blob: c649a3b19ccf35d5fb2f94d4ebf27476315a2e4d [file] [log] [blame]
Chris Lattnerd6b65252001-10-24 01:15:12 +00001//===- Reader.cpp - Code to read bytecode files ---------------------------===//
Misha Brukman8a96c532005-04-21 21:44:41 +00002//
John Criswellb576c942003-10-20 19:43:21 +00003// The LLVM Compiler Infrastructure
4//
5// This file was developed by the LLVM research group and is distributed under
6// the University of Illinois Open Source License. See LICENSE.TXT for details.
Misha Brukman8a96c532005-04-21 21:44:41 +00007//
John Criswellb576c942003-10-20 19:43:21 +00008//===----------------------------------------------------------------------===//
Chris Lattner00950542001-06-06 20:29:01 +00009//
10// This library implements the functionality defined in llvm/Bytecode/Reader.h
11//
Misha Brukman8a96c532005-04-21 21:44:41 +000012// Note that this library should be as fast as possible, reentrant, and
Chris Lattner00950542001-06-06 20:29:01 +000013// threadsafe!!
14//
Chris Lattner00950542001-06-06 20:29:01 +000015// TODO: Allow passing in an option to ignore the symbol table
16//
Chris Lattnerd6b65252001-10-24 01:15:12 +000017//===----------------------------------------------------------------------===//
Chris Lattner00950542001-06-06 20:29:01 +000018
Reid Spencer060d25d2004-06-29 23:29:38 +000019#include "Reader.h"
20#include "llvm/Bytecode/BytecodeHandler.h"
21#include "llvm/BasicBlock.h"
Chris Lattnerdee199f2005-05-06 22:34:01 +000022#include "llvm/CallingConv.h"
Reid Spencer060d25d2004-06-29 23:29:38 +000023#include "llvm/Constants.h"
Chris Lattner3bc5a602006-01-25 23:08:15 +000024#include "llvm/InlineAsm.h"
Reid Spencer04cde2c2004-07-04 11:33:49 +000025#include "llvm/Instructions.h"
26#include "llvm/SymbolTable.h"
Reid Spencer78d033e2007-01-06 07:24:44 +000027#include "llvm/TypeSymbolTable.h"
Chris Lattner00950542001-06-06 20:29:01 +000028#include "llvm/Bytecode/Format.h"
Chris Lattnerdee199f2005-05-06 22:34:01 +000029#include "llvm/Config/alloca.h"
Reid Spencer060d25d2004-06-29 23:29:38 +000030#include "llvm/Support/GetElementPtrTypeIterator.h"
Reid Spencer17f52c52004-11-06 23:17:23 +000031#include "llvm/Support/Compressor.h"
Jim Laskeycb6682f2005-08-17 19:34:49 +000032#include "llvm/Support/MathExtras.h"
Chris Lattner4c3d3a92007-01-31 19:56:15 +000033#include "llvm/ADT/SmallVector.h"
Reid Spencer551ccae2004-09-01 22:55:40 +000034#include "llvm/ADT/StringExtras.h"
Reid Spencer060d25d2004-06-29 23:29:38 +000035#include <sstream>
Alkis Evlogimenos20aa4742004-09-03 18:19:51 +000036#include <algorithm>
Chris Lattner29b789b2003-11-19 17:27:18 +000037using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000038
Reid Spencer46b002c2004-07-11 17:28:43 +000039namespace {
Chris Lattnercad28bd2005-01-29 00:36:19 +000040 /// @brief A class for maintaining the slot number definition
41 /// as a placeholder for the actual definition for forward constants defs.
42 class ConstantPlaceHolder : public ConstantExpr {
43 ConstantPlaceHolder(); // DO NOT IMPLEMENT
44 void operator=(const ConstantPlaceHolder &); // DO NOT IMPLEMENT
45 public:
Chris Lattner61323322005-01-31 01:11:13 +000046 Use Op;
Misha Brukman8a96c532005-04-21 21:44:41 +000047 ConstantPlaceHolder(const Type *Ty)
Chris Lattner61323322005-01-31 01:11:13 +000048 : ConstantExpr(Ty, Instruction::UserOp1, &Op, 1),
Reid Spencer88cfda22006-12-31 05:44:24 +000049 Op(UndefValue::get(Type::Int32Ty), this) {
Chris Lattner61323322005-01-31 01:11:13 +000050 }
Chris Lattnercad28bd2005-01-29 00:36:19 +000051 };
Reid Spencer46b002c2004-07-11 17:28:43 +000052}
Reid Spencer060d25d2004-06-29 23:29:38 +000053
Reid Spencer24399722004-07-09 22:21:33 +000054// Provide some details on error
Reid Spencer233fe722006-08-22 16:09:19 +000055inline void BytecodeReader::error(const std::string& err) {
56 ErrorMsg = err + " (Vers=" + itostr(RevisionNum) + ", Pos="
57 + itostr(At-MemStart) + ")";
58 longjmp(context,1);
Reid Spencer24399722004-07-09 22:21:33 +000059}
60
Reid Spencer060d25d2004-06-29 23:29:38 +000061//===----------------------------------------------------------------------===//
62// Bytecode Reading Methods
63//===----------------------------------------------------------------------===//
64
Reid Spencer04cde2c2004-07-04 11:33:49 +000065/// Determine if the current block being read contains any more data.
Reid Spencer060d25d2004-06-29 23:29:38 +000066inline bool BytecodeReader::moreInBlock() {
67 return At < BlockEnd;
Chris Lattner00950542001-06-06 20:29:01 +000068}
69
Reid Spencer04cde2c2004-07-04 11:33:49 +000070/// Throw an error if we've read past the end of the current block
Reid Spencer060d25d2004-06-29 23:29:38 +000071inline void BytecodeReader::checkPastBlockEnd(const char * block_name) {
Reid Spencer46b002c2004-07-11 17:28:43 +000072 if (At > BlockEnd)
Chris Lattnera79e7cc2004-10-16 18:18:16 +000073 error(std::string("Attempt to read past the end of ") + block_name +
74 " block.");
Reid Spencer060d25d2004-06-29 23:29:38 +000075}
Chris Lattner36392bc2003-10-08 21:18:57 +000076
Reid Spencer04cde2c2004-07-04 11:33:49 +000077/// Read a whole unsigned integer
Reid Spencer060d25d2004-06-29 23:29:38 +000078inline unsigned BytecodeReader::read_uint() {
Misha Brukman8a96c532005-04-21 21:44:41 +000079 if (At+4 > BlockEnd)
Reid Spencer24399722004-07-09 22:21:33 +000080 error("Ran out of data reading uint!");
Reid Spencer060d25d2004-06-29 23:29:38 +000081 At += 4;
82 return At[-4] | (At[-3] << 8) | (At[-2] << 16) | (At[-1] << 24);
83}
84
Reid Spencer04cde2c2004-07-04 11:33:49 +000085/// Read a variable-bit-rate encoded unsigned integer
Reid Spencer060d25d2004-06-29 23:29:38 +000086inline unsigned BytecodeReader::read_vbr_uint() {
87 unsigned Shift = 0;
88 unsigned Result = 0;
89 BufPtr Save = At;
Misha Brukman8a96c532005-04-21 21:44:41 +000090
Reid Spencer060d25d2004-06-29 23:29:38 +000091 do {
Misha Brukman8a96c532005-04-21 21:44:41 +000092 if (At == BlockEnd)
Reid Spencer24399722004-07-09 22:21:33 +000093 error("Ran out of data reading vbr_uint!");
Reid Spencer060d25d2004-06-29 23:29:38 +000094 Result |= (unsigned)((*At++) & 0x7F) << Shift;
95 Shift += 7;
96 } while (At[-1] & 0x80);
Reid Spencer04cde2c2004-07-04 11:33:49 +000097 if (Handler) Handler->handleVBR32(At-Save);
Reid Spencer060d25d2004-06-29 23:29:38 +000098 return Result;
99}
100
Reid Spencer04cde2c2004-07-04 11:33:49 +0000101/// Read a variable-bit-rate encoded unsigned 64-bit integer.
Reid Spencer060d25d2004-06-29 23:29:38 +0000102inline uint64_t BytecodeReader::read_vbr_uint64() {
103 unsigned Shift = 0;
104 uint64_t Result = 0;
105 BufPtr Save = At;
Misha Brukman8a96c532005-04-21 21:44:41 +0000106
Reid Spencer060d25d2004-06-29 23:29:38 +0000107 do {
Misha Brukman8a96c532005-04-21 21:44:41 +0000108 if (At == BlockEnd)
Reid Spencer24399722004-07-09 22:21:33 +0000109 error("Ran out of data reading vbr_uint64!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000110 Result |= (uint64_t)((*At++) & 0x7F) << Shift;
111 Shift += 7;
112 } while (At[-1] & 0x80);
Reid Spencer04cde2c2004-07-04 11:33:49 +0000113 if (Handler) Handler->handleVBR64(At-Save);
Reid Spencer060d25d2004-06-29 23:29:38 +0000114 return Result;
115}
116
Reid Spencer04cde2c2004-07-04 11:33:49 +0000117/// Read a variable-bit-rate encoded signed 64-bit integer.
Reid Spencer060d25d2004-06-29 23:29:38 +0000118inline int64_t BytecodeReader::read_vbr_int64() {
119 uint64_t R = read_vbr_uint64();
120 if (R & 1) {
121 if (R != 1)
122 return -(int64_t)(R >> 1);
123 else // There is no such thing as -0 with integers. "-0" really means
124 // 0x8000000000000000.
125 return 1LL << 63;
126 } else
127 return (int64_t)(R >> 1);
128}
129
Reid Spencer04cde2c2004-07-04 11:33:49 +0000130/// Read a pascal-style string (length followed by text)
Reid Spencer060d25d2004-06-29 23:29:38 +0000131inline std::string BytecodeReader::read_str() {
132 unsigned Size = read_vbr_uint();
133 const unsigned char *OldAt = At;
134 At += Size;
135 if (At > BlockEnd) // Size invalid?
Reid Spencer24399722004-07-09 22:21:33 +0000136 error("Ran out of data reading a string!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000137 return std::string((char*)OldAt, Size);
138}
139
Reid Spencer04cde2c2004-07-04 11:33:49 +0000140/// Read an arbitrary block of data
Reid Spencer060d25d2004-06-29 23:29:38 +0000141inline void BytecodeReader::read_data(void *Ptr, void *End) {
142 unsigned char *Start = (unsigned char *)Ptr;
143 unsigned Amount = (unsigned char *)End - Start;
Misha Brukman8a96c532005-04-21 21:44:41 +0000144 if (At+Amount > BlockEnd)
Reid Spencer24399722004-07-09 22:21:33 +0000145 error("Ran out of data!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000146 std::copy(At, At+Amount, Start);
147 At += Amount;
148}
149
Reid Spencer46b002c2004-07-11 17:28:43 +0000150/// Read a float value in little-endian order
151inline void BytecodeReader::read_float(float& FloatVal) {
Reid Spencerada16182004-07-25 21:36:26 +0000152 /// FIXME: This isn't optimal, it has size problems on some platforms
153 /// where FP is not IEEE.
Jim Laskeycb6682f2005-08-17 19:34:49 +0000154 FloatVal = BitsToFloat(At[0] | (At[1] << 8) | (At[2] << 16) | (At[3] << 24));
Reid Spencerada16182004-07-25 21:36:26 +0000155 At+=sizeof(uint32_t);
Reid Spencer46b002c2004-07-11 17:28:43 +0000156}
157
158/// Read a double value in little-endian order
159inline void BytecodeReader::read_double(double& DoubleVal) {
Reid Spencerada16182004-07-25 21:36:26 +0000160 /// FIXME: This isn't optimal, it has size problems on some platforms
161 /// where FP is not IEEE.
Jim Laskeycb6682f2005-08-17 19:34:49 +0000162 DoubleVal = BitsToDouble((uint64_t(At[0]) << 0) | (uint64_t(At[1]) << 8) |
163 (uint64_t(At[2]) << 16) | (uint64_t(At[3]) << 24) |
164 (uint64_t(At[4]) << 32) | (uint64_t(At[5]) << 40) |
165 (uint64_t(At[6]) << 48) | (uint64_t(At[7]) << 56));
Reid Spencerada16182004-07-25 21:36:26 +0000166 At+=sizeof(uint64_t);
Reid Spencer46b002c2004-07-11 17:28:43 +0000167}
168
Reid Spencer04cde2c2004-07-04 11:33:49 +0000169/// Read a block header and obtain its type and size
Reid Spencer060d25d2004-06-29 23:29:38 +0000170inline void BytecodeReader::read_block(unsigned &Type, unsigned &Size) {
Reid Spencerd798a512006-11-14 04:47:22 +0000171 Size = read_uint(); // Read the header
172 Type = Size & 0x1F; // mask low order five bits to get type
173 Size >>= 5; // high order 27 bits is the size
Reid Spencer060d25d2004-06-29 23:29:38 +0000174 BlockStart = At;
Reid Spencer46b002c2004-07-11 17:28:43 +0000175 if (At + Size > BlockEnd)
Reid Spencer24399722004-07-09 22:21:33 +0000176 error("Attempt to size a block past end of memory");
Reid Spencer060d25d2004-06-29 23:29:38 +0000177 BlockEnd = At + Size;
Reid Spencer46b002c2004-07-11 17:28:43 +0000178 if (Handler) Handler->handleBlock(Type, BlockStart, Size);
Reid Spencer04cde2c2004-07-04 11:33:49 +0000179}
180
Reid Spencer060d25d2004-06-29 23:29:38 +0000181//===----------------------------------------------------------------------===//
182// IR Lookup Methods
183//===----------------------------------------------------------------------===//
184
Reid Spencer04cde2c2004-07-04 11:33:49 +0000185/// Determine if a type id has an implicit null value
Reid Spencer46b002c2004-07-11 17:28:43 +0000186inline bool BytecodeReader::hasImplicitNull(unsigned TyID) {
Reid Spencerd798a512006-11-14 04:47:22 +0000187 return TyID != Type::LabelTyID && TyID != Type::VoidTyID;
Reid Spencer060d25d2004-06-29 23:29:38 +0000188}
189
Reid Spencerd2bb8872007-01-30 19:36:46 +0000190/// Obtain a type given a typeid and account for things like function level vs
191/// module level, and the offsetting for the primitive types.
Reid Spencer060d25d2004-06-29 23:29:38 +0000192const Type *BytecodeReader::getType(unsigned ID) {
Reid Spencera54b7cb2007-01-12 07:05:14 +0000193 if (ID <= Type::LastPrimitiveTyID)
Chris Lattnerf70c22b2004-06-17 18:19:28 +0000194 if (const Type *T = Type::getPrimitiveType((Type::TypeID)ID))
Chris Lattner927b1852003-10-09 20:22:47 +0000195 return T; // Asked for a primitive type...
Chris Lattner36392bc2003-10-08 21:18:57 +0000196
197 // Otherwise, derived types need offset...
Chris Lattner89e02532004-01-18 21:08:15 +0000198 ID -= Type::FirstDerivedTyID;
199
Chris Lattner36392bc2003-10-08 21:18:57 +0000200 // Is it a module-level type?
Reid Spencer46b002c2004-07-11 17:28:43 +0000201 if (ID < ModuleTypes.size())
202 return ModuleTypes[ID].get();
Chris Lattner36392bc2003-10-08 21:18:57 +0000203
Reid Spencer46b002c2004-07-11 17:28:43 +0000204 // Nope, is it a function-level type?
205 ID -= ModuleTypes.size();
206 if (ID < FunctionTypes.size())
207 return FunctionTypes[ID].get();
Chris Lattner36392bc2003-10-08 21:18:57 +0000208
Reid Spencer46b002c2004-07-11 17:28:43 +0000209 error("Illegal type reference!");
210 return Type::VoidTy;
Chris Lattner00950542001-06-06 20:29:01 +0000211}
212
Reid Spencer3795ad12006-12-03 05:47:10 +0000213/// This method just saves some coding. It uses read_vbr_uint to read in a
214/// type id, errors that its not the type type, and then calls getType to
215/// return the type value.
Reid Spencerd798a512006-11-14 04:47:22 +0000216inline const Type* BytecodeReader::readType() {
217 return getType(read_vbr_uint());
Reid Spencer04cde2c2004-07-04 11:33:49 +0000218}
219
220/// Get the slot number associated with a type accounting for primitive
Reid Spencerd2bb8872007-01-30 19:36:46 +0000221/// types and function level vs module level.
Reid Spencer060d25d2004-06-29 23:29:38 +0000222unsigned BytecodeReader::getTypeSlot(const Type *Ty) {
223 if (Ty->isPrimitiveType())
224 return Ty->getTypeID();
225
Reid Spencer060d25d2004-06-29 23:29:38 +0000226 // Check the function level types first...
Chris Lattnera79e7cc2004-10-16 18:18:16 +0000227 TypeListTy::iterator I = std::find(FunctionTypes.begin(),
228 FunctionTypes.end(), Ty);
Reid Spencer060d25d2004-06-29 23:29:38 +0000229
230 if (I != FunctionTypes.end())
Misha Brukman8a96c532005-04-21 21:44:41 +0000231 return Type::FirstDerivedTyID + ModuleTypes.size() +
Reid Spencer46b002c2004-07-11 17:28:43 +0000232 (&*I - &FunctionTypes[0]);
Reid Spencer060d25d2004-06-29 23:29:38 +0000233
Chris Lattnereebac5f2005-10-03 21:26:53 +0000234 // If we don't have our cache yet, build it now.
