Home | History | Annotate | Download | only in TableGen
      1 //===------------ FixedLenDecoderEmitter.cpp - Decoder Generator ----------===//
      2 //
      3 //                     The LLVM Compiler Infrastructure
      4 //
      5 // This file is distributed under the University of Illinois Open Source
      6 // License. See LICENSE.TXT for details.
      7 //
      8 //===----------------------------------------------------------------------===//
      9 //
     10 // It contains the tablegen backend that emits the decoder functions for
     11 // targets with fixed length instruction set.
     12 //
     13 //===----------------------------------------------------------------------===//
     14 
     15 #include "CodeGenTarget.h"
     16 #include "llvm/ADT/APInt.h"
     17 #include "llvm/ADT/SmallString.h"
     18 #include "llvm/ADT/StringExtras.h"
     19 #include "llvm/ADT/StringRef.h"
     20 #include "llvm/ADT/Twine.h"
     21 #include "llvm/MC/MCFixedLenDisassembler.h"
     22 #include "llvm/Support/DataTypes.h"
     23 #include "llvm/Support/Debug.h"
     24 #include "llvm/Support/FormattedStream.h"
     25 #include "llvm/Support/LEB128.h"
     26 #include "llvm/Support/raw_ostream.h"
     27 #include "llvm/TableGen/Error.h"
     28 #include "llvm/TableGen/Record.h"
     29 #include <map>
     30 #include <string>
     31 #include <utility>
     32 #include <vector>
     33 
     34 using namespace llvm;
     35 
     36 #define DEBUG_TYPE "decoder-emitter"
     37 
     38 namespace {
     39 struct EncodingField {
     40   unsigned Base, Width, Offset;
     41   EncodingField(unsigned B, unsigned W, unsigned O)
     42     : Base(B), Width(W), Offset(O) { }
     43 };
     44 
     45 struct OperandInfo {
     46   std::vector<EncodingField> Fields;
     47   std::string Decoder;
     48   bool HasCompleteDecoder;
     49 
     50   OperandInfo(std::string D, bool HCD)
     51       : Decoder(std::move(D)), HasCompleteDecoder(HCD) {}
     52 
     53   void addField(unsigned Base, unsigned Width, unsigned Offset) {
     54     Fields.push_back(EncodingField(Base, Width, Offset));
     55   }
     56 
     57   unsigned numFields() const { return Fields.size(); }
     58 
     59   typedef std::vector<EncodingField>::const_iterator const_iterator;
     60 
     61   const_iterator begin() const { return Fields.begin(); }
     62   const_iterator end() const   { return Fields.end();   }
     63 };
     64 
     65 typedef std::vector<uint8_t> DecoderTable;
     66 typedef uint32_t DecoderFixup;
     67 typedef std::vector<DecoderFixup> FixupList;
     68 typedef std::vector<FixupList> FixupScopeList;
     69 typedef SmallSetVector<std::string, 16> PredicateSet;
     70 typedef SmallSetVector<std::string, 16> DecoderSet;
     71 struct DecoderTableInfo {
     72   DecoderTable Table;
     73   FixupScopeList FixupStack;
     74   PredicateSet Predicates;
     75   DecoderSet Decoders;
     76 };
     77 
     78 } // End anonymous namespace
     79 
     80 namespace {
     81 class FixedLenDecoderEmitter {
     82   ArrayRef<const CodeGenInstruction *> NumberedInstructions;
     83 public:
     84 
     85   // Defaults preserved here for documentation, even though they aren't
     86   // strictly necessary given the way that this is currently being called.
     87   FixedLenDecoderEmitter(RecordKeeper &R, std::string PredicateNamespace,
     88                          std::string GPrefix = "if (",
     89                          std::string GPostfix = " == MCDisassembler::Fail)",
     90                          std::string ROK = "MCDisassembler::Success",
     91                          std::string RFail = "MCDisassembler::Fail",
     92                          std::string L = "")
     93       : Target(R), PredicateNamespace(std::move(PredicateNamespace)),
     94         GuardPrefix(std::move(GPrefix)), GuardPostfix(std::move(GPostfix)),
     95         ReturnOK(std::move(ROK)), ReturnFail(std::move(RFail)),
     96         Locals(std::move(L)) {}
     97 
     98   // Emit the decoder state machine table.
     99   void emitTable(formatted_raw_ostream &o, DecoderTable &Table,
    100                  unsigned Indentation, unsigned BitWidth,
    101                  StringRef Namespace) const;
    102   void emitPredicateFunction(formatted_raw_ostream &OS,
    103                              PredicateSet &Predicates,
    104                              unsigned Indentation) const;
    105   void emitDecoderFunction(formatted_raw_ostream &OS,
    106                            DecoderSet &Decoders,
    107                            unsigned Indentation) const;
    108 
    109   // run - Output the code emitter
    110   void run(raw_ostream &o);
    111 
    112 private:
    113   CodeGenTarget Target;
    114 public:
    115   std::string PredicateNamespace;
    116   std::string GuardPrefix, GuardPostfix;
    117   std::string ReturnOK, ReturnFail;
    118   std::string Locals;
    119 };
    120 } // End anonymous namespace
    121 
    122 // The set (BIT_TRUE, BIT_FALSE, BIT_UNSET) represents a ternary logic system
    123 // for a bit value.
    124 //
    125 // BIT_UNFILTERED is used as the init value for a filter position.  It is used
    126 // only for filter processings.
    127 typedef enum {
    128   BIT_TRUE,      // '1'
    129   BIT_FALSE,     // '0'
    130   BIT_UNSET,     // '?'
    131   BIT_UNFILTERED // unfiltered
    132 } bit_value_t;
    133 
    134 static bool ValueSet(bit_value_t V) {
    135   return (V == BIT_TRUE || V == BIT_FALSE);
    136 }
    137 static bool ValueNotSet(bit_value_t V) {
    138   return (V == BIT_UNSET);
    139 }
    140 static int Value(bit_value_t V) {
    141   return ValueNotSet(V) ? -1 : (V == BIT_FALSE ? 0 : 1);
    142 }
    143 static bit_value_t bitFromBits(const BitsInit &bits, unsigned index) {
    144   if (BitInit *bit = dyn_cast<BitInit>(bits.getBit(index)))
    145     return bit->getValue() ? BIT_TRUE : BIT_FALSE;
    146 
    147   // The bit is uninitialized.
    148   return BIT_UNSET;
    149 }
    150 // Prints the bit value for each position.
    151 static void dumpBits(raw_ostream &o, const BitsInit &bits) {
    152   for (unsigned index = bits.getNumBits(); index > 0; --index) {
    153     switch (bitFromBits(bits, index - 1)) {
    154     case BIT_TRUE:
    155       o << "1";
    156       break;
    157     case BIT_FALSE:
    158       o << "0";
    159       break;
    160     case BIT_UNSET:
    161       o << "_";
    162       break;
    163     default:
    164       llvm_unreachable("unexpected return value from bitFromBits");
    165     }
    166   }
    167 }
    168 
    169 static BitsInit &getBitsField(const Record &def, const char *str) {
    170   BitsInit *bits = def.getValueAsBitsInit(str);
    171   return *bits;
    172 }
    173 
    174 // Forward declaration.
    175 namespace {
    176 class FilterChooser;
    177 } // End anonymous namespace
    178 
    179 // Representation of the instruction to work on.
    180 typedef std::vector<bit_value_t> insn_t;
    181 
    182 /// Filter - Filter works with FilterChooser to produce the decoding tree for
    183 /// the ISA.
    184 ///
    185 /// It is useful to think of a Filter as governing the switch stmts of the
    186 /// decoding tree in a certain level.  Each case stmt delegates to an inferior
    187 /// FilterChooser to decide what further decoding logic to employ, or in another
    188 /// words, what other remaining bits to look at.  The FilterChooser eventually
    189 /// chooses a best Filter to do its job.
    190 ///
    191 /// This recursive scheme ends when the number of Opcodes assigned to the
    192 /// FilterChooser becomes 1 or if there is a conflict.  A conflict happens when
    193 /// the Filter/FilterChooser combo does not know how to distinguish among the
    194 /// Opcodes assigned.
    195 ///
    196 /// An example of a conflict is
    197 ///
    198 /// Conflict:
    199 ///                     111101000.00........00010000....
    200 ///                     111101000.00........0001........
    201 ///                     1111010...00........0001........
    202 ///                     1111010...00....................
    203 ///                     1111010.........................
    204 ///                     1111............................
    205 ///                     ................................
    206 ///     VST4q8a         111101000_00________00010000____
    207 ///     VST4q8b         111101000_00________00010000____
    208 ///
    209 /// The Debug output shows the path that the decoding tree follows to reach the
    210 /// the conclusion that there is a conflict.  VST4q8a is a vst4 to double-spaced
    211 /// even registers, while VST4q8b is a vst4 to double-spaced odd registers.
    212 ///
    213 /// The encoding info in the .td files does not specify this meta information,
    214 /// which could have been used by the decoder to resolve the conflict.  The
    215 /// decoder could try to decode the even/odd register numbering and assign to
    216 /// VST4q8a or VST4q8b, but for the time being, the decoder chooses the "a"
    217 /// version and return the Opcode since the two have the same Asm format string.
    218 namespace {
    219 class Filter {
    220 protected:
    221   const FilterChooser *Owner;// points to the FilterChooser who owns this filter
    222   unsigned StartBit; // the starting bit position
    223   unsigned NumBits; // number of bits to filter
    224   bool Mixed; // a mixed region contains both set and unset bits
    225 
    226   // Map of well-known segment value to the set of uid's with that value.
    227   std::map<uint64_t, std::vector<unsigned> > FilteredInstructions;
    228 
    229   // Set of uid's with non-constant segment values.
    230   std::vector<unsigned> VariableInstructions;
    231 
    232   // Map of well-known segment value to its delegate.
    233   std::map<unsigned, std::unique_ptr<const FilterChooser>> FilterChooserMap;
    234 
    235   // Number of instructions which fall under FilteredInstructions category.
    236   unsigned NumFiltered;
    237 
    238   // Keeps track of the last opcode in the filtered bucket.
    239   unsigned LastOpcFiltered;
    240 
    241 public:
    242   unsigned getNumFiltered() const { return NumFiltered; }
    243   unsigned getSingletonOpc() const {
    244     assert(NumFiltered == 1);
    245     return LastOpcFiltered;
    246   }
    247   // Return the filter chooser for the group of instructions without constant
    248   // segment values.
    249   const FilterChooser &getVariableFC() const {
    250     assert(NumFiltered == 1);
    251     assert(FilterChooserMap.size() == 1);
    252     return *(FilterChooserMap.find((unsigned)-1)->second);
    253   }
    254 
    255   Filter(Filter &&f);
    256   Filter(FilterChooser &owner, unsigned startBit, unsigned numBits, bool mixed);
    257 
    258   ~Filter();
    259 
    260   // Divides the decoding task into sub tasks and delegates them to the
    261   // inferior FilterChooser's.
    262   //
    263   // A special case arises when there's only one entry in the filtered
    264   // instructions.  In order to unambiguously decode the singleton, we need to
    265   // match the remaining undecoded encoding bits against the singleton.
    266   void recurse();
    267 
    268   // Emit table entries to decode instructions given a segment or segments of
    269   // bits.
    270   void emitTableEntry(DecoderTableInfo &TableInfo) const;
    271 
    272   // Returns the number of fanout produced by the filter.  More fanout implies
    273   // the filter distinguishes more categories of instructions.
    274   unsigned usefulness() const;
    275 }; // End of class Filter
    276 } // End anonymous namespace
    277 
    278 // These are states of our finite state machines used in FilterChooser's
    279 // filterProcessor() which produces the filter candidates to use.
    280 typedef enum {
    281   ATTR_NONE,
    282   ATTR_FILTERED,
    283   ATTR_ALL_SET,
    284   ATTR_ALL_UNSET,
    285   ATTR_MIXED
    286 } bitAttr_t;
    287 
    288 /// FilterChooser - FilterChooser chooses the best filter among a set of Filters
    289 /// in order to perform the decoding of instructions at the current level.
    290 ///
    291 /// Decoding proceeds from the top down.  Based on the well-known encoding bits
    292 /// of instructions available, FilterChooser builds up the possible Filters that
    293 /// can further the task of decoding by distinguishing among the remaining
    294 /// candidate instructions.
    295 ///
    296 /// Once a filter has been chosen, it is called upon to divide the decoding task
    297 /// into sub-tasks and delegates them to its inferior FilterChoosers for further
    298 /// processings.
    299 ///
    300 /// It is useful to think of a Filter as governing the switch stmts of the
    301 /// decoding tree.  And each case is delegated to an inferior FilterChooser to
    302 /// decide what further remaining bits to look at.
    303 namespace {
    304 class FilterChooser {
    305 protected:
    306   friend class Filter;
    307 
    308   // Vector of codegen instructions to choose our filter.
    309   ArrayRef<const CodeGenInstruction *> AllInstructions;
    310 
    311   // Vector of uid's for this filter chooser to work on.
    312   const std::vector<unsigned> &Opcodes;
    313 
    314   // Lookup table for the operand decoding of instructions.
    315   const std::map<unsigned, std::vector<OperandInfo> > &Operands;
    316 
    317   // Vector of candidate filters.
    318   std::vector<Filter> Filters;
    319 
    320   // Array of bit values passed down from our parent.
