src/arch/x86/predecoder.cc

   1 /*
   2  * Copyright (c) 2007-2008 The Hewlett-Packard Development Company
   3  * All rights reserved.
   4  *
   5  * Redistribution and use of this software in source and binary forms,
   6  * with or without modification, are permitted provided that the
   7  * following conditions are met:
   8  *
   9  * The software must be used only for Non-Commercial Use which means any
  10  * use which is NOT directed to receiving any direct monetary
  11  * compensation for, or commercial advantage from such use.  Illustrative
  12  * examples of non-commercial use are academic research, personal study,
  13  * teaching, education and corporate research & development.
  14  * Illustrative examples of commercial use are distributing products for
  15  * commercial advantage and providing services using the software for
  16  * commercial advantage.
  17  *
  18  * If you wish to use this software or functionality therein that may be
  19  * covered by patents for commercial use, please contact:
  20  *     Director of Intellectual Property Licensing
  21  *     Office of Strategy and Technology
  22  *     Hewlett-Packard Company
  23  *     1501 Page Mill Road
  24  *     Palo Alto, California  94304
  25  *
  26  * Redistributions of source code must retain the above copyright notice,
  27  * this list of conditions and the following disclaimer.  Redistributions
  28  * in binary form must reproduce the above copyright notice, this list of
  29  * conditions and the following disclaimer in the documentation and/or
  30  * other materials provided with the distribution.  Neither the name of
  31  * the COPYRIGHT HOLDER(s), HEWLETT-PACKARD COMPANY, nor the names of its
  32  * contributors may be used to endorse or promote products derived from
  33  * this software without specific prior written permission.  No right of
  34  * sublicense is granted herewith.  Derivatives of the software and
  35  * output created using the software may be prepared, but only for
  36  * Non-Commercial Uses.  Derivatives of the software may be shared with
  37  * others provided: (i) the others agree to abide by the list of
  38  * conditions herein which includes the Non-Commercial Use restrictions;
  39  * and (ii) such Derivatives of the software include the above copyright
  40  * notice to acknowledge the contribution from this software where
  41  * applicable, this list of conditions and the disclaimer below.
  42  *
  43  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  44  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  45  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  46  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  47  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  48  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  49  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  50  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  51  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  52  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  53  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  54  *
  55  * Authors: Gabe Black
  56  */
  57
  58 #include "arch/x86/miscregs.hh"
  59 #include "arch/x86/predecoder.hh"
  60 #include "base/misc.hh"
  61 #include "base/trace.hh"
  62 #include "cpu/thread_context.hh"
  63 #include "sim/host.hh"
  64
  65 namespace X86ISA
  66 {
  67     void Predecoder::doReset()
  68     {
  69         origPC = basePC + offset;
  70         DPRINTF(Predecoder, "Setting origPC to %#x\n", origPC);
  71         emi.rex = 0;
  72         emi.legacy = 0;
  73         emi.opcode.num = 0;
  74         emi.opcode.op = 0;
  75         emi.opcode.prefixA = emi.opcode.prefixB = 0;
  76
  77         immediateCollected = 0;
  78         emi.immediate = 0;
  79         emi.displacement = 0;
  80
  81         emi.modRM = 0;
  82         emi.sib = 0;
  83         m5Reg = tc->readMiscRegNoEffect(MISCREG_M5_REG);
  84         emi.mode.mode = m5Reg.mode;
  85         emi.mode.submode = m5Reg.submode;
  86     }
  87
  88     void Predecoder::process()
  89     {
  90         //This function drives the predecoder state machine.
  91
  92         //Some sanity checks. You shouldn't try to process more bytes if
  93         //there aren't any, and you shouldn't overwrite an already
  94         //predecoder ExtMachInst.
  95         assert(!outOfBytes);
  96         assert(!emiIsReady);
  97
  98         //While there's still something to do...
