[x86] Allow segment and address-size overrides for LODS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199803 91177308-0d34-0410-b5e6-96231b3b80d8

17 files changed, 280 insertions, 45 deletions

diff --git a/lib/Target/X86/AsmParser/X86AsmParser.cpp b/lib/Target/X86/AsmParser/X86AsmParser.cpp
index 5d6048616b..f354c95c3a 100644
--- a/lib/Target/X86/AsmParser/X86AsmParser.cpp
+++ b/lib/Target/X86/AsmParser/X86AsmParser.cpp
@@ -551,6 +551,7 @@ private:
     return 0;
   }
 
+  X86Operand *DefaultMemSIOperand(SMLoc Loc);
   X86Operand *ParseOperand();
   X86Operand *ParseATTOperand();
   X86Operand *ParseIntelOperand();
@@ -922,6 +923,25 @@ struct X86Operand : public MCParsedAsmOperand {
       !getMemIndexReg() && getMemScale() == 1;
   }
 
+  bool isSrcIdx() const {
+    return !getMemIndexReg() && getMemScale() == 1 &&
+      (getMemBaseReg() == X86::RSI || getMemBaseReg() == X86::ESI ||
+       getMemBaseReg() == X86::SI) && isa<MCConstantExpr>(getMemDisp()) &&
+      cast<MCConstantExpr>(getMemDisp())->getValue() == 0;
+  }
+  bool isSrcIdx8() const {
+    return isMem8() && isSrcIdx();
+  }
+  bool isSrcIdx16() const {
+    return isMem16() && isSrcIdx();
+  }
+  bool isSrcIdx32() const {
+    return isMem32() && isSrcIdx();
+  }
+  bool isSrcIdx64() const {
+    return isMem64() && isSrcIdx();
+  }
+
   bool isMemOffs8() const {
     return Kind == Memory && !getMemBaseReg() &&
       !getMemIndexReg() && getMemScale() == 1 && (!Mem.Size || Mem.Size == 8);
@@ -1014,6 +1034,12 @@ struct X86Operand : public MCParsedAsmOperand {
       Inst.addOperand(MCOperand::CreateExpr(getMemDisp()));
   }
 
+  void addSrcIdxOperands(MCInst &Inst, unsigned N) const {
+    assert((N == 2) && "Invalid number of operands!");
+    Inst.addOperand(MCOperand::CreateReg(getMemBaseReg()));
+    Inst.addOperand(MCOperand::CreateReg(getMemSegReg()));
+  }
+
   void addMemOffsOperands(MCInst &Inst, unsigned N) const {
     assert((N == 2) && "Invalid number of operands!");
     // Add as immediates when possible.
@@ -1230,6 +1256,14 @@ bool X86AsmParser::ParseRegister(unsigned &RegNo,
   return false;
 }
 