    321   // Set to all BIT_UNFILTERED's for Parent == NULL.
    322   std::vector<bit_value_t> FilterBitValues;
    323 
    324   // Links to the FilterChooser above us in the decoding tree.
    325   const FilterChooser *Parent;
    326 
    327   // Index of the best filter from Filters.
    328   int BestIndex;
    329 
    330   // Width of instructions
    331   unsigned BitWidth;
    332 
    333   // Parent emitter
    334   const FixedLenDecoderEmitter *Emitter;
    335 
    336   FilterChooser(const FilterChooser &) = delete;
    337   void operator=(const FilterChooser &) = delete;
    338 public:
    339 
    340   FilterChooser(ArrayRef<const CodeGenInstruction *> Insts,
    341                 const std::vector<unsigned> &IDs,
    342                 const std::map<unsigned, std::vector<OperandInfo> > &Ops,
    343                 unsigned BW,
    344                 const FixedLenDecoderEmitter *E)
    345     : AllInstructions(Insts), Opcodes(IDs), Operands(Ops), Filters(),
    346       FilterBitValues(BW, BIT_UNFILTERED), Parent(nullptr), BestIndex(-1),
    347       BitWidth(BW), Emitter(E) {
    348     doFilter();
    349   }
    350 
    351   FilterChooser(ArrayRef<const CodeGenInstruction *> Insts,
    352                 const std::vector<unsigned> &IDs,
    353                 const std::map<unsigned, std::vector<OperandInfo> > &Ops,
    354                 const std::vector<bit_value_t> &ParentFilterBitValues,
    355                 const FilterChooser &parent)
    356     : AllInstructions(Insts), Opcodes(IDs), Operands(Ops),
    357       Filters(), FilterBitValues(ParentFilterBitValues),
    358       Parent(&parent), BestIndex(-1), BitWidth(parent.BitWidth),
    359       Emitter(parent.Emitter) {
    360     doFilter();
    361   }
    362 
    363   unsigned getBitWidth() const { return BitWidth; }
    364 
    365 protected:
    366   // Populates the insn given the uid.
    367   void insnWithID(insn_t &Insn, unsigned Opcode) const {
    368     BitsInit &Bits = getBitsField(*AllInstructions[Opcode]->TheDef, "Inst");
    369 
    370     // We may have a SoftFail bitmask, which specifies a mask where an encoding
    371     // may differ from the value in "Inst" and yet still be valid, but the
    372     // disassembler should return SoftFail instead of Success.
    373     //
    374     // This is used for marking UNPREDICTABLE instructions in the ARM world.
    375     BitsInit *SFBits =
    376       AllInstructions[Opcode]->TheDef->getValueAsBitsInit("SoftFail");
    377 
    378     for (unsigned i = 0; i < BitWidth; ++i) {
    379       if (SFBits && bitFromBits(*SFBits, i) == BIT_TRUE)
    380         Insn.push_back(BIT_UNSET);
    381       else
    382         Insn.push_back(bitFromBits(Bits, i));
    383     }
    384   }
    385 
    386   // Returns the record name.
    387   const std::string &nameWithID(unsigned Opcode) const {
    388     return AllInstructions[Opcode]->TheDef->getName();
    389   }
    390 
    391   // Populates the field of the insn given the start position and the number of
    392   // consecutive bits to scan for.
    393   //
    394   // Returns false if there exists any uninitialized bit value in the range.
    395   // Returns true, otherwise.
    396   bool fieldFromInsn(uint64_t &Field, insn_t &Insn, unsigned StartBit,
    397                      unsigned NumBits) const;
    398 
    399   /// dumpFilterArray - dumpFilterArray prints out debugging info for the given
    400   /// filter array as a series of chars.
    401   void dumpFilterArray(raw_ostream &o,
    402                        const std::vector<bit_value_t> & filter) const;
    403 
    404   /// dumpStack - dumpStack traverses the filter chooser chain and calls
    405   /// dumpFilterArray on each filter chooser up to the top level one.
    406   void dumpStack(raw_ostream &o, const char *prefix) const;
    407 
    408   Filter &bestFilter() {
    409     assert(BestIndex != -1 && "BestIndex not set");
    410     return Filters[BestIndex];
    411   }
    412 
    413   bool PositionFiltered(unsigned i) const {
    414     return ValueSet(FilterBitValues[i]);
    415   }
    416 
    417   // Calculates the island(s) needed to decode the instruction.
    418   // This returns a lit of undecoded bits of an instructions, for example,
    419   // Inst{20} = 1 && Inst{3-0} == 0b1111 represents two islands of yet-to-be
    420   // decoded bits in order to verify that the instruction matches the Opcode.
    421   unsigned getIslands(std::vector<unsigned> &StartBits,
    422                       std::vector<unsigned> &EndBits,
    423                       std::vector<uint64_t> &FieldVals,
    424                       const insn_t &Insn) const;
    425 
    426   // Emits code to check the Predicates member of an instruction are true.
    427   // Returns true if predicate matches were emitted, false otherwise.
    428   bool emitPredicateMatch(raw_ostream &o, unsigned &Indentation,
    429                           unsigned Opc) const;
    430 
    431   bool doesOpcodeNeedPredicate(unsigned Opc) const;
    432   unsigned getPredicateIndex(DecoderTableInfo &TableInfo, StringRef P) const;
    433   void emitPredicateTableEntry(DecoderTableInfo &TableInfo,
    434                                unsigned Opc) const;
    435 
    436   void emitSoftFailTableEntry(DecoderTableInfo &TableInfo,
    437                               unsigned Opc) const;
    438 
    439   // Emits table entries to decode the singleton.
    440   void emitSingletonTableEntry(DecoderTableInfo &TableInfo,
    441                                unsigned Opc) const;
    442 
    443   // Emits code to decode the singleton, and then to decode the rest.
    444   void emitSingletonTableEntry(DecoderTableInfo &TableInfo,
    445                                const Filter &Best) const;
    446 
    447   void emitBinaryParser(raw_ostream &o, unsigned &Indentation,
    448                         const OperandInfo &OpInfo,
    449                         bool &OpHasCompleteDecoder) const;
    450 
    451   void emitDecoder(raw_ostream &OS, unsigned Indentation, unsigned Opc,
    452                    bool &HasCompleteDecoder) const;
    453   unsigned getDecoderIndex(DecoderSet &Decoders, unsigned Opc,
    454                            bool &HasCompleteDecoder) const;
    455 
    456   // Assign a single filter and run with it.
    457   void runSingleFilter(unsigned startBit, unsigned numBit, bool mixed);
    458 
    459   // reportRegion is a helper function for filterProcessor to mark a region as
    460   // eligible for use as a filter region.
    461   void reportRegion(bitAttr_t RA, unsigned StartBit, unsigned BitIndex,
    462                     bool AllowMixed);
    463 
    464   // FilterProcessor scans the well-known encoding bits of the instructions and
    465   // builds up a list of candidate filters.  It chooses the best filter and
    466   // recursively descends down the decoding tree.
    467   bool filterProcessor(bool AllowMixed, bool Greedy = true);
    468 
    469   // Decides on the best configuration of filter(s) to use in order to decode
    470   // the instructions.  A conflict of instructions may occur, in which case we
    471   // dump the conflict set to the standard error.
    472   void doFilter();
    473 
    474 public:
    475   // emitTableEntries - Emit state machine entries to decode our share of
    476   // instructions.
    477   void emitTableEntries(DecoderTableInfo &TableInfo) const;
    478 };
    479 } // End anonymous namespace
    480 
    481 ///////////////////////////
    482 //                       //
    483 // Filter Implementation //
    484 //                       //
    485 ///////////////////////////
    486 
    487 Filter::Filter(Filter &&f)
    488   : Owner(f.Owner), StartBit(f.StartBit), NumBits(f.NumBits), Mixed(f.Mixed),
    489     FilteredInstructions(std::move(f.FilteredInstructions)),
    490     VariableInstructions(std::move(f.VariableInstructions)),
    491     FilterChooserMap(std::move(f.FilterChooserMap)), NumFiltered(f.NumFiltered),
    492     LastOpcFiltered(f.LastOpcFiltered) {
    493 }
    494 
    495 Filter::Filter(FilterChooser &owner, unsigned startBit, unsigned numBits,
    496                bool mixed)
    497   : Owner(&owner), StartBit(startBit), NumBits(numBits), Mixed(mixed) {
    498   assert(StartBit + NumBits - 1 < Owner->BitWidth);
    499 
    500   NumFiltered = 0;
    501   LastOpcFiltered = 0;
    502 
    503   for (unsigned i = 0, e = Owner->Opcodes.size(); i != e; ++i) {
    504     insn_t Insn;
    505 
    506     // Populates the insn given the uid.
    507     Owner->insnWithID(Insn, Owner->Opcodes[i]);
    508 
    509     uint64_t Field;
    510     // Scans the segment for possibly well-specified encoding bits.
    511     bool ok = Owner->fieldFromInsn(Field, Insn, StartBit, NumBits);
    512 
    513     if (ok) {
    514       // The encoding bits are well-known.  Lets add the uid of the
    515       // instruction into the bucket keyed off the constant field value.
    516       LastOpcFiltered = Owner->Opcodes[i];
    517       FilteredInstructions[Field].push_back(LastOpcFiltered);
    518       ++NumFiltered;
    519     } else {
    520       // Some of the encoding bit(s) are unspecified.  This contributes to
    521       // one additional member of "Variable" instructions.
    522       VariableInstructions.push_back(Owner->Opcodes[i]);
    523     }
    524   }
    525 
    526   assert((FilteredInstructions.size() + VariableInstructions.size() > 0)
    527          && "Filter returns no instruction categories");
    528 }
    529 
    530 Filter::~Filter() {
    531 }
    532 
    533 // Divides the decoding task into sub tasks and delegates them to the
    534 // inferior FilterChooser's.
    535 //
    536 // A special case arises when there's only one entry in the filtered
    537 // instructions.  In order to unambiguously decode the singleton, we need to
    538 // match the remaining undecoded encoding bits against the singleton.
    539 void Filter::recurse() {
    540   // Starts by inheriting our parent filter chooser's filter bit values.
    541   std::vector<bit_value_t> BitValueArray(Owner->FilterBitValues);
    542 
    543   if (!VariableInstructions.empty()) {
    544     // Conservatively marks each segment position as BIT_UNSET.
    545     for (unsigned bitIndex = 0; bitIndex < NumBits; ++bitIndex)
    546       BitValueArray[StartBit + bitIndex] = BIT_UNSET;
    547 
    548     // Delegates to an inferior filter chooser for further processing on this
    549     // group of instructions whose segment values are variable.
    550     FilterChooserMap.insert(
    551         std::make_pair(-1U, llvm::make_unique<FilterChooser>(
    552                                 Owner->AllInstructions, VariableInstructions,
    553                                 Owner->Operands, BitValueArray, *Owner)));
    554   }
    555 
    556   // No need to recurse for a singleton filtered instruction.
    557   // See also Filter::emit*().
    558   if (getNumFiltered() == 1) {
    559     assert(FilterChooserMap.size() == 1);
    560     return;
    561   }
    562 
    563   // Otherwise, create sub choosers.
    564   for (const auto &Inst : FilteredInstructions) {
    565 
    566     // Marks all the segment positions with either BIT_TRUE or BIT_FALSE.
    567     for (unsigned bitIndex = 0; bitIndex < NumBits; ++bitIndex) {
    568       if (Inst.first & (1ULL << bitIndex))
    569         BitValueArray[StartBit + bitIndex] = BIT_TRUE;
    570       else
    571         BitValueArray[StartBit + bitIndex] = BIT_FALSE;
    572     }
    573 
    574     // Delegates to an inferior filter chooser for further processing on this
    575     // category of instructions.
    576     FilterChooserMap.insert(std::make_pair(
    577         Inst.first, llvm::make_unique<FilterChooser>(
    578                                 Owner->AllInstructions, Inst.second,
    579                                 Owner->Operands, BitValueArray, *Owner)));
    580   }
    581 }
    582 
    583 static void resolveTableFixups(DecoderTable &Table, const FixupList &Fixups,
    584                                uint32_t DestIdx) {
    585   // Any NumToSkip fixups in the current scope can resolve to the
    586   // current location.
    587   for (FixupList::const_reverse_iterator I = Fixups.rbegin(),
    588                                          E = Fixups.rend();
    589        I != E; ++I) {
    590     // Calculate the distance from the byte following the fixup entry byte
    591     // to the destination. The Target is calculated from after the 16-bit
    592     // NumToSkip entry itself, so subtract two  from the displacement here
    593     // to account for that.
    594     uint32_t FixupIdx = *I;
    595     uint32_t Delta = DestIdx - FixupIdx - 2;
    596     // Our NumToSkip entries are 16-bits. Make sure our table isn't too
    597     // big.
    598     assert(Delta < 65536U && "disassembler decoding table too large!");
    599     Table[FixupIdx] = (uint8_t)Delta;
    600     Table[FixupIdx + 1] = (uint8_t)(Delta >> 8);
    601   }
    602 }
    603 
    604 // Emit table entries to decode instructions given a segment or segments
    605 // of bits.
    606 void Filter::emitTableEntry(DecoderTableInfo &TableInfo) const {
    607   TableInfo.Table.push_back(MCD::OPC_ExtractField);
    608   TableInfo.Table.push_back(StartBit);
    609   TableInfo.Table.push_back(NumBits);
    610 
    611   // A new filter entry begins a new scope for fixup resolution.
