Merge remote-tracking branch 'origin/GP-7126_caheckman_SmallFields'

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

@@ -6075,6 +6075,7 @@ void ActionDatabase::universalAction(Architecture *conf)
     actcleanup->addRule( new RulePtrsubCharConstant("cleanup") );
     actcleanup->addRule( new RuleExtensionPush("cleanup") );
     actcleanup->addRule( new RulePieceStructure("cleanup") );
+    actcleanup->addRule( new RuleAndStructure("cleanup") );
     actcleanup->addRule( new RuleSplitCopy("splitcopy") );
     actcleanup->addRule( new RuleSplitLoad("splitpointer") );
     actcleanup->addRule( new RuleSplitStore("splitpointer") );

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

@@ -7685,6 +7685,63 @@ int4 RulePieceStructure::applyOp(PcodeOp *op,Funcdata &data)
   return 1;
 }
 
+/// \class RuleAndStructure
+/// \brief Convert `and(x,#mask) => zext(sub(x,#c)` when the mask corresponds to a field access
+void RuleAndStructure::getOpList(vector<uint4> &oplist) const
+
+{
+  oplist.push_back(CPUI_INT_AND);
+}
+
+int4 RuleAndStructure::applyOp(PcodeOp *op,Funcdata &data)
+
+{
+  Varnode *cvn = op->getIn(1);
+  if (!cvn->isConstant()) return 0;
+  Varnode *vn = op->getIn(0);
+  Datatype *baseType = vn->getTypeReadFacing(op);
+  type_metatype meta = baseType->getMetatype();
+  if (meta != TYPE_STRUCT && meta != TYPE_PARTIALSTRUCT && meta != TYPE_ARRAY)
+    return 0;
+  uintb off = cvn->getOffset();
+  int4 count = sizeof(uintb)*8 - count_leading_zeros(off);
+  int4 size = count / 8;	// Number of bytes being masked off
+  if (count % 8 != 0)
+    size += 1;
+  int8 structOff = vn->getSpace()->isBigEndian() ? vn->getSize() - size : 0;
+  Datatype *ct = baseType->findSmallestContainer(structOff, size, &structOff);
+  if (ct->getSize() >= vn->getSize() || structOff != 0)
+    return 0;
+  PcodeOp *subOp = data.newOp(2, op->getAddr());
+  data.opSetOpcode(subOp, CPUI_SUBPIECE);
+  data.opSetInput(subOp,data.newConstant(4, 0),1);
+  data.opMarkSpecialPrint(subOp);	// Mark SUBPIECE as a field extraction
+  Address addr = op->getOut()->getAddr();
+  if (addr.isBigEndian())
+    addr = addr + (vn->getSize() - ct->getSize());
+  addr.renormalize(vn->getSize());		// Allow for possible join address
+  Varnode *outvn = data.newVarnodeOut(ct->getSize(), addr, subOp);
+  outvn->updateType(ct);
+  if (calc_mask(size) == off) {
+    data.opRemoveInput(op, 1);
+    data.opSetInput(op, outvn, 0);
+    data.opSetOpcode(op, CPUI_INT_ZEXT);
+    data.opSetInput(subOp,vn,0);
+    data.opInsertBefore(subOp, op);
+  }
+  else {
+    PcodeOp *newZext = data.newOp(1, op->getAddr());
+    data.opSetOpcode(newZext, CPUI_INT_ZEXT);
+    Varnode *outzext = data.newUniqueOut(vn->getSize(), newZext);
+    data.opSetInput(op, outzext, 0);
+    data.opSetInput(subOp, vn, 0);
+    data.opSetInput(newZext,outvn,0);
+    data.opInsertBefore(newZext,op);
+    data.opInsertBefore(subOp,newZext);
+  }
+  return 1;
+}
+
 /// \class RuleSubNormal
 /// \brief Pull-back SUBPIECE through INT_RIGHT and INT_SRIGHT
 ///

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

@@ -1212,6 +1212,17 @@ public:
   virtual int4 applyOp(PcodeOp *op,Funcdata &data);
 };
 
+class RuleAndStructure : public Rule {
+public:
+  RuleAndStructure(const string &g) : Rule( g, 0, "andstructure") {}		///< Constructor
+  virtual Rule *clone(const ActionGroupList &grouplist) const {
+    if (!grouplist.contains(getGroup())) return (Rule *)0;
+    return new RuleAndStructure(getGroup());
+  }
+  virtual void getOpList(vector<uint4> &oplist) const;
+  virtual int4 applyOp(PcodeOp *op,Funcdata &data);
+};
+
 class RuleSubNormal : public Rule {
 public:
   RuleSubNormal(const string &g) : Rule( g, 0, "subnormal") {}	///< Constructor