235 if (ModuleTypeIDCache.empty()) {
236 unsigned N = 0;
237 ModuleTypeIDCache.reserve(ModuleTypes.size());
238 for (TypeListTy::iterator I = ModuleTypes.begin(), E = ModuleTypes.end();
239 I != E; ++I, ++N)
240 ModuleTypeIDCache.push_back(std::make_pair(*I, N));
241
242 std::sort(ModuleTypeIDCache.begin(), ModuleTypeIDCache.end());
243 }
244
245 // Binary search the cache for the entry.
246 std::vector<std::pair<const Type*, unsigned> >::iterator IT =
247 std::lower_bound(ModuleTypeIDCache.begin(), ModuleTypeIDCache.end(),
248 std::make_pair(Ty, 0U));
249 if (IT == ModuleTypeIDCache.end() || IT->first != Ty)
Reid Spencer24399722004-07-09 22:21:33 +0000250 error("Didn't find type in ModuleTypes.");
Chris Lattnereebac5f2005-10-03 21:26:53 +0000251
252 return Type::FirstDerivedTyID + IT->second;
Chris Lattner80b97342004-01-17 23:25:43 +0000253}
254
Misha Brukman8a96c532005-04-21 21:44:41 +0000255/// Retrieve a value of a given type and slot number, possibly creating
256/// it if it doesn't already exist.
Reid Spencer060d25d2004-06-29 23:29:38 +0000257Value * BytecodeReader::getValue(unsigned type, unsigned oNum, bool Create) {
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000258 assert(type != Type::LabelTyID && "getValue() cannot get blocks!");
Chris Lattner00950542001-06-06 20:29:01 +0000259 unsigned Num = oNum;
Chris Lattner00950542001-06-06 20:29:01 +0000260
Reid Spencerd2bb8872007-01-30 19:36:46 +0000261 // By default, the global type id is the type id passed in
262 unsigned GlobalTyID = type;
Reid Spencer060d25d2004-06-29 23:29:38 +0000263
Reid Spencerd2bb8872007-01-30 19:36:46 +0000264 if (hasImplicitNull(GlobalTyID)) {
265 const Type *Ty = getType(type);
266 if (!isa<OpaqueType>(Ty)) {
267 if (Num == 0)
268 return Constant::getNullValue(Ty);
269 --Num;
Chris Lattner89e02532004-01-18 21:08:15 +0000270 }
Reid Spencerd2bb8872007-01-30 19:36:46 +0000271 }
Chris Lattner89e02532004-01-18 21:08:15 +0000272
Reid Spencerd2bb8872007-01-30 19:36:46 +0000273 if (GlobalTyID < ModuleValues.size() && ModuleValues[GlobalTyID]) {
274 if (Num < ModuleValues[GlobalTyID]->size())
275 return ModuleValues[GlobalTyID]->getOperand(Num);
276 Num -= ModuleValues[GlobalTyID]->size();
Chris Lattner52e20b02003-03-19 20:54:26 +0000277 }
278
Misha Brukman8a96c532005-04-21 21:44:41 +0000279 if (FunctionValues.size() > type &&
280 FunctionValues[type] &&
Reid Spencer060d25d2004-06-29 23:29:38 +0000281 Num < FunctionValues[type]->size())
282 return FunctionValues[type]->getOperand(Num);
Chris Lattner00950542001-06-06 20:29:01 +0000283
Chris Lattner74734132002-08-17 22:01:27 +0000284 if (!Create) return 0; // Do not create a placeholder?
Chris Lattner00950542001-06-06 20:29:01 +0000285
Reid Spencer551ccae2004-09-01 22:55:40 +0000286 // Did we already create a place holder?
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000287 std::pair<unsigned,unsigned> KeyValue(type, oNum);
Reid Spencer060d25d2004-06-29 23:29:38 +0000288 ForwardReferenceMap::iterator I = ForwardReferences.lower_bound(KeyValue);
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000289 if (I != ForwardReferences.end() && I->first == KeyValue)
290 return I->second; // We have already created this placeholder
291
Reid Spencer551ccae2004-09-01 22:55:40 +0000292 // If the type exists (it should)
293 if (const Type* Ty = getType(type)) {
294 // Create the place holder
295 Value *Val = new Argument(Ty);
296 ForwardReferences.insert(I, std::make_pair(KeyValue, Val));
297 return Val;
298 }
Reid Spencer233fe722006-08-22 16:09:19 +0000299 error("Can't create placeholder for value of type slot #" + utostr(type));
300 return 0; // just silence warning, error calls longjmp
Chris Lattner00950542001-06-06 20:29:01 +0000301}
302
Reid Spencer060d25d2004-06-29 23:29:38 +0000303
Reid Spencer04cde2c2004-07-04 11:33:49 +0000304/// Just like getValue, except that it returns a null pointer
305/// only on error. It always returns a constant (meaning that if the value is
306/// defined, but is not a constant, that is an error). If the specified
Misha Brukman8a96c532005-04-21 21:44:41 +0000307/// constant hasn't been parsed yet, a placeholder is defined and used.
Reid Spencer04cde2c2004-07-04 11:33:49 +0000308/// Later, after the real value is parsed, the placeholder is eliminated.
Reid Spencer060d25d2004-06-29 23:29:38 +0000309Constant* BytecodeReader::getConstantValue(unsigned TypeSlot, unsigned Slot) {
310 if (Value *V = getValue(TypeSlot, Slot, false))
311 if (Constant *C = dyn_cast<Constant>(V))
312 return C; // If we already have the value parsed, just return it
Reid Spencer060d25d2004-06-29 23:29:38 +0000313 else
Misha Brukman8a96c532005-04-21 21:44:41 +0000314 error("Value for slot " + utostr(Slot) +
Reid Spencera86037e2004-07-18 00:12:03 +0000315 " is expected to be a constant!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000316
Chris Lattner389bd042004-12-09 06:19:44 +0000317 std::pair<unsigned, unsigned> Key(TypeSlot, Slot);
Reid Spencer060d25d2004-06-29 23:29:38 +0000318 ConstantRefsType::iterator I = ConstantFwdRefs.lower_bound(Key);
319
320 if (I != ConstantFwdRefs.end() && I->first == Key) {
321 return I->second;
322 } else {
323 // Create a placeholder for the constant reference and
324 // keep track of the fact that we have a forward ref to recycle it
Chris Lattner389bd042004-12-09 06:19:44 +0000325 Constant *C = new ConstantPlaceHolder(getType(TypeSlot));
Misha Brukman8a96c532005-04-21 21:44:41 +0000326
Reid Spencer060d25d2004-06-29 23:29:38 +0000327 // Keep track of the fact that we have a forward ref to recycle it
328 ConstantFwdRefs.insert(I, std::make_pair(Key, C));
329 return C;
330 }
331}
332
333//===----------------------------------------------------------------------===//
334// IR Construction Methods
335//===----------------------------------------------------------------------===//
336
Reid Spencer04cde2c2004-07-04 11:33:49 +0000337/// As values are created, they are inserted into the appropriate place
338/// with this method. The ValueTable argument must be one of ModuleValues
339/// or FunctionValues data members of this class.
Misha Brukman8a96c532005-04-21 21:44:41 +0000340unsigned BytecodeReader::insertValue(Value *Val, unsigned type,
Reid Spencer46b002c2004-07-11 17:28:43 +0000341 ValueTable &ValueTab) {
Reid Spencer060d25d2004-06-29 23:29:38 +0000342 if (ValueTab.size() <= type)
343 ValueTab.resize(type+1);
344
345 if (!ValueTab[type]) ValueTab[type] = new ValueList();
346
347 ValueTab[type]->push_back(Val);
348
Chris Lattneraba5ff52005-05-05 20:57:00 +0000349 bool HasOffset = hasImplicitNull(type) && !isa<OpaqueType>(Val->getType());
Reid Spencer060d25d2004-06-29 23:29:38 +0000350 return ValueTab[type]->size()-1 + HasOffset;
351}
352
Reid Spencer04cde2c2004-07-04 11:33:49 +0000353/// Insert the arguments of a function as new values in the reader.
Reid Spencer46b002c2004-07-11 17:28:43 +0000354void BytecodeReader::insertArguments(Function* F) {
Reid Spencer060d25d2004-06-29 23:29:38 +0000355 const FunctionType *FT = F->getFunctionType();
Chris Lattnere4d5c442005-03-15 04:54:21 +0000356 Function::arg_iterator AI = F->arg_begin();
Reid Spencer060d25d2004-06-29 23:29:38 +0000357 for (FunctionType::param_iterator It = FT->param_begin();
358 It != FT->param_end(); ++It, ++AI)
359 insertValue(AI, getTypeSlot(AI->getType()), FunctionValues);
360}
361
362//===----------------------------------------------------------------------===//
363// Bytecode Parsing Methods
364//===----------------------------------------------------------------------===//
365
Reid Spencer04cde2c2004-07-04 11:33:49 +0000366/// This method parses a single instruction. The instruction is
367/// inserted at the end of the \p BB provided. The arguments of
Misha Brukman44666b12004-09-28 16:57:46 +0000368/// the instruction are provided in the \p Oprnds vector.
Reid Spencer060d25d2004-06-29 23:29:38 +0000369void BytecodeReader::ParseInstruction(std::vector<unsigned> &Oprnds,
Reid Spencer46b002c2004-07-11 17:28:43 +0000370 BasicBlock* BB) {
Reid Spencer060d25d2004-06-29 23:29:38 +0000371 BufPtr SaveAt = At;
372
373 // Clear instruction data
374 Oprnds.clear();
375 unsigned iType = 0;
376 unsigned Opcode = 0;
377 unsigned Op = read_uint();
378
379 // bits Instruction format: Common to all formats
380 // --------------------------
381 // 01-00: Opcode type, fixed to 1.
382 // 07-02: Opcode
383 Opcode = (Op >> 2) & 63;
384 Oprnds.resize((Op >> 0) & 03);
385
386 // Extract the operands
387 switch (Oprnds.size()) {
388 case 1:
389 // bits Instruction format:
390 // --------------------------
391 // 19-08: Resulting type plane
392 // 31-20: Operand #1 (if set to (2^12-1), then zero operands)
393 //
394 iType = (Op >> 8) & 4095;
395 Oprnds[0] = (Op >> 20) & 4095;
396 if (Oprnds[0] == 4095) // Handle special encoding for 0 operands...
397 Oprnds.resize(0);
398 break;
399 case 2:
400 // bits Instruction format:
401 // --------------------------
402 // 15-08: Resulting type plane
403 // 23-16: Operand #1
Misha Brukman8a96c532005-04-21 21:44:41 +0000404 // 31-24: Operand #2
Reid Spencer060d25d2004-06-29 23:29:38 +0000405 //
406 iType = (Op >> 8) & 255;
407 Oprnds[0] = (Op >> 16) & 255;
408 Oprnds[1] = (Op >> 24) & 255;
409 break;
410 case 3:
411 // bits Instruction format:
412 // --------------------------
413 // 13-08: Resulting type plane
414 // 19-14: Operand #1
415 // 25-20: Operand #2
416 // 31-26: Operand #3
417 //
418 iType = (Op >> 8) & 63;
419 Oprnds[0] = (Op >> 14) & 63;
420 Oprnds[1] = (Op >> 20) & 63;
421 Oprnds[2] = (Op >> 26) & 63;
422 break;
423 case 0:
424 At -= 4; // Hrm, try this again...
425 Opcode = read_vbr_uint();
426 Opcode >>= 2;
427 iType = read_vbr_uint();
428
429 unsigned NumOprnds = read_vbr_uint();
430 Oprnds.resize(NumOprnds);
431
432 if (NumOprnds == 0)
Reid Spencer24399722004-07-09 22:21:33 +0000433 error("Zero-argument instruction found; this is invalid.");
Reid Spencer060d25d2004-06-29 23:29:38 +0000434
435 for (unsigned i = 0; i != NumOprnds; ++i)
436 Oprnds[i] = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +0000437 break;
438 }
439
Reid Spencerd798a512006-11-14 04:47:22 +0000440 const Type *InstTy = getType(iType);
Reid Spencer060d25d2004-06-29 23:29:38 +0000441
Reid Spencer1628cec2006-10-26 06:15:43 +0000442 // Make the necessary adjustments for dealing with backwards compatibility
443 // of opcodes.
Reid Spencer3795ad12006-12-03 05:47:10 +0000444 Instruction* Result = 0;
Reid Spencer1628cec2006-10-26 06:15:43 +0000445
Reid Spencer46b002c2004-07-11 17:28:43 +0000446 // We have enough info to inform the handler now.
Reid Spencer1628cec2006-10-26 06:15:43 +0000447 if (Handler)
448 Handler->handleInstruction(Opcode, InstTy, Oprnds, At-SaveAt);
Reid Spencer060d25d2004-06-29 23:29:38 +0000449
Reid Spencer3795ad12006-12-03 05:47:10 +0000450 // First, handle the easy binary operators case
451 if (Opcode >= Instruction::BinaryOpsBegin &&
Reid Spencerc8dab492006-12-03 06:28:54 +0000452 Opcode < Instruction::BinaryOpsEnd && Oprnds.size() == 2) {
Reid Spencer3795ad12006-12-03 05:47:10 +0000453 Result = BinaryOperator::create(Instruction::BinaryOps(Opcode),
454 getValue(iType, Oprnds[0]),
455 getValue(iType, Oprnds[1]));
Reid Spencerc8dab492006-12-03 06:28:54 +0000456 } else {
Reid Spencer1628cec2006-10-26 06:15:43 +0000457 // Indicate that we don't think this is a call instruction (yet).
458 // Process based on the Opcode read
459 switch (Opcode) {
460 default: // There was an error, this shouldn't happen.