+X86Operand *X86AsmParser::DefaultMemSIOperand(SMLoc Loc) {
+  unsigned basereg =
+    is64BitMode() ? X86::RSI : (is32BitMode() ? X86::ESI : X86::SI);
+  const MCExpr *Disp = MCConstantExpr::Create(0, getContext());
+  return X86Operand::CreateMem(/*SegReg=*/0, Disp, /*BaseReg=*/basereg,
+                               /*IndexReg=*/0, /*Scale=*/1, Loc, Loc, 0);
+}
+
 X86Operand *X86AsmParser::ParseOperand() {
   if (isParsingIntelSyntax())
     return ParseIntelOperand();
@@ -2278,36 +2312,14 @@ ParseInstruction(ParseInstructionInfo &Info, StringRef Name, SMLoc NameLoc,
       delete &Op2;
     }
   }
-  // Transform "lods[bwl] %ds:(%esi),{%al,%ax,%eax,%rax}" into "lods[bwl]"
-  if (Name.startswith("lods") && Operands.size() == 3 &&
+  // Transform "lods[bwlq]" into "lods[bwlq] ($SIREG)" for appropriate
+  // values of $SIREG according to the mode. It would be nice if this
+  // could be achieved with InstAlias in the tables.
+  if (Name.startswith("lods") && Operands.size() == 1 &&
       (Name == "lods" || Name == "lodsb" || Name == "lodsw" ||
-       Name == "lodsl" || (is64BitMode() && Name == "lodsq"))) {
-    X86Operand *Op1 = static_cast<X86Operand*>(Operands[1]);
-    X86Operand *Op2 = static_cast<X86Operand*>(Operands[2]);
-    if (isSrcOp(*Op1) && Op2->isReg()) {
-      const char *ins;
-      unsigned reg = Op2->getReg();
-      bool isLods = Name == "lods";
-      if (reg == X86::AL && (isLods || Name == "lodsb"))
-        ins = "lodsb";
-      else if (reg == X86::AX && (isLods || Name == "lodsw"))
-        ins = "lodsw";
-      else if (reg == X86::EAX && (isLods || Name == "lodsl"))
-        ins = "lodsl";
-      else if (reg == X86::RAX && (isLods || Name == "lodsq"))
-        ins = "lodsq";
-      else
-        ins = NULL;
-      if (ins != NULL) {
-        Operands.pop_back();
-        Operands.pop_back();
-        delete Op1;
-        delete Op2;
-        if (Name != ins)
-          static_cast<X86Operand*>(Operands[0])->setTokenValue(ins);
-      }
-    }
-  }
+       Name == "lodsl" || Name == "lodsd" || Name == "lodsq"))
+    Operands.push_back(DefaultMemSIOperand(NameLoc));
+
   // Transform "stos[bwl] {%al,%ax,%eax,%rax},%es:(%edi)" into "stos[bwl]"
   if (Name.startswith("stos") && Operands.size() == 3 &&
       (Name == "stos" || Name == "stosb" || Name == "stosw" ||
diff --git a/lib/Target/X86/Disassembler/X86Disassembler.cpp b/lib/Target/X86/Disassembler/X86Disassembler.cpp
index acfe88dd8f..dae1345f70 100644
--- a/lib/Target/X86/Disassembler/X86Disassembler.cpp
+++ b/lib/Target/X86/Disassembler/X86Disassembler.cpp
@@ -233,6 +233,29 @@ static const uint8_t segmentRegnums[SEG_OVERRIDE_max] = {
   X86::GS
 };
 
+/// translateSrcIndex   - Appends a source index operand to an MCInst.
+///
+/// @param mcInst       - The MCInst to append to.
+/// @param operand      - The operand, as stored in the descriptor table.
+/// @param insn         - The internal instruction.
+static bool translateSrcIndex(MCInst &mcInst, InternalInstruction &insn) {
+  unsigned baseRegNo;
+
+  if (insn.mode == MODE_64BIT)
+    baseRegNo = insn.prefixPresent[0x67] ? X86::ESI : X86::RSI;
+  else if (insn.mode == MODE_32BIT)
+    baseRegNo = insn.prefixPresent[0x67] ? X86::SI : X86::ESI;
+  else if (insn.mode == MODE_16BIT)
+    baseRegNo = insn.prefixPresent[0x67] ? X86::ESI : X86::SI;
+  MCOperand baseReg = MCOperand::CreateReg(baseRegNo);
+  mcInst.addOperand(baseReg);
+
+  MCOperand segmentReg;
+  segmentReg = MCOperand::CreateReg(segmentRegnums[insn.segmentOverride]);
+  mcInst.addOperand(segmentReg);
+  return false;
+}
+
 /// translateImmediate  - Appends an immediate operand to an MCInst.
 ///
 /// @param mcInst       - The MCInst to append to.
@@ -694,6 +717,8 @@ static bool translateOperand(MCInst &mcInst, const OperandSpecifier &operand,
                        insn,
                        Dis);
     return false;
+  case ENCODING_SI:
+    return translateSrcIndex(mcInst, insn);
   case ENCODING_RB:
   case ENCODING_RW:
   case ENCODING_RD:
diff --git a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c
index 48c1697713..8e591b8fdd 100644
--- a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c
+++ b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c
@@ -1682,6 +1682,7 @@ static int readOperands(struct InternalInstruction* insn) {
   for (index = 0; index < X86_MAX_OPERANDS; ++index) {
     switch (x86OperandSets[insn->spec->operands][index].encoding) {
     case ENCODING_NONE:
+    case ENCODING_SI:
       break;
     case ENCODING_REG:
     case ENCODING_RM:
diff --git a/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h b/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h
index 1acaef1b94..c7fcb0ca20 100644
--- a/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h
+++ b/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h
@@ -409,7 +409,8 @@ struct ContextDecision {
   ENUM_ENTRY(ENCODING_Rv,     "Register code of operand size added to the "    \
                               "opcode byte")                                   \
   ENUM_ENTRY(ENCODING_DUP,    "Duplicate of another operand; ID is encoded "   \
-                              "in type")
+                              "in type")                                       \
+  ENUM_ENTRY(ENCODING_SI,     "Source index; encoded in OpSize/Adsize prefix")
 