  99         while(!emiIsReady && !outOfBytes)
 100         {
 101             uint8_t nextByte = getNextByte();
 102             switch(state)
 103             {
 104               case ResetState:
 105                 doReset();
 106                 state = PrefixState;
 107               case PrefixState:
 108                 state = doPrefixState(nextByte);
 109                 break;
 110               case OpcodeState:
 111                 state = doOpcodeState(nextByte);
 112                 break;
 113               case ModRMState:
 114                 state = doModRMState(nextByte);
 115                 break;
 116               case SIBState:
 117                 state = doSIBState(nextByte);
 118                 break;
 119               case DisplacementState:
 120                 state = doDisplacementState();
 121                 break;
 122               case ImmediateState:
 123                 state = doImmediateState();
 124                 break;
 125               case ErrorState:
 126                 panic("Went to the error state in the predecoder.\n");
 127               default:
 128                 panic("Unrecognized state! %d\n", state);
 129             }
 130         }
 131     }
 132
 133     //Either get a prefix and record it in the ExtMachInst, or send the
 134     //state machine on to get the opcode(s).
 135     Predecoder::State Predecoder::doPrefixState(uint8_t nextByte)
 136     {
 137         uint8_t prefix = Prefixes[nextByte];
 138         State nextState = PrefixState;
 139         // REX prefixes are only recognized in 64 bit mode.
 140         if (prefix == RexPrefix && emi.mode.submode != SixtyFourBitMode)
 141             prefix = 0;
 142         if (prefix)
 143             consumeByte();
 144         switch(prefix)
 145         {
 146             //Operand size override prefixes
 147           case OperandSizeOverride:
 148             DPRINTF(Predecoder, "Found operand size override prefix.\n");
 149             emi.legacy.op = true;
 150             break;
 151           case AddressSizeOverride:
 152             DPRINTF(Predecoder, "Found address size override prefix.\n");
 153             emi.legacy.addr = true;
 154             break;
 155             //Segment override prefixes
 156           case CSOverride:
 157           case DSOverride:
 158           case ESOverride:
 159           case FSOverride:
 160           case GSOverride:
 161           case SSOverride:
 162             DPRINTF(Predecoder, "Found segment override.\n");
 163             emi.legacy.seg = prefix;
 164             break;
 165           case Lock:
 166             DPRINTF(Predecoder, "Found lock prefix.\n");
 167             emi.legacy.lock = true;
 168             break;
 169           case Rep:
 170             DPRINTF(Predecoder, "Found rep prefix.\n");
 171             emi.legacy.rep = true;
 172             break;
 173           case Repne:
 174             DPRINTF(Predecoder, "Found repne prefix.\n");
 175             emi.legacy.repne = true;
 176             break;
 177           case RexPrefix:
 178             DPRINTF(Predecoder, "Found Rex prefix %#x.\n", nextByte);
 179             emi.rex = nextByte;
 180             break;
 181           case 0:
 182             nextState = OpcodeState;
 183             break;
 184           default:
 185             panic("Unrecognized prefix %#x\n", nextByte);
 186         }
 187         return nextState;
 188     }
 189
 190     //Load all the opcodes (currently up to 2) and then figure out
 191     //what immediate and/or ModRM is needed.
 192     Predecoder::State Predecoder::doOpcodeState(uint8_t nextByte)
 193     {
 194         State nextState = ErrorState;
 195         emi.opcode.num++;
 196         //We can't handle 3+ byte opcodes right now
 197         assert(emi.opcode.num < 3);
 198         consumeByte();
 199         if(emi.opcode.num == 1 && nextByte == 0x0f)
 200         {
 201             nextState = OpcodeState;
 202             DPRINTF(Predecoder, "Found two byte opcode.\n");
 203             emi.opcode.prefixA = nextByte;
 204         }
 205         else if(emi.opcode.num == 2 &&
 206                 (nextByte == 0x0f ||
 207                  (nextByte & 0xf8) == 0x38))
 208         {
 209             panic("Three byte opcodes aren't yet supported!\n");
 210             nextState = OpcodeState;
 211             DPRINTF(Predecoder, "Found three byte opcode.\n");
 212             emi.opcode.prefixB = nextByte;
 213         }
 214         else
 215         {
 216             DPRINTF(Predecoder, "Found opcode %#x.\n", nextByte);
 217             emi.opcode.op = nextByte;
 218
 219             //Figure out the effective operand size. This can be overriden to
 220             //a fixed value at the decoder level.
