[AMDGPU] Disassembler: Added basic disassembler for AMDGPU target

Changes:

- Added disassembler project
- Fixed all decoding conflicts in .td files
- Added DecoderMethod=“NONE” option to Target.td that allows to
  disable decoder generation for an instruction.
- Created decoding functions for VS_32 and VReg_32 register classes.
- Added stubs for decoding all register classes.
- Added several tests for disassembler

Disassembler only supports:

- VI subtarget
- VOP1 instruction encoding
- 32-bit register operands and inline constants

[Valery]

One of the point that requires to pay attention to is how decoder
conflicts were resolved:

- Groups of target instructions were separated by using different
  DecoderNamespace (SICI, VI, CI) using similar to AssemblerPredicate
  approach.

- There were conflicts in IMAGE_<> instructions caused by two
  different reasons:

1. dmask wasn’t specified for the output (fixed)
2. There are image instructions that differ only by the number of
   the address components but have the same encoding by the HW spec. The
   actual number of address components is determined by the HW at runtime
   using image resource descriptor starting from the VGPR encoded in an
   IMAGE instruction. This means that we should choose only one instruction
   from conflicting group to be the rule for decoder. I didn’t find the way
   to disable decoder generation for an arbitrary instruction and therefore
   made a onelinear fix to tablegen generator that would suppress decoder
   generation when DecoderMethod is set to “NONE”. This is a change that
   should be reviewed and submitted first. Otherwise I would need to
   specify different DecoderNamespace for every instruction in the
   conflicting group. I haven’t checked yet if DecoderMethod=“NONE” is not
   used in other targets.
3. IMAGE_GATHER decoder generation is for now disabled and to be
   done later.

[/Valery]

Patch By: Sam Kolton

Differential Revision: http://reviews.llvm.org/D16723

llvm-svn: 261185

llvm/lib/Target/AMDGPU/AMDGPUInstructions.td

@@ -23,6 +23,14 @@ class AMDGPUInst <dag outs, dag ins, string asm, list<dag> pattern> : Instructio
   let Pattern = pattern;
   let Itinerary = NullALU;
 
+  // SoftFail is a field the disassembler can use to provide a way for
+  // instructions to not match without killing the whole decode process. It is
+  // mainly used for ARM, but Tablegen expects this field to exist or it fails
+  // to build the decode table.
+  field bits<64> SoftFail = 0;
+
+  let DecoderNamespace = Namespace;
+
   let TSFlags{63} = isRegisterLoad;
   let TSFlags{62} = isRegisterStore;
 }

llvm/lib/Target/AMDGPU/CIInstructions.td

@@ -100,9 +100,11 @@ defm S_DCACHE_INV_VOL : SMRD_Inval <smrd<0x1d, 0x22>,
 // MUBUF Instructions
 //===----------------------------------------------------------------------===//
 
+let DisableSIDecoder = 1 in {
 defm BUFFER_WBINVL1_VOL : MUBUF_Invalidate <mubuf<0x70, 0x3f>,
   "buffer_wbinvl1_vol", int_amdgcn_buffer_wbinvl1_vol
 >;
+}
 
 //===----------------------------------------------------------------------===//
 // Flat Instructions
@@ -233,7 +235,7 @@ defm FLAT_ATOMIC_DEC_X2 : FLAT_ATOMIC <
 
 // CI Only flat instructions
 
-let SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst in {
+let SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst, DisableVIDecoder = 1 in {
 
 defm FLAT_ATOMIC_FCMPSWAP : FLAT_ATOMIC <
   flat<0x3e>, "flat_atomic_fcmpswap", VGPR_32, VReg_64
@@ -254,7 +256,7 @@ defm FLAT_ATOMIC_FMAX_X2 : FLAT_ATOMIC <
   flat<0x60>, "flat_atomic_fmax_x2", VReg_64
 >;
 