 #define ENUM_ENTRY(n, d) n,
   typedef enum {
@@ -460,6 +461,10 @@ struct ContextDecision {
   ENUM_ENTRY(TYPE_M16_16,     "2+2-byte (BOUND)")                              \
   ENUM_ENTRY(TYPE_M32_32,     "4+4-byte (BOUND)")                              \
   ENUM_ENTRY(TYPE_M16_64,     "2+8-byte (LIDT, LGDT)")                         \
+  ENUM_ENTRY(TYPE_SRCIDX8,    "1-byte memory at source index")                 \
+  ENUM_ENTRY(TYPE_SRCIDX16,   "2-byte memory at source index")                 \
+  ENUM_ENTRY(TYPE_SRCIDX32,   "4-byte memory at source index")                 \
+  ENUM_ENTRY(TYPE_SRCIDX64,   "8-byte memory at source index")                 \
   ENUM_ENTRY(TYPE_MOFFS8,     "1-byte memory offset (relative to segment "     \
                               "base)")                                         \
   ENUM_ENTRY(TYPE_MOFFS16,    "2-byte")                                        \
diff --git a/lib/Target/X86/InstPrinter/X86ATTInstPrinter.cpp b/lib/Target/X86/InstPrinter/X86ATTInstPrinter.cpp
index 11a9ada41e..4cd445765a 100644
--- a/lib/Target/X86/InstPrinter/X86ATTInstPrinter.cpp
+++ b/lib/Target/X86/InstPrinter/X86ATTInstPrinter.cpp
@@ -226,6 +226,25 @@ void X86ATTInstPrinter::printMemReference(const MCInst *MI, unsigned Op,
   O << markup(">");
 }
 
+void X86ATTInstPrinter::printSrcIdx(const MCInst *MI, unsigned Op,
+                                    raw_ostream &O) {
+  const MCOperand &SegReg = MI->getOperand(Op+1);
+
+  O << markup("<mem:");
+
+  // If this has a segment register, print it.
+  if (SegReg.getReg()) {
+    printOperand(MI, Op+1, O);
+    O << ':';
+  }
+
+  O << "(";
+  printOperand(MI, Op, O);
+  O << ")";
+
+  O << markup(">");
+}
+
 void X86ATTInstPrinter::printMemOffset(const MCInst *MI, unsigned Op,
                                        raw_ostream &O) {
   const MCOperand &DispSpec = MI->getOperand(Op);
diff --git a/lib/Target/X86/InstPrinter/X86ATTInstPrinter.h b/lib/Target/X86/InstPrinter/X86ATTInstPrinter.h
index 4dc4fe6eab..8c7c838ae4 100644
--- a/lib/Target/X86/InstPrinter/X86ATTInstPrinter.h
+++ b/lib/Target/X86/InstPrinter/X86ATTInstPrinter.h
@@ -42,6 +42,7 @@ public:
   void printSSECC(const MCInst *MI, unsigned Op, raw_ostream &OS);
   void printAVXCC(const MCInst *MI, unsigned Op, raw_ostream &OS);
   void printPCRelImm(const MCInst *MI, unsigned OpNo, raw_ostream &OS);
+  void printSrcIdx(const MCInst *MI, unsigned OpNo, raw_ostream &OS);
   void printMemOffset(const MCInst *MI, unsigned OpNo, raw_ostream &OS);
   void printRoundingControl(const MCInst *MI, unsigned Op, raw_ostream &OS);
 
@@ -89,6 +90,18 @@ public:
     printMemReference(MI, OpNo, O);
   }
 
+  void printSrcIdx8(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx16(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx32(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx64(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    printSrcIdx(MI, OpNo, O);
+  }
   void printMemOffs8(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
     printMemOffset(MI, OpNo, O);
   }
diff --git a/lib/Target/X86/InstPrinter/X86IntelInstPrinter.cpp b/lib/Target/X86/InstPrinter/X86IntelInstPrinter.cpp
index 59634f9ad8..8ea070b39f 100644
--- a/lib/Target/X86/InstPrinter/X86IntelInstPrinter.cpp
+++ b/lib/Target/X86/InstPrinter/X86IntelInstPrinter.cpp
@@ -212,6 +212,20 @@ void X86IntelInstPrinter::printMemReference(const MCInst *MI, unsigned Op,
   O << ']';
 }
 