 221             int logOpSize;
 222             if (emi.rex.w)
 223                 logOpSize = 3; // 64 bit operand size
 224             else if (emi.legacy.op)
 225                 logOpSize = m5Reg.altOp;
 226             else
 227                 logOpSize = m5Reg.defOp;
 228
 229             //Set the actual op size
 230             emi.opSize = 1 << logOpSize;
 231
 232             //Figure out the effective address size. This can be overriden to
 233             //a fixed value at the decoder level.
 234             int logAddrSize;
 235             if(emi.legacy.addr)
 236                 logAddrSize = m5Reg.altAddr;
 237             else
 238                 logAddrSize = m5Reg.defAddr;
 239
 240             //Set the actual address size
 241             emi.addrSize = 1 << logAddrSize;
 242
 243             //Figure out the effective stack width. This can be overriden to
 244             //a fixed value at the decoder level.
 245             emi.stackSize = 1 << m5Reg.stack;
 246
 247             //Figure out how big of an immediate we'll retreive based
 248             //on the opcode.
 249             int immType = ImmediateType[emi.opcode.num - 1][nextByte];
 250             if (emi.opcode.num == 1 && nextByte >= 0xA0 && nextByte <= 0xA3)
 251                 immediateSize = SizeTypeToSize[logAddrSize - 1][immType];
 252             else
 253                 immediateSize = SizeTypeToSize[logOpSize - 1][immType];
 254
 255             //Determine what to expect next
 256             if (UsesModRM[emi.opcode.num - 1][nextByte]) {
 257                 nextState = ModRMState;
 258             } else {
 259                 if(immediateSize) {
 260                     nextState = ImmediateState;
 261                 } else {
 262                     emiIsReady = true;
 263                     nextState = ResetState;
 264                 }
 265             }
 266         }
 267         return nextState;
 268     }
 269
 270     //Get the ModRM byte and determine what displacement, if any, there is.
 271     //Also determine whether or not to get the SIB byte, displacement, or
 272     //immediate next.
 273     Predecoder::State Predecoder::doModRMState(uint8_t nextByte)
 274     {
 275         State nextState = ErrorState;
 276         ModRM modRM;
 277         modRM = nextByte;
 278         DPRINTF(Predecoder, "Found modrm byte %#x.\n", nextByte);
 279         if (m5Reg.defOp == 1) {
 280             //figure out 16 bit displacement size
 281             if ((modRM.mod == 0 && modRM.rm == 6) || modRM.mod == 2)
 282                 displacementSize = 2;
 283             else if (modRM.mod == 1)
 284                 displacementSize = 1;
 285             else
 286                 displacementSize = 0;
 287         } else {
 288             //figure out 32/64 bit displacement size
 289             if ((modRM.mod == 0 && modRM.rm == 5) || modRM.mod == 2)
 290                 displacementSize = 4;
 291             else if (modRM.mod == 1)
 292                 displacementSize = 1;
 293             else
 294                 displacementSize = 0;
 295         }
 296
 297         // The "test" instruction in group 3 needs an immediate, even though
 298         // the other instructions with the same actual opcode don't.
 299         if (emi.opcode.num == 1 && (modRM.reg & 0x6) == 0) {
 300            if (emi.opcode.op == 0xF6)
 301                immediateSize = 1;
 302            else if (emi.opcode.op == 0xF7)
 303                immediateSize = (emi.opSize == 8) ? 4 : emi.opSize;
 304         }
 305
 306         //If there's an SIB, get that next.
 307         //There is no SIB in 16 bit mode.