    612   TableInfo.FixupStack.emplace_back();
    613 
    614   DecoderTable &Table = TableInfo.Table;
    615 
    616   size_t PrevFilter = 0;
    617   bool HasFallthrough = false;
    618   for (auto &Filter : FilterChooserMap) {
    619     // Field value -1 implies a non-empty set of variable instructions.
    620     // See also recurse().
    621     if (Filter.first == (unsigned)-1) {
    622       HasFallthrough = true;
    623 
    624       // Each scope should always have at least one filter value to check
    625       // for.
    626       assert(PrevFilter != 0 && "empty filter set!");
    627       FixupList &CurScope = TableInfo.FixupStack.back();
    628       // Resolve any NumToSkip fixups in the current scope.
    629       resolveTableFixups(Table, CurScope, Table.size());
    630       CurScope.clear();
    631       PrevFilter = 0;  // Don't re-process the filter's fallthrough.
    632     } else {
    633       Table.push_back(MCD::OPC_FilterValue);
    634       // Encode and emit the value to filter against.
    635       uint8_t Buffer[8];
    636       unsigned Len = encodeULEB128(Filter.first, Buffer);
    637       Table.insert(Table.end(), Buffer, Buffer + Len);
    638       // Reserve space for the NumToSkip entry. We'll backpatch the value
    639       // later.
    640       PrevFilter = Table.size();
    641       Table.push_back(0);
    642       Table.push_back(0);
    643     }
    644 
    645     // We arrive at a category of instructions with the same segment value.
    646     // Now delegate to the sub filter chooser for further decodings.
    647     // The case may fallthrough, which happens if the remaining well-known
    648     // encoding bits do not match exactly.
    649     Filter.second->emitTableEntries(TableInfo);
    650 
    651     // Now that we've emitted the body of the handler, update the NumToSkip
    652     // of the filter itself to be able to skip forward when false. Subtract
    653     // two as to account for the width of the NumToSkip field itself.
    654     if (PrevFilter) {
    655       uint32_t NumToSkip = Table.size() - PrevFilter - 2;
    656       assert(NumToSkip < 65536U && "disassembler decoding table too large!");
    657       Table[PrevFilter] = (uint8_t)NumToSkip;
    658       Table[PrevFilter + 1] = (uint8_t)(NumToSkip >> 8);
    659     }
    660   }
    661 
    662   // Any remaining unresolved fixups bubble up to the parent fixup scope.
    663   assert(TableInfo.FixupStack.size() > 1 && "fixup stack underflow!");
    664   FixupScopeList::iterator Source = TableInfo.FixupStack.end() - 1;
    665   FixupScopeList::iterator Dest = Source - 1;
    666   Dest->insert(Dest->end(), Source->begin(), Source->end());
    667   TableInfo.FixupStack.pop_back();
    668 
    669   // If there is no fallthrough, then the final filter should get fixed
    670   // up according to the enclosing scope rather than the current position.
    671   if (!HasFallthrough)
    672     TableInfo.FixupStack.back().push_back(PrevFilter);
    673 }
    674 
    675 // Returns the number of fanout produced by the filter.  More fanout implies
    676 // the filter distinguishes more categories of instructions.
    677 unsigned Filter::usefulness() const {
    678   if (!VariableInstructions.empty())
    679     return FilteredInstructions.size();
    680   else
    681     return FilteredInstructions.size() + 1;
    682 }
    683 
    684 //////////////////////////////////
    685 //                              //
    686 // Filterchooser Implementation //
    687 //                              //
    688 //////////////////////////////////
    689 
    690 // Emit the decoder state machine table.
    691 void FixedLenDecoderEmitter::emitTable(formatted_raw_ostream &OS,
    692                                        DecoderTable &Table,
    693                                        unsigned Indentation,
    694                                        unsigned BitWidth,
    695                                        StringRef Namespace) const {
    696   OS.indent(Indentation) << "static const uint8_t DecoderTable" << Namespace
    697     << BitWidth << "[] = {\n";
    698 
    699   Indentation += 2;
    700 
    701   // FIXME: We may be able to use the NumToSkip values to recover
    702   // appropriate indentation levels.
    703   DecoderTable::const_iterator I = Table.begin();
    704   DecoderTable::const_iterator E = Table.end();
    705   while (I != E) {
    706     assert (I < E && "incomplete decode table entry!");
    707 
    708     uint64_t Pos = I - Table.begin();
    709     OS << "/* " << Pos << " */";
    710     OS.PadToColumn(12);
    711 
    712     switch (*I) {
    713     default:
    714       PrintFatalError("invalid decode table opcode");
    715     case MCD::OPC_ExtractField: {
    716       ++I;
    717       unsigned Start = *I++;
    718       unsigned Len = *I++;
    719       OS.indent(Indentation) << "MCD::OPC_ExtractField, " << Start << ", "
    720         << Len << ",  // Inst{";
    721       if (Len > 1)
    722         OS << (Start + Len - 1) << "-";
    723       OS << Start << "} ...\n";
    724       break;
    725     }
    726     case MCD::OPC_FilterValue: {
    727       ++I;
    728       OS.indent(Indentation) << "MCD::OPC_FilterValue, ";
    729       // The filter value is ULEB128 encoded.
    730       while (*I >= 128)
    731         OS << (unsigned)*I++ << ", ";
    732       OS << (unsigned)*I++ << ", ";
    733 
    734       // 16-bit numtoskip value.
    735       uint8_t Byte = *I++;
    736       uint32_t NumToSkip = Byte;
    737       OS << (unsigned)Byte << ", ";
    738       Byte = *I++;
    739       OS << (unsigned)Byte << ", ";
    740       NumToSkip |= Byte << 8;
    741       OS << "// Skip to: " << ((I - Table.begin()) + NumToSkip) << "\n";
    742       break;
    743     }
    744     case MCD::OPC_CheckField: {
    745       ++I;
    746       unsigned Start = *I++;
    747       unsigned Len = *I++;
    748       OS.indent(Indentation) << "MCD::OPC_CheckField, " << Start << ", "
    749         << Len << ", ";// << Val << ", " << NumToSkip << ",\n";
    750       // ULEB128 encoded field value.
    751       for (; *I >= 128; ++I)
    752         OS << (unsigned)*I << ", ";
    753       OS << (unsigned)*I++ << ", ";
    754       // 16-bit numtoskip value.
    755       uint8_t Byte = *I++;
    756       uint32_t NumToSkip = Byte;
    757       OS << (unsigned)Byte << ", ";
    758       Byte = *I++;
    759       OS << (unsigned)Byte << ", ";
    760       NumToSkip |= Byte << 8;
    761       OS << "// Skip to: " << ((I - Table.begin()) + NumToSkip) << "\n";
    762       break;
    763     }
    764     case MCD::OPC_CheckPredicate: {
    765       ++I;
    766       OS.indent(Indentation) << "MCD::OPC_CheckPredicate, ";
    767       for (; *I >= 128; ++I)
    768         OS << (unsigned)*I << ", ";
    769       OS << (unsigned)*I++ << ", ";
    770 
    771       // 16-bit numtoskip value.
    772       uint8_t Byte = *I++;
    773       uint32_t NumToSkip = Byte;
    774       OS << (unsigned)Byte << ", ";
    775       Byte = *I++;
    776       OS << (unsigned)Byte << ", ";
    777       NumToSkip |= Byte << 8;
    778       OS << "// Skip to: " << ((I - Table.begin()) + NumToSkip) << "\n";
    779       break;
    780     }
    781     case MCD::OPC_Decode:
    782     case MCD::OPC_TryDecode: {
    783       bool IsTry = *I == MCD::OPC_TryDecode;
    784       ++I;
    785       // Extract the ULEB128 encoded Opcode to a buffer.
    786       uint8_t Buffer[8], *p = Buffer;
    787       while ((*p++ = *I++) >= 128)
    788         assert((p - Buffer) <= (ptrdiff_t)sizeof(Buffer)
    789                && "ULEB128 value too large!");
    790       // Decode the Opcode value.
    791       unsigned Opc = decodeULEB128(Buffer);
    792       OS.indent(Indentation) << "MCD::OPC_" << (IsTry ? "Try" : "")
    793         << "Decode, ";
    794       for (p = Buffer; *p >= 128; ++p)
    795         OS << (unsigned)*p << ", ";
    796       OS << (unsigned)*p << ", ";
    797 
    798       // Decoder index.
    799       for (; *I >= 128; ++I)
    800         OS << (unsigned)*I << ", ";
    801       OS << (unsigned)*I++ << ", ";
    802 
    803       if (!IsTry) {
    804         OS << "// Opcode: "
    805            << NumberedInstructions[Opc]->TheDef->getName() << "\n";
    806         break;
    807       }
    808 
    809       // Fallthrough for OPC_TryDecode.
    810 
    811       // 16-bit numtoskip value.
    812       uint8_t Byte = *I++;
    813       uint32_t NumToSkip = Byte;
    814       OS << (unsigned)Byte << ", ";
    815       Byte = *I++;
    816       OS << (unsigned)Byte << ", ";
    817       NumToSkip |= Byte << 8;
    818 
    819       OS << "// Opcode: "
    820          << NumberedInstructions[Opc]->TheDef->getName()
    821          << ", skip to: " << ((I - Table.begin()) + NumToSkip) << "\n";
    822       break;
    823     }
    824     case MCD::OPC_SoftFail: {
    825       ++I;
    826       OS.indent(Indentation) << "MCD::OPC_SoftFail";
    827       // Positive mask
    828       uint64_t Value = 0;
    829       unsigned Shift = 0;
    830       do {
    831         OS << ", " << (unsigned)*I;
    832         Value += (*I & 0x7f) << Shift;
    833         Shift += 7;
    834       } while (*I++ >= 128);
    835       if (Value > 127) {
    836         OS << " /* 0x";
    837         OS.write_hex(Value);
    838         OS << " */";
    839       }
    840       // Negative mask
    841       Value = 0;
    842       Shift = 0;
    843       do {
    844         OS << ", " << (unsigned)*I;
    845         Value += (*I & 0x7f) << Shift;
    846         Shift += 7;
    847       } while (*I++ >= 128);
    848       if (Value > 127) {
    849         OS << " /* 0x";
    850         OS.write_hex(Value);
    851         OS << " */";
    852       }
    853       OS << ",\n";
    854       break;
    855     }
    856     case MCD::OPC_Fail: {
    857       ++I;
    858       OS.indent(Indentation) << "MCD::OPC_Fail,\n";
    859       break;
    860     }
    861     }
    862   }
    863   OS.indent(Indentation) << "0\n";
    864 
    865   Indentation -= 2;
    866 
    867   OS.indent(Indentation) << "};\n\n";
    868 }
    869 
    870 void FixedLenDecoderEmitter::
    871 emitPredicateFunction(formatted_raw_ostream &OS, PredicateSet &Predicates,
    872                       unsigned Indentation) const {
    873   // The predicate function is just a big switch statement based on the
    874   // input predicate index.
    875   OS.indent(Indentation) << "static bool checkDecoderPredicate(unsigned Idx, "
    876     << "const FeatureBitset& Bits) {\n";
    877   Indentation += 2;
    878   if (!Predicates.empty()) {
    879     OS.indent(Indentation) << "switch (Idx) {\n";
    880     OS.indent(Indentation) << "default: llvm_unreachable(\"Invalid index!\");\n";
    881     unsigned Index = 0;
    882     for (const auto &Predicate : Predicates) {
    883       OS.indent(Indentation) << "case " << Index++ << ":\n";
    884       OS.indent(Indentation+2) << "return (" << Predicate << ");\n";
    885     }
    886     OS.indent(Indentation) << "}\n";
    887   } else {
    888     // No case statement to emit
    889     OS.indent(Indentation) << "llvm_unreachable(\"Invalid index!\");\n";
    890   }
    891   Indentation -= 2;
    892   OS.indent(Indentation) << "}\n\n";
    893 }
    894 
    895 void FixedLenDecoderEmitter::
    896 emitDecoderFunction(formatted_raw_ostream &OS, DecoderSet &Decoders,
    897                     unsigned Indentation) const {
    898   // The decoder function is just a big switch statement based on the
    899   // input decoder index.
    900   OS.indent(Indentation) << "template<typename InsnType>\n";
    901   OS.indent(Indentation) << "static DecodeStatus decodeToMCInst(DecodeStatus S,"
    902     << " unsigned Idx, InsnType insn, MCInst &MI,\n";
    903   OS.indent(Indentation) << "                                   uint64_t "
    904     << "Address, const void *Decoder, bool &DecodeComplete) {\n";
    905   Indentation += 2;
    906   OS.indent(Indentation) << "DecodeComplete = true;\n";
    907   OS.indent(Indentation) << "InsnType tmp;\n";
    908   OS.indent(Indentation) << "switch (Idx) {\n";
    909   OS.indent(Indentation) << "default: llvm_unreachable(\"Invalid index!\");\n";
    910   unsigned Index = 0;
    911   for (const auto &Decoder : Decoders) {
    912     OS.indent(Indentation) << "case " << Index++ << ":\n";
    913     OS << Decoder;
    914     OS.indent(Indentation+2) << "return S;\n";
    915   }
    916   OS.indent(Indentation) << "}\n";
    917   Indentation -= 2;
    918   OS.indent(Indentation) << "}\n\n";
    919 }
    920 
    921 // Populates the field of the insn given the start position and the number of
    922 // consecutive bits to scan for.
