIGH constructors.
+
+///
+/// A SLEIGH specification may contain many source files.  This class is
+/// used to associate each constructor in a SLEIGH language to the source
+/// file where it is defined. This information is useful when debugging
+/// SLEIGH specifications.  Sourcefiles are assigned a numeric index and
+/// the mapping from indices to filenames is written to the generated .sla
+/// file.  For each constructor, the data written to the .sla file includes
+/// the source file index.
+class SourceFileIndexer {
+public:
+    SourceFileIndexer() {leastUnusedIndex = 0;}
+    ~SourceFileIndexer(void) { }
+    ///Returns the index of the file.  If the file is not in the index it is added.
+    int4 index(const string filename);
+	int4 getIndex(const string);  ///< get the index of a file.  Error if the file is not in the index.
+	string getFilename(int4);     ///< get the filename corresponding to an index
+    void restoreXml(const Element *el); ///< read a stored index mapping from an XML file
+	void saveXml(ostream&) const;       ///< save the index mapping to an XML file
+
+private:
+	int4 leastUnusedIndex;  ///< one-up count for assigning indices to files
+	map<int4, string> indexToFile;  ///< map from indices to files
+	map<string, int4> fileToIndex;  ///< map from files to indices
+};
+
 /// \brief Common core of classes that read or write SLEIGH specification files natively.
+
 ///
 /// This class represents what's in common across the SLEIGH infrastructure between:
 ///   - Reading the various SLEIGH specification files
@@ -36,6 +64,7 @@ protected:
   uint4 maxdelayslotbytes;	///< Maximum number of bytes in a delay-slot directive
   uint4 unique_allocatemask;	///< Bits that are guaranteed to be zero in the unique allocation scheme
   uint4 numSections;		///< Number of \e named sections
+  SourceFileIndexer indexer;    ///< source file index used when generating SLEIGH constructor debug info
   void buildXrefs(vector<string> &errorPairs);	///< Build register map. Collect user-ops and context-fields.
   void reregisterContext(void);	///< Reregister context fields for a new executable
   void restoreXml(const Element *el);	///< Read a SLEIGH specification from XML

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

@@ -2502,6 +2502,8 @@ Constructor *SleighCompile::createConstructor(SubtableSymbol *sym)
   sym->addConstructor(curct);
   symtab.addScope();		// Make a new symbol scope for our constructor
   pcode.resetLabelCount();
+  int4 index = indexer.index(ctorLocationMap[curct].getFilename());
+  curct->setSrcIndex(index);
   return curct;
 }
 

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

@@ -1596,7 +1596,7 @@ void Constructor::saveXml(ostream &s) const
   s << " parent=\"0x" << hex << parent->getId() << "\"";
   s << " first=\"" << dec << firstwhitespace << "\"";
   s << " length=\"" << minimumlength << "\"";
-  s << " line=\"" << lineno << "\">\n";
+  s << " line=\"" << src_index << ":" << lineno << "\">\n";
   for(int4 i=0;i<operands.size();++i)
     s << "<oper id=\"0x" << hex << operands[i]->getId() << "\"/>\n";
   for(int4 i=0;i<printpiece.size();++i) {
@@ -1643,9 +1643,10 @@ void Constructor::restoreXml(const Element *el,SleighBase *trans)
     s >> minimumlength;
   }
   {
-    istringstream s(el->getAttributeValue("line"));
-    s.unsetf(ios::dec | ios::hex | ios::oct);
-    s >> lineno;
+   	string src_and_line = el->getAttributeValue("line");
+    size_t pos = src_and_line.find(":");
+    src_index = stoi(src_and_line.substr(0, pos),NULL,10);
+    lineno = stoi(src_and_line.substr(pos+1,src_and_line.length()),NULL,10);
   }
   const List &list(el->getChildren());
   List::const_iterator iter;

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

@@ -477,6 +477,7 @@ class Constructor {		// This is NOT a symbol
   int4 firstwhitespace;		// Index of first whitespace piece in -printpiece-
   int4 flowthruindex;		// if >=0 then print only a single operand no markup
   int4 lineno;
+  int4 src_index;           //source file index
   mutable bool inerror;                 // An error is associated with this Constructor
   void orderOperands(void);
 public:
@@ -491,6 +492,8 @@ public:
   uintm getId(void) const { return id; }
   void setLineno(int4 ln) { lineno = ln; }
   int4 getLineno(void) const { return lineno; }
+  void setSrcIndex(int4 index) {src_index = index;}
+  int4 getSrcIndex(void) {return src_index;}
   void addContext(const vector<ContextChange *> &vec) { context = vec; }
   void addOperand(OperandSymbol *sym);
   void addInvisibleOperand(OperandSymbol *sym);