461 if (Result == 0)
462 error("Illegal instruction read!");
463 break;
464 case Instruction::VAArg:
465 if (Oprnds.size() != 2)
466 error("Invalid VAArg instruction!");
467 Result = new VAArgInst(getValue(iType, Oprnds[0]),
Reid Spencerd798a512006-11-14 04:47:22 +0000468 getType(Oprnds[1]));
Reid Spencer1628cec2006-10-26 06:15:43 +0000469 break;
470 case Instruction::ExtractElement: {
471 if (Oprnds.size() != 2)
472 error("Invalid extractelement instruction!");
473 Value *V1 = getValue(iType, Oprnds[0]);
Reid Spencera54b7cb2007-01-12 07:05:14 +0000474 Value *V2 = getValue(Int32TySlot, Oprnds[1]);
Chris Lattner59fecec2006-04-08 04:09:19 +0000475
Reid Spencer1628cec2006-10-26 06:15:43 +0000476 if (!ExtractElementInst::isValidOperands(V1, V2))
477 error("Invalid extractelement instruction!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000478
Reid Spencer1628cec2006-10-26 06:15:43 +0000479 Result = new ExtractElementInst(V1, V2);
480 break;
Chris Lattnera65371e2006-05-26 18:42:34 +0000481 }
Reid Spencer1628cec2006-10-26 06:15:43 +0000482 case Instruction::InsertElement: {
483 const PackedType *PackedTy = dyn_cast<PackedType>(InstTy);
484 if (!PackedTy || Oprnds.size() != 3)
485 error("Invalid insertelement instruction!");
486
487 Value *V1 = getValue(iType, Oprnds[0]);
488 Value *V2 = getValue(getTypeSlot(PackedTy->getElementType()),Oprnds[1]);
Reid Spencera54b7cb2007-01-12 07:05:14 +0000489 Value *V3 = getValue(Int32TySlot, Oprnds[2]);
Reid Spencer1628cec2006-10-26 06:15:43 +0000490
491 if (!InsertElementInst::isValidOperands(V1, V2, V3))
492 error("Invalid insertelement instruction!");
493 Result = new InsertElementInst(V1, V2, V3);
494 break;
495 }
496 case Instruction::ShuffleVector: {
497 const PackedType *PackedTy = dyn_cast<PackedType>(InstTy);
498 if (!PackedTy || Oprnds.size() != 3)
499 error("Invalid shufflevector instruction!");
500 Value *V1 = getValue(iType, Oprnds[0]);
501 Value *V2 = getValue(iType, Oprnds[1]);
502 const PackedType *EltTy =
Reid Spencer88cfda22006-12-31 05:44:24 +0000503 PackedType::get(Type::Int32Ty, PackedTy->getNumElements());
Reid Spencer1628cec2006-10-26 06:15:43 +0000504 Value *V3 = getValue(getTypeSlot(EltTy), Oprnds[2]);
505 if (!ShuffleVectorInst::isValidOperands(V1, V2, V3))
506 error("Invalid shufflevector instruction!");
507 Result = new ShuffleVectorInst(V1, V2, V3);
508 break;
509 }
Reid Spencer3da59db2006-11-27 01:05:10 +0000510 case Instruction::Trunc:
511 if (Oprnds.size() != 2)
512 error("Invalid cast instruction!");
513 Result = new TruncInst(getValue(iType, Oprnds[0]),
514 getType(Oprnds[1]));
515 break;
516 case Instruction::ZExt:
517 if (Oprnds.size() != 2)
518 error("Invalid cast instruction!");
519 Result = new ZExtInst(getValue(iType, Oprnds[0]),
520 getType(Oprnds[1]));
521 break;
522 case Instruction::SExt:
Reid Spencer1628cec2006-10-26 06:15:43 +0000523 if (Oprnds.size() != 2)
524 error("Invalid Cast instruction!");
Reid Spencer3da59db2006-11-27 01:05:10 +0000525 Result = new SExtInst(getValue(iType, Oprnds[0]),
Reid Spencerd798a512006-11-14 04:47:22 +0000526 getType(Oprnds[1]));
Reid Spencer1628cec2006-10-26 06:15:43 +0000527 break;
Reid Spencer3da59db2006-11-27 01:05:10 +0000528 case Instruction::FPTrunc:
529 if (Oprnds.size() != 2)
530 error("Invalid cast instruction!");
531 Result = new FPTruncInst(getValue(iType, Oprnds[0]),
532 getType(Oprnds[1]));
533 break;
534 case Instruction::FPExt:
535 if (Oprnds.size() != 2)
536 error("Invalid cast instruction!");
537 Result = new FPExtInst(getValue(iType, Oprnds[0]),
538 getType(Oprnds[1]));
539 break;
540 case Instruction::UIToFP:
541 if (Oprnds.size() != 2)
542 error("Invalid cast instruction!");
543 Result = new UIToFPInst(getValue(iType, Oprnds[0]),
544 getType(Oprnds[1]));
545 break;
546 case Instruction::SIToFP:
547 if (Oprnds.size() != 2)
548 error("Invalid cast instruction!");
549 Result = new SIToFPInst(getValue(iType, Oprnds[0]),
550 getType(Oprnds[1]));
551 break;
552 case Instruction::FPToUI:
553 if (Oprnds.size() != 2)
554 error("Invalid cast instruction!");
555 Result = new FPToUIInst(getValue(iType, Oprnds[0]),
556 getType(Oprnds[1]));
557 break;
558 case Instruction::FPToSI:
559 if (Oprnds.size() != 2)
560 error("Invalid cast instruction!");
561 Result = new FPToSIInst(getValue(iType, Oprnds[0]),
562 getType(Oprnds[1]));
563 break;
564 case Instruction::IntToPtr:
565 if (Oprnds.size() != 2)
566 error("Invalid cast instruction!");
567 Result = new IntToPtrInst(getValue(iType, Oprnds[0]),
568 getType(Oprnds[1]));
569 break;
570 case Instruction::PtrToInt:
571 if (Oprnds.size() != 2)
572 error("Invalid cast instruction!");
573 Result = new PtrToIntInst(getValue(iType, Oprnds[0]),
574 getType(Oprnds[1]));
575 break;
576 case Instruction::BitCast:
577 if (Oprnds.size() != 2)
578 error("Invalid cast instruction!");
579 Result = new BitCastInst(getValue(iType, Oprnds[0]),
580 getType(Oprnds[1]));
581 break;
Reid Spencer1628cec2006-10-26 06:15:43 +0000582 case Instruction::Select:
583 if (Oprnds.size() != 3)
584 error("Invalid Select instruction!");
Reid Spencera54b7cb2007-01-12 07:05:14 +0000585 Result = new SelectInst(getValue(BoolTySlot, Oprnds[0]),
Reid Spencer1628cec2006-10-26 06:15:43 +0000586 getValue(iType, Oprnds[1]),
587 getValue(iType, Oprnds[2]));
588 break;
589 case Instruction::PHI: {
590 if (Oprnds.size() == 0 || (Oprnds.size() & 1))
591 error("Invalid phi node encountered!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000592
Reid Spencer1628cec2006-10-26 06:15:43 +0000593 PHINode *PN = new PHINode(InstTy);
594 PN->reserveOperandSpace(Oprnds.size());
595 for (unsigned i = 0, e = Oprnds.size(); i != e; i += 2)
596 PN->addIncoming(
597 getValue(iType, Oprnds[i]), getBasicBlock(Oprnds[i+1]));
598 Result = PN;
599 break;
600 }
Reid Spencerc8dab492006-12-03 06:28:54 +0000601 case Instruction::ICmp:
602 case Instruction::FCmp:
Reid Spencer9f132762006-12-03 17:17:02 +0000603 if (Oprnds.size() != 3)
604 error("Cmp instructions requires 3 operands");
Reid Spencerc8dab492006-12-03 06:28:54 +0000605 // These instructions encode the comparison predicate as the 3rd operand.
606 Result = CmpInst::create(Instruction::OtherOps(Opcode),
607 static_cast<unsigned short>(Oprnds[2]),
608 getValue(iType, Oprnds[0]), getValue(iType, Oprnds[1]));
609 break;
Reid Spencer1628cec2006-10-26 06:15:43 +0000610 case Instruction::Shl:
Reid Spencer3822ff52006-11-08 06:47:33 +0000611 case Instruction::LShr:
612 case Instruction::AShr:
Reid Spencer1628cec2006-10-26 06:15:43 +0000613 Result = new ShiftInst(Instruction::OtherOps(Opcode),
614 getValue(iType, Oprnds[0]),
Reid Spencera54b7cb2007-01-12 07:05:14 +0000615 getValue(Int8TySlot, Oprnds[1]));
Reid Spencer1628cec2006-10-26 06:15:43 +0000616 break;
617 case Instruction::Ret:
618 if (Oprnds.size() == 0)
619 Result = new ReturnInst();
620 else if (Oprnds.size() == 1)
621 Result = new ReturnInst(getValue(iType, Oprnds[0]));
622 else
623 error("Unrecognized instruction!");
624 break;
625
626 case Instruction::Br:
627 if (Oprnds.size() == 1)
628 Result = new BranchInst(getBasicBlock(Oprnds[0]));
629 else if (Oprnds.size() == 3)
630 Result = new BranchInst(getBasicBlock(Oprnds[0]),
Reid Spencera54b7cb2007-01-12 07:05:14 +0000631 getBasicBlock(Oprnds[1]), getValue(BoolTySlot, Oprnds[2]));
Reid Spencer1628cec2006-10-26 06:15:43 +0000632 else
633 error("Invalid number of operands for a 'br' instruction!");
634 break;
635 case Instruction::Switch: {
636 if (Oprnds.size() & 1)
637 error("Switch statement with odd number of arguments!");
638
639 SwitchInst *I = new SwitchInst(getValue(iType, Oprnds[0]),
640 getBasicBlock(Oprnds[1]),
641 Oprnds.size()/2-1);
642 for (unsigned i = 2, e = Oprnds.size(); i != e; i += 2)
643 I->addCase(cast<ConstantInt>(getValue(iType, Oprnds[i])),
644 getBasicBlock(Oprnds[i+1]));
645 Result = I;
646 break;
647 }
648 case 58: // Call with extra operand for calling conv
649 case 59: // tail call, Fast CC
650 case 60: // normal call, Fast CC
651 case 61: // tail call, C Calling Conv
652 case Instruction::Call: { // Normal Call, C Calling Convention
653 if (Oprnds.size() == 0)
654 error("Invalid call instruction encountered!");
Reid Spencer1628cec2006-10-26 06:15:43 +0000655 Value *F = getValue(iType, Oprnds[0]);
656
657 unsigned CallingConv = CallingConv::C;
658 bool isTailCall = false;
659
660 if (Opcode == 61 || Opcode == 59)
661 isTailCall = true;
662
663 if (Opcode == 58) {
664 isTailCall = Oprnds.back() & 1;
665 CallingConv = Oprnds.back() >> 1;
666 Oprnds.pop_back();
667 } else if (Opcode == 59 || Opcode == 60) {
668 CallingConv = CallingConv::Fast;
669 }
670
671 // Check to make sure we have a pointer to function type
672 const PointerType *PTy = dyn_cast<PointerType>(F->getType());
673 if (PTy == 0) error("Call to non function pointer value!");
674 const FunctionType *FTy = dyn_cast<FunctionType>(PTy->getElementType());
675 if (FTy == 0) error("Call to non function pointer value!");
676
677 std::vector<Value *> Params;
678 if (!FTy->isVarArg()) {
679 FunctionType::param_iterator It = FTy->param_begin();
680
681 for (unsigned i = 1, e = Oprnds.size(); i != e; ++i) {
682 if (It == FTy->param_end())
683 error("Invalid call instruction!");
684 Params.push_back(getValue(getTypeSlot(*It++), Oprnds[i]));
685 }
686 if (It != FTy->param_end())
Reid Spencer24399722004-07-09 22:21:33 +0000687 error("Invalid call instruction!");
Reid Spencer1628cec2006-10-26 06:15:43 +0000688 } else {
689 Oprnds.erase(Oprnds.begin(), Oprnds.begin()+1);
690
691 unsigned FirstVariableOperand;
692 if (Oprnds.size() < FTy->getNumParams())
693 error("Call instruction missing operands!");
694
695 // Read all of the fixed arguments
696 for (unsigned i = 0, e = FTy->getNumParams(); i != e; ++i)
697 Params.push_back(
698 getValue(getTypeSlot(FTy->getParamType(i)),Oprnds[i]));
699
700 FirstVariableOperand = FTy->getNumParams();
701
702 if ((Oprnds.size()-FirstVariableOperand) & 1)
703 error("Invalid call instruction!"); // Must be pairs of type/value
704
705 for (unsigned i = FirstVariableOperand, e = Oprnds.size();
706 i != e; i += 2)
707 Params.push_back(getValue(Oprnds[i], Oprnds[i+1]));
Reid Spencer060d25d2004-06-29 23:29:38 +0000708 }
Reid Spencer060d25d2004-06-29 23:29:38 +0000709
Reid Spencer1628cec2006-10-26 06:15:43 +0000710 Result = new CallInst(F, Params);
711 if (isTailCall) cast<CallInst>(Result)->setTailCall();
712 if (CallingConv) cast<CallInst>(Result)->setCallingConv(CallingConv);
713 break;
Reid Spencer060d25d2004-06-29 23:29:38 +0000714 }
Reid Spencer1628cec2006-10-26 06:15:43 +0000715 case Instruction::Invoke: { // Invoke C CC
716 if (Oprnds.size() < 3)
717 error("Invalid invoke instruction!");
718 Value *F = getValue(iType, Oprnds[0]);
Reid Spencer060d25d2004-06-29 23:29:38 +0000719
Reid Spencer1628cec2006-10-26 06:15:43 +0000720 // Check to make sure we have a pointer to function type
721 const PointerType *PTy = dyn_cast<PointerType>(F->getType());
722 if (PTy == 0)
723 error("Invoke to non function pointer value!");
724 const FunctionType *FTy = dyn_cast<FunctionType>(PTy->getElementType());
725 if (FTy == 0)
726 error("Invoke to non function pointer value!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000727
Reid Spencer1628cec2006-10-26 06:15:43 +0000728 std::vector<Value *> Params;
729 BasicBlock *Normal, *Except;
Reid Spencer3da59db2006-11-27 01:05:10 +0000730 unsigned CallingConv = Oprnds.back();
731 Oprnds.pop_back();
Chris Lattnerdee199f2005-05-06 22:34:01 +0000732
Reid Spencer1628cec2006-10-26 06:15:43 +0000733 if (!FTy->isVarArg()) {
734 Normal = getBasicBlock(Oprnds[1]);
735 Except = getBasicBlock(Oprnds[2]);
Reid Spencer060d25d2004-06-29 23:29:38 +0000736
Reid Spencer1628cec2006-10-26 06:15:43 +0000737 FunctionType::param_iterator It = FTy->param_begin();
738 for (unsigned i = 3, e = Oprnds.size(); i != e; ++i) {
739 if (It == FTy->param_end())
740 error("Invalid invoke instruction!");
741 Params.push_back(getValue(getTypeSlot(*It++), Oprnds[i]));
742 }
743 if (It != FTy->param_end())
Reid Spencer24399722004-07-09 22:21:33 +0000744 error("Invalid invoke instruction!");
Reid Spencer1628cec2006-10-26 06:15:43 +0000745 } else {
746 Oprnds.erase(Oprnds.begin(), Oprnds.begin()+1);
747
748 Normal = getBasicBlock(Oprnds[0]);
749 Except = getBasicBlock(Oprnds[1]);
750
751 unsigned FirstVariableArgument = FTy->getNumParams()+2;
752 for (unsigned i = 2; i != FirstVariableArgument; ++i)
753 Params.push_back(getValue(getTypeSlot(FTy->getParamType(i-2)),
754 Oprnds[i]));
755
756 // Must be type/value pairs. If not, error out.
757 if (Oprnds.size()-FirstVariableArgument & 1)
758 error("Invalid invoke instruction!");
759
760 for (unsigned i = FirstVariableArgument; i < Oprnds.size(); i += 2)
761 Params.push_back(getValue(Oprnds[i], Oprnds[i+1]));
Reid Spencer060d25d2004-06-29 23:29:38 +0000762 }
Reid Spencer060d25d2004-06-29 23:29:38 +0000763
Reid Spencer1628cec2006-10-26 06:15:43 +0000764 Result = new InvokeInst(F, Normal, Except, Params);
765 if (CallingConv) cast<InvokeInst>(Result)->setCallingConv(CallingConv);
766 break;
Reid Spencer060d25d2004-06-29 23:29:38 +0000767 }
Reid Spencer1628cec2006-10-26 06:15:43 +0000768 case Instruction::Malloc: {
769 unsigned Align = 0;
770 if (Oprnds.size() == 2)
771 Align = (1 << Oprnds[1]) >> 1;
772 else if (Oprnds.size() > 2)
773 error("Invalid malloc instruction!");
774 if (!isa<PointerType>(InstTy))
775 error("Invalid malloc instruction!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000776
Reid Spencer1628cec2006-10-26 06:15:43 +0000777 Result = new MallocInst(cast<PointerType>(InstTy)->getElementType(),
Reid Spencera54b7cb2007-01-12 07:05:14 +0000778 getValue(Int32TySlot, Oprnds[0]), Align);
Reid Spencer1628cec2006-10-26 06:15:43 +0000779 break;
780 }
781 case Instruction::Alloca: {
782 unsigned Align = 0;
783 if (Oprnds.size() == 2)
784 Align = (1 << Oprnds[1]) >> 1;
785 else if (Oprnds.size() > 2)
786 error("Invalid alloca instruction!");
787 if (!isa<PointerType>(InstTy))
788 error("Invalid alloca instruction!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000789
Reid Spencer1628cec2006-10-26 06:15:43 +0000790 Result = new AllocaInst(cast<PointerType>(InstTy)->getElementType(),
Reid Spencera54b7cb2007-01-12 07:05:14 +0000791 getValue(Int32TySlot, Oprnds[0]), Align);
Reid Spencer1628cec2006-10-26 06:15:43 +0000792 break;
793 }
794 case Instruction::Free:
795 if (!isa<PointerType>(InstTy))
796 error("Invalid free instruction!");
797 Result = new FreeInst(getValue(iType, Oprnds[0]));
798 break;
799 case Instruction::GetElementPtr: {
800 if (Oprnds.size() == 0 || !isa<PointerType>(InstTy))
Misha Brukman8a96c532005-04-21 21:44:41 +0000801 error("Invalid getelementptr instruction!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000802
Chris Lattner4c3d3a92007-01-31 19:56:15 +0000803 SmallVector<Value*, 8> Idx;
Reid Spencer1628cec2006-10-26 06:15:43 +0000804
805 const Type *NextTy = InstTy;
806 for (unsigned i = 1, e = Oprnds.size(); i != e; ++i) {
807 const CompositeType *TopTy = dyn_cast_or_null<CompositeType>(NextTy);
808 if (!TopTy)
809 error("Invalid getelementptr instruction!");
810
811 unsigned ValIdx = Oprnds[i];
812 unsigned IdxTy = 0;
Reid Spencerd798a512006-11-14 04:47:22 +0000813 // Struct indices are always uints, sequential type indices can be
814 // any of the 32 or 64-bit integer types. The actual choice of
Reid Spencer88cfda22006-12-31 05:44:24 +0000815 // type is encoded in the low bit of the slot number.