    923 //
    924 // Returns false if and on the first uninitialized bit value encountered.
    925 // Returns true, otherwise.
    926 bool FilterChooser::fieldFromInsn(uint64_t &Field, insn_t &Insn,
    927                                   unsigned StartBit, unsigned NumBits) const {
    928   Field = 0;
    929 
    930   for (unsigned i = 0; i < NumBits; ++i) {
    931     if (Insn[StartBit + i] == BIT_UNSET)
    932       return false;
    933 
    934     if (Insn[StartBit + i] == BIT_TRUE)
    935       Field = Field | (1ULL << i);
    936   }
    937 
    938   return true;
    939 }
    940 
    941 /// dumpFilterArray - dumpFilterArray prints out debugging info for the given
    942 /// filter array as a series of chars.
    943 void FilterChooser::dumpFilterArray(raw_ostream &o,
    944                                  const std::vector<bit_value_t> &filter) const {
    945   for (unsigned bitIndex = BitWidth; bitIndex > 0; bitIndex--) {
    946     switch (filter[bitIndex - 1]) {
    947     case BIT_UNFILTERED:
    948       o << ".";
    949       break;
    950     case BIT_UNSET:
    951       o << "_";
    952       break;
    953     case BIT_TRUE:
    954       o << "1";
    955       break;
    956     case BIT_FALSE:
    957       o << "0";
    958       break;
    959     }
    960   }
    961 }
    962 
    963 /// dumpStack - dumpStack traverses the filter chooser chain and calls
    964 /// dumpFilterArray on each filter chooser up to the top level one.
    965 void FilterChooser::dumpStack(raw_ostream &o, const char *prefix) const {
    966   const FilterChooser *current = this;
    967 
    968   while (current) {
    969     o << prefix;
    970     dumpFilterArray(o, current->FilterBitValues);
    971     o << '\n';
    972     current = current->Parent;
    973   }
    974 }
    975 
    976 // Calculates the island(s) needed to decode the instruction.
    977 // This returns a list of undecoded bits of an instructions, for example,
    978 // Inst{20} = 1 && Inst{3-0} == 0b1111 represents two islands of yet-to-be
    979 // decoded bits in order to verify that the instruction matches the Opcode.
    980 unsigned FilterChooser::getIslands(std::vector<unsigned> &StartBits,
    981                                    std::vector<unsigned> &EndBits,
    982                                    std::vector<uint64_t> &FieldVals,
    983                                    const insn_t &Insn) const {
    984   unsigned Num, BitNo;
    985   Num = BitNo = 0;
    986 
    987   uint64_t FieldVal = 0;
    988 
    989   // 0: Init
    990   // 1: Water (the bit value does not affect decoding)
    991   // 2: Island (well-known bit value needed for decoding)
    992   int State = 0;
    993   int Val = -1;
    994 
    995   for (unsigned i = 0; i < BitWidth; ++i) {
    996     Val = Value(Insn[i]);
    997     bool Filtered = PositionFiltered(i);
    998     switch (State) {
    999     default: llvm_unreachable("Unreachable code!");
   1000     case 0:
   1001     case 1:
   1002       if (Filtered || Val == -1)
   1003         State = 1; // Still in Water
   1004       else {
   1005         State = 2; // Into the Island
   1006         BitNo = 0;
   1007         StartBits.push_back(i);
   1008         FieldVal = Val;
   1009       }
   1010       break;
   1011     case 2:
   1012       if (Filtered || Val == -1) {
   1013         State = 1; // Into the Water
   1014         EndBits.push_back(i - 1);
   1015         FieldVals.push_back(FieldVal);
   1016         ++Num;
   1017       } else {
   1018         State = 2; // Still in Island
   1019         ++BitNo;
   1020         FieldVal = FieldVal | Val << BitNo;
   1021       }
   1022       break;
   1023     }
   1024   }
   1025   // If we are still in Island after the loop, do some housekeeping.
   1026   if (State == 2) {
   1027     EndBits.push_back(BitWidth - 1);
   1028     FieldVals.push_back(FieldVal);
   1029     ++Num;
   1030   }
   1031 
   1032   assert(StartBits.size() == Num && EndBits.size() == Num &&
   1033          FieldVals.size() == Num);
   1034   return Num;
   1035 }
   1036 
   1037 void FilterChooser::emitBinaryParser(raw_ostream &o, unsigned &Indentation,
   1038                                      const OperandInfo &OpInfo,
   1039                                      bool &OpHasCompleteDecoder) const {
   1040   const std::string &Decoder = OpInfo.Decoder;
   1041 
   1042   if (OpInfo.numFields() != 1)
   1043     o.indent(Indentation) << "tmp = 0;\n";
   1044 
   1045   for (const EncodingField &EF : OpInfo) {
   1046     o.indent(Indentation) << "tmp ";
   1047     if (OpInfo.numFields() != 1) o << '|';
   1048     o << "= fieldFromInstruction"
   1049       << "(insn, " << EF.Base << ", " << EF.Width << ')';
   1050     if (OpInfo.numFields() != 1 || EF.Offset != 0)
   1051       o << " << " << EF.Offset;
   1052     o << ";\n";
   1053   }
   1054 
   1055   if (Decoder != "") {
   1056     OpHasCompleteDecoder = OpInfo.HasCompleteDecoder;
   1057     o.indent(Indentation) << Emitter->GuardPrefix << Decoder
   1058       << "(MI, tmp, Address, Decoder)"
   1059       << Emitter->GuardPostfix
   1060       << " { " << (OpHasCompleteDecoder ? "" : "DecodeComplete = false; ")
   1061       << "return MCDisassembler::Fail; }\n";
   1062   } else {
   1063     OpHasCompleteDecoder = true;
   1064     o.indent(Indentation) << "MI.addOperand(MCOperand::createImm(tmp));\n";
   1065   }
   1066 }
   1067 
   1068 void FilterChooser::emitDecoder(raw_ostream &OS, unsigned Indentation,
   1069                                 unsigned Opc, bool &HasCompleteDecoder) const {
   1070   HasCompleteDecoder = true;
   1071 
   1072   for (const auto &Op : Operands.find(Opc)->second) {
   1073     // If a custom instruction decoder was specified, use that.
   1074     if (Op.numFields() == 0 && Op.Decoder.size()) {
   1075       HasCompleteDecoder = Op.HasCompleteDecoder;
   1076       OS.indent(Indentation) << Emitter->GuardPrefix << Op.Decoder
   1077         << "(MI, insn, Address, Decoder)"
   1078         << Emitter->GuardPostfix
   1079         << " { " << (HasCompleteDecoder ? "" : "DecodeComplete = false; ")
   1080         << "return MCDisassembler::Fail; }\n";
   1081       break;
   1082     }
   1083 
   1084     bool OpHasCompleteDecoder;
   1085     emitBinaryParser(OS, Indentation, Op, OpHasCompleteDecoder);
   1086     if (!OpHasCompleteDecoder)
   1087       HasCompleteDecoder = false;
   1088   }
   1089 }
   1090 
   1091 unsigned FilterChooser::getDecoderIndex(DecoderSet &Decoders,
   1092                                         unsigned Opc,
   1093                                         bool &HasCompleteDecoder) const {
   1094   // Build up the predicate string.
   1095   SmallString<256> Decoder;
   1096   // FIXME: emitDecoder() function can take a buffer directly rather than
   1097   // a stream.
   1098   raw_svector_ostream S(Decoder);
   1099   unsigned I = 4;
   1100   emitDecoder(S, I, Opc, HasCompleteDecoder);
   1101 
   1102   // Using the full decoder string as the key value here is a bit
   1103   // heavyweight, but is effective. If the string comparisons become a
   1104   // performance concern, we can implement a mangling of the predicate
   1105   // data easily enough with a map back to the actual string. That's
   1106   // overkill for now, though.
   1107 
   1108   // Make sure the predicate is in the table.
   1109   Decoders.insert(StringRef(Decoder));
   1110   // Now figure out the index for when we write out the table.
   1111   DecoderSet::const_iterator P = std::find(Decoders.begin(),
   1112                                            Decoders.end(),
   1113                                            Decoder.str());
   1114   return (unsigned)(P - Decoders.begin());
   1115 }
   1116 
   1117 static void emitSinglePredicateMatch(raw_ostream &o, StringRef str,
   1118                                      const std::string &PredicateNamespace) {
   1119   if (str[0] == '!')
   1120     o << "!Bits[" << PredicateNamespace << "::"
   1121       << str.slice(1,str.size()) << "]";
   1122   else
   1123     o << "Bits[" << PredicateNamespace << "::" << str << "]";
   1124 }
   1125 
   1126 bool FilterChooser::emitPredicateMatch(raw_ostream &o, unsigned &Indentation,
   1127                                        unsigned Opc) const {
   1128   ListInit *Predicates =
   1129     AllInstructions[Opc]->TheDef->getValueAsListInit("Predicates");
   1130   bool IsFirstEmission = true;
   1131   for (unsigned i = 0; i < Predicates->size(); ++i) {
   1132     Record *Pred = Predicates->getElementAsRecord(i);
   1133     if (!Pred->getValue("AssemblerMatcherPredicate"))
   1134       continue;
   1135 
   1136     std::string P = Pred->getValueAsString("AssemblerCondString");
   1137 
   1138     if (!P.length())
   1139       continue;
   1140 
   1141     if (!IsFirstEmission)
   1142       o << " && ";
   1143 
   1144     StringRef SR(P);
   1145     std::pair<StringRef, StringRef> pairs = SR.split(',');
   1146     while (pairs.second.size()) {
   1147       emitSinglePredicateMatch(o, pairs.first, Emitter->PredicateNamespace);
   1148       o << " && ";
   1149       pairs = pairs.second.split(',');
   1150     }
   1151     emitSinglePredicateMatch(o, pairs.first, Emitter->PredicateNamespace);
   1152     IsFirstEmission = false;
   1153   }
   1154   return !Predicates->empty();
   1155 }
   1156 
   1157 bool FilterChooser::doesOpcodeNeedPredicate(unsigned Opc) const {
   1158   ListInit *Predicates =
   1159     AllInstructions[Opc]->TheDef->getValueAsListInit("Predicates");
   1160   for (unsigned i = 0; i < Predicates->size(); ++i) {
   1161     Record *Pred = Predicates->getElementAsRecord(i);
   1162     if (!Pred->getValue("AssemblerMatcherPredicate"))
   1163       continue;
   1164 
   1165     std::string P = Pred->getValueAsString("AssemblerCondString");
   1166 
   1167     if (!P.length())
   1168       continue;
   1169 
   1170     return true;
   1171   }
   1172   return false;
   1173 }
   1174 
   1175 unsigned FilterChooser::getPredicateIndex(DecoderTableInfo &TableInfo,
   1176                                           StringRef Predicate) const {
   1177   // Using the full predicate string as the key value here is a bit
   1178   // heavyweight, but is effective. If the string comparisons become a
   1179   // performance concern, we can implement a mangling of the predicate
   1180   // data easily enough with a map back to the actual string. That's
   1181   // overkill for now, though.
   1182 
   1183   // Make sure the predicate is in the table.
   1184   TableInfo.Predicates.insert(Predicate.str());
   1185   // Now figure out the index for when we write out the table.
   1186   PredicateSet::const_iterator P = std::find(TableInfo.Predicates.begin(),
   1187                                              TableInfo.Predicates.end(),
   1188                                              Predicate.str());
   1189   return (unsigned)(P - TableInfo.Predicates.begin());
   1190 }
   1191 
   1192 void FilterChooser::emitPredicateTableEntry(DecoderTableInfo &TableInfo,
   1193                                             unsigned Opc) const {
   1194   if (!doesOpcodeNeedPredicate(Opc))
   1195     return;
   1196 
   1197   // Build up the predicate string.
   1198   SmallString<256> Predicate;
   1199   // FIXME: emitPredicateMatch() functions can take a buffer directly rather
   1200   // than a stream.
   1201   raw_svector_ostream PS(Predicate);
   1202   unsigned I = 0;
   1203   emitPredicateMatch(PS, I, Opc);
   1204 
   1205   // Figure out the index into the predicate table for the predicate just
   1206   // computed.
   1207   unsigned PIdx = getPredicateIndex(TableInfo, PS.str());
   1208   SmallString<16> PBytes;
   1209   raw_svector_ostream S(PBytes);
   1210   encodeULEB128(PIdx, S);
   1211 
   1212   TableInfo.Table.push_back(MCD::OPC_CheckPredicate);
   1213   // Predicate index
   1214   for (unsigned i = 0, e = PBytes.size(); i != e; ++i)
   1215     TableInfo.Table.push_back(PBytes[i]);
   1216   // Push location for NumToSkip backpatching.
