1 //===-- IRForTarget.cpp ---------------------------------------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8
9 #include "IRForTarget.h"
10
11 #include "ClangExpressionDeclMap.h"
12 #include "ClangUtil.h"
13
14 #include "Plugins/TypeSystem/Clang/TypeSystemClang.h"
15 #include "llvm/IR/Constants.h"
16 #include "llvm/IR/DataLayout.h"
17 #include "llvm/IR/Operator.h"
18 #include "llvm/IR/InstrTypes.h"
19 #include "llvm/IR/Instructions.h"
20 #include "llvm/IR/Intrinsics.h"
21 #include "llvm/IR/LegacyPassManager.h"
22 #include "llvm/IR/Metadata.h"
23 #include "llvm/IR/Module.h"
24 #include "llvm/IR/ValueSymbolTable.h"
25 #include "llvm/Support/raw_ostream.h"
26 #include "llvm/Transforms/IPO.h"
27
28 #include "clang/AST/ASTContext.h"
29
30 #include "lldb/Core/dwarf.h"
31 #include "lldb/Expression/IRExecutionUnit.h"
32 #include "lldb/Expression/IRInterpreter.h"
33 #include "lldb/Symbol/CompilerType.h"
34 #include "lldb/Utility/ConstString.h"
35 #include "lldb/Utility/DataBufferHeap.h"
36 #include "lldb/Utility/Endian.h"
37 #include "lldb/Utility/LLDBLog.h"
38 #include "lldb/Utility/Log.h"
39 #include "lldb/Utility/Scalar.h"
40 #include "lldb/Utility/StreamString.h"
41
42 #include <map>
43 #include <optional>
44
45 using namespace llvm;
46 using lldb_private::LLDBLog;
47
48 typedef SmallVector<Instruction *, 2> InstrList;
49
FunctionValueCache(Maker const & maker)50 IRForTarget::FunctionValueCache::FunctionValueCache(Maker const &maker)
51 : m_maker(maker), m_values() {}
52
53 IRForTarget::FunctionValueCache::~FunctionValueCache() = default;
54
55 llvm::Value *
GetValue(llvm::Function * function)56 IRForTarget::FunctionValueCache::GetValue(llvm::Function *function) {
57 if (!m_values.count(function)) {
58 llvm::Value *ret = m_maker(function);
59 m_values[function] = ret;
60 return ret;
61 }
62 return m_values[function];
63 }
64
FindEntryInstruction(llvm::Function * function)65 static llvm::Value *FindEntryInstruction(llvm::Function *function) {
66 if (function->empty())
67 return nullptr;
68
69 return function->getEntryBlock().getFirstNonPHIOrDbg();
70 }
71
IRForTarget(lldb_private::ClangExpressionDeclMap * decl_map,bool resolve_vars,lldb_private::IRExecutionUnit & execution_unit,lldb_private::Stream & error_stream,const char * func_name)72 IRForTarget::IRForTarget(lldb_private::ClangExpressionDeclMap *decl_map,
73 bool resolve_vars,
74 lldb_private::IRExecutionUnit &execution_unit,
75 lldb_private::Stream &error_stream,
76 const char *func_name)
77 : m_resolve_vars(resolve_vars), m_func_name(func_name),
78 m_decl_map(decl_map), m_error_stream(error_stream),
79 m_execution_unit(execution_unit),
80 m_entry_instruction_finder(FindEntryInstruction) {}
81
82 /* Handy utility functions used at several places in the code */
83
PrintValue(const Value * value,bool truncate=false)84 static std::string PrintValue(const Value *value, bool truncate = false) {
85 std::string s;
86 if (value) {
87 raw_string_ostream rso(s);
88 value->print(rso);
89 rso.flush();
90 if (truncate)