Reid Spencerd798a512006-11-14 04:47:22 +0000816 if (isa<StructType>(TopTy))
Reid Spencera54b7cb2007-01-12 07:05:14 +0000817 IdxTy = Int32TySlot;
Reid Spencerd798a512006-11-14 04:47:22 +0000818 else {
Reid Spencer88cfda22006-12-31 05:44:24 +0000819 switch (ValIdx & 1) {
Reid Spencerd798a512006-11-14 04:47:22 +0000820 default:
Reid Spencera54b7cb2007-01-12 07:05:14 +0000821 case 0: IdxTy = Int32TySlot; break;
822 case 1: IdxTy = Int64TySlot; break;
Reid Spencer060d25d2004-06-29 23:29:38 +0000823 }
Reid Spencer88cfda22006-12-31 05:44:24 +0000824 ValIdx >>= 1;
Reid Spencer060d25d2004-06-29 23:29:38 +0000825 }
Reid Spencer1628cec2006-10-26 06:15:43 +0000826 Idx.push_back(getValue(IdxTy, ValIdx));
Chris Lattner4c3d3a92007-01-31 19:56:15 +0000827 NextTy = GetElementPtrInst::getIndexedType(InstTy, &Idx[0], Idx.size(),
828 true);
Reid Spencer060d25d2004-06-29 23:29:38 +0000829 }
830
Chris Lattner4c3d3a92007-01-31 19:56:15 +0000831 Result = new GetElementPtrInst(getValue(iType, Oprnds[0]),
832 &Idx[0], Idx.size());
Reid Spencer1628cec2006-10-26 06:15:43 +0000833 break;
Reid Spencer060d25d2004-06-29 23:29:38 +0000834 }
Reid Spencer1628cec2006-10-26 06:15:43 +0000835 case 62: // volatile load
836 case Instruction::Load:
837 if (Oprnds.size() != 1 || !isa<PointerType>(InstTy))
838 error("Invalid load instruction!");
839 Result = new LoadInst(getValue(iType, Oprnds[0]), "", Opcode == 62);
840 break;
841 case 63: // volatile store
842 case Instruction::Store: {
843 if (!isa<PointerType>(InstTy) || Oprnds.size() != 2)
844 error("Invalid store instruction!");
Reid Spencer060d25d2004-06-29 23:29:38 +0000845
Reid Spencer1628cec2006-10-26 06:15:43 +0000846 Value *Ptr = getValue(iType, Oprnds[1]);
847 const Type *ValTy = cast<PointerType>(Ptr->getType())->getElementType();
848 Result = new StoreInst(getValue(getTypeSlot(ValTy), Oprnds[0]), Ptr,
849 Opcode == 63);
850 break;
851 }
852 case Instruction::Unwind:
853 if (Oprnds.size() != 0) error("Invalid unwind instruction!");
854 Result = new UnwindInst();
855 break;
856 case Instruction::Unreachable:
857 if (Oprnds.size() != 0) error("Invalid unreachable instruction!");
858 Result = new UnreachableInst();
859 break;
860 } // end switch(Opcode)
Reid Spencer3795ad12006-12-03 05:47:10 +0000861 } // end if !Result
Reid Spencer060d25d2004-06-29 23:29:38 +0000862
Reid Spencere1e96c02006-01-19 07:02:16 +0000863 BB->getInstList().push_back(Result);
864
Reid Spencer060d25d2004-06-29 23:29:38 +0000865 unsigned TypeSlot;
866 if (Result->getType() == InstTy)
867 TypeSlot = iType;
868 else
869 TypeSlot = getTypeSlot(Result->getType());
870
871 insertValue(Result, TypeSlot, FunctionValues);
Reid Spencer060d25d2004-06-29 23:29:38 +0000872}
873
Reid Spencer04cde2c2004-07-04 11:33:49 +0000874/// Get a particular numbered basic block, which might be a forward reference.
Reid Spencerd798a512006-11-14 04:47:22 +0000875/// This works together with ParseInstructionList to handle these forward
876/// references in a clean manner. This function is used when constructing
877/// phi, br, switch, and other instructions that reference basic blocks.
878/// Blocks are numbered sequentially as they appear in the function.
Reid Spencer060d25d2004-06-29 23:29:38 +0000879BasicBlock *BytecodeReader::getBasicBlock(unsigned ID) {
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000880 // Make sure there is room in the table...
881 if (ParsedBasicBlocks.size() <= ID) ParsedBasicBlocks.resize(ID+1);
882
Reid Spencerd798a512006-11-14 04:47:22 +0000883 // First check to see if this is a backwards reference, i.e. this block
884 // has already been created, or if the forward reference has already
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000885 // been created.
886 if (ParsedBasicBlocks[ID])
887 return ParsedBasicBlocks[ID];
888
889 // Otherwise, the basic block has not yet been created. Do so and add it to
890 // the ParsedBasicBlocks list.
891 return ParsedBasicBlocks[ID] = new BasicBlock();
892}
893
Reid Spencer04cde2c2004-07-04 11:33:49 +0000894/// Parse all of the BasicBlock's & Instruction's in the body of a function.
Misha Brukman8a96c532005-04-21 21:44:41 +0000895/// In post 1.0 bytecode files, we no longer emit basic block individually,
Reid Spencer04cde2c2004-07-04 11:33:49 +0000896/// in order to avoid per-basic-block overhead.
Reid Spencerd798a512006-11-14 04:47:22 +0000897/// @returns the number of basic blocks encountered.
Reid Spencer060d25d2004-06-29 23:29:38 +0000898unsigned BytecodeReader::ParseInstructionList(Function* F) {
Chris Lattner8d1dbd22003-12-01 07:05:31 +0000899 unsigned BlockNo = 0;
900 std::vector<unsigned> Args;
901
Reid Spencer46b002c2004-07-11 17:28:43 +0000902 while (moreInBlock()) {
903 if (Handler) Handler->handleBasicBlockBegin(BlockNo);
Chris Lattner8d1dbd22003-12-01 07:05:31 +0000904 BasicBlock *BB;
905 if (ParsedBasicBlocks.size() == BlockNo)
906 ParsedBasicBlocks.push_back(BB = new BasicBlock());
907 else if (ParsedBasicBlocks[BlockNo] == 0)
908 BB = ParsedBasicBlocks[BlockNo] = new BasicBlock();
909 else
910 BB = ParsedBasicBlocks[BlockNo];
911 ++BlockNo;
912 F->getBasicBlockList().push_back(BB);
913
914 // Read instructions into this basic block until we get to a terminator
Reid Spencer46b002c2004-07-11 17:28:43 +0000915 while (moreInBlock() && !BB->getTerminator())
Reid Spencer060d25d2004-06-29 23:29:38 +0000916 ParseInstruction(Args, BB);
Chris Lattner8d1dbd22003-12-01 07:05:31 +0000917
918 if (!BB->getTerminator())
Reid Spencer24399722004-07-09 22:21:33 +0000919 error("Non-terminated basic block found!");
Reid Spencer5c15fe52004-07-05 00:57:50 +0000920
Reid Spencer46b002c2004-07-11 17:28:43 +0000921 if (Handler) Handler->handleBasicBlockEnd(BlockNo-1);
Chris Lattner8d1dbd22003-12-01 07:05:31 +0000922 }
923
924 return BlockNo;
925}
926
Reid Spencer78d033e2007-01-06 07:24:44 +0000927/// Parse a type symbol table.
928void BytecodeReader::ParseTypeSymbolTable(TypeSymbolTable *TST) {
929 // Type Symtab block header: [num entries]
930 unsigned NumEntries = read_vbr_uint();
931 for (unsigned i = 0; i < NumEntries; ++i) {
932 // Symtab entry: [type slot #][name]
933 unsigned slot = read_vbr_uint();
934 std::string Name = read_str();
935 const Type* T = getType(slot);
936 TST->insert(Name, T);
937 }
938}
939
940/// Parse a value symbol table. This works for both module level and function
Reid Spencer04cde2c2004-07-04 11:33:49 +0000941/// level symbol tables. For function level symbol tables, the CurrentFunction
942/// parameter must be non-zero and the ST parameter must correspond to
943/// CurrentFunction's symbol table. For Module level symbol tables, the
944/// CurrentFunction argument must be zero.
Reid Spencer78d033e2007-01-06 07:24:44 +0000945void BytecodeReader::ParseValueSymbolTable(Function *CurrentFunction,
946 SymbolTable *ST) {
947
Reid Spencer04cde2c2004-07-04 11:33:49 +0000948 if (Handler) Handler->handleSymbolTableBegin(CurrentFunction,ST);
Reid Spencer060d25d2004-06-29 23:29:38 +0000949
Chris Lattner39cacce2003-10-10 05:43:47 +0000950 // Allow efficient basic block lookup by number.
951 std::vector<BasicBlock*> BBMap;
952 if (CurrentFunction)
953 for (Function::iterator I = CurrentFunction->begin(),
954 E = CurrentFunction->end(); I != E; ++I)
955 BBMap.push_back(I);
956
Reid Spencer46b002c2004-07-11 17:28:43 +0000957 while (moreInBlock()) {
Chris Lattner00950542001-06-06 20:29:01 +0000958 // Symtab block header: [num entries][type id number]
Reid Spencer060d25d2004-06-29 23:29:38 +0000959 unsigned NumEntries = read_vbr_uint();
Reid Spencerd798a512006-11-14 04:47:22 +0000960 unsigned Typ = read_vbr_uint();
Chris Lattner1d670cc2001-09-07 16:37:43 +0000961
Chris Lattner7dc3a2e2003-10-13 14:57:53 +0000962 for (unsigned i = 0; i != NumEntries; ++i) {
Chris Lattner00950542001-06-06 20:29:01 +0000963 // Symtab entry: [def slot #][name]
Reid Spencer060d25d2004-06-29 23:29:38 +0000964 unsigned slot = read_vbr_uint();
965 std::string Name = read_str();
Reid Spencerd798a512006-11-14 04:47:22 +0000966 Value *V = 0;
Reid Spencera54b7cb2007-01-12 07:05:14 +0000967 if (Typ == LabelTySlot) {
Reid Spencerd798a512006-11-14 04:47:22 +0000968 if (slot < BBMap.size())
969 V = BBMap[slot];
Chris Lattner39cacce2003-10-10 05:43:47 +0000970 } else {
Reid Spencerd798a512006-11-14 04:47:22 +0000971 V = getValue(Typ, slot, false); // Find mapping...
Chris Lattner39cacce2003-10-10 05:43:47 +0000972 }
Reid Spencerd798a512006-11-14 04:47:22 +0000973 if (V == 0)
974 error("Failed value look-up for name '" + Name + "'");
975 V->setName(Name);
Chris Lattner00950542001-06-06 20:29:01 +0000976 }
977 }
Reid Spencer060d25d2004-06-29 23:29:38 +0000978 checkPastBlockEnd("Symbol Table");
Reid Spencer04cde2c2004-07-04 11:33:49 +0000979 if (Handler) Handler->handleSymbolTableEnd();
Chris Lattner00950542001-06-06 20:29:01 +0000980}
981
Reid Spencer46b002c2004-07-11 17:28:43 +0000982// Parse a single type. The typeid is read in first. If its a primitive type
983// then nothing else needs to be read, we know how to instantiate it. If its
Misha Brukman8a96c532005-04-21 21:44:41 +0000984// a derived type, then additional data is read to fill out the type
Reid Spencer46b002c2004-07-11 17:28:43 +0000985// definition.
986const Type *BytecodeReader::ParseType() {
Reid Spencerd798a512006-11-14 04:47:22 +0000987 unsigned PrimType = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +0000988 const Type *Result = 0;
989 if ((Result = Type::getPrimitiveType((Type::TypeID)PrimType)))
990 return Result;
Misha Brukman8a96c532005-04-21 21:44:41 +0000991
Reid Spencer060d25d2004-06-29 23:29:38 +0000992 switch (PrimType) {
Reid Spencera54b7cb2007-01-12 07:05:14 +0000993 case Type::IntegerTyID: {
994 unsigned NumBits = read_vbr_uint();
995 Result = IntegerType::get(NumBits);
996 break;
997 }
Reid Spencer060d25d2004-06-29 23:29:38 +0000998 case Type::FunctionTyID: {
Reid Spencerd798a512006-11-14 04:47:22 +0000999 const Type *RetType = readType();
Reid Spencer88cfda22006-12-31 05:44:24 +00001000 unsigned RetAttr = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +00001001
1002 unsigned NumParams = read_vbr_uint();
1003
1004 std::vector<const Type*> Params;
Reid Spencer88cfda22006-12-31 05:44:24 +00001005 std::vector<FunctionType::ParameterAttributes> Attrs;
1006 Attrs.push_back(FunctionType::ParameterAttributes(RetAttr));
1007 while (NumParams--) {
Reid Spencerd798a512006-11-14 04:47:22 +00001008 Params.push_back(readType());
Reid Spencer88cfda22006-12-31 05:44:24 +00001009 if (Params.back() != Type::VoidTy)
1010 Attrs.push_back(FunctionType::ParameterAttributes(read_vbr_uint()));
1011 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001012
1013 bool isVarArg = Params.size() && Params.back() == Type::VoidTy;
1014 if (isVarArg) Params.pop_back();
1015
Reid Spencer88cfda22006-12-31 05:44:24 +00001016 Result = FunctionType::get(RetType, Params, isVarArg, Attrs);
Reid Spencer060d25d2004-06-29 23:29:38 +00001017 break;
1018 }
1019 case Type::ArrayTyID: {
Reid Spencerd798a512006-11-14 04:47:22 +00001020 const Type *ElementType = readType();
Reid Spencer060d25d2004-06-29 23:29:38 +00001021 unsigned NumElements = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +00001022 Result = ArrayType::get(ElementType, NumElements);
1023 break;
1024 }
Brian Gaeke715c90b2004-08-20 06:00:58 +00001025 case Type::PackedTyID: {
Reid Spencerd798a512006-11-14 04:47:22 +00001026 const Type *ElementType = readType();
Brian Gaeke715c90b2004-08-20 06:00:58 +00001027 unsigned NumElements = read_vbr_uint();
1028 Result = PackedType::get(ElementType, NumElements);
1029 break;
1030 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001031 case Type::StructTyID: {
1032 std::vector<const Type*> Elements;
Reid Spencerd798a512006-11-14 04:47:22 +00001033 unsigned Typ = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +00001034 while (Typ) { // List is terminated by void/0 typeid
1035 Elements.push_back(getType(Typ));
Reid Spencerd798a512006-11-14 04:47:22 +00001036 Typ = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +00001037 }
1038
Andrew Lenharth38ecbf12006-12-08 18:06:16 +00001039 Result = StructType::get(Elements, false);
1040 break;
1041 }
Reid Spencera54b7cb2007-01-12 07:05:14 +00001042 case Type::PackedStructTyID: {
Andrew Lenharth38ecbf12006-12-08 18:06:16 +00001043 std::vector<const Type*> Elements;
1044 unsigned Typ = read_vbr_uint();
1045 while (Typ) { // List is terminated by void/0 typeid
1046 Elements.push_back(getType(Typ));
1047 Typ = read_vbr_uint();
1048 }
1049
1050 Result = StructType::get(Elements, true);
Reid Spencer060d25d2004-06-29 23:29:38 +00001051 break;
1052 }
1053 case Type::PointerTyID: {
Reid Spencerd798a512006-11-14 04:47:22 +00001054 Result = PointerType::get(readType());
Reid Spencer060d25d2004-06-29 23:29:38 +00001055 break;
1056 }
1057
1058 case Type::OpaqueTyID: {
1059 Result = OpaqueType::get();
1060 break;
1061 }
1062
1063 default:
Reid Spencer24399722004-07-09 22:21:33 +00001064 error("Don't know how to deserialize primitive type " + utostr(PrimType));
Reid Spencer060d25d2004-06-29 23:29:38 +00001065 break;
1066 }
Reid Spencer46b002c2004-07-11 17:28:43 +00001067 if (Handler) Handler->handleType(Result);
Reid Spencer060d25d2004-06-29 23:29:38 +00001068 return Result;
1069}
1070
Reid Spencer5b472d92004-08-21 20:49:23 +00001071// ParseTypes - We have to use this weird code to handle recursive
Reid Spencer060d25d2004-06-29 23:29:38 +00001072// types. We know that recursive types will only reference the current slab of
1073// values in the type plane, but they can forward reference types before they
1074// have been read. For example, Type #0 might be '{ Ty#1 }' and Type #1 might
1075// be 'Ty#0*'. When reading Type #0, type number one doesn't exist. To fix
1076// this ugly problem, we pessimistically insert an opaque type for each type we
1077// are about to read. This means that forward references will resolve to
1078// something and when we reread the type later, we can replace the opaque type
1079// with a new resolved concrete type.
1080//
Reid Spencer46b002c2004-07-11 17:28:43 +00001081void BytecodeReader::ParseTypes(TypeListTy &Tab, unsigned NumEntries){
Reid Spencer060d25d2004-06-29 23:29:38 +00001082 assert(Tab.size() == 0 && "should not have read type constants in before!");
1083
1084 // Insert a bunch of opaque types to be resolved later...
1085 Tab.reserve(NumEntries);
1086 for (unsigned i = 0; i != NumEntries; ++i)
1087 Tab.push_back(OpaqueType::get());
1088
Misha Brukman8a96c532005-04-21 21:44:41 +00001089 if (Handler)
Reid Spencer5b472d92004-08-21 20:49:23 +00001090 Handler->handleTypeList(NumEntries);
1091
Chris Lattnereebac5f2005-10-03 21:26:53 +00001092 // If we are about to resolve types, make sure the type cache is clear.
1093 if (NumEntries)
1094 ModuleTypeIDCache.clear();
1095
Reid Spencer060d25d2004-06-29 23:29:38 +00001096 // Loop through reading all of the types. Forward types will make use of the
1097 // opaque types just inserted.