   1217   TableInfo.FixupStack.back().push_back(TableInfo.Table.size());
   1218   TableInfo.Table.push_back(0);
   1219   TableInfo.Table.push_back(0);
   1220 }
   1221 
   1222 void FilterChooser::emitSoftFailTableEntry(DecoderTableInfo &TableInfo,
   1223                                            unsigned Opc) const {
   1224   BitsInit *SFBits =
   1225     AllInstructions[Opc]->TheDef->getValueAsBitsInit("SoftFail");
   1226   if (!SFBits) return;
   1227   BitsInit *InstBits = AllInstructions[Opc]->TheDef->getValueAsBitsInit("Inst");
   1228 
   1229   APInt PositiveMask(BitWidth, 0ULL);
   1230   APInt NegativeMask(BitWidth, 0ULL);
   1231   for (unsigned i = 0; i < BitWidth; ++i) {
   1232     bit_value_t B = bitFromBits(*SFBits, i);
   1233     bit_value_t IB = bitFromBits(*InstBits, i);
   1234 
   1235     if (B != BIT_TRUE) continue;
   1236 
   1237     switch (IB) {
   1238     case BIT_FALSE:
   1239       // The bit is meant to be false, so emit a check to see if it is true.
   1240       PositiveMask.setBit(i);
   1241       break;
   1242     case BIT_TRUE:
   1243       // The bit is meant to be true, so emit a check to see if it is false.
   1244       NegativeMask.setBit(i);
   1245       break;
   1246     default:
   1247       // The bit is not set; this must be an error!
   1248       StringRef Name = AllInstructions[Opc]->TheDef->getName();
   1249       errs() << "SoftFail Conflict: bit SoftFail{" << i << "} in " << Name
   1250              << " is set but Inst{" << i << "} is unset!\n"
   1251              << "  - You can only mark a bit as SoftFail if it is fully defined"
   1252              << " (1/0 - not '?') in Inst\n";
   1253       return;
   1254     }
   1255   }
   1256 
   1257   bool NeedPositiveMask = PositiveMask.getBoolValue();
   1258   bool NeedNegativeMask = NegativeMask.getBoolValue();
   1259 
   1260   if (!NeedPositiveMask && !NeedNegativeMask)
   1261     return;
   1262 
   1263   TableInfo.Table.push_back(MCD::OPC_SoftFail);
   1264 
   1265   SmallString<16> MaskBytes;
   1266   raw_svector_ostream S(MaskBytes);
   1267   if (NeedPositiveMask) {
   1268     encodeULEB128(PositiveMask.getZExtValue(), S);
   1269     for (unsigned i = 0, e = MaskBytes.size(); i != e; ++i)
   1270       TableInfo.Table.push_back(MaskBytes[i]);
   1271   } else
   1272     TableInfo.Table.push_back(0);
   1273   if (NeedNegativeMask) {
   1274     MaskBytes.clear();
   1275     encodeULEB128(NegativeMask.getZExtValue(), S);
   1276     for (unsigned i = 0, e = MaskBytes.size(); i != e; ++i)
   1277       TableInfo.Table.push_back(MaskBytes[i]);
   1278   } else
   1279     TableInfo.Table.push_back(0);
   1280 }
   1281 
   1282 // Emits table entries to decode the singleton.
   1283 void FilterChooser::emitSingletonTableEntry(DecoderTableInfo &TableInfo,
   1284                                             unsigned Opc) const {
   1285   std::vector<unsigned> StartBits;
   1286   std::vector<unsigned> EndBits;
   1287   std::vector<uint64_t> FieldVals;
   1288   insn_t Insn;
   1289   insnWithID(Insn, Opc);
   1290 
   1291   // Look for islands of undecoded bits of the singleton.
   1292   getIslands(StartBits, EndBits, FieldVals, Insn);
   1293 
   1294   unsigned Size = StartBits.size();
   1295 
   1296   // Emit the predicate table entry if one is needed.
   1297   emitPredicateTableEntry(TableInfo, Opc);
   1298 
   1299   // Check any additional encoding fields needed.
   1300   for (unsigned I = Size; I != 0; --I) {
   1301     unsigned NumBits = EndBits[I-1] - StartBits[I-1] + 1;
   1302     TableInfo.Table.push_back(MCD::OPC_CheckField);
   1303     TableInfo.Table.push_back(StartBits[I-1]);
   1304     TableInfo.Table.push_back(NumBits);
   1305     uint8_t Buffer[8], *p;
   1306     encodeULEB128(FieldVals[I-1], Buffer);
   1307     for (p = Buffer; *p >= 128 ; ++p)
   1308       TableInfo.Table.push_back(*p);
   1309     TableInfo.Table.push_back(*p);
   1310     // Push location for NumToSkip backpatching.
   1311     TableInfo.FixupStack.back().push_back(TableInfo.Table.size());
   1312     // The fixup is always 16-bits, so go ahead and allocate the space
   1313     // in the table so all our relative position calculations work OK even
   1314     // before we fully resolve the real value here.
   1315     TableInfo.Table.push_back(0);
   1316     TableInfo.Table.push_back(0);
   1317   }
   1318 
   1319   // Check for soft failure of the match.
   1320   emitSoftFailTableEntry(TableInfo, Opc);
   1321 
   1322   bool HasCompleteDecoder;
   1323   unsigned DIdx = getDecoderIndex(TableInfo.Decoders, Opc, HasCompleteDecoder);
   1324 
   1325   // Produce OPC_Decode or OPC_TryDecode opcode based on the information
   1326   // whether the instruction decoder is complete or not. If it is complete
   1327   // then it handles all possible values of remaining variable/unfiltered bits
   1328   // and for any value can determine if the bitpattern is a valid instruction
   1329   // or not. This means OPC_Decode will be the final step in the decoding
   1330   // process. If it is not complete, then the Fail return code from the
   1331   // decoder method indicates that additional processing should be done to see
   1332   // if there is any other instruction that also matches the bitpattern and
   1333   // can decode it.
   1334   TableInfo.Table.push_back(HasCompleteDecoder ? MCD::OPC_Decode :
   1335       MCD::OPC_TryDecode);
   1336   uint8_t Buffer[8], *p;
   1337   encodeULEB128(Opc, Buffer);
   1338   for (p = Buffer; *p >= 128 ; ++p)
   1339     TableInfo.Table.push_back(*p);
   1340   TableInfo.Table.push_back(*p);
   1341 
   1342   SmallString<16> Bytes;
   1343   raw_svector_ostream S(Bytes);
   1344   encodeULEB128(DIdx, S);
   1345 
   1346   // Decoder index
   1347   for (unsigned i = 0, e = Bytes.size(); i != e; ++i)
   1348     TableInfo.Table.push_back(Bytes[i]);
   1349 
   1350   if (!HasCompleteDecoder) {
   1351     // Push location for NumToSkip backpatching.
   1352     TableInfo.FixupStack.back().push_back(TableInfo.Table.size());
   1353     // Allocate the space for the fixup.
   1354     TableInfo.Table.push_back(0);
   1355     TableInfo.Table.push_back(0);
   1356   }
   1357 }
   1358 
   1359 // Emits table entries to decode the singleton, and then to decode the rest.
   1360 void FilterChooser::emitSingletonTableEntry(DecoderTableInfo &TableInfo,
   1361                                             const Filter &Best) const {
   1362   unsigned Opc = Best.getSingletonOpc();
   1363 
   1364   // complex singletons need predicate checks from the first singleton
   1365   // to refer forward to the variable filterchooser that follows.
   1366   TableInfo.FixupStack.emplace_back();
   1367 
   1368   emitSingletonTableEntry(TableInfo, Opc);
   1369 
   1370   resolveTableFixups(TableInfo.Table, TableInfo.FixupStack.back(),
   1371                      TableInfo.Table.size());
   1372   TableInfo.FixupStack.pop_back();
   1373 
   1374   Best.getVariableFC().emitTableEntries(TableInfo);
   1375 }
   1376 
   1377 
   1378 // Assign a single filter and run with it.  Top level API client can initialize
   1379 // with a single filter to start the filtering process.
   1380 void FilterChooser::runSingleFilter(unsigned startBit, unsigned numBit,
   1381                                     bool mixed) {
   1382   Filters.clear();
   1383   Filters.emplace_back(*this, startBit, numBit, true);
   1384   BestIndex = 0; // Sole Filter instance to choose from.
   1385   bestFilter().recurse();
   1386 }
   1387 
   1388 // reportRegion is a helper function for filterProcessor to mark a region as
   1389 // eligible for use as a filter region.
   1390 void FilterChooser::reportRegion(bitAttr_t RA, unsigned StartBit,
   1391                                  unsigned BitIndex, bool AllowMixed) {
   1392   if (RA == ATTR_MIXED && AllowMixed)
   1393     Filters.emplace_back(*this, StartBit, BitIndex - StartBit, true);
   1394   else if (RA == ATTR_ALL_SET && !AllowMixed)
   1395     Filters.emplace_back(*this, StartBit, BitIndex - StartBit, false);
   1396 }
   1397 
   1398 // FilterProcessor scans the well-known encoding bits of the instructions and
   1399 // builds up a list of candidate filters.  It chooses the best filter and
   1400 // recursively descends down the decoding tree.
   1401 bool FilterChooser::filterProcessor(bool AllowMixed, bool Greedy) {
   1402   Filters.clear();
   1403   BestIndex = -1;
   1404   unsigned numInstructions = Opcodes.size();
   1405 
   1406   assert(numInstructions && "Filter created with no instructions");
   1407 
   1408   // No further filtering is necessary.
   1409   if (numInstructions == 1)
   1410     return true;
   1411 
   1412   // Heuristics.  See also doFilter()'s "Heuristics" comment when num of
   1413   // instructions is 3.
   1414   if (AllowMixed && !Greedy) {
   1415     assert(numInstructions == 3);
   1416 
   1417     for (unsigned i = 0; i < Opcodes.size(); ++i) {
   1418       std::vector<unsigned> StartBits;
   1419       std::vector<unsigned> EndBits;
   1420       std::vector<uint64_t> FieldVals;
   1421       insn_t Insn;
   1422 
   1423       insnWithID(Insn, Opcodes[i]);
   1424 
   1425       // Look for islands of undecoded bits of any instruction.
   1426       if (getIslands(StartBits, EndBits, FieldVals, Insn) > 0) {
   1427         // Found an instruction with island(s).  Now just assign a filter.
   1428         runSingleFilter(StartBits[0], EndBits[0] - StartBits[0] + 1, true);
   1429         return true;
   1430       }
   1431     }
   1432   }
   1433 
   1434   unsigned BitIndex;
   1435 
   1436   // We maintain BIT_WIDTH copies of the bitAttrs automaton.
   1437   // The automaton consumes the corresponding bit from each
   1438   // instruction.
   1439   //
   1440   //   Input symbols: 0, 1, and _ (unset).
   1441   //   States:        NONE, FILTERED, ALL_SET, ALL_UNSET, and MIXED.
   1442   //   Initial state: NONE.
   1443   //
   1444   // (NONE) ------- [01] -> (ALL_SET)
   1445   // (NONE) ------- _ ----> (ALL_UNSET)
   1446   // (ALL_SET) ---- [01] -> (ALL_SET)
   1447   // (ALL_SET) ---- _ ----> (MIXED)
   1448   // (ALL_UNSET) -- [01] -> (MIXED)
   1449   // (ALL_UNSET) -- _ ----> (ALL_UNSET)
   1450   // (MIXED) ------ . ----> (MIXED)
   1451   // (FILTERED)---- . ----> (FILTERED)
   1452 
   1453   std::vector<bitAttr_t> bitAttrs;
   1454 
   1455   // FILTERED bit positions provide no entropy and are not worthy of pursuing.
   1456   // Filter::recurse() set either BIT_TRUE or BIT_FALSE for each position.
   1457   for (BitIndex = 0; BitIndex < BitWidth; ++BitIndex)
   1458     if (FilterBitValues[BitIndex] == BIT_TRUE ||
   1459         FilterBitValues[BitIndex] == BIT_FALSE)
   1460       bitAttrs.push_back(ATTR_FILTERED);
   1461     else
   1462       bitAttrs.push_back(ATTR_NONE);
   1463 
   1464   for (unsigned InsnIndex = 0; InsnIndex < numInstructions; ++InsnIndex) {
   1465     insn_t insn;
   1466 
   1467     insnWithID(insn, Opcodes[InsnIndex]);
   1468 
   1469     for (BitIndex = 0; BitIndex < BitWidth; ++BitIndex) {
   1470       switch (bitAttrs[BitIndex]) {
   1471       case ATTR_NONE:
   1472         if (insn[BitIndex] == BIT_UNSET)
   1473           bitAttrs[BitIndex] = ATTR_ALL_UNSET;
   1474         else
   1475           bitAttrs[BitIndex] = ATTR_ALL_SET;
   1476         break;
   1477       case ATTR_ALL_SET:
   1478         if (insn[BitIndex] == BIT_UNSET)
   1479           bitAttrs[BitIndex] = ATTR_MIXED;
   1480         break;
   1481       case ATTR_ALL_UNSET:
   1482         if (insn[BitIndex] != BIT_UNSET)
   1483           bitAttrs[BitIndex] = ATTR_MIXED;
   1484         break;
   1485       case ATTR_MIXED:
   1486       case ATTR_FILTERED:
   1487         break;
   1488       }
   1489     }
   1490   }
   1491 
   1492   // The regionAttr automaton consumes the bitAttrs automatons' state,
   1493   // lowest-to-highest.