91 s.resize(s.length() - 1);
92 }
93 return s;
94 }
95
PrintType(const llvm::Type * type,bool truncate=false)96 static std::string PrintType(const llvm::Type *type, bool truncate = false) {
97 std::string s;
98 raw_string_ostream rso(s);
99 type->print(rso);
100 rso.flush();
101 if (truncate)
102 s.resize(s.length() - 1);
103 return s;
104 }
105
FixFunctionLinkage(llvm::Function & llvm_function)106 bool IRForTarget::FixFunctionLinkage(llvm::Function &llvm_function) {
107 llvm_function.setLinkage(GlobalValue::ExternalLinkage);
108
109 return true;
110 }
111
DeclForGlobal(const GlobalValue * global_val,Module * module)112 clang::NamedDecl *IRForTarget::DeclForGlobal(const GlobalValue *global_val,
113 Module *module) {
114 NamedMDNode *named_metadata =
115 module->getNamedMetadata("clang.global.decl.ptrs");
116
117 if (!named_metadata)
118 return nullptr;
119
120 unsigned num_nodes = named_metadata->getNumOperands();
121 unsigned node_index;
122
123 for (node_index = 0; node_index < num_nodes; ++node_index) {
124 llvm::MDNode *metadata_node =
125 dyn_cast<llvm::MDNode>(named_metadata->getOperand(node_index));
126 if (!metadata_node)
127 return nullptr;
128
129 if (metadata_node->getNumOperands() != 2)
130 continue;
131
132 if (mdconst::dyn_extract_or_null<GlobalValue>(
133 metadata_node->getOperand(0)) != global_val)
134 continue;
135
136 ConstantInt *constant_int =
137 mdconst::dyn_extract<ConstantInt>(metadata_node->getOperand(1));
138
139 if (!constant_int)
140 return nullptr;
141
142 uintptr_t ptr = constant_int->getZExtValue();
143
144 return reinterpret_cast<clang::NamedDecl *>(ptr);
145 }
146
147 return nullptr;
148 }
149
DeclForGlobal(GlobalValue * global_val)150 clang::NamedDecl *IRForTarget::DeclForGlobal(GlobalValue *global_val) {
151 return DeclForGlobal(global_val, m_module);
152 }
153
154 /// Returns true iff the mangled symbol is for a static guard variable.
isGuardVariableSymbol(llvm::StringRef mangled_symbol,bool check_ms_abi=true)155 static bool isGuardVariableSymbol(llvm::StringRef mangled_symbol,
156 bool check_ms_abi = true) {
157 bool result = mangled_symbol.startswith("_ZGV"); // Itanium ABI guard variable
158 if (check_ms_abi)
159 result |= mangled_symbol.endswith("@4IA"); // Microsoft ABI
160 return result;
161 }
162
CreateResultVariable(llvm::Function & llvm_function)163 bool IRForTarget::CreateResultVariable(llvm::Function &llvm_function) {
164 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
165
166 if (!m_resolve_vars)
167 return true;
168
169 // Find the result variable. If it doesn't exist, we can give up right here.
170
171 ValueSymbolTable &value_symbol_table = m_module->getValueSymbolTable();
172
173 llvm::StringRef result_name;
174 bool found_result = false;
175
176 for (StringMapEntry<llvm::Value *> &value_symbol : value_symbol_table) {
177 result_name = value_symbol.first();
178
179 // Check if this is a guard variable. It seems this causes some hiccups
180 // on Windows, so let's only check for Itanium guard variables.