-} // End SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst
+} // End SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst, DisableVIDecoder = 1
 
 let Predicates = [isCI] in {
 

llvm/lib/Target/AMDGPU/CMakeLists.txt

@@ -10,6 +10,7 @@ tablegen(LLVM AMDGPUGenMCCodeEmitter.inc -gen-emitter)
 tablegen(LLVM AMDGPUGenDFAPacketizer.inc -gen-dfa-packetizer)
 tablegen(LLVM AMDGPUGenAsmWriter.inc -gen-asm-writer)
 tablegen(LLVM AMDGPUGenAsmMatcher.inc -gen-asm-matcher)
+tablegen(LLVM AMDGPUGenDisassemblerTables.inc -gen-disassembler)
 add_public_tablegen_target(AMDGPUCommonTableGen)
 
 add_llvm_target(AMDGPUCodeGen
@@ -65,6 +66,7 @@ add_llvm_target(AMDGPUCodeGen
 
 add_subdirectory(AsmParser)
 add_subdirectory(InstPrinter)
+add_subdirectory(Disassembler)
 add_subdirectory(TargetInfo)
 add_subdirectory(MCTargetDesc)
 add_subdirectory(Utils)

llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp

@@ -0,0 +1,302 @@
+//===-- AMDGPUDisassembler.cpp - Disassembler for AMDGPU ISA --------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+//===----------------------------------------------------------------------===//
+//
+/// \file
+///
+/// This file contains definition for AMDGPU ISA disassembler
+//
+//===----------------------------------------------------------------------===//
+
+// ToDo: What to do with instruction suffixes (v_mov_b32 vs v_mov_b32_e32)?
+
+#include "AMDGPUDisassembler.h"
+#include "AMDGPU.h"
+#include "AMDGPURegisterInfo.h"
+#include "Utils/AMDGPUBaseInfo.h"
+
+#include "llvm/MC/MCFixedLenDisassembler.h"
+#include "llvm/MC/MCInst.h"
+#include "llvm/MC/MCInstrDesc.h"
+#include "llvm/MC/MCSubtargetInfo.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/TargetRegistry.h"
+
+
+using namespace llvm;
+
+#define DEBUG_TYPE "amdgpu-disassembler"
+
+typedef llvm::MCDisassembler::DecodeStatus DecodeStatus;
+
+
+static DecodeStatus DecodeVGPR_32RegisterClass(MCInst &Inst, unsigned Imm,
+                                               uint64_t Addr, const void *Decoder) {
+  const AMDGPUDisassembler *Dis = 
+    static_cast<const AMDGPUDisassembler *>(Decoder);
+  return Dis->DecodeVGPR_32RegisterClass(Inst, Imm, Addr);
+}
+
+static DecodeStatus DecodeVS_32RegisterClass(MCInst &Inst, unsigned Imm,
+                                             uint64_t Addr, const void *Decoder) {
+  const AMDGPUDisassembler *Dis = 
+    static_cast<const AMDGPUDisassembler *>(Decoder);
+  return Dis->DecodeVS_32RegisterClass(Inst, Imm, Addr);
+}
+
+static DecodeStatus DecodeVS_64RegisterClass(MCInst &Inst, unsigned Imm, 
+                                             uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeVReg_64RegisterClass(MCInst &Inst, unsigned Imm, 
+                                               uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeVReg_96RegisterClass(MCInst &Inst, unsigned Imm, 
+                                               uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeVReg_128RegisterClass(MCInst &Inst, unsigned Imm, 
+                                                uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeSReg_32RegisterClass(MCInst &Inst, unsigned Imm, 
+                                               uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeSReg_64RegisterClass(MCInst &Inst, unsigned Imm, 
+                                               uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeSReg_128RegisterClass(MCInst &Inst, unsigned Imm, 
+                                                uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+static DecodeStatus DecodeSReg_256RegisterClass(MCInst &Inst, unsigned Imm, 
+                                                uint64_t Addr, const void *Decoder) {
+  // ToDo
+  return MCDisassembler::Fail;