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

@@ -3032,7 +3032,22 @@ int4 RuleDumptyHumpLate::applyOp(PcodeOp *op,Funcdata &data)
   Varnode *vn = op->getIn(0);
   if (!vn->isWritten()) return 0;
   PcodeOp *pieceOp = vn->getDef();
-  if (pieceOp->code() != CPUI_PIECE) return 0;
+  OpCode opc = pieceOp->code();
+  if (opc == CPUI_SUBPIECE) {
+    // SUB(SUB(base,#c),#d)   =>  SUB(base,#c+#d)
+    Varnode *base = pieceOp->getIn(0);
+    data.opSetInput(op,base,0);
+    uintb trunc = op->getIn(1)->getOffset() + pieceOp->getIn(1)->getOffset();
+    if (trunc != op->getIn(1)->getOffset())
+      data.opSetInput(op,data.newConstant(4, trunc),1);
+    if (vn->hasNoDescend() && !vn->isAutoLive()) {
+      vector<PcodeOp *> scratch;
+      data.opDestroyRecursive(pieceOp, scratch);
+    }
+    return 1;
+  }
+  else if (opc != CPUI_PIECE)
+    return 0;
   Varnode *out = op->getOut();
   int4 outSize = out->getSize();
   int4 trunc = (int4)op->getIn(1)->getOffset();

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

@@ -360,6 +360,7 @@ public:
 /// Simplify expressions like:
 ///  - `sub( concat(V,W), 0)  =>  W`
 ///  - `sub( concat(V,W), c)  =>  V`
+///  - `sub( sub(V, c), d) =>  SUB(V,#c+#d)`
 ///
 /// preserving the data-types and removing the SUBPIECE and PIECE operations that are discarded.
 class RuleDumptyHumpLate : public Rule {

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

@@ -589,6 +589,29 @@ bool Datatype::testForArraySlack(int8 off)
   return nearestArrayedComponentBackward(off, comp);
 }
 
+/// Return the component data-type, which may be \b this if there is no smaller component, and pass back
+/// the relative offset of the start of the range into the returned component.
+/// \param off is the starting byte offset of the given range, within \b this
+/// \param sz is the number of bytes in the range
+/// \param newoff is used to pass back the relative offset into the returned component
+/// \return the component data-type contain the range
+Datatype *Datatype::findSmallestContainer(int8 off,int8 sz,int8 *newoff)
+
+{
+  Datatype *res = this;
+  Datatype *next = res;
+  int8 curOff = off;
+  for(;;) {
+    next = next->getSubType(curOff, &curOff);
+    if (next == (Datatype *)0) break;		// No smaller component
+    if (curOff + sz > next->getSize()) break;	// Next component down does not contain
+    res = next;
+    off = curOff;
+  }
+  *newoff = off;
+  return res;
+}
+
 /// Called only if the \b typedefImm field is non-null.  Encode the data-type to the
 /// stream as a simple \<typedef> element including only the names and ids of \b this and
 /// the data-type it typedefs.

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

@@ -312,6 +312,7 @@ public:
   bool isPrimitiveWhole(void) const;		///< Is \b this made up of a single primitive
   bool nearestArrayedComponent(int8 off,uint4 arrayHint,int8 *newoff) const;
   bool testForArraySlack(int8 off);	///< Test if an \e out-of-bounds offset makes sense as array slack
+  Datatype *findSmallestContainer(int8 off,int8 sz,int8 *newoff);	///< Find the smallest component containing the given range
 
   static uint4 encodeIntegerFormat(const string &val);
   static string decodeIntegerFormat(uint4 val);

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

@@ -308,8 +308,8 @@ void HighVariable::stripType(void) const
   if (meta == TYPE_PARTIALUNION || meta == TYPE_PARTIALSTRUCT) {
     if (symbol != (Symbol *)0 && symboloffset != -1) {	// If there is a bigger backing symbol
 	type_metatype submeta = symbol->getType()->getMetatype();
-	if (submeta == TYPE_STRUCT || submeta == TYPE_UNION)
-	  return;			// Don't strip the partial union
+	if (submeta == TYPE_STRUCT || submeta == TYPE_UNION || submeta == TYPE_ARRAY)
+	  return;			// Don't strip the partial
     }
   }
   else if (type->isEnumType()) {