+void X86IntelInstPrinter::printSrcIdx(const MCInst *MI, unsigned Op,
+                                      raw_ostream &O) {
+  const MCOperand &SegReg   = MI->getOperand(Op+1);
+
+  // If this has a segment register, print it.
+  if (SegReg.getReg()) {
+    printOperand(MI, Op+1, O);
+    O << ':';
+  }
+  O << '[';
+  printOperand(MI, Op, O);
+  O << ']';
+}
+
 void X86IntelInstPrinter::printMemOffset(const MCInst *MI, unsigned Op,
                                          raw_ostream &O) {
   const MCOperand &DispSpec = MI->getOperand(Op);
diff --git a/lib/Target/X86/InstPrinter/X86IntelInstPrinter.h b/lib/Target/X86/InstPrinter/X86IntelInstPrinter.h
index 90d4e39784..067df8129a 100644
--- a/lib/Target/X86/InstPrinter/X86IntelInstPrinter.h
+++ b/lib/Target/X86/InstPrinter/X86IntelInstPrinter.h
@@ -40,6 +40,7 @@ public:
   void printAVXCC(const MCInst *MI, unsigned Op, raw_ostream &O);
   void printPCRelImm(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printMemOffset(const MCInst *MI, unsigned OpNo, raw_ostream &O);
+  void printSrcIdx(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printRoundingControl(const MCInst *MI, unsigned Op, raw_ostream &OS);
 
   void printopaquemem(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
@@ -100,6 +101,23 @@ public:
     printMemReference(MI, OpNo, O);
   }
 
+
+  void printSrcIdx8(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    O << "byte ptr ";
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx16(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    O << "word ptr ";
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx32(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    O << "dword ptr ";
+    printSrcIdx(MI, OpNo, O);
+  }
+  void printSrcIdx64(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
+    O << "qword ptr ";
+    printSrcIdx(MI, OpNo, O);
+  }
   void printMemOffs8(const MCInst *MI, unsigned OpNo, raw_ostream &O) {
     O << "byte ptr ";
     printMemOffset(MI, OpNo, O);
diff --git a/lib/Target/X86/MCTargetDesc/X86BaseInfo.h b/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
index b25a35e321..1ca0bd656d 100644
--- a/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
+++ b/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
@@ -259,6 +259,10 @@ namespace X86II {
     /// memory offset as an immediate with a possible segment override.
     RawFrmMemOffs  = 7,
 
+    /// RawFrmSrc - This form is for instructions that use the source index
+    /// register SI/ESI/RSI with a possible segment override.
+    RawFrmSrc      = 8,
+
     /// MRM[0-7][rm] - These forms are used to represent instructions that use
     /// a Mod/RM byte, and use the middle field to hold extended opcode
     /// information.  In the intel manual these are represented as /0, /1, ...
@@ -612,6 +616,7 @@ namespace X86II {
     case X86II::RawFrmImm8:
     case X86II::RawFrmImm16:
     case X86II::RawFrmMemOffs:
+    case X86II::RawFrmSrc:
        return -1;
     case X86II::MRMDestMem:
       return 0;
diff --git a/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp b/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
index 813691b09e..ac72fce032 100644
--- a/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
+++ b/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
@@ -1317,6 +1317,19 @@ EncodeInstruction(const MCInst &MI, raw_ostream &OS,
     llvm_unreachable("Unknown FormMask value in X86MCCodeEmitter!");
   case X86II::Pseudo:
     llvm_unreachable("Pseudo instruction shouldn't be emitted");
+  case X86II::RawFrmSrc: {
+    unsigned siReg = MI.getOperand(0).getReg();
+    // Emit segment override opcode prefix as needed (not for %ds).
+    if (MI.getOperand(1).getReg() != X86::DS)
+      EmitSegmentOverridePrefix(CurByte, 1, MI, OS);
+    // Emit OpSize prefix as needed.
+    if ((!is32BitMode() && siReg == X86::ESI) ||
+        (is32BitMode() && siReg == X86::SI))
+      EmitByte(0x67, CurByte, OS);
+    CurOp += 2; // Consume operands.
+    EmitByte(BaseOpcode, CurByte, OS);
+    break;
+  }
   case X86II::RawFrm:
     EmitByte(BaseOpcode, CurByte, OS);
     break;
diff --git a/lib/Target/X86/X86InstrFormats.td b/lib/Target/X86/X86InstrFormats.td
index eed361958f..a239d1f523 100644
--- a/lib/Target/X86/X86InstrFormats.td
+++ b/lib/Target/X86/X86InstrFormats.td
@@ -22,6 +22,7 @@ def Pseudo     : Format<0>; def RawFrm     : Format<1>;
 def AddRegFrm  : Format<2>; def MRMDestReg : Format<3>;
 def MRMDestMem : Format<4>; def MRMSrcReg  : Format<5>;
 def MRMSrcMem  : Format<6>; def RawFrmMemOffs : Format<7>;
+def RawFrmSrc  : Format<8>;
 def MRM0r  : Format<16>; def MRM1r  : Format<17>; def MRM2r  : Format<18>;
 def MRM3r  : Format<19>; def MRM4r  : Format<20>; def MRM5r  : Format<21>;
 def MRM6r  : Format<22>; def MRM7r  : Format<23>;
diff --git a/lib/Target/X86/X86InstrInfo.td b/lib/Target/X86/X86InstrInfo.td
index 486435d35d..d3e102747e 100644
--- a/lib/Target/X86/X86InstrInfo.td
+++ b/lib/Target/X86/X86InstrInfo.td
@@ -445,6 +445,26 @@ def brtarget8 : Operand<OtherVT>;
 