+}
+
+
+#define GET_SUBTARGETINFO_ENUM
+#include "AMDGPUGenSubtargetInfo.inc"
+#undef GET_SUBTARGETINFO_ENUM
+
+#include "AMDGPUGenDisassemblerTables.inc"
+
+//===----------------------------------------------------------------------===//
+//
+//===----------------------------------------------------------------------===//
+
+DecodeStatus AMDGPUDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
+                                                ArrayRef<uint8_t> Bytes,
+                                                uint64_t Address, 
+                                                raw_ostream &WS,
+                                                raw_ostream &CS) const {
+  CommentStream = &CS;
+
+  // ToDo: AMDGPUDisassembler supports only VI ISA.
+  assert(AMDGPU::isVI(STI) && "Can disassemble only VI ISA.");
+
+  // Try decode 32-bit instruction
+  if (Bytes.size() < 4) {
+    Size = 0;
+    return MCDisassembler::Fail;
+  }
+  uint32_t Insn =
+      (Bytes[3] << 24) | (Bytes[2] << 16) | (Bytes[1] << 8) | (Bytes[0] << 0);
+
+  // Calling the auto-generated decoder function.
+  DecodeStatus Result =
+      decodeInstruction(DecoderTableVI32, MI, Insn, Address, this, STI);
+  if (Result != MCDisassembler::Success) {
+      Size = 0;
+      return MCDisassembler::Fail;
+  }
+  Size = 4;
+
+  return MCDisassembler::Success;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeLitFloat(unsigned Imm, uint32_t& F) const {
+  // ToDo: case 248: 1/(2*PI) - is allowed only on VI
+  // ToDo: AMDGPUInstPrinter does not support 1/(2*PI). It consider 1/(2*PI) as 
+  // literal constant.
+  switch(Imm) {
+  case 240: F = FloatToBits(0.5f); return MCDisassembler::Success;
+  case 241: F = FloatToBits(-0.5f); return MCDisassembler::Success;
+  case 242: F = FloatToBits(1.0f); return MCDisassembler::Success;
+  case 243: F = FloatToBits(-1.0f); return MCDisassembler::Success;
+  case 244: F = FloatToBits(2.0f); return MCDisassembler::Success;
+  case 245: F = FloatToBits(-2.0f); return MCDisassembler::Success;
+  case 246: F = FloatToBits(4.0f); return MCDisassembler::Success;
+  case 247: F = FloatToBits(-4.0f); return MCDisassembler::Success;
+  case 248: F = 0x3e22f983; return MCDisassembler::Success; // 1/(2*PI)
+  default: return MCDisassembler::Fail;
+  }
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeLitInteger(unsigned Imm, 
+                                                  int64_t& I) const {
+  if ((Imm >= 128) && (Imm <= 192)) {
+    I = Imm - 128;
+    return MCDisassembler::Success;
+  } else if ((Imm >= 193) && (Imm <= 208)) {
+    I = 192 - Imm;
+    return MCDisassembler::Success;
+  }
+  return MCDisassembler::Fail;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeVgprRegister(unsigned Val, 
+                                                    unsigned& RegID) const {
+  if (Val > 255) {
+    return MCDisassembler::Fail;
+  }
+  RegID = AMDGPUMCRegisterClasses[AMDGPU::VGPR_32RegClassID].getRegister(Val);
+  return MCDisassembler::Success;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeSgprRegister(unsigned Val, 
+                                                    unsigned& RegID) const {
+  // ToDo: SI/CI have 104 SGPRs, VI - 102
+  if (Val > 101) {
+    return MCDisassembler::Fail;
+  }
+  RegID = AMDGPUMCRegisterClasses[AMDGPU::SGPR_32RegClassID].getRegister(Val);
+  return MCDisassembler::Success;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeSrcRegister(unsigned Val, 
+                                                   unsigned& RegID) const {
+  // ToDo: deal with out-of range registers  
+  using namespace AMDGPU;
+  if (Val <= 101) {
+    return DecodeSgprRegister(Val, RegID);
+  } else if ((Val >= 256) && (Val <= 511)) {
+    return DecodeVgprRegister(Val - 256, RegID);
+  } else {