1098 //
1099 for (unsigned i = 0; i != NumEntries; ++i) {
Reid Spencer46b002c2004-07-11 17:28:43 +00001100 const Type* NewTy = ParseType();
Reid Spencer04cde2c2004-07-04 11:33:49 +00001101 const Type* OldTy = Tab[i].get();
Misha Brukman8a96c532005-04-21 21:44:41 +00001102 if (NewTy == 0)
Reid Spencer24399722004-07-09 22:21:33 +00001103 error("Couldn't parse type!");
Reid Spencer060d25d2004-06-29 23:29:38 +00001104
Misha Brukman8a96c532005-04-21 21:44:41 +00001105 // Don't directly push the new type on the Tab. Instead we want to replace
Reid Spencer060d25d2004-06-29 23:29:38 +00001106 // the opaque type we previously inserted with the new concrete value. This
1107 // approach helps with forward references to types. The refinement from the
1108 // abstract (opaque) type to the new type causes all uses of the abstract
1109 // type to use the concrete type (NewTy). This will also cause the opaque
1110 // type to be deleted.
1111 cast<DerivedType>(const_cast<Type*>(OldTy))->refineAbstractTypeTo(NewTy);
1112
1113 // This should have replaced the old opaque type with the new type in the
1114 // value table... or with a preexisting type that was already in the system.
1115 // Let's just make sure it did.
1116 assert(Tab[i] != OldTy && "refineAbstractType didn't work!");
1117 }
1118}
1119
Reid Spencer04cde2c2004-07-04 11:33:49 +00001120/// Parse a single constant value
Chris Lattner3bc5a602006-01-25 23:08:15 +00001121Value *BytecodeReader::ParseConstantPoolValue(unsigned TypeID) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001122 // We must check for a ConstantExpr before switching by type because
1123 // a ConstantExpr can be of any type, and has no explicit value.
Misha Brukman8a96c532005-04-21 21:44:41 +00001124 //
Reid Spencer060d25d2004-06-29 23:29:38 +00001125 // 0 if not expr; numArgs if is expr
1126 unsigned isExprNumArgs = read_vbr_uint();
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001127
Reid Spencer060d25d2004-06-29 23:29:38 +00001128 if (isExprNumArgs) {
Reid Spencerd798a512006-11-14 04:47:22 +00001129 // 'undef' is encoded with 'exprnumargs' == 1.
1130 if (isExprNumArgs == 1)
1131 return UndefValue::get(getType(TypeID));
Misha Brukman8a96c532005-04-21 21:44:41 +00001132
Reid Spencerd798a512006-11-14 04:47:22 +00001133 // Inline asm is encoded with exprnumargs == ~0U.
1134 if (isExprNumArgs == ~0U) {
1135 std::string AsmStr = read_str();
1136 std::string ConstraintStr = read_str();
1137 unsigned Flags = read_vbr_uint();
Chris Lattner3bc5a602006-01-25 23:08:15 +00001138
Reid Spencerd798a512006-11-14 04:47:22 +00001139 const PointerType *PTy = dyn_cast<PointerType>(getType(TypeID));
1140 const FunctionType *FTy =
1141 PTy ? dyn_cast<FunctionType>(PTy->getElementType()) : 0;
1142
1143 if (!FTy || !InlineAsm::Verify(FTy, ConstraintStr))
1144 error("Invalid constraints for inline asm");
1145 if (Flags & ~1U)
1146 error("Invalid flags for inline asm");
1147 bool HasSideEffects = Flags & 1;
1148 return InlineAsm::get(FTy, AsmStr, ConstraintStr, HasSideEffects);
Chris Lattner3bc5a602006-01-25 23:08:15 +00001149 }
Reid Spencerd798a512006-11-14 04:47:22 +00001150
1151 --isExprNumArgs;
Chris Lattner3bc5a602006-01-25 23:08:15 +00001152
Reid Spencer060d25d2004-06-29 23:29:38 +00001153 // FIXME: Encoding of constant exprs could be much more compact!
1154 std::vector<Constant*> ArgVec;
1155 ArgVec.reserve(isExprNumArgs);
1156 unsigned Opcode = read_vbr_uint();
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001157
Reid Spencer060d25d2004-06-29 23:29:38 +00001158 // Read the slot number and types of each of the arguments
1159 for (unsigned i = 0; i != isExprNumArgs; ++i) {
1160 unsigned ArgValSlot = read_vbr_uint();
Reid Spencerd798a512006-11-14 04:47:22 +00001161 unsigned ArgTypeSlot = read_vbr_uint();
Misha Brukman8a96c532005-04-21 21:44:41 +00001162
Reid Spencer060d25d2004-06-29 23:29:38 +00001163 // Get the arg value from its slot if it exists, otherwise a placeholder
1164 ArgVec.push_back(getConstantValue(ArgTypeSlot, ArgValSlot));
1165 }
Misha Brukman8a96c532005-04-21 21:44:41 +00001166
Reid Spencer060d25d2004-06-29 23:29:38 +00001167 // Construct a ConstantExpr of the appropriate kind
1168 if (isExprNumArgs == 1) { // All one-operand expressions
Reid Spencer3da59db2006-11-27 01:05:10 +00001169 if (!Instruction::isCast(Opcode))
Chris Lattner02dce162004-12-04 05:28:27 +00001170 error("Only cast instruction has one argument for ConstantExpr");
Reid Spencer46b002c2004-07-11 17:28:43 +00001171
Reid Spencera77fa7e2006-12-11 23:20:20 +00001172 Constant *Result = ConstantExpr::getCast(Opcode, ArgVec[0],
1173 getType(TypeID));
Reid Spencer04cde2c2004-07-04 11:33:49 +00001174 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
Reid Spencer060d25d2004-06-29 23:29:38 +00001175 return Result;
1176 } else if (Opcode == Instruction::GetElementPtr) { // GetElementPtr
Chris Lattnere0135402007-01-31 04:43:46 +00001177 Constant *Result = ConstantExpr::getGetElementPtr(ArgVec[0], &ArgVec[1],
1178 ArgVec.size()-1);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001179 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
Reid Spencer060d25d2004-06-29 23:29:38 +00001180 return Result;
1181 } else if (Opcode == Instruction::Select) {
Reid Spencer46b002c2004-07-11 17:28:43 +00001182 if (ArgVec.size() != 3)
1183 error("Select instruction must have three arguments.");
Misha Brukman8a96c532005-04-21 21:44:41 +00001184 Constant* Result = ConstantExpr::getSelect(ArgVec[0], ArgVec[1],
Reid Spencer04cde2c2004-07-04 11:33:49 +00001185 ArgVec[2]);
1186 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
Reid Spencer060d25d2004-06-29 23:29:38 +00001187 return Result;
Robert Bocchinofee31b32006-01-10 19:04:39 +00001188 } else if (Opcode == Instruction::ExtractElement) {
Chris Lattner59fecec2006-04-08 04:09:19 +00001189 if (ArgVec.size() != 2 ||
1190 !ExtractElementInst::isValidOperands(ArgVec[0], ArgVec[1]))
1191 error("Invalid extractelement constand expr arguments");
Robert Bocchinofee31b32006-01-10 19:04:39 +00001192 Constant* Result = ConstantExpr::getExtractElement(ArgVec[0], ArgVec[1]);
1193 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
1194 return Result;
Robert Bocchinob1f240b2006-01-17 20:06:35 +00001195 } else if (Opcode == Instruction::InsertElement) {
Chris Lattner59fecec2006-04-08 04:09:19 +00001196 if (ArgVec.size() != 3 ||
1197 !InsertElementInst::isValidOperands(ArgVec[0], ArgVec[1], ArgVec[2]))
1198 error("Invalid insertelement constand expr arguments");
1199
1200 Constant *Result =
Robert Bocchinob1f240b2006-01-17 20:06:35 +00001201 ConstantExpr::getInsertElement(ArgVec[0], ArgVec[1], ArgVec[2]);
1202 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
1203 return Result;
Chris Lattner30b44b62006-04-08 01:17:59 +00001204 } else if (Opcode == Instruction::ShuffleVector) {
1205 if (ArgVec.size() != 3 ||
1206 !ShuffleVectorInst::isValidOperands(ArgVec[0], ArgVec[1], ArgVec[2]))
Chris Lattner59fecec2006-04-08 04:09:19 +00001207 error("Invalid shufflevector constant expr arguments.");
Chris Lattner30b44b62006-04-08 01:17:59 +00001208 Constant *Result =
1209 ConstantExpr::getShuffleVector(ArgVec[0], ArgVec[1], ArgVec[2]);
1210 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
1211 return Result;
Reid Spencer9f132762006-12-03 17:17:02 +00001212 } else if (Opcode == Instruction::ICmp) {
1213 if (ArgVec.size() != 2)
Reid Spencer595b4772006-12-04 05:23:49 +00001214 error("Invalid ICmp constant expr arguments.");
1215 unsigned predicate = read_vbr_uint();
1216 Constant *Result = ConstantExpr::getICmp(predicate, ArgVec[0], ArgVec[1]);
1217 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
1218 return Result;
Reid Spencer9f132762006-12-03 17:17:02 +00001219 } else if (Opcode == Instruction::FCmp) {
1220 if (ArgVec.size() != 2)
Reid Spencer595b4772006-12-04 05:23:49 +00001221 error("Invalid FCmp constant expr arguments.");
1222 unsigned predicate = read_vbr_uint();
1223 Constant *Result = ConstantExpr::getFCmp(predicate, ArgVec[0], ArgVec[1]);
1224 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
1225 return Result;
Reid Spencer060d25d2004-06-29 23:29:38 +00001226 } else { // All other 2-operand expressions
1227 Constant* Result = ConstantExpr::get(Opcode, ArgVec[0], ArgVec[1]);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001228 if (Handler) Handler->handleConstantExpression(Opcode, ArgVec, Result);
Reid Spencer060d25d2004-06-29 23:29:38 +00001229 return Result;
1230 }
1231 }
Misha Brukman8a96c532005-04-21 21:44:41 +00001232
Reid Spencer060d25d2004-06-29 23:29:38 +00001233 // Ok, not an ConstantExpr. We now know how to read the given type...
1234 const Type *Ty = getType(TypeID);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001235 Constant *Result = 0;
Reid Spencer060d25d2004-06-29 23:29:38 +00001236 switch (Ty->getTypeID()) {
Reid Spencera54b7cb2007-01-12 07:05:14 +00001237 case Type::IntegerTyID: {
1238 const IntegerType *IT = cast<IntegerType>(Ty);
1239 if (IT->getBitWidth() <= 32) {
1240 uint32_t Val = read_vbr_uint();
Reid Spencerb61c1ce2007-01-13 00:09:12 +00001241 if (!ConstantInt::isValueValidForType(Ty, uint64_t(Val)))
1242 error("Integer value read is invalid for type.");
1243 Result = ConstantInt::get(IT, Val);
1244 if (Handler) Handler->handleConstantValue(Result);
Reid Spencera54b7cb2007-01-12 07:05:14 +00001245 } else if (IT->getBitWidth() <= 64) {
1246 uint64_t Val = read_vbr_uint64();
1247 if (!ConstantInt::isValueValidForType(Ty, Val))
1248 error("Invalid constant integer read.");
1249 Result = ConstantInt::get(IT, Val);
1250 if (Handler) Handler->handleConstantValue(Result);
1251 } else
1252 assert("Integer types > 64 bits not supported");
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001253 break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001254 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001255 case Type::FloatTyID: {
Reid Spencer46b002c2004-07-11 17:28:43 +00001256 float Val;
1257 read_float(Val);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001258 Result = ConstantFP::get(Ty, Val);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001259 if (Handler) Handler->handleConstantValue(Result);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001260 break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001261 }
1262
1263 case Type::DoubleTyID: {
1264 double Val;
Reid Spencer46b002c2004-07-11 17:28:43 +00001265 read_double(Val);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001266 Result = ConstantFP::get(Ty, Val);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001267 if (Handler) Handler->handleConstantValue(Result);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001268 break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001269 }
1270
Reid Spencer060d25d2004-06-29 23:29:38 +00001271 case Type::ArrayTyID: {
1272 const ArrayType *AT = cast<ArrayType>(Ty);
1273 unsigned NumElements = AT->getNumElements();
1274 unsigned TypeSlot = getTypeSlot(AT->getElementType());
1275 std::vector<Constant*> Elements;
1276 Elements.reserve(NumElements);
1277 while (NumElements--) // Read all of the elements of the constant.
1278 Elements.push_back(getConstantValue(TypeSlot,
1279 read_vbr_uint()));
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001280 Result = ConstantArray::get(AT, Elements);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001281 if (Handler) Handler->handleConstantArray(AT, Elements, TypeSlot, Result);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001282 break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001283 }
1284
1285 case Type::StructTyID: {
1286 const StructType *ST = cast<StructType>(Ty);
1287
1288 std::vector<Constant *> Elements;
1289 Elements.reserve(ST->getNumElements());
1290 for (unsigned i = 0; i != ST->getNumElements(); ++i)
1291 Elements.push_back(getConstantValue(ST->getElementType(i),
1292 read_vbr_uint()));
1293
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001294 Result = ConstantStruct::get(ST, Elements);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001295 if (Handler) Handler->handleConstantStruct(ST, Elements, Result);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001296 break;
Misha Brukman8a96c532005-04-21 21:44:41 +00001297 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001298
Brian Gaeke715c90b2004-08-20 06:00:58 +00001299 case Type::PackedTyID: {
1300 const PackedType *PT = cast<PackedType>(Ty);
1301 unsigned NumElements = PT->getNumElements();
1302 unsigned TypeSlot = getTypeSlot(PT->getElementType());
1303 std::vector<Constant*> Elements;
1304 Elements.reserve(NumElements);
1305 while (NumElements--) // Read all of the elements of the constant.
1306 Elements.push_back(getConstantValue(TypeSlot,
1307 read_vbr_uint()));
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001308 Result = ConstantPacked::get(PT, Elements);
Brian Gaeke715c90b2004-08-20 06:00:58 +00001309 if (Handler) Handler->handleConstantPacked(PT, Elements, TypeSlot, Result);
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001310 break;
Brian Gaeke715c90b2004-08-20 06:00:58 +00001311 }
1312
Chris Lattner638c3812004-11-19 16:24:05 +00001313 case Type::PointerTyID: { // ConstantPointerRef value (backwards compat).
Reid Spencer060d25d2004-06-29 23:29:38 +00001314 const PointerType *PT = cast<PointerType>(Ty);
1315 unsigned Slot = read_vbr_uint();
Misha Brukman8a96c532005-04-21 21:44:41 +00001316
Reid Spencer060d25d2004-06-29 23:29:38 +00001317 // Check to see if we have already read this global variable...
1318 Value *Val = getValue(TypeID, Slot, false);
Reid Spencer060d25d2004-06-29 23:29:38 +00001319 if (Val) {
Chris Lattnerbcb11cf2004-07-27 02:34:49 +00001320 GlobalValue *GV = dyn_cast<GlobalValue>(Val);
1321 if (!GV) error("GlobalValue not in ValueTable!");
1322 if (Handler) Handler->handleConstantPointer(PT, Slot, GV);
1323 return GV;
Reid Spencer060d25d2004-06-29 23:29:38 +00001324 } else {
Reid Spencer24399722004-07-09 22:21:33 +00001325 error("Forward references are not allowed here.");
Reid Spencer060d25d2004-06-29 23:29:38 +00001326 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001327 }
1328
1329 default:
Reid Spencer24399722004-07-09 22:21:33 +00001330 error("Don't know how to deserialize constant value of type '" +
Reid Spencer060d25d2004-06-29 23:29:38 +00001331 Ty->getDescription());
1332 break;
1333 }
Chris Lattnerd2cfb7a2006-04-07 05:00:02 +00001334
1335 // Check that we didn't read a null constant if they are implicit for this
1336 // type plane. Do not do this check for constantexprs, as they may be folded
1337 // to a null value in a way that isn't predicted when a .bc file is initially
1338 // produced.
1339 assert((!isa<Constant>(Result) || !cast<Constant>(Result)->isNullValue()) ||
1340 !hasImplicitNull(TypeID) &&
1341 "Cannot read null values from bytecode!");
1342 return Result;
Reid Spencer060d25d2004-06-29 23:29:38 +00001343}
1344
Misha Brukman8a96c532005-04-21 21:44:41 +00001345/// Resolve references for constants. This function resolves the forward
1346/// referenced constants in the ConstantFwdRefs map. It uses the
Reid Spencer04cde2c2004-07-04 11:33:49 +00001347/// replaceAllUsesWith method of Value class to substitute the placeholder
1348/// instance with the actual instance.
Chris Lattner389bd042004-12-09 06:19:44 +00001349void BytecodeReader::ResolveReferencesToConstant(Constant *NewV, unsigned Typ,
1350 unsigned Slot) {
Chris Lattner29b789b2003-11-19 17:27:18 +00001351 ConstantRefsType::iterator I =
Chris Lattner389bd042004-12-09 06:19:44 +00001352 ConstantFwdRefs.find(std::make_pair(Typ, Slot));
Chris Lattner29b789b2003-11-19 17:27:18 +00001353 if (I == ConstantFwdRefs.end()) return; // Never forward referenced?
Chris Lattner00950542001-06-06 20:29:01 +00001354
Chris Lattner29b789b2003-11-19 17:27:18 +00001355 Value *PH = I->second; // Get the placeholder...