   1494   //
   1495   //   Input symbols: F(iltered), (all_)S(et), (all_)U(nset), M(ixed)
   1496   //   States:        NONE, ALL_SET, MIXED
   1497   //   Initial state: NONE
   1498   //
   1499   // (NONE) ----- F --> (NONE)
   1500   // (NONE) ----- S --> (ALL_SET)     ; and set region start
   1501   // (NONE) ----- U --> (NONE)
   1502   // (NONE) ----- M --> (MIXED)       ; and set region start
   1503   // (ALL_SET) -- F --> (NONE)        ; and report an ALL_SET region
   1504   // (ALL_SET) -- S --> (ALL_SET)
   1505   // (ALL_SET) -- U --> (NONE)        ; and report an ALL_SET region
   1506   // (ALL_SET) -- M --> (MIXED)       ; and report an ALL_SET region
   1507   // (MIXED) ---- F --> (NONE)        ; and report a MIXED region
   1508   // (MIXED) ---- S --> (ALL_SET)     ; and report a MIXED region
   1509   // (MIXED) ---- U --> (NONE)        ; and report a MIXED region
   1510   // (MIXED) ---- M --> (MIXED)
   1511 
   1512   bitAttr_t RA = ATTR_NONE;
   1513   unsigned StartBit = 0;
   1514 
   1515   for (BitIndex = 0; BitIndex < BitWidth; ++BitIndex) {
   1516     bitAttr_t bitAttr = bitAttrs[BitIndex];
   1517 
   1518     assert(bitAttr != ATTR_NONE && "Bit without attributes");
   1519 
   1520     switch (RA) {
   1521     case ATTR_NONE:
   1522       switch (bitAttr) {
   1523       case ATTR_FILTERED:
   1524         break;
   1525       case ATTR_ALL_SET:
   1526         StartBit = BitIndex;
   1527         RA = ATTR_ALL_SET;
   1528         break;
   1529       case ATTR_ALL_UNSET:
   1530         break;
   1531       case ATTR_MIXED:
   1532         StartBit = BitIndex;
   1533         RA = ATTR_MIXED;
   1534         break;
   1535       default:
   1536         llvm_unreachable("Unexpected bitAttr!");
   1537       }
   1538       break;
   1539     case ATTR_ALL_SET:
   1540       switch (bitAttr) {
   1541       case ATTR_FILTERED:
   1542         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1543         RA = ATTR_NONE;
   1544         break;
   1545       case ATTR_ALL_SET:
   1546         break;
   1547       case ATTR_ALL_UNSET:
   1548         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1549         RA = ATTR_NONE;
   1550         break;
   1551       case ATTR_MIXED:
   1552         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1553         StartBit = BitIndex;
   1554         RA = ATTR_MIXED;
   1555         break;
   1556       default:
   1557         llvm_unreachable("Unexpected bitAttr!");
   1558       }
   1559       break;
   1560     case ATTR_MIXED:
   1561       switch (bitAttr) {
   1562       case ATTR_FILTERED:
   1563         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1564         StartBit = BitIndex;
   1565         RA = ATTR_NONE;
   1566         break;
   1567       case ATTR_ALL_SET:
   1568         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1569         StartBit = BitIndex;
   1570         RA = ATTR_ALL_SET;
   1571         break;
   1572       case ATTR_ALL_UNSET:
   1573         reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1574         RA = ATTR_NONE;
   1575         break;
   1576       case ATTR_MIXED:
   1577         break;
   1578       default:
   1579         llvm_unreachable("Unexpected bitAttr!");
   1580       }
   1581       break;
   1582     case ATTR_ALL_UNSET:
   1583       llvm_unreachable("regionAttr state machine has no ATTR_UNSET state");
   1584     case ATTR_FILTERED:
   1585       llvm_unreachable("regionAttr state machine has no ATTR_FILTERED state");
   1586     }
   1587   }
   1588 
   1589   // At the end, if we're still in ALL_SET or MIXED states, report a region
   1590   switch (RA) {
   1591   case ATTR_NONE:
   1592     break;
   1593   case ATTR_FILTERED:
   1594     break;
   1595   case ATTR_ALL_SET:
   1596     reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1597     break;
   1598   case ATTR_ALL_UNSET:
   1599     break;
   1600   case ATTR_MIXED:
   1601     reportRegion(RA, StartBit, BitIndex, AllowMixed);
   1602     break;
   1603   }
   1604 
   1605   // We have finished with the filter processings.  Now it's time to choose
   1606   // the best performing filter.
   1607   BestIndex = 0;
   1608   bool AllUseless = true;
   1609   unsigned BestScore = 0;
   1610 
   1611   for (unsigned i = 0, e = Filters.size(); i != e; ++i) {
   1612     unsigned Usefulness = Filters[i].usefulness();
   1613 
   1614     if (Usefulness)
   1615       AllUseless = false;
   1616 
   1617     if (Usefulness > BestScore) {
   1618       BestIndex = i;
   1619       BestScore = Usefulness;
   1620     }
   1621   }
   1622 
   1623   if (!AllUseless)
   1624     bestFilter().recurse();
   1625 
   1626   return !AllUseless;
   1627 } // end of FilterChooser::filterProcessor(bool)
   1628 
   1629 // Decides on the best configuration of filter(s) to use in order to decode
   1630 // the instructions.  A conflict of instructions may occur, in which case we
   1631 // dump the conflict set to the standard error.
   1632 void FilterChooser::doFilter() {
   1633   unsigned Num = Opcodes.size();
   1634   assert(Num && "FilterChooser created with no instructions");
   1635 
   1636   // Try regions of consecutive known bit values first.
   1637   if (filterProcessor(false))
   1638     return;
   1639 
   1640   // Then regions of mixed bits (both known and unitialized bit values allowed).
   1641   if (filterProcessor(true))
   1642     return;
   1643 
   1644   // Heuristics to cope with conflict set {t2CMPrs, t2SUBSrr, t2SUBSrs} where
   1645   // no single instruction for the maximum ATTR_MIXED region Inst{14-4} has a
   1646   // well-known encoding pattern.  In such case, we backtrack and scan for the
   1647   // the very first consecutive ATTR_ALL_SET region and assign a filter to it.
   1648   if (Num == 3 && filterProcessor(true, false))
   1649     return;
   1650 
   1651   // If we come to here, the instruction decoding has failed.
   1652   // Set the BestIndex to -1 to indicate so.
   1653   BestIndex = -1;
   1654 }
   1655 
   1656 // emitTableEntries - Emit state machine entries to decode our share of
   1657 // instructions.
   1658 void FilterChooser::emitTableEntries(DecoderTableInfo &TableInfo) const {
   1659   if (Opcodes.size() == 1) {
   1660     // There is only one instruction in the set, which is great!
   1661     // Call emitSingletonDecoder() to see whether there are any remaining
   1662     // encodings bits.
   1663     emitSingletonTableEntry(TableInfo, Opcodes[0]);
   1664     return;
   1665   }
   1666 
   1667   // Choose the best filter to do the decodings!
   1668   if (BestIndex != -1) {
   1669     const Filter &Best = Filters[BestIndex];
   1670     if (Best.getNumFiltered() == 1)
   1671       emitSingletonTableEntry(TableInfo, Best);
   1672     else
   1673       Best.emitTableEntry(TableInfo);
   1674     return;
   1675   }
   1676 
   1677   // We don't know how to decode these instructions!  Dump the
   1678   // conflict set and bail.
   1679 
   1680   // Print out useful conflict information for postmortem analysis.
   1681   errs() << "Decoding Conflict:\n";
   1682 
   1683   dumpStack(errs(), "\t\t");
   1684 
   1685   for (unsigned i = 0; i < Opcodes.size(); ++i) {
   1686     const std::string &Name = nameWithID(Opcodes[i]);
   1687 
   1688     errs() << '\t' << Name << " ";
   1689     dumpBits(errs(),
   1690              getBitsField(*AllInstructions[Opcodes[i]]->TheDef, "Inst"));
   1691     errs() << '\n';
   1692   }
   1693 }
   1694 
   1695 static bool populateInstruction(CodeGenTarget &Target,
   1696                        const CodeGenInstruction &CGI, unsigned Opc,
   1697                        std::map<unsigned, std::vector<OperandInfo> > &Operands){
   1698   const Record &Def = *CGI.TheDef;
   1699   // If all the bit positions are not specified; do not decode this instruction.
   1700   // We are bound to fail!  For proper disassembly, the well-known encoding bits
   1701   // of the instruction must be fully specified.
   1702 
   1703   BitsInit &Bits = getBitsField(Def, "Inst");
   1704   if (Bits.allInComplete()) return false;
   1705 
   1706   std::vector<OperandInfo> InsnOperands;
   1707 
   1708   // If the instruction has specified a custom decoding hook, use that instead
   1709   // of trying to auto-generate the decoder.
   1710   std::string InstDecoder = Def.getValueAsString("DecoderMethod");
   1711   if (InstDecoder != "") {
   1712     bool HasCompleteInstDecoder = Def.getValueAsBit("hasCompleteDecoder");
   1713     InsnOperands.push_back(OperandInfo(InstDecoder, HasCompleteInstDecoder));
   1714     Operands[Opc] = InsnOperands;
   1715     return true;
   1716   }
   1717 
   1718   // Generate a description of the operand of the instruction that we know
   1719   // how to decode automatically.
   1720   // FIXME: We'll need to have a way to manually override this as needed.
   1721 
   1722   // Gather the outputs/inputs of the instruction, so we can find their
   1723   // positions in the encoding.  This assumes for now that they appear in the
   1724   // MCInst in the order that they're listed.
   1725   std::vector<std::pair<Init*, std::string> > InOutOperands;
   1726   DagInit *Out  = Def.getValueAsDag("OutOperandList");
   1727   DagInit *In  = Def.getValueAsDag("InOperandList");
   1728   for (unsigned i = 0; i < Out->getNumArgs(); ++i)
   1729     InOutOperands.push_back(std::make_pair(Out->getArg(i), Out->getArgName(i)));
   1730   for (unsigned i = 0; i < In->getNumArgs(); ++i)
   1731     InOutOperands.push_back(std::make_pair(In->getArg(i), In->getArgName(i)));
   1732 
   1733   // Search for tied operands, so that we can correctly instantiate
   1734   // operands that are not explicitly represented in the encoding.
   1735   std::map<std::string, std::string> TiedNames;
   1736   for (unsigned i = 0; i < CGI.Operands.size(); ++i) {
   1737     int tiedTo = CGI.Operands[i].getTiedRegister();
   1738     if (tiedTo != -1) {
   1739       std::pair<unsigned, unsigned> SO =
   1740         CGI.Operands.getSubOperandNumber(tiedTo);
   1741       TiedNames[InOutOperands[i].second] = InOutOperands[SO.first].second;
   1742       TiedNames[InOutOperands[SO.first].second] = InOutOperands[i].second;
   1743     }
   1744   }
   1745 
   1746   std::map<std::string, std::vector<OperandInfo> > NumberedInsnOperands;
   1747   std::set<std::string> NumberedInsnOperandsNoTie;
   1748   if (Target.getInstructionSet()->
   1749         getValueAsBit("decodePositionallyEncodedOperands")) {
   1750     const std::vector<RecordVal> &Vals = Def.getValues();
   1751     unsigned NumberedOp = 0;
   1752 
   1753     std::set<unsigned> NamedOpIndices;
   1754     if (Target.getInstructionSet()->
   1755          getValueAsBit("noNamedPositionallyEncodedOperands"))
   1756       // Collect the set of operand indices that might correspond to named
   1757       // operand, and skip these when assigning operands based on position.
   1758       for (unsigned i = 0, e = Vals.size(); i != e; ++i) {
   1759         unsigned OpIdx;
   1760         if (!CGI.Operands.hasOperandNamed(Vals[i].getName(), OpIdx))
   1761           continue;
   1762 
   1763         NamedOpIndices.insert(OpIdx);
   1764       }
   1765 
   1766     for (unsigned i = 0, e = Vals.size(); i != e; ++i) {
   1767       // Ignore fixed fields in the record, we're looking for values like:
   1768       //    bits<5> RST = { ?, ?, ?, ?, ? };
   1769       if (Vals[i].getPrefix() || Vals[i].getValue()->isComplete())
   1770         continue;
   1771 
   1772       // Determine if Vals[i] actually contributes to the Inst encoding.
   1773       unsigned bi = 0;
   1774       for (; bi < Bits.getNumBits(); ++bi) {
   1775         VarInit *Var = nullptr;
   1776         VarBitInit *BI = dyn_cast<VarBitInit>(Bits.getBit(bi));
   1777         if (BI)
   1778           Var = dyn_cast<VarInit>(BI->getBitVar());
   1779         else
   1780           Var = dyn_cast<VarInit>(Bits.getBit(bi));
   1781 
   1782         if (Var && Var->getName() == Vals[i].getName())
   1783           break;
   1784       }
   1785 
   1786       if (bi == Bits.getNumBits())
   1787         continue;
   1788 
   1789       // Skip variables that correspond to explicitly-named operands.
   1790       unsigned OpIdx;
   1791       if (CGI.Operands.hasOperandNamed(Vals[i].getName(), OpIdx))
   1792         continue;
   1793 
   1794       // Get the bit range for this operand:
   1795       unsigned bitStart = bi++, bitWidth = 1;
   1796       for (; bi < Bits.getNumBits(); ++bi) {
   1797         VarInit *Var = nullptr;
   1798         VarBitInit *BI = dyn_cast<VarBitInit>(Bits.getBit(bi));
   1799         if (BI)
   1800           Var = dyn_cast<VarInit>(BI->getBitVar());
   1801         else
   1802           Var = dyn_cast<VarInit>(Bits.getBit(bi));
   1803 
   1804         if (!Var)
   1805           break;
   1806 
   1807         if (Var->getName() != Vals[i].getName())
   1808           break;
   1809 
   1810         ++bitWidth;
   1811       }
   1812 
   1813       unsigned NumberOps = CGI.Operands.size();
   1814       while (NumberedOp < NumberOps &&
   1815              (CGI.Operands.isFlatOperandNotEmitted(NumberedOp) ||
   1816               (!NamedOpIndices.empty() && NamedOpIndices.count(
   1817                 CGI.Operands.getSubOperandNumber(NumberedOp).first))))
   1818         ++NumberedOp;
   1819 
   1820       OpIdx = NumberedOp++;
   1821 
   1822       // OpIdx now holds the ordered operand number of Vals[i].