+    switch(Val) {
+    case 102: RegID = getMCReg(FLAT_SCR_LO, STI); return MCDisassembler::Success;
+    case 103: RegID = getMCReg(FLAT_SCR_HI, STI); return MCDisassembler::Success;
+    // ToDo: no support for xnack_mask_lo/_hi register
+    case 104:
+    case 105: return MCDisassembler::Fail;
+    case 106: RegID = getMCReg(VCC_LO, STI); return MCDisassembler::Success;
+    case 107: RegID = getMCReg(VCC_HI, STI); return MCDisassembler::Success;
+    // ToDo: no support for tba_lo/_hi register
+    case 108:
+    case 109: return MCDisassembler::Fail;
+    // ToDo: no support for tma_lo/_hi register
+    case 110:
+    case 111: return MCDisassembler::Fail;
+    // ToDo: no support for ttmp[0:11] register
+    case 112:
+    case 113:
+    case 114:
+    case 115:
+    case 116:
+    case 117:
+    case 118:
+    case 119:
+    case 120:
+    case 121:
+    case 122:
+    case 123: return MCDisassembler::Fail;
+    case 124: RegID = getMCReg(M0, STI); return MCDisassembler::Success;
+    case 126: RegID = getMCReg(EXEC_LO, STI); return MCDisassembler::Success;
+    case 127: RegID = getMCReg(EXEC_HI, STI); return MCDisassembler::Success;
+    // ToDo: no support for vccz register
+    case 251: return MCDisassembler::Fail;
+    // ToDo: no support for execz register
+    case 252: return MCDisassembler::Fail;
+    case 253: RegID = getMCReg(SCC, STI); return MCDisassembler::Success;
+    default: return MCDisassembler::Fail;
+    }
+  }
+  return MCDisassembler::Fail;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeVGPR_32RegisterClass(llvm::MCInst &Inst, 
+                                                            unsigned Imm, 
+                                                            uint64_t Addr) const {
+  unsigned RegID;
+  if (DecodeVgprRegister(Imm, RegID) == MCDisassembler::Success) {
+    Inst.addOperand(MCOperand::createReg(RegID));
+    return MCDisassembler::Success;
+  }
+  return MCDisassembler::Fail;
+}
+
+DecodeStatus AMDGPUDisassembler::DecodeVS_32RegisterClass(MCInst &Inst, 
+                                                          unsigned Imm, 
+                                                          uint64_t Addr) const {
+  // ToDo: different opcodes allow different formats og this operands
+  if ((Imm >= 128) && (Imm <= 208)) {
+    // immediate integer
+    int64_t Val;
+    if (DecodeLitInteger(Imm, Val) == MCDisassembler::Success) {
+      Inst.addOperand(MCOperand::createImm(Val));
+      return MCDisassembler::Success;
+    }
+  } else if ((Imm >= 240) && (Imm <= 248)) {
+    // immediate float
+    uint32_t Val;
+    if (DecodeLitFloat(Imm, Val) == MCDisassembler::Success) {
+      Inst.addOperand(MCOperand::createImm(Val));
+      return MCDisassembler::Success;
+    }
+  } else if (Imm == 254) {
+    // LDS direct
+    // ToDo: implement LDS direct read
+  } else if (Imm == 255) {
+    // literal constant
+  } else if ((Imm == 125) || 
+             ((Imm >= 209) && (Imm <= 239)) || 
+             (Imm == 249) || 
+             (Imm == 250) || 
+             (Imm >= 512)) {
+    // reserved
+    return MCDisassembler::Fail;
+  } else {
+    // register
+    unsigned RegID;
+    if (DecodeSrcRegister(Imm, RegID) == MCDisassembler::Success) {
+      Inst.addOperand(MCOperand::createReg(RegID));
+      return MCDisassembler::Success;
+    }
+  }
+  return MCDisassembler::Fail;
+}
+
+static MCDisassembler *createAMDGPUDisassembler(const Target &T,
+                                                const MCSubtargetInfo &STI,
+                                                MCContext &Ctx) {
+  return new AMDGPUDisassembler(STI, Ctx);
+}
+
+extern "C" void LLVMInitializeAMDGPUDisassembler() {
+  TargetRegistry::RegisterMCDisassembler(TheGCNTarget, createAMDGPUDisassembler);
+}