1356 PH->replaceAllUsesWith(NewV);
1357 delete PH; // Delete the old placeholder
1358 ConstantFwdRefs.erase(I); // Remove the map entry for it
Vikram S. Advec1e4a812002-07-14 23:04:18 +00001359}
1360
Reid Spencer04cde2c2004-07-04 11:33:49 +00001361/// Parse the constant strings section.
Reid Spencer060d25d2004-06-29 23:29:38 +00001362void BytecodeReader::ParseStringConstants(unsigned NumEntries, ValueTable &Tab){
1363 for (; NumEntries; --NumEntries) {
Reid Spencerd798a512006-11-14 04:47:22 +00001364 unsigned Typ = read_vbr_uint();
Reid Spencer060d25d2004-06-29 23:29:38 +00001365 const Type *Ty = getType(Typ);
1366 if (!isa<ArrayType>(Ty))
Reid Spencer24399722004-07-09 22:21:33 +00001367 error("String constant data invalid!");
Misha Brukman8a96c532005-04-21 21:44:41 +00001368
Reid Spencer060d25d2004-06-29 23:29:38 +00001369 const ArrayType *ATy = cast<ArrayType>(Ty);
Reid Spencer88cfda22006-12-31 05:44:24 +00001370 if (ATy->getElementType() != Type::Int8Ty &&
1371 ATy->getElementType() != Type::Int8Ty)
Reid Spencer24399722004-07-09 22:21:33 +00001372 error("String constant data invalid!");
Misha Brukman8a96c532005-04-21 21:44:41 +00001373
Reid Spencer060d25d2004-06-29 23:29:38 +00001374 // Read character data. The type tells us how long the string is.
Misha Brukman8a96c532005-04-21 21:44:41 +00001375 char *Data = reinterpret_cast<char *>(alloca(ATy->getNumElements()));
Reid Spencer060d25d2004-06-29 23:29:38 +00001376 read_data(Data, Data+ATy->getNumElements());
Chris Lattner52e20b02003-03-19 20:54:26 +00001377
Reid Spencer060d25d2004-06-29 23:29:38 +00001378 std::vector<Constant*> Elements(ATy->getNumElements());
Reid Spencerb83eb642006-10-20 07:07:24 +00001379 const Type* ElemType = ATy->getElementType();
1380 for (unsigned i = 0, e = ATy->getNumElements(); i != e; ++i)
1381 Elements[i] = ConstantInt::get(ElemType, (unsigned char)Data[i]);
Misha Brukman12c29d12003-09-22 23:38:23 +00001382
Reid Spencer060d25d2004-06-29 23:29:38 +00001383 // Create the constant, inserting it as needed.
1384 Constant *C = ConstantArray::get(ATy, Elements);
1385 unsigned Slot = insertValue(C, Typ, Tab);
Chris Lattner389bd042004-12-09 06:19:44 +00001386 ResolveReferencesToConstant(C, Typ, Slot);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001387 if (Handler) Handler->handleConstantString(cast<ConstantArray>(C));
Reid Spencer060d25d2004-06-29 23:29:38 +00001388 }
Misha Brukman12c29d12003-09-22 23:38:23 +00001389}
1390
Reid Spencer04cde2c2004-07-04 11:33:49 +00001391/// Parse the constant pool.
Misha Brukman8a96c532005-04-21 21:44:41 +00001392void BytecodeReader::ParseConstantPool(ValueTable &Tab,
Reid Spencer04cde2c2004-07-04 11:33:49 +00001393 TypeListTy &TypeTab,
Reid Spencer46b002c2004-07-11 17:28:43 +00001394 bool isFunction) {
Reid Spencer04cde2c2004-07-04 11:33:49 +00001395 if (Handler) Handler->handleGlobalConstantsBegin();
1396
1397 /// In LLVM 1.3 Type does not derive from Value so the types
1398 /// do not occupy a plane. Consequently, we read the types
1399 /// first in the constant pool.
Reid Spencerd798a512006-11-14 04:47:22 +00001400 if (isFunction) {
Reid Spencer04cde2c2004-07-04 11:33:49 +00001401 unsigned NumEntries = read_vbr_uint();
Reid Spencer46b002c2004-07-11 17:28:43 +00001402 ParseTypes(TypeTab, NumEntries);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001403 }
1404
Reid Spencer46b002c2004-07-11 17:28:43 +00001405 while (moreInBlock()) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001406 unsigned NumEntries = read_vbr_uint();
Reid Spencerd798a512006-11-14 04:47:22 +00001407 unsigned Typ = read_vbr_uint();
Reid Spencer04cde2c2004-07-04 11:33:49 +00001408
Reid Spencerd798a512006-11-14 04:47:22 +00001409 if (Typ == Type::VoidTyID) {
Reid Spencer04cde2c2004-07-04 11:33:49 +00001410 /// Use of Type::VoidTyID is a misnomer. It actually means
1411 /// that the following plane is constant strings
Reid Spencer060d25d2004-06-29 23:29:38 +00001412 assert(&Tab == &ModuleValues && "Cannot read strings in functions!");
1413 ParseStringConstants(NumEntries, Tab);
1414 } else {
1415 for (unsigned i = 0; i < NumEntries; ++i) {
Chris Lattner3bc5a602006-01-25 23:08:15 +00001416 Value *V = ParseConstantPoolValue(Typ);
1417 assert(V && "ParseConstantPoolValue returned NULL!");
1418 unsigned Slot = insertValue(V, Typ, Tab);
Chris Lattner29b789b2003-11-19 17:27:18 +00001419
Reid Spencer060d25d2004-06-29 23:29:38 +00001420 // If we are reading a function constant table, make sure that we adjust
1421 // the slot number to be the real global constant number.
1422 //
1423 if (&Tab != &ModuleValues && Typ < ModuleValues.size() &&
1424 ModuleValues[Typ])
1425 Slot += ModuleValues[Typ]->size();
Chris Lattner3bc5a602006-01-25 23:08:15 +00001426 if (Constant *C = dyn_cast<Constant>(V))
1427 ResolveReferencesToConstant(C, Typ, Slot);
Reid Spencer060d25d2004-06-29 23:29:38 +00001428 }
1429 }
1430 }
Chris Lattner02dce162004-12-04 05:28:27 +00001431
1432 // After we have finished parsing the constant pool, we had better not have
1433 // any dangling references left.
Reid Spencer3c391272004-12-04 22:19:53 +00001434 if (!ConstantFwdRefs.empty()) {
Reid Spencer3c391272004-12-04 22:19:53 +00001435 ConstantRefsType::const_iterator I = ConstantFwdRefs.begin();
Reid Spencer3c391272004-12-04 22:19:53 +00001436 Constant* missingConst = I->second;
Misha Brukman8a96c532005-04-21 21:44:41 +00001437 error(utostr(ConstantFwdRefs.size()) +
1438 " unresolved constant reference exist. First one is '" +
1439 missingConst->getName() + "' of type '" +
Chris Lattner389bd042004-12-09 06:19:44 +00001440 missingConst->getType()->getDescription() + "'.");
Reid Spencer3c391272004-12-04 22:19:53 +00001441 }
Chris Lattner02dce162004-12-04 05:28:27 +00001442
Reid Spencer060d25d2004-06-29 23:29:38 +00001443 checkPastBlockEnd("Constant Pool");
Reid Spencer04cde2c2004-07-04 11:33:49 +00001444 if (Handler) Handler->handleGlobalConstantsEnd();
Reid Spencer060d25d2004-06-29 23:29:38 +00001445}
Chris Lattner00950542001-06-06 20:29:01 +00001446
Reid Spencer04cde2c2004-07-04 11:33:49 +00001447/// Parse the contents of a function. Note that this function can be
1448/// called lazily by materializeFunction
1449/// @see materializeFunction
Reid Spencer46b002c2004-07-11 17:28:43 +00001450void BytecodeReader::ParseFunctionBody(Function* F) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001451
1452 unsigned FuncSize = BlockEnd - At;
Chris Lattnere3869c82003-04-16 21:16:05 +00001453 GlobalValue::LinkageTypes Linkage = GlobalValue::ExternalLinkage;
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001454 GlobalValue::VisibilityTypes Visibility = GlobalValue::DefaultVisibility;
Chris Lattnere3869c82003-04-16 21:16:05 +00001455
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001456 unsigned rWord = read_vbr_uint();
1457 unsigned LinkageID = rWord & 65535;
1458 unsigned VisibilityID = rWord >> 16;
1459 switch (LinkageID) {
Chris Lattnerc08912f2004-01-14 16:44:44 +00001460 case 0: Linkage = GlobalValue::ExternalLinkage; break;
1461 case 1: Linkage = GlobalValue::WeakLinkage; break;
1462 case 2: Linkage = GlobalValue::AppendingLinkage; break;
1463 case 3: Linkage = GlobalValue::InternalLinkage; break;
1464 case 4: Linkage = GlobalValue::LinkOnceLinkage; break;
Anton Korobeynikovb74ed072006-09-14 18:23:27 +00001465 case 5: Linkage = GlobalValue::DLLImportLinkage; break;
1466 case 6: Linkage = GlobalValue::DLLExportLinkage; break;
1467 case 7: Linkage = GlobalValue::ExternalWeakLinkage; break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001468 default:
Reid Spencer24399722004-07-09 22:21:33 +00001469 error("Invalid linkage type for Function.");
Reid Spencer060d25d2004-06-29 23:29:38 +00001470 Linkage = GlobalValue::InternalLinkage;
1471 break;
Chris Lattnere3869c82003-04-16 21:16:05 +00001472 }
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001473 switch (VisibilityID) {
1474 case 0: Visibility = GlobalValue::DefaultVisibility; break;
1475 case 1: Visibility = GlobalValue::HiddenVisibility; break;
1476 default:
1477 error("Unknown visibility type: " + utostr(VisibilityID));
1478 Visibility = GlobalValue::DefaultVisibility;
1479 break;
1480 }
Chris Lattnerd23b1d32001-11-26 18:56:10 +00001481
Reid Spencer46b002c2004-07-11 17:28:43 +00001482 F->setLinkage(Linkage);
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001483 F->setVisibility(Visibility);
Reid Spencer04cde2c2004-07-04 11:33:49 +00001484 if (Handler) Handler->handleFunctionBegin(F,FuncSize);
Chris Lattner00950542001-06-06 20:29:01 +00001485
Chris Lattner4ee8ef22003-10-08 22:52:54 +00001486 // Keep track of how many basic blocks we have read in...
1487 unsigned BlockNum = 0;
Chris Lattner89e02532004-01-18 21:08:15 +00001488 bool InsertedArguments = false;
Chris Lattner4ee8ef22003-10-08 22:52:54 +00001489
Reid Spencer060d25d2004-06-29 23:29:38 +00001490 BufPtr MyEnd = BlockEnd;
Reid Spencer46b002c2004-07-11 17:28:43 +00001491 while (At < MyEnd) {
Chris Lattner00950542001-06-06 20:29:01 +00001492 unsigned Type, Size;
Reid Spencer060d25d2004-06-29 23:29:38 +00001493 BufPtr OldAt = At;
1494 read_block(Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +00001495
1496 switch (Type) {
Reid Spencerad89bd62004-07-25 18:07:36 +00001497 case BytecodeFormat::ConstantPoolBlockID:
Chris Lattner89e02532004-01-18 21:08:15 +00001498 if (!InsertedArguments) {
1499 // Insert arguments into the value table before we parse the first basic
Reid Spencerd2bb8872007-01-30 19:36:46 +00001500 // block in the function
Reid Spencer04cde2c2004-07-04 11:33:49 +00001501 insertArguments(F);
Chris Lattner89e02532004-01-18 21:08:15 +00001502 InsertedArguments = true;
1503 }
1504
Reid Spencer04cde2c2004-07-04 11:33:49 +00001505 ParseConstantPool(FunctionValues, FunctionTypes, true);
Chris Lattner00950542001-06-06 20:29:01 +00001506 break;
1507
Reid Spencerad89bd62004-07-25 18:07:36 +00001508 case BytecodeFormat::InstructionListBlockID: {
Chris Lattner89e02532004-01-18 21:08:15 +00001509 // Insert arguments into the value table before we parse the instruction
Reid Spencerd2bb8872007-01-30 19:36:46 +00001510 // list for the function
Chris Lattner89e02532004-01-18 21:08:15 +00001511 if (!InsertedArguments) {
Reid Spencer04cde2c2004-07-04 11:33:49 +00001512 insertArguments(F);
Chris Lattner89e02532004-01-18 21:08:15 +00001513 InsertedArguments = true;
1514 }
1515
Misha Brukman8a96c532005-04-21 21:44:41 +00001516 if (BlockNum)
Reid Spencer24399722004-07-09 22:21:33 +00001517 error("Already parsed basic blocks!");
Reid Spencer060d25d2004-06-29 23:29:38 +00001518 BlockNum = ParseInstructionList(F);
Chris Lattner8d1dbd22003-12-01 07:05:31 +00001519 break;
1520 }
1521
Reid Spencer78d033e2007-01-06 07:24:44 +00001522 case BytecodeFormat::ValueSymbolTableBlockID:
1523 ParseValueSymbolTable(F, &F->getValueSymbolTable());
1524 break;
1525
1526 case BytecodeFormat::TypeSymbolTableBlockID:
1527 error("Functions don't have type symbol tables");
Chris Lattner00950542001-06-06 20:29:01 +00001528 break;
1529
1530 default:
Reid Spencer060d25d2004-06-29 23:29:38 +00001531 At += Size;
Misha Brukman8a96c532005-04-21 21:44:41 +00001532 if (OldAt > At)
Reid Spencer24399722004-07-09 22:21:33 +00001533 error("Wrapped around reading bytecode.");
Chris Lattner00950542001-06-06 20:29:01 +00001534 break;
1535 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001536 BlockEnd = MyEnd;
Chris Lattner00950542001-06-06 20:29:01 +00001537 }
1538
Chris Lattner4ee8ef22003-10-08 22:52:54 +00001539 // Make sure there were no references to non-existant basic blocks.
1540 if (BlockNum != ParsedBasicBlocks.size())
Reid Spencer24399722004-07-09 22:21:33 +00001541 error("Illegal basic block operand reference");
Reid Spencer060d25d2004-06-29 23:29:38 +00001542
Chris Lattner4ee8ef22003-10-08 22:52:54 +00001543 ParsedBasicBlocks.clear();
1544
Chris Lattner97330cf2003-10-09 23:10:14 +00001545 // Resolve forward references. Replace any uses of a forward reference value
1546 // with the real value.
Chris Lattner8eb10ce2003-10-09 06:05:40 +00001547 while (!ForwardReferences.empty()) {
Chris Lattnerc4d69162004-12-09 04:51:50 +00001548 std::map<std::pair<unsigned,unsigned>, Value*>::iterator
1549 I = ForwardReferences.begin();
1550 Value *V = getValue(I->first.first, I->first.second, false);
Chris Lattner8eb10ce2003-10-09 06:05:40 +00001551 Value *PlaceHolder = I->second;
Chris Lattnerc4d69162004-12-09 04:51:50 +00001552 PlaceHolder->replaceAllUsesWith(V);
Chris Lattner8eb10ce2003-10-09 06:05:40 +00001553 ForwardReferences.erase(I);
Chris Lattner8eb10ce2003-10-09 06:05:40 +00001554 delete PlaceHolder;
Chris Lattner6e448022003-10-08 21:51:46 +00001555 }
Chris Lattner00950542001-06-06 20:29:01 +00001556
Misha Brukman12c29d12003-09-22 23:38:23 +00001557 // Clear out function-level types...
Reid Spencer060d25d2004-06-29 23:29:38 +00001558 FunctionTypes.clear();
Reid Spencer060d25d2004-06-29 23:29:38 +00001559 freeTable(FunctionValues);
1560
Reid Spencer04cde2c2004-07-04 11:33:49 +00001561 if (Handler) Handler->handleFunctionEnd(F);
Chris Lattner00950542001-06-06 20:29:01 +00001562}
1563
Reid Spencer04cde2c2004-07-04 11:33:49 +00001564/// This function parses LLVM functions lazily. It obtains the type of the
1565/// function and records where the body of the function is in the bytecode
Misha Brukman8a96c532005-04-21 21:44:41 +00001566/// buffer. The caller can then use the ParseNextFunction and
Reid Spencer04cde2c2004-07-04 11:33:49 +00001567/// ParseAllFunctionBodies to get handler events for the functions.
Reid Spencer060d25d2004-06-29 23:29:38 +00001568void BytecodeReader::ParseFunctionLazily() {
1569 if (FunctionSignatureList.empty())
Reid Spencer24399722004-07-09 22:21:33 +00001570 error("FunctionSignatureList empty!");
Chris Lattner89e02532004-01-18 21:08:15 +00001571
Reid Spencer060d25d2004-06-29 23:29:38 +00001572 Function *Func = FunctionSignatureList.back();
1573 FunctionSignatureList.pop_back();
Chris Lattner24102432004-01-18 22:35:34 +00001574
Reid Spencer060d25d2004-06-29 23:29:38 +00001575 // Save the information for future reading of the function
1576 LazyFunctionLoadMap[Func] = LazyFunctionInfo(BlockStart, BlockEnd);
Chris Lattner89e02532004-01-18 21:08:15 +00001577
Misha Brukmana3e6ad62004-11-14 21:02:55 +00001578 // This function has a body but it's not loaded so it appears `External'.
