Merge remote-tracking branch 'origin/patch'

Ghidra/Features/Base/src/main/java/ghidra/app/util/bin/format/macho/commands/SymbolTableCommand.java

@@ -86,8 +86,8 @@ public class SymbolTableCommand extends LoadCommand {
 		// sort the entries by the index in the string table, so don't jump around reading
 		List<NList> sortedList = nlistList
 				.stream()
-				.sorted((o1, o2) -> Integer.valueOf(o1.getStringTableIndex())
+				.sorted(
-						.compareTo(Integer.valueOf(o2.getStringTableIndex())))
+					(o1, o2) -> Integer.compare(o1.getStringTableIndex(), o2.getStringTableIndex()))
 				.collect(Collectors.toList());
 
 		// initialize the NList strings from string table

Ghidra/Features/Base/src/main/java/ghidra/app/util/bin/format/macho/dyld/DyldCacheLocalSymbolsInfo.java

@@ -161,8 +161,8 @@ public class DyldCacheLocalSymbolsInfo implements StructConverter {
 			// sort the entries by the index in the string table, so don't jump around reading
 			List<NList> sortedList = nlistList
 					.stream()
-					.sorted((o1, o2) -> Integer.valueOf(o1.getStringTableIndex())
+					.sorted((o1, o2) -> Integer.compare(o1.getStringTableIndex(),
-							.compareTo(Integer.valueOf(o2.getStringTableIndex())))
+						o2.getStringTableIndex()))
 					.collect(Collectors.toList());
 
 			// initialize the NList strings from string table

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

@@ -1142,7 +1142,7 @@ Address SegmentedResolver::resolve(uintb val,int4 sz,const Address &point,uintb
 
 {
   int4 innersz = segop->getInnerSize();
-  if (sz <= innersz) { // If -sz- matches the inner size, consider the value a "near" pointer
+  if (sz >= 0 && sz <= innersz) { // If -sz- matches the inner size, consider the value a "near" pointer
   // In this case the address offset is not fully specified
   // we check if the rest is stored in a context variable
   // (as with near pointers)

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

@@ -490,9 +490,12 @@ AddrSpace *AddrSpaceManager::getSpaceByShortcut(char sc) const
 /// If there is a special resolver for the AddrSpace, this is invoked, otherwise
 /// basic wordsize conversion and wrapping is performed. If the address encoding is
 /// partial (as in a \e near pointer) and the full encoding can be recovered, it is passed back.
+/// The \e sz parameter indicates the number of bytes in constant and is used to determine if
+/// the constant is a partial or full pointer encoding. A value of -1 indicates the value is
+/// known to be a full encoding.
 /// \param spc is the space to generate the address from
 /// \param val is the constant encoding of the address
-/// \param sz is the size of the constant encoding
+/// \param sz is the size of the constant encoding (or -1)
 /// \param point is the context address (for recovering full encoding info if necessary)
 /// \param fullEncoding is used to pass back the recovered full encoding of the pointer
 /// \return the formal Address associated with the encoding

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

@@ -144,8 +144,12 @@ public:
   /// \brief The main resolver method.
   ///
   /// Given a native constant in a specific context, resolve what address is being referred to.
+  /// The constant can be a partially encoded pointer, in which case the full pointer encoding
+  /// is recovered as well as the address.  Whether or not a pointer is partially encoded or not
+  /// is determined by the \e sz parameter, indicating the number of bytes in the pointer. A value
+  /// of -1 here indicates that the pointer is known to be a full encoding.
   /// \param val is constant to be resolved to an address
-  /// \param sz is the size of \e val in context.
+  /// \param sz is the size of \e val in context (or -1).
   /// \param point is the address at which this constant is being used
   /// \param fullEncoding is used to hold the full pointer encoding if \b val is a partial encoding
   /// \return the resolved Address

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

@@ -1125,15 +1125,15 @@ Datatype *TypeSpacebase::getSubType(uintb off,uintb *newoff) const
 {
   Scope *scope = getMap();
   off = AddrSpace::byteToAddress(off, spaceid->getWordSize());	// Convert from byte offset to address unit
-  // It should always be the case that given offset represents a full encoding of the
+  // It should always be the case that the given offset represents a full encoding of the
-  // pointer, so the point of context is unused
+  // pointer, so the point of context is unused and the size is given as -1
   Address nullPoint;
   uintb fullEncoding;
-  Address addr = glb->resolveConstant(spaceid, off, spaceid->getAddrSize(), nullPoint, fullEncoding);
+  Address addr = glb->resolveConstant(spaceid, off, -1, nullPoint, fullEncoding);
   SymbolEntry *smallest;
 
-  // Assume symbol being referenced is address tied,
+  // Assume symbol being referenced is address tied so we use a null point of context
-  // so we use empty usepoint
+  // FIXME: A valid point of context may be necessary in the future
   smallest = scope->queryContainer(addr,1,nullPoint);
   
   if (smallest == (SymbolEntry *)0) {
@@ -1172,6 +1172,9 @@ Address TypeSpacebase::getAddress(uintb off,int4 sz,const Address &point) const
 
 {
   uintb fullEncoding;
+  // Currently a constant off of a global spacebase must be a full pointer encoding
+  if (localframe.isInvalid())
+    sz = -1;	// Set size to -1 to guarantee that full encoding recovery isn't launched
   return glb->resolveConstant(spaceid,off,sz,point,fullEncoding);
 }