1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
|
//===- llvm-ir2vec.cpp - IR2Vec Embedding Generation Tool -----------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
///
/// \file
/// This file implements the IR2Vec embedding generation tool.
///
/// This tool provides three main modes:
///
/// 1. Triplet Generation Mode (--mode=triplets):
/// Generates numeric triplets (head, tail, relation) for vocabulary
/// training. Output format: MAX_RELATION=N header followed by
/// head\ttail\trelation lines. Relations: 0=Type, 1=Next, 2+=Arg0,Arg1,...
/// Usage: llvm-ir2vec --mode=triplets input.bc -o train2id.txt
///
/// 2. Entities Generation Mode (--mode=entities):
/// Generates entity mappings for vocabulary training.
/// Output format: <total_entities> header followed by entity\tid lines.
/// Usage: llvm-ir2vec --mode=entities input.bc -o entity2id.txt
///
/// 3. Embedding Generation Mode (--mode=embeddings):
/// Generates IR2Vec embeddings using a trained vocabulary.
/// Usage: llvm-ir2vec --mode=embeddings --ir2vec-vocab-path=vocab.json
/// --level=func input.bc -o embeddings.txt Levels: --level=inst
/// (instructions), --level=bb (basic blocks), --level=func (functions)
/// (See IR2Vec.cpp for more embedding generation options)
///
//===----------------------------------------------------------------------===//
#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassInstrumentation.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/Type.h"
#include "llvm/IRReader/IRReader.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/InitLLVM.h"
#include "llvm/Support/SourceMgr.h"
#include "llvm/Support/raw_ostream.h"
#define DEBUG_TYPE "ir2vec"
namespace llvm {
namespace ir2vec {
static cl::OptionCategory IR2VecToolCategory("IR2Vec Tool Options");
static cl::opt<std::string>
InputFilename(cl::Positional,
cl::desc("<input bitcode file or '-' for stdin>"),
cl::init("-"), cl::cat(IR2VecToolCategory));
static cl::opt<std::string> OutputFilename("o", cl::desc("Output filename"),
cl::value_desc("filename"),
cl::init("-"),
cl::cat(IR2VecToolCategory));
enum ToolMode {
TripletMode, // Generate triplets for vocabulary training
EntityMode, // Generate entity mappings for vocabulary training
EmbeddingMode // Generate embeddings using trained vocabulary
};
static cl::opt<ToolMode> Mode(
"mode", cl::desc("Tool operation mode:"),
cl::values(clEnumValN(TripletMode, "triplets",
"Generate triplets for vocabulary training"),
clEnumValN(EntityMode, "entities",
"Generate entity mappings for vocabulary training"),
clEnumValN(EmbeddingMode, "embeddings",
"Generate embeddings using trained vocabulary")),
cl::init(EmbeddingMode), cl::cat(IR2VecToolCategory));
static cl::opt<std::string>
FunctionName("function", cl::desc("Process specific function only"),
cl::value_desc("name"), cl::Optional, cl::init(""),
cl::cat(IR2VecToolCategory));
enum EmbeddingLevel {
InstructionLevel, // Generate instruction-level embeddings
BasicBlockLevel, // Generate basic block-level embeddings
FunctionLevel // Generate function-level embeddings
};
static cl::opt<EmbeddingLevel>
Level("level", cl::desc("Embedding generation level (for embedding mode):"),
cl::values(clEnumValN(InstructionLevel, "inst",
"Generate instruction-level embeddings"),
clEnumValN(BasicBlockLevel, "bb",
"Generate basic block-level embeddings"),
clEnumValN(FunctionLevel, "func",
"Generate function-level embeddings")),
cl::init(FunctionLevel), cl::cat(IR2VecToolCategory));
namespace {
/// Relation types for triplet generation