1579 // Mark it as a `Ghost' instead to notify the users that it has a body.
1580 Func->setLinkage(GlobalValue::GhostLinkage);
1581
Reid Spencer060d25d2004-06-29 23:29:38 +00001582 // Pretend we've `parsed' this function
1583 At = BlockEnd;
1584}
Chris Lattner89e02532004-01-18 21:08:15 +00001585
Misha Brukman8a96c532005-04-21 21:44:41 +00001586/// The ParserFunction method lazily parses one function. Use this method to
1587/// casue the parser to parse a specific function in the module. Note that
1588/// this will remove the function from what is to be included by
Reid Spencer04cde2c2004-07-04 11:33:49 +00001589/// ParseAllFunctionBodies.
1590/// @see ParseAllFunctionBodies
1591/// @see ParseBytecode
Reid Spencer99655e12006-08-25 19:54:53 +00001592bool BytecodeReader::ParseFunction(Function* Func, std::string* ErrMsg) {
1593
Reid Spencer9b84ad12006-12-15 19:49:23 +00001594 if (setjmp(context)) {
1595 // Set caller's error message, if requested
1596 if (ErrMsg)
1597 *ErrMsg = ErrorMsg;
1598 // Indicate an error occurred
Reid Spencer99655e12006-08-25 19:54:53 +00001599 return true;
Reid Spencer9b84ad12006-12-15 19:49:23 +00001600 }
Reid Spencer99655e12006-08-25 19:54:53 +00001601
Reid Spencer060d25d2004-06-29 23:29:38 +00001602 // Find {start, end} pointers and slot in the map. If not there, we're done.
1603 LazyFunctionMap::iterator Fi = LazyFunctionLoadMap.find(Func);
Chris Lattner89e02532004-01-18 21:08:15 +00001604
Reid Spencer060d25d2004-06-29 23:29:38 +00001605 // Make sure we found it
Reid Spencer46b002c2004-07-11 17:28:43 +00001606 if (Fi == LazyFunctionLoadMap.end()) {
Reid Spencer24399722004-07-09 22:21:33 +00001607 error("Unrecognized function of type " + Func->getType()->getDescription());
Reid Spencer99655e12006-08-25 19:54:53 +00001608 return true;
Chris Lattner89e02532004-01-18 21:08:15 +00001609 }
1610
Reid Spencer060d25d2004-06-29 23:29:38 +00001611 BlockStart = At = Fi->second.Buf;
1612 BlockEnd = Fi->second.EndBuf;
Reid Spencer24399722004-07-09 22:21:33 +00001613 assert(Fi->first == Func && "Found wrong function?");
Reid Spencer060d25d2004-06-29 23:29:38 +00001614
1615 LazyFunctionLoadMap.erase(Fi);
1616
Reid Spencer46b002c2004-07-11 17:28:43 +00001617 this->ParseFunctionBody(Func);
Reid Spencer99655e12006-08-25 19:54:53 +00001618 return false;
Chris Lattner89e02532004-01-18 21:08:15 +00001619}
1620
Reid Spencer04cde2c2004-07-04 11:33:49 +00001621/// The ParseAllFunctionBodies method parses through all the previously
1622/// unparsed functions in the bytecode file. If you want to completely parse
1623/// a bytecode file, this method should be called after Parsebytecode because
1624/// Parsebytecode only records the locations in the bytecode file of where
1625/// the function definitions are located. This function uses that information
1626/// to materialize the functions.
1627/// @see ParseBytecode
Reid Spencer99655e12006-08-25 19:54:53 +00001628bool BytecodeReader::ParseAllFunctionBodies(std::string* ErrMsg) {
Reid Spencer9b84ad12006-12-15 19:49:23 +00001629 if (setjmp(context)) {
1630 // Set caller's error message, if requested
1631 if (ErrMsg)
1632 *ErrMsg = ErrorMsg;
1633 // Indicate an error occurred
Reid Spencer99655e12006-08-25 19:54:53 +00001634 return true;
Reid Spencer9b84ad12006-12-15 19:49:23 +00001635 }
Reid Spencer99655e12006-08-25 19:54:53 +00001636
Reid Spencer060d25d2004-06-29 23:29:38 +00001637 LazyFunctionMap::iterator Fi = LazyFunctionLoadMap.begin();
1638 LazyFunctionMap::iterator Fe = LazyFunctionLoadMap.end();
Chris Lattner89e02532004-01-18 21:08:15 +00001639
Reid Spencer46b002c2004-07-11 17:28:43 +00001640 while (Fi != Fe) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001641 Function* Func = Fi->first;
1642 BlockStart = At = Fi->second.Buf;
1643 BlockEnd = Fi->second.EndBuf;
Chris Lattnerb52f1c22005-02-13 17:48:18 +00001644 ParseFunctionBody(Func);
Reid Spencer060d25d2004-06-29 23:29:38 +00001645 ++Fi;
1646 }
Chris Lattnerb52f1c22005-02-13 17:48:18 +00001647 LazyFunctionLoadMap.clear();
Reid Spencer99655e12006-08-25 19:54:53 +00001648 return false;
Reid Spencer060d25d2004-06-29 23:29:38 +00001649}
Chris Lattner89e02532004-01-18 21:08:15 +00001650
Reid Spencer04cde2c2004-07-04 11:33:49 +00001651/// Parse the global type list
Reid Spencer060d25d2004-06-29 23:29:38 +00001652void BytecodeReader::ParseGlobalTypes() {
Reid Spencer04cde2c2004-07-04 11:33:49 +00001653 // Read the number of types
1654 unsigned NumEntries = read_vbr_uint();
Reid Spencer46b002c2004-07-11 17:28:43 +00001655 ParseTypes(ModuleTypes, NumEntries);
Reid Spencer060d25d2004-06-29 23:29:38 +00001656}
1657
Reid Spencer04cde2c2004-07-04 11:33:49 +00001658/// Parse the Global info (types, global vars, constants)
Reid Spencer060d25d2004-06-29 23:29:38 +00001659void BytecodeReader::ParseModuleGlobalInfo() {
1660
Reid Spencer04cde2c2004-07-04 11:33:49 +00001661 if (Handler) Handler->handleModuleGlobalsBegin();
Chris Lattner00950542001-06-06 20:29:01 +00001662
Chris Lattner404cddf2005-11-12 01:33:40 +00001663 // SectionID - If a global has an explicit section specified, this map
1664 // remembers the ID until we can translate it into a string.
1665 std::map<GlobalValue*, unsigned> SectionID;
1666
Chris Lattner70cc3392001-09-10 07:58:01 +00001667 // Read global variables...
Reid Spencer060d25d2004-06-29 23:29:38 +00001668 unsigned VarType = read_vbr_uint();
Chris Lattner70cc3392001-09-10 07:58:01 +00001669 while (VarType != Type::VoidTyID) { // List is terminated by Void
Chris Lattner9dd87702004-04-03 23:43:42 +00001670 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer, bit2,3,4 =
1671 // Linkage, bit4+ = slot#
1672 unsigned SlotNo = VarType >> 5;
1673 unsigned LinkageID = (VarType >> 2) & 7;
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001674 unsigned VisibilityID = 0;
Reid Spencer060d25d2004-06-29 23:29:38 +00001675 bool isConstant = VarType & 1;
Chris Lattnerce5e04e2005-11-06 08:23:17 +00001676 bool hasInitializer = (VarType & 2) != 0;
Chris Lattner8eb52dd2005-11-06 07:11:04 +00001677 unsigned Alignment = 0;
Chris Lattner404cddf2005-11-12 01:33:40 +00001678 unsigned GlobalSectionID = 0;
Chris Lattner8eb52dd2005-11-06 07:11:04 +00001679
1680 // An extension word is present when linkage = 3 (internal) and hasinit = 0.
1681 if (LinkageID == 3 && !hasInitializer) {
1682 unsigned ExtWord = read_vbr_uint();
1683 // The extension word has this format: bit 0 = has initializer, bit 1-3 =
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001684 // linkage, bit 4-8 = alignment (log2), bit 9 = has section,
1685 // bits 10-12 = visibility, bits 13+ = future use.
Chris Lattner8eb52dd2005-11-06 07:11:04 +00001686 hasInitializer = ExtWord & 1;
1687 LinkageID = (ExtWord >> 1) & 7;
1688 Alignment = (1 << ((ExtWord >> 4) & 31)) >> 1;
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001689 VisibilityID = (ExtWord >> 10) & 7;
Chris Lattner404cddf2005-11-12 01:33:40 +00001690
1691 if (ExtWord & (1 << 9)) // Has a section ID.
1692 GlobalSectionID = read_vbr_uint();
Chris Lattner8eb52dd2005-11-06 07:11:04 +00001693 }
Chris Lattnere3869c82003-04-16 21:16:05 +00001694
Chris Lattnerce5e04e2005-11-06 08:23:17 +00001695 GlobalValue::LinkageTypes Linkage;
Chris Lattnerc08912f2004-01-14 16:44:44 +00001696 switch (LinkageID) {
Chris Lattnerc08912f2004-01-14 16:44:44 +00001697 case 0: Linkage = GlobalValue::ExternalLinkage; break;
1698 case 1: Linkage = GlobalValue::WeakLinkage; break;
1699 case 2: Linkage = GlobalValue::AppendingLinkage; break;
1700 case 3: Linkage = GlobalValue::InternalLinkage; break;
1701 case 4: Linkage = GlobalValue::LinkOnceLinkage; break;
Anton Korobeynikovb74ed072006-09-14 18:23:27 +00001702 case 5: Linkage = GlobalValue::DLLImportLinkage; break;
1703 case 6: Linkage = GlobalValue::DLLExportLinkage; break;
1704 case 7: Linkage = GlobalValue::ExternalWeakLinkage; break;
Misha Brukman8a96c532005-04-21 21:44:41 +00001705 default:
Reid Spencer24399722004-07-09 22:21:33 +00001706 error("Unknown linkage type: " + utostr(LinkageID));
Reid Spencer060d25d2004-06-29 23:29:38 +00001707 Linkage = GlobalValue::InternalLinkage;
1708 break;
Chris Lattnere3869c82003-04-16 21:16:05 +00001709 }
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001710 GlobalValue::VisibilityTypes Visibility;
1711 switch (VisibilityID) {
1712 case 0: Visibility = GlobalValue::DefaultVisibility; break;
1713 case 1: Visibility = GlobalValue::HiddenVisibility; break;
1714 default:
1715 error("Unknown visibility type: " + utostr(VisibilityID));
1716 Visibility = GlobalValue::DefaultVisibility;
1717 break;
1718 }
1719
Chris Lattnere3869c82003-04-16 21:16:05 +00001720 const Type *Ty = getType(SlotNo);
Chris Lattnere73bd452005-11-06 07:43:39 +00001721 if (!Ty)
Reid Spencer24399722004-07-09 22:21:33 +00001722 error("Global has no type! SlotNo=" + utostr(SlotNo));
Reid Spencer060d25d2004-06-29 23:29:38 +00001723
Chris Lattnere73bd452005-11-06 07:43:39 +00001724 if (!isa<PointerType>(Ty))
Reid Spencer24399722004-07-09 22:21:33 +00001725 error("Global not a pointer type! Ty= " + Ty->getDescription());
Chris Lattner70cc3392001-09-10 07:58:01 +00001726
Chris Lattner52e20b02003-03-19 20:54:26 +00001727 const Type *ElTy = cast<PointerType>(Ty)->getElementType();
Chris Lattnerd70684f2001-09-18 04:01:05 +00001728
Chris Lattner70cc3392001-09-10 07:58:01 +00001729 // Create the global variable...
Reid Spencer060d25d2004-06-29 23:29:38 +00001730 GlobalVariable *GV = new GlobalVariable(ElTy, isConstant, Linkage,
Chris Lattner52e20b02003-03-19 20:54:26 +00001731 0, "", TheModule);
Chris Lattner8eb52dd2005-11-06 07:11:04 +00001732 GV->setAlignment(Alignment);
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001733 GV->setVisibility(Visibility);
Chris Lattner29b789b2003-11-19 17:27:18 +00001734 insertValue(GV, SlotNo, ModuleValues);
Chris Lattner05950c32001-10-13 06:47:01 +00001735
Chris Lattner404cddf2005-11-12 01:33:40 +00001736 if (GlobalSectionID != 0)
1737 SectionID[GV] = GlobalSectionID;
1738
Reid Spencer060d25d2004-06-29 23:29:38 +00001739 unsigned initSlot = 0;
Misha Brukman8a96c532005-04-21 21:44:41 +00001740 if (hasInitializer) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001741 initSlot = read_vbr_uint();
1742 GlobalInits.push_back(std::make_pair(GV, initSlot));
1743 }
1744
1745 // Notify handler about the global value.
Chris Lattner4a242b32004-10-14 01:39:18 +00001746 if (Handler)
Anton Korobeynikov7f705592007-01-12 19:20:47 +00001747 Handler->handleGlobalVariable(ElTy, isConstant, Linkage, Visibility,
1748 SlotNo, initSlot);
Reid Spencer060d25d2004-06-29 23:29:38 +00001749
1750 // Get next item
1751 VarType = read_vbr_uint();
Chris Lattner70cc3392001-09-10 07:58:01 +00001752 }
1753
Chris Lattner52e20b02003-03-19 20:54:26 +00001754 // Read the function objects for all of the functions that are coming
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001755 unsigned FnSignature = read_vbr_uint();
Reid Spencer24399722004-07-09 22:21:33 +00001756
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001757 // List is terminated by VoidTy.
Chris Lattnere73bd452005-11-06 07:43:39 +00001758 while (((FnSignature & (~0U >> 1)) >> 5) != Type::VoidTyID) {
1759 const Type *Ty = getType((FnSignature & (~0U >> 1)) >> 5);
Chris Lattner927b1852003-10-09 20:22:47 +00001760 if (!isa<PointerType>(Ty) ||
Reid Spencer060d25d2004-06-29 23:29:38 +00001761 !isa<FunctionType>(cast<PointerType>(Ty)->getElementType())) {
Misha Brukman8a96c532005-04-21 21:44:41 +00001762 error("Function not a pointer to function type! Ty = " +
Reid Spencer46b002c2004-07-11 17:28:43 +00001763 Ty->getDescription());
Reid Spencer060d25d2004-06-29 23:29:38 +00001764 }
Chris Lattner8cdc6b72002-10-23 00:51:54 +00001765
Chris Lattner2a7b6ba2003-03-06 17:15:19 +00001766 // We create functions by passing the underlying FunctionType to create...
Misha Brukman8a96c532005-04-21 21:44:41 +00001767 const FunctionType* FTy =
Reid Spencer060d25d2004-06-29 23:29:38 +00001768 cast<FunctionType>(cast<PointerType>(Ty)->getElementType());
Chris Lattner00950542001-06-06 20:29:01 +00001769
Chris Lattner18549c22004-11-15 21:43:03 +00001770 // Insert the place holder.
Chris Lattner404cddf2005-11-12 01:33:40 +00001771 Function *Func = new Function(FTy, GlobalValue::ExternalLinkage,
Reid Spencer04cde2c2004-07-04 11:33:49 +00001772 "", TheModule);
Reid Spencere1e96c02006-01-19 07:02:16 +00001773
Chris Lattnere73bd452005-11-06 07:43:39 +00001774 insertValue(Func, (FnSignature & (~0U >> 1)) >> 5, ModuleValues);
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001775
1776 // Flags are not used yet.
Chris Lattner97fbc502004-11-15 22:38:52 +00001777 unsigned Flags = FnSignature & 31;
Chris Lattner00950542001-06-06 20:29:01 +00001778
Chris Lattner97fbc502004-11-15 22:38:52 +00001779 // Save this for later so we know type of lazily instantiated functions.
1780 // Note that known-external functions do not have FunctionInfo blocks, so we
1781 // do not add them to the FunctionSignatureList.
1782 if ((Flags & (1 << 4)) == 0)
1783 FunctionSignatureList.push_back(Func);
Chris Lattner52e20b02003-03-19 20:54:26 +00001784
Chris Lattnere73bd452005-11-06 07:43:39 +00001785 // Get the calling convention from the low bits.
1786 unsigned CC = Flags & 15;
1787 unsigned Alignment = 0;
1788 if (FnSignature & (1 << 31)) { // Has extension word?
1789 unsigned ExtWord = read_vbr_uint();
1790 Alignment = (1 << (ExtWord & 31)) >> 1;
1791 CC |= ((ExtWord >> 5) & 15) << 4;
Chris Lattner404cddf2005-11-12 01:33:40 +00001792
1793 if (ExtWord & (1 << 10)) // Has a section ID.