181 bool is_guard_var = isGuardVariableSymbol(result_name, /*MS ABI*/ false);
182
183 if (result_name.contains("$__lldb_expr_result_ptr") && !is_guard_var) {
184 found_result = true;
185 m_result_is_pointer = true;
186 break;
187 }
188
189 if (result_name.contains("$__lldb_expr_result") && !is_guard_var) {
190 found_result = true;
191 m_result_is_pointer = false;
192 break;
193 }
194 }
195
196 if (!found_result) {
197 LLDB_LOG(log, "Couldn't find result variable");
198
199 return true;
200 }
201
202 LLDB_LOG(log, "Result name: \"{0}\"", result_name);
203
204 Value *result_value = m_module->getNamedValue(result_name);
205
206 if (!result_value) {
207 LLDB_LOG(log, "Result variable had no data");
208
209 m_error_stream.Format("Internal error [IRForTarget]: Result variable's "
210 "name ({0}) exists, but not its definition\n",
211 result_name);
212
213 return false;
214 }
215
216 LLDB_LOG(log, "Found result in the IR: \"{0}\"",
217 PrintValue(result_value, false));
218
219 GlobalVariable *result_global = dyn_cast<GlobalVariable>(result_value);
220
221 if (!result_global) {
222 LLDB_LOG(log, "Result variable isn't a GlobalVariable");
223
224 m_error_stream.Format("Internal error [IRForTarget]: Result variable ({0}) "
225 "is defined, but is not a global variable\n",
226 result_name);
227
228 return false;
229 }
230
231 clang::NamedDecl *result_decl = DeclForGlobal(result_global);
232 if (!result_decl) {
233 LLDB_LOG(log, "Result variable doesn't have a corresponding Decl");
234
235 m_error_stream.Format("Internal error [IRForTarget]: Result variable ({0}) "
236 "does not have a corresponding Clang entity\n",
237 result_name);
238
239 return false;
240 }
241
242 if (log) {
243 std::string decl_desc_str;
244 raw_string_ostream decl_desc_stream(decl_desc_str);
245 result_decl->print(decl_desc_stream);
246 decl_desc_stream.flush();
247
248 LLDB_LOG(log, "Found result decl: \"{0}\"", decl_desc_str);
249 }
250
251 clang::VarDecl *result_var = dyn_cast<clang::VarDecl>(result_decl);
252 if (!result_var) {
253 LLDB_LOG(log, "Result variable Decl isn't a VarDecl");
254
255 m_error_stream.Format("Internal error [IRForTarget]: Result variable "
256 "({0})'s corresponding Clang entity isn't a "
257 "variable\n",
258 result_name);
259
260 return false;
261 }
262
263 // Get the next available result name from m_decl_map and create the
264 // persistent variable for it
265
266 // If the result is an Lvalue, it is emitted as a pointer; see
267 // ASTResultSynthesizer::SynthesizeBodyResult.
268 if (m_result_is_pointer) {
269 clang::QualType pointer_qual_type = result_var->getType();
270 const clang::Type *pointer_type = pointer_qual_type.getTypePtr();
271
272 const clang::PointerType *pointer_pointertype =
273 pointer_type->getAs<clang::PointerType>();
274 const clang::ObjCObjectPointerType *pointer_objcobjpointertype =
275 pointer_type->getAs<clang::ObjCObjectPointerType>();
276
277 if (pointer_pointertype) {
278 clang::QualType element_qual_type = pointer_pointertype->getPointeeType();
279
280 m_result_type = lldb_private::TypeFromParser(
281 m_decl_map->GetTypeSystem()->GetType(element_qual_type));
282 } else if (pointer_objcobjpointertype) {
283 clang::QualType element_qual_type =
284 clang::QualType(pointer_objcobjpointertype->getObjectType(), 0);
285
286 m_result_type = lldb_private::TypeFromParser(
287 m_decl_map->GetTypeSystem()->GetType(element_qual_type));
288 } else {
289 LLDB_LOG(log, "Expected result to have pointer type, but it did not");
290
291 m_error_stream.Format("Internal error [IRForTarget]: Lvalue result ({0}) "
292 "is not a pointer variable\n",
293 result_name);
294
295 return false;
296 }
297 } else {
298 m_result_type = lldb_private::TypeFromParser(
299 m_decl_map->GetTypeSystem()->GetType(result_var->getType()));
300 }
301
302 lldb::TargetSP target_sp(m_execution_unit.GetTarget());
303 std::optional<uint64_t> bit_size = m_result_type.GetBitSize(target_sp.get());
304 if (!bit_size) {
305 lldb_private::StreamString type_desc_stream;
306 m_result_type.DumpTypeDescription(&type_desc_stream);
307
308 LLDB_LOG(log, "Result type has unknown size");
309
310 m_error_stream.Printf("Error [IRForTarget]: Size of result type '%s' "
311 "couldn't be determined\n",
312 type_desc_stream.GetData());
313 return false;
314 }
315
316 if (log) {
317 lldb_private::StreamString type_desc_stream;
318 m_result_type.DumpTypeDescription(&type_desc_stream);
319
320 LLDB_LOG(log, "Result decl type: \"{0}\"", type_desc_stream.GetData());
321 }
322
323 m_result_name = lldb_private::ConstString("$RESULT_NAME");
324
325 LLDB_LOG(log, "Creating a new result global: \"{0}\" with size {1}",
326 m_result_name,
327 m_result_type.GetByteSize(target_sp.get()).value_or(0));
328
329 // Construct a new result global and set up its metadata
330
331 GlobalVariable *new_result_global = new GlobalVariable(
332 (*m_module), result_global->getValueType(), false, /* not constant */
333 GlobalValue::ExternalLinkage, nullptr, /* no initializer */
334 m_result_name.GetCString());
335
336 // It's too late in compilation to create a new VarDecl for this, but we
337 // don't need to. We point the metadata at the old VarDecl. This creates an
338 // odd anomaly: a variable with a Value whose name is something like $0 and a
339 // Decl whose name is $__lldb_expr_result. This condition is handled in
340 // ClangExpressionDeclMap::DoMaterialize, and the name of the variable is
341 // fixed up.