enum RelationType {
TypeRelation = 0, ///< Instruction to type relationship
NextRelation = 1, ///< Sequential instruction relationship
ArgRelation = 2 ///< Instruction to operand relationship (ArgRelation + N)
};
/// Helper class for collecting IR triplets and generating embeddings
class IR2VecTool {
private:
Module &M;
ModuleAnalysisManager MAM;
const Vocabulary *Vocab = nullptr;
public:
explicit IR2VecTool(Module &M) : M(M) {}
/// Initialize the IR2Vec vocabulary analysis
bool initializeVocabulary() {
// Register and run the IR2Vec vocabulary analysis
// The vocabulary file path is specified via --ir2vec-vocab-path global
// option
MAM.registerPass([&] { return PassInstrumentationAnalysis(); });
MAM.registerPass([&] { return IR2VecVocabAnalysis(); });
// This will throw an error if vocab is not found or invalid
Vocab = &MAM.getResult<IR2VecVocabAnalysis>(M);
return Vocab->isValid();
}
/// Generate triplets for the module
/// Output format: MAX_RELATION=N header followed by relationships
void generateTriplets(raw_ostream &OS) const {
unsigned MaxRelation = NextRelation; // Track maximum relation ID
std::string Relationships;
raw_string_ostream RelOS(Relationships);
for (const Function &F : M) {
unsigned FuncMaxRelation = generateTriplets(F, RelOS);
MaxRelation = std::max(MaxRelation, FuncMaxRelation);
}
RelOS.flush();
// Write metadata header followed by relationships
OS << "MAX_RELATION=" << MaxRelation << '\n';
OS << Relationships;
}
/// Generate triplets for a single function
/// Returns the maximum relation ID used in this function
unsigned generateTriplets(const Function &F, raw_ostream &OS) const {
if (F.isDeclaration())
return 0;
unsigned MaxRelation = 1;
unsigned PrevOpcode = 0;
bool HasPrevOpcode = false;
for (const BasicBlock &BB : F) {
for (const auto &I : BB.instructionsWithoutDebug()) {
unsigned Opcode = Vocabulary::getNumericID(I.getOpcode());
unsigned TypeID = Vocabulary::getNumericID(I.getType()->getTypeID());
// Add "Next" relationship with previous instruction
if (HasPrevOpcode) {
OS << PrevOpcode << '\t' << Opcode << '\t' << NextRelation << '\n';
LLVM_DEBUG(dbgs()
<< Vocabulary::getVocabKeyForOpcode(PrevOpcode + 1) << '\t'
<< Vocabulary::getVocabKeyForOpcode(Opcode + 1) << '\t'
<< "Next\n");
}
// Add "Type" relationship
OS << Opcode << '\t' << TypeID << '\t' << TypeRelation << '\n';
LLVM_DEBUG(
dbgs() << Vocabulary::getVocabKeyForOpcode(Opcode + 1) << '\t'
<< Vocabulary::getVocabKeyForTypeID(I.getType()->getTypeID())
<< '\t' << "Type\n");
// Add "Arg" relationships
unsigned ArgIndex = 0;
for (const Use &U : I.operands()) {
unsigned OperandID = Vocabulary::getNumericID(U.get());
unsigned RelationID = ArgRelation + ArgIndex;
OS << Opcode << '\t' << OperandID << '\t' << RelationID << '\n';
LLVM_DEBUG({
StringRef OperandStr = Vocabulary::getVocabKeyForOperandKind(
Vocabulary::getOperandKind(U.get()));
dbgs() << Vocabulary::getVocabKeyForOpcode(Opcode + 1) << '\t'
<< OperandStr << '\t' << "Arg" << ArgIndex << '\n';
});
++ArgIndex;
}
// Only update MaxRelation if there were operands
if (ArgIndex > 0) {
MaxRelation = std::max(MaxRelation, ArgRelation + ArgIndex - 1);
}
PrevOpcode = Opcode;
HasPrevOpcode = true;
}
}
return MaxRelation;
}
/// Dump entity ID to string mappings
static void generateEntityMappings(raw_ostream &OS) {
// FIXME: Currently, the generated entity mappings are not one-to-one;
// Multiple TypeIDs map to same string key (Like Half, BFloat, etc. map to
// FloatTy). This would hinder learning good seed embeddings.