1794 SectionID[Func] = read_vbr_uint();
Anton Korobeynikovb74ed072006-09-14 18:23:27 +00001795
1796 // Parse external declaration linkage
1797 switch ((ExtWord >> 11) & 3) {
1798 case 0: break;
1799 case 1: Func->setLinkage(Function::DLLImportLinkage); break;
1800 case 2: Func->setLinkage(Function::ExternalWeakLinkage); break;
1801 default: assert(0 && "Unsupported external linkage");
1802 }
Chris Lattnere73bd452005-11-06 07:43:39 +00001803 }
1804
Chris Lattner54b369e2005-11-06 07:46:13 +00001805 Func->setCallingConv(CC-1);
Chris Lattnere73bd452005-11-06 07:43:39 +00001806 Func->setAlignment(Alignment);
Chris Lattner479ffeb2005-05-06 20:42:57 +00001807
Reid Spencer04cde2c2004-07-04 11:33:49 +00001808 if (Handler) Handler->handleFunctionDeclaration(Func);
Reid Spencer060d25d2004-06-29 23:29:38 +00001809
Chris Lattnera79e7cc2004-10-16 18:18:16 +00001810 // Get the next function signature.
1811 FnSignature = read_vbr_uint();
Chris Lattner00950542001-06-06 20:29:01 +00001812 }
1813
Misha Brukman8a96c532005-04-21 21:44:41 +00001814 // Now that the function signature list is set up, reverse it so that we can
Chris Lattner74734132002-08-17 22:01:27 +00001815 // remove elements efficiently from the back of the vector.
1816 std::reverse(FunctionSignatureList.begin(), FunctionSignatureList.end());
Chris Lattner00950542001-06-06 20:29:01 +00001817
Chris Lattner404cddf2005-11-12 01:33:40 +00001818 /// SectionNames - This contains the list of section names encoded in the
1819 /// moduleinfoblock. Functions and globals with an explicit section index
1820 /// into this to get their section name.
1821 std::vector<std::string> SectionNames;
1822
Reid Spencerd798a512006-11-14 04:47:22 +00001823 // Read in the dependent library information.
1824 unsigned num_dep_libs = read_vbr_uint();
1825 std::string dep_lib;
1826 while (num_dep_libs--) {
1827 dep_lib = read_str();
1828 TheModule->addLibrary(dep_lib);
Reid Spencer5b472d92004-08-21 20:49:23 +00001829 if (Handler)
Reid Spencerd798a512006-11-14 04:47:22 +00001830 Handler->handleDependentLibrary(dep_lib);
Reid Spencerad89bd62004-07-25 18:07:36 +00001831 }
1832
Reid Spencerd798a512006-11-14 04:47:22 +00001833 // Read target triple and place into the module.
1834 std::string triple = read_str();
1835 TheModule->setTargetTriple(triple);
1836 if (Handler)
1837 Handler->handleTargetTriple(triple);
1838
Reid Spenceraacc35a2007-01-26 08:10:24 +00001839 // Read the data layout string and place into the module.
1840 std::string datalayout = read_str();
1841 TheModule->setDataLayout(datalayout);
1842 // FIXME: Implement
1843 // if (Handler)
1844 // Handler->handleDataLayout(datalayout);
1845
Reid Spencerd798a512006-11-14 04:47:22 +00001846 if (At != BlockEnd) {
1847 // If the file has section info in it, read the section names now.
1848 unsigned NumSections = read_vbr_uint();
1849 while (NumSections--)
1850 SectionNames.push_back(read_str());
1851 }
1852
1853 // If the file has module-level inline asm, read it now.
1854 if (At != BlockEnd)
1855 TheModule->setModuleInlineAsm(read_str());
1856
Chris Lattner404cddf2005-11-12 01:33:40 +00001857 // If any globals are in specified sections, assign them now.
1858 for (std::map<GlobalValue*, unsigned>::iterator I = SectionID.begin(), E =
1859 SectionID.end(); I != E; ++I)
1860 if (I->second) {
1861 if (I->second > SectionID.size())
1862 error("SectionID out of range for global!");
1863 I->first->setSection(SectionNames[I->second-1]);
1864 }
Reid Spencerad89bd62004-07-25 18:07:36 +00001865
Chris Lattner00950542001-06-06 20:29:01 +00001866 // This is for future proofing... in the future extra fields may be added that
1867 // we don't understand, so we transparently ignore them.
1868 //
Reid Spencer060d25d2004-06-29 23:29:38 +00001869 At = BlockEnd;
1870
Reid Spencer04cde2c2004-07-04 11:33:49 +00001871 if (Handler) Handler->handleModuleGlobalsEnd();
Chris Lattner00950542001-06-06 20:29:01 +00001872}
1873
Reid Spencer04cde2c2004-07-04 11:33:49 +00001874/// Parse the version information and decode it by setting flags on the
1875/// Reader that enable backward compatibility of the reader.
Reid Spencer060d25d2004-06-29 23:29:38 +00001876void BytecodeReader::ParseVersionInfo() {
Reid Spenceraacc35a2007-01-26 08:10:24 +00001877 unsigned RevisionNum = read_vbr_uint();
Chris Lattnere3869c82003-04-16 21:16:05 +00001878
Reid Spencer3795ad12006-12-03 05:47:10 +00001879 // We don't provide backwards compatibility in the Reader any more. To
1880 // upgrade, the user should use llvm-upgrade.
1881 if (RevisionNum < 7)
1882 error("Bytecode formats < 7 are no longer supported. Use llvm-upgrade.");
Chris Lattner036b8aa2003-03-06 17:55:45 +00001883
Reid Spenceraacc35a2007-01-26 08:10:24 +00001884 if (Handler) Handler->handleVersionInfo(RevisionNum);
Chris Lattner036b8aa2003-03-06 17:55:45 +00001885}
1886
Reid Spencer04cde2c2004-07-04 11:33:49 +00001887/// Parse a whole module.
Reid Spencer060d25d2004-06-29 23:29:38 +00001888void BytecodeReader::ParseModule() {
Chris Lattner00950542001-06-06 20:29:01 +00001889 unsigned Type, Size;
Chris Lattner00950542001-06-06 20:29:01 +00001890
Reid Spencer060d25d2004-06-29 23:29:38 +00001891 FunctionSignatureList.clear(); // Just in case...
Chris Lattner00950542001-06-06 20:29:01 +00001892
1893 // Read into instance variables...
Reid Spencer060d25d2004-06-29 23:29:38 +00001894 ParseVersionInfo();
Chris Lattner00950542001-06-06 20:29:01 +00001895
Reid Spencer060d25d2004-06-29 23:29:38 +00001896 bool SeenModuleGlobalInfo = false;
1897 bool SeenGlobalTypePlane = false;
1898 BufPtr MyEnd = BlockEnd;
1899 while (At < MyEnd) {
1900 BufPtr OldAt = At;
1901 read_block(Type, Size);
1902
Chris Lattner00950542001-06-06 20:29:01 +00001903 switch (Type) {
Reid Spencer060d25d2004-06-29 23:29:38 +00001904
Reid Spencerad89bd62004-07-25 18:07:36 +00001905 case BytecodeFormat::GlobalTypePlaneBlockID:
Reid Spencer46b002c2004-07-11 17:28:43 +00001906 if (SeenGlobalTypePlane)
Reid Spencer24399722004-07-09 22:21:33 +00001907 error("Two GlobalTypePlane Blocks Encountered!");
Reid Spencer060d25d2004-06-29 23:29:38 +00001908
Reid Spencer5b472d92004-08-21 20:49:23 +00001909 if (Size > 0)
1910 ParseGlobalTypes();
Reid Spencer060d25d2004-06-29 23:29:38 +00001911 SeenGlobalTypePlane = true;
Chris Lattner52e20b02003-03-19 20:54:26 +00001912 break;
1913
Misha Brukman8a96c532005-04-21 21:44:41 +00001914 case BytecodeFormat::ModuleGlobalInfoBlockID:
Reid Spencer46b002c2004-07-11 17:28:43 +00001915 if (SeenModuleGlobalInfo)
Reid Spencer24399722004-07-09 22:21:33 +00001916 error("Two ModuleGlobalInfo Blocks Encountered!");
Reid Spencer060d25d2004-06-29 23:29:38 +00001917 ParseModuleGlobalInfo();
1918 SeenModuleGlobalInfo = true;
Chris Lattner52e20b02003-03-19 20:54:26 +00001919 break;
1920
Reid Spencerad89bd62004-07-25 18:07:36 +00001921 case BytecodeFormat::ConstantPoolBlockID:
Reid Spencer04cde2c2004-07-04 11:33:49 +00001922 ParseConstantPool(ModuleValues, ModuleTypes,false);
Chris Lattner00950542001-06-06 20:29:01 +00001923 break;
1924
Reid Spencerad89bd62004-07-25 18:07:36 +00001925 case BytecodeFormat::FunctionBlockID:
Reid Spencer060d25d2004-06-29 23:29:38 +00001926 ParseFunctionLazily();
Chris Lattner00950542001-06-06 20:29:01 +00001927 break;
Chris Lattner00950542001-06-06 20:29:01 +00001928
Reid Spencer78d033e2007-01-06 07:24:44 +00001929 case BytecodeFormat::ValueSymbolTableBlockID:
1930 ParseValueSymbolTable(0, &TheModule->getValueSymbolTable());
1931 break;
1932
1933 case BytecodeFormat::TypeSymbolTableBlockID:
1934 ParseTypeSymbolTable(&TheModule->getTypeSymbolTable());
Chris Lattner00950542001-06-06 20:29:01 +00001935 break;
Reid Spencer060d25d2004-06-29 23:29:38 +00001936
Chris Lattner00950542001-06-06 20:29:01 +00001937 default:
Reid Spencer060d25d2004-06-29 23:29:38 +00001938 At += Size;
1939 if (OldAt > At) {
Reid Spencer46b002c2004-07-11 17:28:43 +00001940 error("Unexpected Block of Type #" + utostr(Type) + " encountered!");
Reid Spencer060d25d2004-06-29 23:29:38 +00001941 }
Chris Lattner00950542001-06-06 20:29:01 +00001942 break;
1943 }
Reid Spencer060d25d2004-06-29 23:29:38 +00001944 BlockEnd = MyEnd;
Chris Lattner00950542001-06-06 20:29:01 +00001945 }
1946
Chris Lattner52e20b02003-03-19 20:54:26 +00001947 // After the module constant pool has been read, we can safely initialize
1948 // global variables...
1949 while (!GlobalInits.empty()) {
1950 GlobalVariable *GV = GlobalInits.back().first;
1951 unsigned Slot = GlobalInits.back().second;
1952 GlobalInits.pop_back();
1953
1954 // Look up the initializer value...
Chris Lattner29b789b2003-11-19 17:27:18 +00001955 // FIXME: Preserve this type ID!
Reid Spencer060d25d2004-06-29 23:29:38 +00001956
1957 const llvm::PointerType* GVType = GV->getType();
1958 unsigned TypeSlot = getTypeSlot(GVType->getElementType());
Chris Lattner93361992004-01-15 18:45:25 +00001959 if (Constant *CV = getConstantValue(TypeSlot, Slot)) {
Misha Brukman8a96c532005-04-21 21:44:41 +00001960 if (GV->hasInitializer())
Reid Spencer24399722004-07-09 22:21:33 +00001961 error("Global *already* has an initializer?!");
Reid Spencer04cde2c2004-07-04 11:33:49 +00001962 if (Handler) Handler->handleGlobalInitializer(GV,CV);
Chris Lattner93361992004-01-15 18:45:25 +00001963 GV->setInitializer(CV);
Chris Lattner52e20b02003-03-19 20:54:26 +00001964 } else
Reid Spencer24399722004-07-09 22:21:33 +00001965 error("Cannot find initializer value.");
Chris Lattner52e20b02003-03-19 20:54:26 +00001966 }
1967
Chris Lattneraba5ff52005-05-05 20:57:00 +00001968 if (!ConstantFwdRefs.empty())
1969 error("Use of undefined constants in a module");
1970
Reid Spencer060d25d2004-06-29 23:29:38 +00001971 /// Make sure we pulled them all out. If we didn't then there's a declaration
1972 /// but a missing body. That's not allowed.
Misha Brukman12c29d12003-09-22 23:38:23 +00001973 if (!FunctionSignatureList.empty())
Reid Spencer24399722004-07-09 22:21:33 +00001974 error("Function declared, but bytecode stream ended before definition");
Chris Lattner00950542001-06-06 20:29:01 +00001975}
1976
Reid Spencer04cde2c2004-07-04 11:33:49 +00001977/// This function completely parses a bytecode buffer given by the \p Buf
1978/// and \p Length parameters.
Anton Korobeynikov7d515442006-09-01 20:35:17 +00001979bool BytecodeReader::ParseBytecode(volatile BufPtr Buf, unsigned Length,
Reid Spencer233fe722006-08-22 16:09:19 +00001980 const std::string &ModuleID,
1981 std::string* ErrMsg) {
Misha Brukmane0dd0d42003-09-23 16:15:29 +00001982
Reid Spencer233fe722006-08-22 16:09:19 +00001983 /// We handle errors by
1984 if (setjmp(context)) {
1985 // Cleanup after error
1986 if (Handler) Handler->handleError(ErrorMsg);
Reid Spencer060d25d2004-06-29 23:29:38 +00001987 freeState();
Chris Lattner2a7b6ba2003-03-06 17:15:19 +00001988 delete TheModule;
1989 TheModule = 0;
Chris Lattner3bdad692004-11-15 21:55:33 +00001990 if (decompressedBlock != 0 ) {
Reid Spencer61aaf2e2004-11-14 21:59:21 +00001991 ::free(decompressedBlock);
Chris Lattner3bdad692004-11-15 21:55:33 +00001992 decompressedBlock = 0;
1993 }
Reid Spencer233fe722006-08-22 16:09:19 +00001994 // Set caller's error message, if requested
1995 if (ErrMsg)
1996 *ErrMsg = ErrorMsg;
1997 // Indicate an error occurred
1998 return true;
Chris Lattner2a7b6ba2003-03-06 17:15:19 +00001999 }
Reid Spencer233fe722006-08-22 16:09:19 +00002000
2001 RevisionNum = 0;
2002 At = MemStart = BlockStart = Buf;
2003 MemEnd = BlockEnd = Buf + Length;
2004
2005 // Create the module
2006 TheModule = new Module(ModuleID);
2007
2008 if (Handler) Handler->handleStart(TheModule, Length);
2009
2010 // Read the four bytes of the signature.
2011 unsigned Sig = read_uint();
2012
2013 // If this is a compressed file
2014 if (Sig == ('l' | ('l' << 8) | ('v' << 16) | ('c' << 24))) {
2015
2016 // Invoke the decompression of the bytecode. Note that we have to skip the
2017 // file's magic number which is not part of the compressed block. Hence,
2018 // the Buf+4 and Length-4. The result goes into decompressedBlock, a data
2019 // member for retention until BytecodeReader is destructed.
2020 unsigned decompressedLength = Compressor::decompressToNewBuffer(
2021 (char*)Buf+4,Length-4,decompressedBlock);
2022
2023 // We must adjust the buffer pointers used by the bytecode reader to point
2024 // into the new decompressed block. After decompression, the
2025 // decompressedBlock will point to a contiguous memory area that has
2026 // the decompressed data.
2027 At = MemStart = BlockStart = Buf = (BufPtr) decompressedBlock;
2028 MemEnd = BlockEnd = Buf + decompressedLength;
2029
2030 // else if this isn't a regular (uncompressed) bytecode file, then its
2031 // and error, generate that now.
2032 } else if (Sig != ('l' | ('l' << 8) | ('v' << 16) | ('m' << 24))) {
2033 error("Invalid bytecode signature: " + utohexstr(Sig));
2034 }
2035
2036 // Tell the handler we're starting a module
2037 if (Handler) Handler->handleModuleBegin(ModuleID);
2038
2039 // Get the module block and size and verify. This is handled specially
2040 // because the module block/size is always written in long format. Other
2041 // blocks are written in short format so the read_block method is used.
2042 unsigned Type, Size;
2043 Type = read_uint();
2044 Size = read_uint();
2045 if (Type != BytecodeFormat::ModuleBlockID) {
2046 error("Expected Module Block! Type:" + utostr(Type) + ", Size:"
2047 + utostr(Size));
2048 }
2049
2050 // It looks like the darwin ranlib program is broken, and adds trailing
2051 // garbage to the end of some bytecode files. This hack allows the bc
2052 // reader to ignore trailing garbage on bytecode files.
2053 if (At + Size < MemEnd)
2054 MemEnd = BlockEnd = At+Size;
2055
2056 if (At + Size != MemEnd)
2057 error("Invalid Top Level Block Length! Type:" + utostr(Type)
2058 + ", Size:" + utostr(Size));
2059
2060 // Parse the module contents
2061 this->ParseModule();
2062
2063 // Check for missing functions
2064 if (hasFunctions())
2065 error("Function expected, but bytecode stream ended!");
2066
Reid Spencer233fe722006-08-22 16:09:19 +00002067 // Tell the handler we're done with the module
2068 if (Handler)
2069 Handler->handleModuleEnd(ModuleID);
2070
2071 // Tell the handler we're finished the parse
2072 if (Handler) Handler->handleFinish();
2073
2074 return false;
2075
Chris Lattner00950542001-06-06 20:29:01 +00002076}
Reid Spencer060d25d2004-06-29 23:29:38 +00002077
2078//===----------------------------------------------------------------------===//
2079//=== Default Implementations of Handler Methods
2080//===----------------------------------------------------------------------===//
2081
2082BytecodeHandler::~BytecodeHandler() {}
Reid Spencer060d25d2004-06-29 23:29:38 +00002083