342
343 ConstantInt *new_constant_int =
344 ConstantInt::get(llvm::Type::getInt64Ty(m_module->getContext()),
345 reinterpret_cast<uintptr_t>(result_decl), false);
346
347 llvm::Metadata *values[2];
348 values[0] = ConstantAsMetadata::get(new_result_global);
349 values[1] = ConstantAsMetadata::get(new_constant_int);
350
351 ArrayRef<Metadata *> value_ref(values, 2);
352
353 MDNode *persistent_global_md = MDNode::get(m_module->getContext(), value_ref);
354 NamedMDNode *named_metadata =
355 m_module->getNamedMetadata("clang.global.decl.ptrs");
356 named_metadata->addOperand(persistent_global_md);
357
358 LLDB_LOG(log, "Replacing \"{0}\" with \"{1}\"", PrintValue(result_global),
359 PrintValue(new_result_global));
360
361 if (result_global->use_empty()) {
362 // We need to synthesize a store for this variable, because otherwise
363 // there's nothing to put into its equivalent persistent variable.
364
365 BasicBlock &entry_block(llvm_function.getEntryBlock());
366 Instruction *first_entry_instruction(entry_block.getFirstNonPHIOrDbg());
367
368 if (!first_entry_instruction)
369 return false;
370
371 if (!result_global->hasInitializer()) {
372 LLDB_LOG(log, "Couldn't find initializer for unused variable");
373
374 m_error_stream.Format("Internal error [IRForTarget]: Result variable "
375 "({0}) has no writes and no initializer\n",
376 result_name);
377
378 return false;
379 }
380
381 Constant *initializer = result_global->getInitializer();
382
383 StoreInst *synthesized_store =
384 new StoreInst(initializer, new_result_global, first_entry_instruction);
385
386 LLDB_LOG(log, "Synthesized result store \"{0}\"\n",
387 PrintValue(synthesized_store));
388 } else {
389 result_global->replaceAllUsesWith(new_result_global);
390 }
391
392 if (!m_decl_map->AddPersistentVariable(
393 result_decl, m_result_name, m_result_type, true, m_result_is_pointer))
394 return false;
395
396 result_global->eraseFromParent();
397
398 return true;
399 }
400
RewriteObjCConstString(llvm::GlobalVariable * ns_str,llvm::GlobalVariable * cstr)401 bool IRForTarget::RewriteObjCConstString(llvm::GlobalVariable *ns_str,
402 llvm::GlobalVariable *cstr) {
403 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
404
405 Type *ns_str_ty = ns_str->getType();
406
407 Type *i8_ptr_ty = Type::getInt8PtrTy(m_module->getContext());
408 Type *i32_ty = Type::getInt32Ty(m_module->getContext());
409 Type *i8_ty = Type::getInt8Ty(m_module->getContext());
410
411 if (!m_CFStringCreateWithBytes) {
412 lldb::addr_t CFStringCreateWithBytes_addr;
413
414 static lldb_private::ConstString g_CFStringCreateWithBytes_str(
415 "CFStringCreateWithBytes");
416
417 bool missing_weak = false;
418 CFStringCreateWithBytes_addr =
419 m_execution_unit.FindSymbol(g_CFStringCreateWithBytes_str,
420 missing_weak);
421 if (CFStringCreateWithBytes_addr == LLDB_INVALID_ADDRESS || missing_weak) {
422 LLDB_LOG(log, "Couldn't find CFStringCreateWithBytes in the target");
423
424 m_error_stream.Printf("Error [IRForTarget]: Rewriting an Objective-C "
425 "constant string requires "
426 "CFStringCreateWithBytes\n");
427
428 return false;
429 }
430
431 LLDB_LOG(log, "Found CFStringCreateWithBytes at {0}",
432 CFStringCreateWithBytes_addr);
433
434 // Build the function type:
435 //
436 // CFStringRef CFStringCreateWithBytes (
437 // CFAllocatorRef alloc,
438 // const UInt8 *bytes,
439 // CFIndex numBytes,
440 // CFStringEncoding encoding,