 }
 
+def X86SrcIdx8Operand : AsmOperandClass {
+  let Name = "SrcIdx8";
+  let RenderMethod = "addSrcIdxOperands";
+  let SuperClasses = [X86Mem8AsmOperand];
+}
+def X86SrcIdx16Operand : AsmOperandClass {
+  let Name = "SrcIdx16";
+  let RenderMethod = "addSrcIdxOperands";
+  let SuperClasses = [X86Mem16AsmOperand];
+}
+def X86SrcIdx32Operand : AsmOperandClass {
+  let Name = "SrcIdx32";
+  let RenderMethod = "addSrcIdxOperands";
+  let SuperClasses = [X86Mem32AsmOperand];
+}
+def X86SrcIdx64Operand : AsmOperandClass {
+  let Name = "SrcIdx64";
+  let RenderMethod = "addSrcIdxOperands";
+  let SuperClasses = [X86Mem64AsmOperand];
+}
 def X86MemOffs8AsmOperand : AsmOperandClass {
   let Name = "MemOffs8";
   let RenderMethod = "addMemOffsOperands";
@@ -465,8 +485,23 @@ def X86MemOffs64AsmOperand : AsmOperandClass {
   let RenderMethod = "addMemOffsOperands";
   let SuperClasses = [X86Mem64AsmOperand];
 }
-
 let OperandType = "OPERAND_MEMORY" in {
+def srcidx8 : Operand<iPTR> {
+  let ParserMatchClass = X86SrcIdx8Operand;
+  let MIOperandInfo = (ops ptr_rc, i8imm);
+  let PrintMethod = "printSrcIdx8"; }
+def srcidx16 : Operand<iPTR> {
+  let ParserMatchClass = X86SrcIdx16Operand;
+  let MIOperandInfo = (ops ptr_rc, i8imm);
+  let PrintMethod = "printSrcIdx16"; }
+def srcidx32 : Operand<iPTR> {
+  let ParserMatchClass = X86SrcIdx32Operand;
+  let MIOperandInfo = (ops ptr_rc, i8imm);
+  let PrintMethod = "printSrcIdx32"; }
+def srcidx64 : Operand<iPTR> {
+  let ParserMatchClass = X86SrcIdx64Operand;
+  let MIOperandInfo = (ops ptr_rc, i8imm);
+  let PrintMethod = "printSrcIdx64"; }
 def offset8 : Operand<iPTR> {
   let ParserMatchClass = X86MemOffs8AsmOperand;
   let MIOperandInfo = (ops i64imm, i8imm);
@@ -1676,10 +1711,14 @@ def REPNE_PREFIX : I<0xF2, RawFrm, (outs),  (ins), "repne", []>;
 