// We should fix this in the future by ensuring unique string keys either by
// post-processing here without changing the mapping in ir2vec::Vocabulary,
// or by changing the Vocabulary generation logic to ensure unique keys.
auto EntityLen = Vocabulary::expectedSize();
OS << EntityLen << "\n";
for (unsigned EntityID = 0; EntityID < EntityLen; ++EntityID)
OS << Vocabulary::getStringKey(EntityID) << '\t' << EntityID << '\n';
}
/// Generate embeddings for the entire module
void generateEmbeddings(raw_ostream &OS) const {
if (!Vocab->isValid()) {
OS << "Error: Vocabulary is not valid. IR2VecTool not initialized.\n";
return;
}
for (const Function &F : M)
generateEmbeddings(F, OS);
}
/// Generate embeddings for a single function
void generateEmbeddings(const Function &F, raw_ostream &OS) const {
if (F.isDeclaration()) {
OS << "Function " << F.getName() << " is a declaration, skipping.\n";
return;
}
// Create embedder for this function
assert(Vocab->isValid() && "Vocabulary is not valid");
auto Emb = Embedder::create(IR2VecKind::Symbolic, F, *Vocab);
if (!Emb) {
OS << "Error: Failed to create embedder for function " << F.getName()
<< "\n";
return;
}
OS << "Function: " << F.getName() << "\n";
// Generate embeddings based on the specified level
switch (Level) {
case FunctionLevel: {
Emb->getFunctionVector().print(OS);
break;
}
case BasicBlockLevel: {
const auto &BBVecMap = Emb->getBBVecMap();
for (const BasicBlock &BB : F) {
auto It = BBVecMap.find(&BB);
if (It != BBVecMap.end()) {
OS << BB.getName() << ":";
It->second.print(OS);
}
}
break;
}
case InstructionLevel: {
const auto &InstMap = Emb->getInstVecMap();
for (const BasicBlock &BB : F) {
for (const Instruction &I : BB) {
auto It = InstMap.find(&I);
if (It != InstMap.end()) {
I.print(OS);
It->second.print(OS);
}
}
}
break;
}
}
}
};
Error processModule(Module &M, raw_ostream &OS) {
IR2VecTool Tool(M);
if (Mode == EmbeddingMode) {
// Initialize vocabulary for embedding generation
// Note: Requires --ir2vec-vocab-path option to be set
auto VocabStatus = Tool.initializeVocabulary();
assert(VocabStatus && "Failed to initialize IR2Vec vocabulary");
(void)VocabStatus;
if (!FunctionName.empty()) {
// Process single function
if (const Function *F = M.getFunction(FunctionName))
Tool.generateEmbeddings(*F, OS);
else
return createStringError(errc::invalid_argument,
"Function '%s' not found",
FunctionName.c_str());
} else {
// Process all functions
Tool.generateEmbeddings(OS);
}
} else {
Tool.generateTriplets(OS);
}
return Error::success();
}
} // namespace
} // namespace ir2vec
} // namespace llvm
int main(int argc, char **argv) {
using namespace llvm;
using namespace llvm::ir2vec;
InitLLVM X(argc, argv);
cl::HideUnrelatedOptions(IR2VecToolCategory);
cl::ParseCommandLineOptions(
argc, argv,
"IR2Vec - Embedding Generation Tool\n"
"Generates embeddings for a given LLVM IR and "
"supports triplet generation for vocabulary "
"training and embedding generation.\n\n"
"See https://llvm.org/docs/CommandGuide/llvm-ir2vec.html for more "
"information.\n");
// Validate command line options
if (Mode != EmbeddingMode) {
if (Level.getNumOccurrences() > 0)
errs() << "Warning: --level option is ignored\n";
if (FunctionName.getNumOccurrences() > 0)
errs() << "Warning: --function option is ignored\n";
}
std::error_code EC;
raw_fd_ostream OS(OutputFilename, EC);
if (EC) {
errs() << "Error opening output file: " << EC.message() << "\n";
return 1;
}
if (Mode == EntityMode) {
// Just dump entity mappings without processing any IR
IR2VecTool::generateEntityMappings(OS);
return 0;
}
// Parse the input LLVM IR file or stdin
SMDiagnostic Err;
LLVMContext Context;
std::unique_ptr<Module> M = parseIRFile(InputFilename, Err, Context);
if (!M) {
Err.print(argv[0], errs());
return 1;
}
if (Error Err = processModule(*M, OS)) {
handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EIB) {
errs() << "Error: " << EIB.message() << "\n";
});
return 1;
}
return 0;
}
|