441 // Boolean isExternalRepresentation
442 // );
443 //
444 // We make the following substitutions:
445 //
446 // CFStringRef -> i8*
447 // CFAllocatorRef -> i8*
448 // UInt8 * -> i8*
449 // CFIndex -> long (i32 or i64, as appropriate; we ask the module for its
450 // pointer size for now) CFStringEncoding -> i32 Boolean -> i8
451
452 Type *arg_type_array[5];
453
454 arg_type_array[0] = i8_ptr_ty;
455 arg_type_array[1] = i8_ptr_ty;
456 arg_type_array[2] = m_intptr_ty;
457 arg_type_array[3] = i32_ty;
458 arg_type_array[4] = i8_ty;
459
460 ArrayRef<Type *> CFSCWB_arg_types(arg_type_array, 5);
461
462 llvm::FunctionType *CFSCWB_ty =
463 FunctionType::get(ns_str_ty, CFSCWB_arg_types, false);
464
465 // Build the constant containing the pointer to the function
466 PointerType *CFSCWB_ptr_ty = PointerType::getUnqual(CFSCWB_ty);
467 Constant *CFSCWB_addr_int =
468 ConstantInt::get(m_intptr_ty, CFStringCreateWithBytes_addr, false);
469 m_CFStringCreateWithBytes = {
470 CFSCWB_ty, ConstantExpr::getIntToPtr(CFSCWB_addr_int, CFSCWB_ptr_ty)};
471 }
472
473 ConstantDataSequential *string_array = nullptr;
474
475 if (cstr)
476 string_array = dyn_cast<ConstantDataSequential>(cstr->getInitializer());
477
478 Constant *alloc_arg = Constant::getNullValue(i8_ptr_ty);
479 Constant *bytes_arg = cstr ? ConstantExpr::getBitCast(cstr, i8_ptr_ty)
480 : Constant::getNullValue(i8_ptr_ty);
481 Constant *numBytes_arg = ConstantInt::get(
482 m_intptr_ty, cstr ? (string_array->getNumElements() - 1) * string_array->getElementByteSize() : 0, false);
483 int encoding_flags = 0;
484 switch (cstr ? string_array->getElementByteSize() : 1) {
485 case 1:
486 encoding_flags = 0x08000100; /* 0x08000100 is kCFStringEncodingUTF8 */
487 break;
488 case 2:
489 encoding_flags = 0x0100; /* 0x0100 is kCFStringEncodingUTF16 */
490 break;
491 case 4:
492 encoding_flags = 0x0c000100; /* 0x0c000100 is kCFStringEncodingUTF32 */
493 break;
494 default:
495 encoding_flags = 0x0600; /* fall back to 0x0600, kCFStringEncodingASCII */
496 LLDB_LOG(log, "Encountered an Objective-C constant string with unusual "
497 "element size {0}",
498 string_array->getElementByteSize());
499 }
500 Constant *encoding_arg = ConstantInt::get(i32_ty, encoding_flags, false);
501 Constant *isExternal_arg =
502 ConstantInt::get(i8_ty, 0x0, false); /* 0x0 is false */
503
504 Value *argument_array[5];
505
506 argument_array[0] = alloc_arg;
507 argument_array[1] = bytes_arg;
508 argument_array[2] = numBytes_arg;
509 argument_array[3] = encoding_arg;
510 argument_array[4] = isExternal_arg;
511
512 ArrayRef<Value *> CFSCWB_arguments(argument_array, 5);
513
514 FunctionValueCache CFSCWB_Caller(
515 [this, &CFSCWB_arguments](llvm::Function *function) -> llvm::Value * {
516 return CallInst::Create(
517 m_CFStringCreateWithBytes, CFSCWB_arguments,
518 "CFStringCreateWithBytes",
519 llvm::cast<Instruction>(
520 m_entry_instruction_finder.GetValue(function)));
521 });
522
523 if (!UnfoldConstant(ns_str, nullptr, CFSCWB_Caller, m_entry_instruction_finder,
524 m_error_stream)) {
525 LLDB_LOG(log, "Couldn't replace the NSString with the result of the call");
526
527 m_error_stream.Printf("error [IRForTarget internal]: Couldn't replace an "
528 "Objective-C constant string with a dynamic "
529 "string\n");
530
531 return false;
532 }
533
534 ns_str->eraseFromParent();
535
536 return true;
537 }
538