 // String manipulation instructions
 let SchedRW = [WriteMicrocoded] in {
-def LODSB : I<0xAC, RawFrm, (outs), (ins), "lodsb", [], IIC_LODS>;
-def LODSW : I<0xAD, RawFrm, (outs), (ins), "lodsw", [], IIC_LODS>, OpSize;
-def LODSL : I<0xAD, RawFrm, (outs), (ins), "lods{l|d}", [], IIC_LODS>, OpSize16;
-def LODSQ : RI<0xAD, RawFrm, (outs), (ins), "lodsq", [], IIC_LODS>;
+def LODSB : I<0xAC, RawFrmSrc, (outs), (ins srcidx8:$src),
+              "lodsb\t{$src, %al|al, $src}", [], IIC_LODS>;
+def LODSW : I<0xAD, RawFrmSrc, (outs), (ins srcidx16:$src),
+              "lodsw\t{$src, %ax|ax, $src}", [], IIC_LODS>, OpSize;
+def LODSL : I<0xAD, RawFrmSrc, (outs), (ins srcidx32:$src),
+              "lods{l|d}\t{$src, %eax|eax, $src}", [], IIC_LODS>, OpSize16;
+def LODSQ : RI<0xAD, RawFrmSrc, (outs), (ins srcidx64:$src),
+               "lodsq\t{$src, %rax|rax, $src}", [], IIC_LODS>;
 }
 
 let SchedRW = [WriteSystem] in {
@@ -2329,6 +2368,18 @@ def : InstAlias<"clrw $reg", (XOR16rr GR16:$reg, GR16:$reg), 0>;
 def : InstAlias<"clrl $reg", (XOR32rr GR32:$reg, GR32:$reg), 0>;
 def : InstAlias<"clrq $reg", (XOR64rr GR64:$reg, GR64:$reg), 0>;
 
+// lods aliases. Accept the destination being omitted because it's implicit
+// in the mnemonic, or the mnemonic suffix being omitted because it's implicit
+// in the destination.
+def : InstAlias<"lodsb $src", (LODSB srcidx8:$src),  0>;
+def : InstAlias<"lodsw $src", (LODSW srcidx16:$src), 0>;
+def : InstAlias<"lods{l|d} $src", (LODSL srcidx32:$src), 0>;
+def : InstAlias<"lodsq $src", (LODSQ srcidx64:$src), 0>, Requires<[In64BitMode]>;
+def : InstAlias<"lods {$src, %al|al, $src}", (LODSB srcidx8:$src),  0>;
+def : InstAlias<"lods {$src, %ax|ax, $src}", (LODSW srcidx16:$src), 0>;
+def : InstAlias<"lods {$src, %eax|eax, $src}", (LODSL srcidx32:$src), 0>;
+def : InstAlias<"lods {$src, %rax|rax, $src}", (LODSQ srcidx64:$src), 0>, Requires<[In64BitMode]>;
+
 // div and idiv aliases for explicit A register.
 def : InstAlias<"div{b}\t{$src, %al|al, $src}", (DIV8r  GR8 :$src)>;
 def : InstAlias<"div{w}\t{$src, %ax|ax, $src}", (DIV16r GR16:$src)>;
diff --git a/test/MC/X86/index-operations.s b/test/MC/X86/index-operations.s
new file mode 100644
index 0000000000..a5dd88b1d4
--- /dev/null
+++ b/test/MC/X86/index-operations.s
@@ -0,0 +1,46 @@
+// RUN: not llvm-mc -triple x86_64-unknown-unknown --show-encoding %s 2> %t.err | FileCheck --check-prefix=64 %s
+// RUN: FileCheck --check-prefix=ERR64 < %t.err %s
+// RUN: not llvm-mc -triple i386-unknown-unknown --show-encoding %s 2> %t.err | FileCheck --check-prefix=32 %s
+// RUN: FileCheck --check-prefix=ERR32 < %t.err %s
+// RUN: not llvm-mc -triple i386-unknown-unknown-code16 --show-encoding %s 2> %t.err | FileCheck --check-prefix=16 %s
+// RUN: FileCheck --check-prefix=ERR16 < %t.err %s
+
+lodsb
+// 64: lodsb (%rsi), %al # encoding: [0xac]
+// 32: lodsb (%esi), %al # encoding: [0xac]
+// 16: lodsb (%si), %al # encoding: [0xac]
+
+lodsb (%rsi), %al
+// 64: lodsb (%rsi), %al # encoding: [0xac]
+// ERR32: 64-bit
+// ERR16: 64-bit
+
+lodsb (%esi), %al
+// 64: lodsb (%esi), %al # encoding: [0x67,0xac]
+// 32: lodsb (%esi), %al # encoding: [0xac]
+// 16: lodsb (%esi), %al # encoding: [0x67,0xac]
+
+lodsb (%si), %al
+// ERR64: invalid 16-bit base register
+// 32: lodsb (%si), %al # encoding: [0x67,0xac]
+// 16: lodsb (%si), %al # encoding: [0xac]
+
+lodsl %gs:(%esi)
+// 64: lodsl %gs:(%esi), %eax # encoding: [0x65,0x67,0xad]
+// 32: lodsl %gs:(%esi), %eax # encoding: [0x65,0xad]
+// 16: lodsl %gs:(%esi), %eax # encoding: [0x66,0x65,0x67,0xad]
+
+lodsl (%edi), %eax
+// ERR64: invalid operand
+// ERR32: invalid operand
+// ERR16: invalid operand
+
+lodsl 44(%edi), %eax
+// ERR64: invalid operand
+// ERR32: invalid operand
+// ERR16: invalid operand
+
+lods (%esi), %ax
+// 64: lodsw (%esi), %ax # encoding: [0x66,0x67,0xad]
+// 32: lodsw (%esi), %ax # encoding: [0x66,0xad]
+// 16: lodsw (%esi), %ax # encoding: [0x67,0xad]
diff --git a/test/MC/X86/x86-16.s b/test/MC/X86/x86-16.s
index 41d068e3d0..4b54a027a8 100644
--- a/test/MC/X86/x86-16.s
+++ b/test/MC/X86/x86-16.s
@@ -845,7 +845,7 @@ pshufw $90, %mm4, %mm0
 	movsl	%ds:(%si), %es:(%di)
 	movsl	(%si), %es:(%di)
 