   1823       std::pair<unsigned, unsigned> SO =
   1824         CGI.Operands.getSubOperandNumber(OpIdx);
   1825       const std::string &Name = CGI.Operands[SO.first].Name;
   1826 
   1827       DEBUG(dbgs() << "Numbered operand mapping for " << Def.getName() << ": " <<
   1828                       Name << "(" << SO.first << ", " << SO.second << ") => " <<
   1829                       Vals[i].getName() << "\n");
   1830 
   1831       std::string Decoder = "";
   1832       Record *TypeRecord = CGI.Operands[SO.first].Rec;
   1833 
   1834       RecordVal *DecoderString = TypeRecord->getValue("DecoderMethod");
   1835       StringInit *String = DecoderString ?
   1836         dyn_cast<StringInit>(DecoderString->getValue()) : nullptr;
   1837       if (String && String->getValue() != "")
   1838         Decoder = String->getValue();
   1839 
   1840       if (Decoder == "" &&
   1841           CGI.Operands[SO.first].MIOperandInfo &&
   1842           CGI.Operands[SO.first].MIOperandInfo->getNumArgs()) {
   1843         Init *Arg = CGI.Operands[SO.first].MIOperandInfo->
   1844                       getArg(SO.second);
   1845         if (TypedInit *TI = cast<TypedInit>(Arg)) {
   1846           RecordRecTy *Type = cast<RecordRecTy>(TI->getType());
   1847           TypeRecord = Type->getRecord();
   1848         }
   1849       }
   1850 
   1851       bool isReg = false;
   1852       if (TypeRecord->isSubClassOf("RegisterOperand"))
   1853         TypeRecord = TypeRecord->getValueAsDef("RegClass");
   1854       if (TypeRecord->isSubClassOf("RegisterClass")) {
   1855         Decoder = "Decode" + TypeRecord->getName() + "RegisterClass";
   1856         isReg = true;
   1857       } else if (TypeRecord->isSubClassOf("PointerLikeRegClass")) {
   1858         Decoder = "DecodePointerLikeRegClass" +
   1859                   utostr(TypeRecord->getValueAsInt("RegClassKind"));
   1860         isReg = true;
   1861       }
   1862 
   1863       DecoderString = TypeRecord->getValue("DecoderMethod");
   1864       String = DecoderString ?
   1865         dyn_cast<StringInit>(DecoderString->getValue()) : nullptr;
   1866       if (!isReg && String && String->getValue() != "")
   1867         Decoder = String->getValue();
   1868 
   1869       RecordVal *HasCompleteDecoderVal =
   1870         TypeRecord->getValue("hasCompleteDecoder");
   1871       BitInit *HasCompleteDecoderBit = HasCompleteDecoderVal ?
   1872         dyn_cast<BitInit>(HasCompleteDecoderVal->getValue()) : nullptr;
   1873       bool HasCompleteDecoder = HasCompleteDecoderBit ?
   1874         HasCompleteDecoderBit->getValue() : true;
   1875 
   1876       OperandInfo OpInfo(Decoder, HasCompleteDecoder);
   1877       OpInfo.addField(bitStart, bitWidth, 0);
   1878 
   1879       NumberedInsnOperands[Name].push_back(OpInfo);
   1880 
   1881       // FIXME: For complex operands with custom decoders we can't handle tied
   1882       // sub-operands automatically. Skip those here and assume that this is
   1883       // fixed up elsewhere.
   1884       if (CGI.Operands[SO.first].MIOperandInfo &&
   1885           CGI.Operands[SO.first].MIOperandInfo->getNumArgs() > 1 &&
   1886           String && String->getValue() != "")
   1887         NumberedInsnOperandsNoTie.insert(Name);
   1888     }
   1889   }
   1890 
   1891   // For each operand, see if we can figure out where it is encoded.
   1892   for (const auto &Op : InOutOperands) {
   1893     if (!NumberedInsnOperands[Op.second].empty()) {
   1894       InsnOperands.insert(InsnOperands.end(),
   1895                           NumberedInsnOperands[Op.second].begin(),
   1896                           NumberedInsnOperands[Op.second].end());
   1897       continue;
   1898     }
   1899     if (!NumberedInsnOperands[TiedNames[Op.second]].empty()) {
   1900       if (!NumberedInsnOperandsNoTie.count(TiedNames[Op.second])) {
   1901         // Figure out to which (sub)operand we're tied.
   1902         unsigned i = CGI.Operands.getOperandNamed(TiedNames[Op.second]);
   1903         int tiedTo = CGI.Operands[i].getTiedRegister();
   1904         if (tiedTo == -1) {
   1905           i = CGI.Operands.getOperandNamed(Op.second);
   1906           tiedTo = CGI.Operands[i].getTiedRegister();
   1907         }
   1908 
   1909         if (tiedTo != -1) {
   1910           std::pair<unsigned, unsigned> SO =
   1911             CGI.Operands.getSubOperandNumber(tiedTo);
   1912 
   1913           InsnOperands.push_back(NumberedInsnOperands[TiedNames[Op.second]]
   1914                                    [SO.second]);
   1915         }
   1916       }
   1917       continue;
   1918     }
   1919 
   1920     std::string Decoder = "";
   1921 
   1922     // At this point, we can locate the field, but we need to know how to
   1923     // interpret it.  As a first step, require the target to provide callbacks
   1924     // for decoding register classes.
   1925     // FIXME: This need to be extended to handle instructions with custom
   1926     // decoder methods, and operands with (simple) MIOperandInfo's.
   1927     TypedInit *TI = cast<TypedInit>(Op.first);
   1928     RecordRecTy *Type = cast<RecordRecTy>(TI->getType());
   1929     Record *TypeRecord = Type->getRecord();
   1930     bool isReg = false;
   1931     if (TypeRecord->isSubClassOf("RegisterOperand"))
   1932       TypeRecord = TypeRecord->getValueAsDef("RegClass");
   1933     if (TypeRecord->isSubClassOf("RegisterClass")) {
   1934       Decoder = "Decode" + TypeRecord->getName() + "RegisterClass";
   1935       isReg = true;
   1936     } else if (TypeRecord->isSubClassOf("PointerLikeRegClass")) {
   1937       Decoder = "DecodePointerLikeRegClass" +
   1938                 utostr(TypeRecord->getValueAsInt("RegClassKind"));
   1939       isReg = true;
   1940     }
   1941 
   1942     RecordVal *DecoderString = TypeRecord->getValue("DecoderMethod");
   1943     StringInit *String = DecoderString ?
   1944       dyn_cast<StringInit>(DecoderString->getValue()) : nullptr;
   1945     if (!isReg && String && String->getValue() != "")
   1946       Decoder = String->getValue();
   1947 
   1948     RecordVal *HasCompleteDecoderVal =
   1949       TypeRecord->getValue("hasCompleteDecoder");
   1950     BitInit *HasCompleteDecoderBit = HasCompleteDecoderVal ?
   1951       dyn_cast<BitInit>(HasCompleteDecoderVal->getValue()) : nullptr;
   1952     bool HasCompleteDecoder = HasCompleteDecoderBit ?
   1953       HasCompleteDecoderBit->getValue() : true;
   1954 
   1955     OperandInfo OpInfo(Decoder, HasCompleteDecoder);
   1956     unsigned Base = ~0U;
   1957     unsigned Width = 0;
   1958     unsigned Offset = 0;
   1959 
   1960     for (unsigned bi = 0; bi < Bits.getNumBits(); ++bi) {
   1961       VarInit *Var = nullptr;
   1962       VarBitInit *BI = dyn_cast<VarBitInit>(Bits.getBit(bi));
   1963       if (BI)
   1964         Var = dyn_cast<VarInit>(BI->getBitVar());
   1965       else
   1966         Var = dyn_cast<VarInit>(Bits.getBit(bi));
   1967 
   1968       if (!Var) {
   1969         if (Base != ~0U) {
   1970           OpInfo.addField(Base, Width, Offset);
   1971           Base = ~0U;
   1972           Width = 0;
   1973           Offset = 0;
   1974         }
   1975         continue;
   1976       }
   1977 
   1978       if (Var->getName() != Op.second &&
   1979           Var->getName() != TiedNames[Op.second]) {
   1980         if (Base != ~0U) {
   1981           OpInfo.addField(Base, Width, Offset);
   1982           Base = ~0U;
   1983           Width = 0;
   1984           Offset = 0;
   1985         }
   1986         continue;
   1987       }
   1988 
   1989       if (Base == ~0U) {
   1990         Base = bi;
   1991         Width = 1;
   1992         Offset = BI ? BI->getBitNum() : 0;
   1993       } else if (BI && BI->getBitNum() != Offset + Width) {
   1994         OpInfo.addField(Base, Width, Offset);
   1995         Base = bi;
   1996         Width = 1;
   1997         Offset = BI->getBitNum();
   1998       } else {
   1999         ++Width;
   2000       }
   2001     }
   2002 
   2003     if (Base != ~0U)
   2004       OpInfo.addField(Base, Width, Offset);
   2005 
   2006     if (OpInfo.numFields() > 0)
   2007       InsnOperands.push_back(OpInfo);
   2008   }
   2009 
   2010   Operands[Opc] = InsnOperands;
   2011 
   2012 
   2013 #if 0
   2014   DEBUG({
   2015       // Dumps the instruction encoding bits.
   2016       dumpBits(errs(), Bits);
   2017 
   2018       errs() << '\n';
   2019 
   2020       // Dumps the list of operand info.
   2021       for (unsigned i = 0, e = CGI.Operands.size(); i != e; ++i) {
   2022         const CGIOperandList::OperandInfo &Info = CGI.Operands[i];
   2023         const std::string &OperandName = Info.Name;
   2024         const Record &OperandDef = *Info.Rec;
   2025 
   2026         errs() << "\t" << OperandName << " (" << OperandDef.getName() << ")\n";
   2027       }
   2028     });
   2029 #endif
   2030 
   2031   return true;
   2032 }
   2033 
   2034 // emitFieldFromInstruction - Emit the templated helper function
   2035 // fieldFromInstruction().
   2036 static void emitFieldFromInstruction(formatted_raw_ostream &OS) {
   2037   OS << "// Helper function for extracting fields from encoded instructions.\n"
   2038      << "template<typename InsnType>\n"
   2039    << "static InsnType fieldFromInstruction(InsnType insn, unsigned startBit,\n"
   2040      << "                                     unsigned numBits) {\n"
   2041      << "    assert(startBit + numBits <= (sizeof(InsnType)*8) &&\n"
   2042      << "           \"Instruction field out of bounds!\");\n"
   2043      << "    InsnType fieldMask;\n"
   2044      << "    if (numBits == sizeof(InsnType)*8)\n"
   2045      << "      fieldMask = (InsnType)(-1LL);\n"
   2046      << "    else\n"
   2047      << "      fieldMask = (((InsnType)1 << numBits) - 1) << startBit;\n"
   2048      << "    return (insn & fieldMask) >> startBit;\n"
   2049      << "}\n\n";
   2050 }
   2051 
   2052 // emitDecodeInstruction - Emit the templated helper function
   2053 // decodeInstruction().