RewriteObjCConstStrings()539 bool IRForTarget::RewriteObjCConstStrings() {
540 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
541
542 ValueSymbolTable &value_symbol_table = m_module->getValueSymbolTable();
543
544 for (StringMapEntry<llvm::Value *> &value_symbol : value_symbol_table) {
545 llvm::StringRef value_name = value_symbol.first();
546
547 if (value_name.contains("_unnamed_cfstring_")) {
548 Value *nsstring_value = value_symbol.second;
549
550 GlobalVariable *nsstring_global =
551 dyn_cast<GlobalVariable>(nsstring_value);
552
553 if (!nsstring_global) {
554 LLDB_LOG(log, "NSString variable is not a GlobalVariable");
555
556 m_error_stream.Printf("Internal error [IRForTarget]: An Objective-C "
557 "constant string is not a global variable\n");
558
559 return false;
560 }
561
562 if (!nsstring_global->hasInitializer()) {
563 LLDB_LOG(log, "NSString variable does not have an initializer");
564
565 m_error_stream.Printf("Internal error [IRForTarget]: An Objective-C "
566 "constant string does not have an initializer\n");
567
568 return false;
569 }
570
571 ConstantStruct *nsstring_struct =
572 dyn_cast<ConstantStruct>(nsstring_global->getInitializer());
573
574 if (!nsstring_struct) {
575 LLDB_LOG(log,
576 "NSString variable's initializer is not a ConstantStruct");
577
578 m_error_stream.Printf("Internal error [IRForTarget]: An Objective-C "
579 "constant string is not a structure constant\n");
580
581 return false;
582 }
583
584 // We expect the following structure:
585 //
586 // struct {
587 // int *isa;
588 // int flags;
589 // char *str;
590 // long length;
591 // };
592
593 if (nsstring_struct->getNumOperands() != 4) {
594
595 LLDB_LOG(log,
596 "NSString variable's initializer structure has an "
597 "unexpected number of members. Should be 4, is {0}",
598 nsstring_struct->getNumOperands());
599
600 m_error_stream.Printf("Internal error [IRForTarget]: The struct for an "
601 "Objective-C constant string is not as "
602 "expected\n");
603
604 return false;
605 }
606
607 Constant *nsstring_member = nsstring_struct->getOperand(2);
608
609 if (!nsstring_member) {
610 LLDB_LOG(log, "NSString initializer's str element was empty");
611
612 m_error_stream.Printf("Internal error [IRForTarget]: An Objective-C "
613 "constant string does not have a string "
614 "initializer\n");
615
616 return false;
617 }
618
619 auto *cstr_global = dyn_cast<GlobalVariable>(nsstring_member);
620 if (!cstr_global) {
621 LLDB_LOG(log,
622 "NSString initializer's str element is not a GlobalVariable");
623
624 m_error_stream.Printf("Internal error [IRForTarget]: Unhandled"
625 "constant string initializer\n");
626
627 return false;
628 }
629
630 if (!cstr_global->hasInitializer()) {
631 LLDB_LOG(log, "NSString initializer's str element does not have an "
632 "initializer");
633
634 m_error_stream.Printf("Internal error [IRForTarget]: An Objective-C "
635 "constant string's string initializer doesn't "
636 "point to initialized data\n");
637
638 return false;
639 }
640
641 /*
642 if (!cstr_array)
643 {
644 if (log)
645 log->PutCString("NSString initializer's str element is not a
646 ConstantArray");
647
648 if (m_error_stream)
649 m_error_stream.Printf("Internal error [IRForTarget]: An
650 Objective-C constant string's string initializer doesn't point to an
651 array\n");
652
653 return false;
654 }
655
656 if (!cstr_array->isCString())
657 {
658 if (log)