-// CHECK: lodsb # encoding: [0xac]
+// CHECK: lodsb (%si), %al # encoding: [0xac]
 // CHECK: lodsb
 // CHECK: lodsb
 // CHECK: lodsb
@@ -856,7 +856,7 @@ pshufw $90, %mm4, %mm0
 	lods	%ds:(%si), %al
 	lods	(%si), %al
 
-// CHECK: lodsw # encoding: [0xad]
+// CHECK: lodsw (%si), %ax # encoding: [0xad]
 // CHECK: lodsw
 // CHECK: lodsw
 // CHECK: lodsw
@@ -867,7 +867,7 @@ pshufw $90, %mm4, %mm0
 	lods	%ds:(%si), %ax
 	lods	(%si), %ax
 
-// CHECK: lodsl # encoding: [0x66,0xad]
+// CHECK: lodsl (%si), %eax # encoding: [0x66,0xad]
 // CHECK: lodsl
 // CHECK: lodsl
 // CHECK: lodsl
diff --git a/test/MC/X86/x86-32.s b/test/MC/X86/x86-32.s
index 140c05af6b..29be52e35c 100644
--- a/test/MC/X86/x86-32.s
+++ b/test/MC/X86/x86-32.s
@@ -921,7 +921,7 @@ pshufw $90, %mm4, %mm0
 	movsl	%ds:(%esi), %es:(%edi)
 	movsl	(%esi), %es:(%edi)
 
-// CHECK: lodsb # encoding: [0xac]
+// CHECK: lodsb (%esi), %al # encoding: [0xac]
 // CHECK: lodsb
 // CHECK: lodsb
 // CHECK: lodsb
@@ -932,7 +932,7 @@ pshufw $90, %mm4, %mm0
 	lods	%ds:(%esi), %al
 	lods	(%esi), %al
 
-// CHECK: lodsw # encoding: [0x66,0xad]
+// CHECK: lodsw (%esi), %ax # encoding: [0x66,0xad]
 // CHECK: lodsw
 // CHECK: lodsw
 // CHECK: lodsw
@@ -943,7 +943,7 @@ pshufw $90, %mm4, %mm0
 	lods	%ds:(%esi), %ax
 	lods	(%esi), %ax
 
-// CHECK: lodsl # encoding: [0xad]
+// CHECK: lodsl (%esi), %eax # encoding: [0xad]
 // CHECK: lodsl
 // CHECK: lodsl
 // CHECK: lodsl
diff --git a/test/MC/X86/x86-64.s b/test/MC/X86/x86-64.s
index 28035c076c..ef798d7f08 100644
--- a/test/MC/X86/x86-64.s
+++ b/test/MC/X86/x86-64.s
@@ -1116,7 +1116,7 @@ xsetbv // CHECK: xsetbv # encoding: [0x0f,0x01,0xd1]
 	movsq	%ds:(%rsi), %es:(%rdi)
 	movsq	(%rsi), %es:(%rdi)
 