 308         if (modRM.rm == 4 && modRM.mod != 3) {
 309                 // && in 32/64 bit mode)
 310             nextState = SIBState;
 311         } else if(displacementSize) {
 312             nextState = DisplacementState;
 313         } else if(immediateSize) {
 314             nextState = ImmediateState;
 315         } else {
 316             emiIsReady = true;
 317             nextState = ResetState;
 318         }
 319         //The ModRM byte is consumed no matter what
 320         consumeByte();
 321         emi.modRM = modRM;
 322         return nextState;
 323     }
 324
 325     //Get the SIB byte. We don't do anything with it at this point, other
 326     //than storing it in the ExtMachInst. Determine if we need to get a
 327     //displacement or immediate next.
 328     Predecoder::State Predecoder::doSIBState(uint8_t nextByte)
 329     {
 330         State nextState = ErrorState;
 331         emi.sib = nextByte;
 332         DPRINTF(Predecoder, "Found SIB byte %#x.\n", nextByte);
 333         consumeByte();
 334         if (emi.modRM.mod == 0 && emi.sib.base == 5)
 335             displacementSize = 4;
 336         if (displacementSize) {
 337             nextState = DisplacementState;
 338         } else if(immediateSize) {
 339             nextState = ImmediateState;
 340         } else {
 341             emiIsReady = true;
 342             nextState = ResetState;
 343         }
 344         return nextState;
 345     }
 346
 347     //Gather up the displacement, or at least as much of it
 348     //as we can get.
 349     Predecoder::State Predecoder::doDisplacementState()
 350     {
 351         State nextState = ErrorState;
 352
 353         getImmediate(immediateCollected,
 354                 emi.displacement,
 355                 displacementSize);
 356
 357         DPRINTF(Predecoder, "Collecting %d byte displacement, got %d bytes.\n",
 358                 displacementSize, immediateCollected);
 359
 360         if(displacementSize == immediateCollected) {
 361             //Reset this for other immediates.
 362             immediateCollected = 0;
 363             //Sign extend the displacement
 364             switch(displacementSize)
 365             {
 366               case 1:
 367                 emi.displacement = sext<8>(emi.displacement);
 368                 break;
 369               case 2:
 370                 emi.displacement = sext<16>(emi.displacement);
 371                 break;
 372               case 4:
 373                 emi.displacement = sext<32>(emi.displacement);
 374                 break;
 375               default:
 376                 panic("Undefined displacement size!\n");
 377             }
 378             DPRINTF(Predecoder, "Collected displacement %#x.\n",
 379                     emi.displacement);
 380             if(immediateSize) {
 381                 nextState = ImmediateState;
 382             } else {
 383                 emiIsReady = true;
 384                 nextState = ResetState;
 385             }
 386         }
 387         else
 388             nextState = DisplacementState;
 389         return nextState;
 390     }
 391
 392     //Gather up the immediate, or at least as much of it
 393     //as we can get
 394     Predecoder::State Predecoder::doImmediateState()
 395     {
 396         State nextState = ErrorState;
 397
 398         getImmediate(immediateCollected,
 399                 emi.immediate,
 400                 immediateSize);
 401
 402         DPRINTF(Predecoder, "Collecting %d byte immediate, got %d bytes.\n",
 403                 immediateSize, immediateCollected);
 404
 405         if(immediateSize == immediateCollected)
 406         {
 407             //Reset this for other immediates.
 408             immediateCollected = 0;
 409
 410             //XXX Warning! The following is an observed pattern and might
 411             //not always be true!
 412
 413             //Instructions which use 64 bit operands but 32 bit immediates
 414             //need to have the immediate sign extended to 64 bits.
 415             //Instructions which use true 64 bit immediates won't be
 416             //affected, and instructions that use true 32 bit immediates
 417             //won't notice.
 418             switch(immediateSize)
 419             {
 420               case 4:
 421                 emi.immediate = sext<32>(emi.immediate);
 422                 break;
 423               case 1:
 424                 emi.immediate = sext<8>(emi.immediate);
 425             }
 426
 427             DPRINTF(Predecoder, "Collected immediate %#x.\n",
 428                     emi.immediate);
 429             emiIsReady = true;
 430             nextState = ResetState;
 431         }
 432         else
 433             nextState = ImmediateState;
 434         return nextState;
 435     }
 436 }