   2054 static void emitDecodeInstruction(formatted_raw_ostream &OS) {
   2055   OS << "template<typename InsnType>\n"
   2056      << "static DecodeStatus decodeInstruction(const uint8_t DecodeTable[], MCInst &MI,\n"
   2057      << "                                      InsnType insn, uint64_t Address,\n"
   2058      << "                                      const void *DisAsm,\n"
   2059      << "                                      const MCSubtargetInfo &STI) {\n"
   2060      << "  const FeatureBitset& Bits = STI.getFeatureBits();\n"
   2061      << "\n"
   2062      << "  const uint8_t *Ptr = DecodeTable;\n"
   2063      << "  uint32_t CurFieldValue = 0;\n"
   2064      << "  DecodeStatus S = MCDisassembler::Success;\n"
   2065      << "  for (;;) {\n"
   2066      << "    ptrdiff_t Loc = Ptr - DecodeTable;\n"
   2067      << "    switch (*Ptr) {\n"
   2068      << "    default:\n"
   2069      << "      errs() << Loc << \": Unexpected decode table opcode!\\n\";\n"
   2070      << "      return MCDisassembler::Fail;\n"
   2071      << "    case MCD::OPC_ExtractField: {\n"
   2072      << "      unsigned Start = *++Ptr;\n"
   2073      << "      unsigned Len = *++Ptr;\n"
   2074      << "      ++Ptr;\n"
   2075      << "      CurFieldValue = fieldFromInstruction(insn, Start, Len);\n"
   2076      << "      DEBUG(dbgs() << Loc << \": OPC_ExtractField(\" << Start << \", \"\n"
   2077      << "                   << Len << \"): \" << CurFieldValue << \"\\n\");\n"
   2078      << "      break;\n"
   2079      << "    }\n"
   2080      << "    case MCD::OPC_FilterValue: {\n"
   2081      << "      // Decode the field value.\n"
   2082      << "      unsigned Len;\n"
   2083      << "      InsnType Val = decodeULEB128(++Ptr, &Len);\n"
   2084      << "      Ptr += Len;\n"
   2085      << "      // NumToSkip is a plain 16-bit integer.\n"
   2086      << "      unsigned NumToSkip = *Ptr++;\n"
   2087      << "      NumToSkip |= (*Ptr++) << 8;\n"
   2088      << "\n"
   2089      << "      // Perform the filter operation.\n"
   2090      << "      if (Val != CurFieldValue)\n"
   2091      << "        Ptr += NumToSkip;\n"
   2092      << "      DEBUG(dbgs() << Loc << \": OPC_FilterValue(\" << Val << \", \" << NumToSkip\n"
   2093      << "                   << \"): \" << ((Val != CurFieldValue) ? \"FAIL:\" : \"PASS:\")\n"
   2094      << "                   << \" continuing at \" << (Ptr - DecodeTable) << \"\\n\");\n"
   2095      << "\n"
   2096      << "      break;\n"
   2097      << "    }\n"
   2098      << "    case MCD::OPC_CheckField: {\n"
   2099      << "      unsigned Start = *++Ptr;\n"
   2100      << "      unsigned Len = *++Ptr;\n"
   2101      << "      InsnType FieldValue = fieldFromInstruction(insn, Start, Len);\n"
   2102      << "      // Decode the field value.\n"
   2103      << "      uint32_t ExpectedValue = decodeULEB128(++Ptr, &Len);\n"
   2104      << "      Ptr += Len;\n"
   2105      << "      // NumToSkip is a plain 16-bit integer.\n"
   2106      << "      unsigned NumToSkip = *Ptr++;\n"
   2107      << "      NumToSkip |= (*Ptr++) << 8;\n"
   2108      << "\n"
   2109      << "      // If the actual and expected values don't match, skip.\n"
   2110      << "      if (ExpectedValue != FieldValue)\n"
   2111      << "        Ptr += NumToSkip;\n"
   2112      << "      DEBUG(dbgs() << Loc << \": OPC_CheckField(\" << Start << \", \"\n"
   2113      << "                   << Len << \", \" << ExpectedValue << \", \" << NumToSkip\n"
   2114      << "                   << \"): FieldValue = \" << FieldValue << \", ExpectedValue = \"\n"
   2115      << "                   << ExpectedValue << \": \"\n"
   2116      << "                   << ((ExpectedValue == FieldValue) ? \"PASS\\n\" : \"FAIL\\n\"));\n"
   2117      << "      break;\n"
   2118      << "    }\n"
   2119      << "    case MCD::OPC_CheckPredicate: {\n"
   2120      << "      unsigned Len;\n"
   2121      << "      // Decode the Predicate Index value.\n"
   2122      << "      unsigned PIdx = decodeULEB128(++Ptr, &Len);\n"
   2123      << "      Ptr += Len;\n"
   2124      << "      // NumToSkip is a plain 16-bit integer.\n"
   2125      << "      unsigned NumToSkip = *Ptr++;\n"
   2126      << "      NumToSkip |= (*Ptr++) << 8;\n"
   2127      << "      // Check the predicate.\n"
   2128      << "      bool Pred;\n"
   2129      << "      if (!(Pred = checkDecoderPredicate(PIdx, Bits)))\n"
   2130      << "        Ptr += NumToSkip;\n"
   2131      << "      (void)Pred;\n"
   2132      << "      DEBUG(dbgs() << Loc << \": OPC_CheckPredicate(\" << PIdx << \"): \"\n"
   2133      << "            << (Pred ? \"PASS\\n\" : \"FAIL\\n\"));\n"
   2134      << "\n"
   2135      << "      break;\n"
   2136      << "    }\n"
   2137      << "    case MCD::OPC_Decode: {\n"
   2138      << "      unsigned Len;\n"
   2139      << "      // Decode the Opcode value.\n"
   2140      << "      unsigned Opc = decodeULEB128(++Ptr, &Len);\n"
   2141      << "      Ptr += Len;\n"
   2142      << "      unsigned DecodeIdx = decodeULEB128(Ptr, &Len);\n"
   2143      << "      Ptr += Len;\n"
   2144      << "\n"
   2145      << "      MI.clear();\n"
   2146      << "      MI.setOpcode(Opc);\n"
   2147      << "      bool DecodeComplete;\n"
   2148      << "      S = decodeToMCInst(S, DecodeIdx, insn, MI, Address, DisAsm, DecodeComplete);\n"
   2149      << "      assert(DecodeComplete);\n"
   2150      << "\n"
   2151      << "      DEBUG(dbgs() << Loc << \": OPC_Decode: opcode \" << Opc\n"
   2152      << "                   << \", using decoder \" << DecodeIdx << \": \"\n"
   2153      << "                   << (S != MCDisassembler::Fail ? \"PASS\" : \"FAIL\") << \"\\n\");\n"
   2154      << "      return S;\n"
   2155      << "    }\n"
   2156      << "    case MCD::OPC_TryDecode: {\n"
   2157      << "      unsigned Len;\n"
   2158      << "      // Decode the Opcode value.\n"
   2159      << "      unsigned Opc = decodeULEB128(++Ptr, &Len);\n"
   2160      << "      Ptr += Len;\n"
   2161      << "      unsigned DecodeIdx = decodeULEB128(Ptr, &Len);\n"
   2162      << "      Ptr += Len;\n"
   2163      << "      // NumToSkip is a plain 16-bit integer.\n"
   2164      << "      unsigned NumToSkip = *Ptr++;\n"
   2165      << "      NumToSkip |= (*Ptr++) << 8;\n"
   2166      << "\n"
   2167      << "      // Perform the decode operation.\n"
   2168      << "      MCInst TmpMI;\n"
   2169      << "      TmpMI.setOpcode(Opc);\n"
   2170      << "      bool DecodeComplete;\n"
   2171      << "      S = decodeToMCInst(S, DecodeIdx, insn, TmpMI, Address, DisAsm, DecodeComplete);\n"
   2172      << "      DEBUG(dbgs() << Loc << \": OPC_TryDecode: opcode \" << Opc\n"
   2173      << "                   << \", using decoder \" << DecodeIdx << \": \");\n"
   2174      << "\n"
   2175      << "      if (DecodeComplete) {\n"
   2176      << "        // Decoding complete.\n"
   2177      << "        DEBUG(dbgs() << (S != MCDisassembler::Fail ? \"PASS\" : \"FAIL\") << \"\\n\");\n"
   2178      << "        MI = TmpMI;\n"
   2179      << "        return S;\n"
   2180      << "      } else {\n"
   2181      << "        assert(S == MCDisassembler::Fail);\n"
   2182      << "        // If the decoding was incomplete, skip.\n"
   2183      << "        Ptr += NumToSkip;\n"
   2184      << "        DEBUG(dbgs() << \"FAIL: continuing at \" << (Ptr - DecodeTable) << \"\\n\");\n"
   2185      << "        // Reset decode status. This also drops a SoftFail status that could be\n"
   2186      << "        // set before the decode attempt.\n"
   2187      << "        S = MCDisassembler::Success;\n"
   2188      << "      }\n"
   2189      << "      break;\n"
   2190      << "    }\n"
   2191      << "    case MCD::OPC_SoftFail: {\n"
   2192      << "      // Decode the mask values.\n"
   2193      << "      unsigned Len;\n"
   2194      << "      InsnType PositiveMask = decodeULEB128(++Ptr, &Len);\n"
   2195      << "      Ptr += Len;\n"
   2196      << "      InsnType NegativeMask = decodeULEB128(Ptr, &Len);\n"
   2197      << "      Ptr += Len;\n"
   2198      << "      bool Fail = (insn & PositiveMask) || (~insn & NegativeMask);\n"
   2199      << "      if (Fail)\n"
   2200      << "        S = MCDisassembler::SoftFail;\n"
   2201      << "      DEBUG(dbgs() << Loc << \": OPC_SoftFail: \" << (Fail ? \"FAIL\\n\":\"PASS\\n\"));\n"
   2202      << "      break;\n"
   2203      << "    }\n"
   2204      << "    case MCD::OPC_Fail: {\n"
   2205      << "      DEBUG(dbgs() << Loc << \": OPC_Fail\\n\");\n"
   2206      << "      return MCDisassembler::Fail;\n"
   2207      << "    }\n"
   2208      << "    }\n"
   2209      << "  }\n"
   2210      << "  llvm_unreachable(\"bogosity detected in disassembler state machine!\");\n"
   2211      << "}\n\n";
   2212 }
   2213 
   2214 // Emits disassembler code for instruction decoding.
   2215 void FixedLenDecoderEmitter::run(raw_ostream &o) {
   2216   formatted_raw_ostream OS(o);
   2217   OS << "#include \"llvm/MC/MCInst.h\"\n";
   2218   OS << "#include \"llvm/Support/Debug.h\"\n";
   2219   OS << "#include \"llvm/Support/DataTypes.h\"\n";
   2220   OS << "#include \"llvm/Support/LEB128.h\"\n";
   2221   OS << "#include \"llvm/Support/raw_ostream.h\"\n";
   2222   OS << "#include <assert.h>\n";
   2223   OS << '\n';
   2224   OS << "namespace llvm {\n\n";
   2225 
   2226   emitFieldFromInstruction(OS);
   2227 
   2228   Target.reverseBitsForLittleEndianEncoding();
   2229 
   2230   // Parameterize the decoders based on namespace and instruction width.
   2231   NumberedInstructions = Target.getInstructionsByEnumValue();
   2232   std::map<std::pair<std::string, unsigned>,
   2233            std::vector<unsigned> > OpcMap;
   2234   std::map<unsigned, std::vector<OperandInfo> > Operands;
   2235 
   2236   for (unsigned i = 0; i < NumberedInstructions.size(); ++i) {
   2237     const CodeGenInstruction *Inst = NumberedInstructions[i];
   2238     const Record *Def = Inst->TheDef;
   2239     unsigned Size = Def->getValueAsInt("Size");
   2240     if (Def->getValueAsString("Namespace") == "TargetOpcode" ||
   2241         Def->getValueAsBit("isPseudo") ||
   2242         Def->getValueAsBit("isAsmParserOnly") ||
   2243         Def->getValueAsBit("isCodeGenOnly"))
   2244       continue;
   2245 
   2246     std::string DecoderNamespace = Def->getValueAsString("DecoderNamespace");
   2247 
   2248     if (Size) {
   2249       if (populateInstruction(Target, *Inst, i, Operands)) {
   2250         OpcMap[std::make_pair(DecoderNamespace, Size)].push_back(i);
   2251       }
   2252     }
   2253   }
   2254 
   2255   DecoderTableInfo TableInfo;
   2256   for (const auto &Opc : OpcMap) {
   2257     // Emit the decoder for this namespace+width combination.
   2258     FilterChooser FC(NumberedInstructions, Opc.second, Operands,
   2259                      8*Opc.first.second, this);
   2260 
   2261     // The decode table is cleared for each top level decoder function. The
   2262     // predicates and decoders themselves, however, are shared across all
   2263     // decoders to give more opportunities for uniqueing.
   2264     TableInfo.Table.clear();
   2265     TableInfo.FixupStack.clear();
   2266     TableInfo.Table.reserve(16384);
   2267     TableInfo.FixupStack.emplace_back();
   2268     FC.emitTableEntries(TableInfo);
   2269     // Any NumToSkip fixups in the top level scope can resolve to the
   2270     // OPC_Fail at the end of the table.
   2271     assert(TableInfo.FixupStack.size() == 1 && "fixup stack phasing error!");
   2272     // Resolve any NumToSkip fixups in the current scope.
   2273     resolveTableFixups(TableInfo.Table, TableInfo.FixupStack.back(),
   2274                        TableInfo.Table.size());
   2275     TableInfo.FixupStack.clear();
   2276 
   2277     TableInfo.Table.push_back(MCD::OPC_Fail);
   2278 
   2279     // Print the table to the output stream.
   2280     emitTable(OS, TableInfo.Table, 0, FC.getBitWidth(), Opc.first.first);
   2281     OS.flush();
   2282   }
   2283 
   2284   // Emit the predicate function.
   2285   emitPredicateFunction(OS, TableInfo.Predicates, 0);
   2286 
   2287   // Emit the decoder function.
   2288   emitDecoderFunction(OS, TableInfo.Decoders, 0);
   2289 
   2290   // Emit the main entry point for the decoder, decodeInstruction().
   2291   emitDecodeInstruction(OS);
   2292 
   2293   OS << "\n} // End llvm namespace\n";
   2294 }
   2295 
   2296 namespace llvm {
   2297 
   2298 void EmitFixedLenDecoder(RecordKeeper &RK, raw_ostream &OS,
   2299                          const std::string &PredicateNamespace,
   2300                          const std::string &GPrefix,
   2301                          const std::string &GPostfix, const std::string &ROK,
   2302                          const std::string &RFail, const std::string &L) {
   2303   FixedLenDecoderEmitter(RK, PredicateNamespace, GPrefix, GPostfix,
   2304                          ROK, RFail, L).run(OS);
   2305 }
   2306 
   2307 } // End llvm namespace
   2308