-// CHECK: lodsb # encoding: [0xac]
+// CHECK: lodsb (%rsi), %al # encoding: [0xac]
 // CHECK: lodsb
 // CHECK: lodsb
 // CHECK: lodsb
@@ -1127,7 +1127,7 @@ xsetbv // CHECK: xsetbv # encoding: [0x0f,0x01,0xd1]
 	lods	%ds:(%rsi), %al
 	lods	(%rsi), %al
 
-// CHECK: lodsw # encoding: [0x66,0xad]
+// CHECK: lodsw (%rsi), %ax # encoding: [0x66,0xad]
 // CHECK: lodsw
 // CHECK: lodsw
 // CHECK: lodsw
@@ -1138,7 +1138,7 @@ xsetbv // CHECK: xsetbv # encoding: [0x0f,0x01,0xd1]
 	lods	%ds:(%rsi), %ax
 	lods	(%rsi), %ax
 
-// CHECK: lodsl # encoding: [0xad]
+// CHECK: lodsl (%rsi), %eax # encoding: [0xad]
 // CHECK: lodsl
 // CHECK: lodsl
 // CHECK: lodsl
@@ -1149,7 +1149,7 @@ xsetbv // CHECK: xsetbv # encoding: [0x0f,0x01,0xd1]
 	lods	%ds:(%rsi), %eax
 	lods	(%rsi), %eax
 
-// CHECK: lodsq # encoding: [0x48,0xad]
+// CHECK: lodsq (%rsi), %rax # encoding: [0x48,0xad]
 // CHECK: lodsq
 // CHECK: lodsq
 // CHECK: lodsq
diff --git a/utils/TableGen/X86RecognizableInstr.cpp b/utils/TableGen/X86RecognizableInstr.cpp
index c09abefeef..1530c42f76 100644
--- a/utils/TableGen/X86RecognizableInstr.cpp
+++ b/utils/TableGen/X86RecognizableInstr.cpp
@@ -60,6 +60,7 @@ namespace X86Local {
     MRMSrcReg   = 5,
     MRMSrcMem   = 6,
     RawFrmMemOffs = 7,
+    RawFrmSrc   = 8,
     MRM0r = 16, MRM1r = 17, MRM2r = 18, MRM3r = 19,
     MRM4r = 20, MRM5r = 21, MRM6r = 22, MRM7r = 23,
     MRM0m = 24, MRM1m = 25, MRM2m = 26, MRM3m = 27,
@@ -630,6 +631,9 @@ void RecognizableInstr::emitInstructionSpecifier() {
 
   switch (Form) {
   default: llvm_unreachable("Unhandled form");
+  case X86Local::RawFrmSrc:
+    HANDLE_OPERAND(relocation);
+    return;
   case X86Local::RawFrm:
     // Operand 1 (optional) is an address or immediate.
     // Operand 2 (optional) is an immediate.
@@ -1262,6 +1266,10 @@ OperandType RecognizableInstr::typeFromString(const std::string &s,
   TYPE("SEGMENT_REG",         TYPE_SEGMENTREG)
   TYPE("DEBUG_REG",           TYPE_DEBUGREG)
   TYPE("CONTROL_REG",         TYPE_CONTROLREG)
+  TYPE("srcidx8",             TYPE_SRCIDX8)
+  TYPE("srcidx16",            TYPE_SRCIDX16)
+  TYPE("srcidx32",            TYPE_SRCIDX32)
+  TYPE("srcidx64",            TYPE_SRCIDX64)
   TYPE("offset8",             TYPE_MOFFS8)
   TYPE("offset16",            TYPE_MOFFS16)
   TYPE("offset32",            TYPE_MOFFS32)
@@ -1474,6 +1482,10 @@ OperandEncoding RecognizableInstr::relocationEncodingFromString
   ENCODING("offset16",        ENCODING_Ia)
   ENCODING("offset32",        ENCODING_Ia)
   ENCODING("offset64",        ENCODING_Ia)
+  ENCODING("srcidx8",         ENCODING_SI)
+  ENCODING("srcidx16",        ENCODING_SI)
+  ENCODING("srcidx32",        ENCODING_SI)
+  ENCODING("srcidx64",        ENCODING_SI)
   errs() << "Unhandled relocation encoding " << s << "\n";
   llvm_unreachable("Unhandled relocation encoding");
 }