659 log->PutCString("NSString initializer's str element is not a C
660 string array");
661
662 if (m_error_stream)
663 m_error_stream.Printf("Internal error [IRForTarget]: An
664 Objective-C constant string's string initializer doesn't point to a C
665 string\n");
666
667 return false;
668 }
669 */
670
671 ConstantDataArray *cstr_array =
672 dyn_cast<ConstantDataArray>(cstr_global->getInitializer());
673
674 if (cstr_array)
675 LLDB_LOG(log, "Found NSString constant {0}, which contains \"{1}\"",
676 value_name, cstr_array->getAsString());
677 else
678 LLDB_LOG(log, "Found NSString constant {0}, which contains \"\"",
679 value_name);
680
681 if (!cstr_array)
682 cstr_global = nullptr;
683
684 if (!RewriteObjCConstString(nsstring_global, cstr_global)) {
685 LLDB_LOG(log, "Error rewriting the constant string");
686
687 // We don't print an error message here because RewriteObjCConstString
688 // has done so for us.
689
690 return false;
691 }
692 }
693 }
694
695 for (StringMapEntry<llvm::Value *> &value_symbol : value_symbol_table) {
696 llvm::StringRef value_name = value_symbol.first();
697
698 if (value_name == "__CFConstantStringClassReference") {
699 GlobalVariable *gv = dyn_cast<GlobalVariable>(value_symbol.second);
700
701 if (!gv) {
702 LLDB_LOG(log,
703 "__CFConstantStringClassReference is not a global variable");
704
705 m_error_stream.Printf("Internal error [IRForTarget]: Found a "
706 "CFConstantStringClassReference, but it is not a "
707 "global object\n");
708
709 return false;
710 }
711
712 gv->eraseFromParent();
713
714 break;
715 }
716 }
717
718 return true;
719 }
720
IsObjCSelectorRef(Value * value)721 static bool IsObjCSelectorRef(Value *value) {
722 GlobalVariable *global_variable = dyn_cast<GlobalVariable>(value);
723
724 return !(!global_variable || !global_variable->hasName() ||
725 !global_variable->getName().startswith("OBJC_SELECTOR_REFERENCES_"));
726 }
727
728 // This function does not report errors; its callers are responsible.
RewriteObjCSelector(Instruction * selector_load)729 bool IRForTarget::RewriteObjCSelector(Instruction *selector_load) {
730 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
731
732 LoadInst *load = dyn_cast<LoadInst>(selector_load);
733
734 if (!load)
735 return false;
736
737 // Unpack the message name from the selector. In LLVM IR, an objc_msgSend
738 // gets represented as
739 //
740 // %sel = load ptr, ptr @OBJC_SELECTOR_REFERENCES_, align 8
741 // call i8 @objc_msgSend(ptr %obj, ptr %sel, ...)
742 //
743 // where %obj is the object pointer and %sel is the selector.
744 //
745 // @"OBJC_SELECTOR_REFERENCES_" is a pointer to a character array called
746 // @"\01L_OBJC_llvm_moduleETH_VAR_NAllvm_moduleE_".
747 // @"\01L_OBJC_llvm_moduleETH_VAR_NAllvm_moduleE_" contains the string.
748
749 // Find the pointer's initializer and get the string from its target.
750
751 GlobalVariable *_objc_selector_references_ =
752 dyn_cast<GlobalVariable>(load->getPointerOperand());
753
754 if (!_objc_selector_references_ ||
755 !_objc_selector_references_->hasInitializer())
756 return false;
757
758 Constant *osr_initializer = _objc_selector_references_->getInitializer();
759 if (!osr_initializer)
760 return false;
761
762 // Find the string's initializer (a ConstantArray) and get the string from it
763
764 GlobalVariable *_objc_meth_var_name_ =
765 dyn_cast<GlobalVariable>(osr_initializer);
766
767 if (!_objc_meth_var_name_ || !_objc_meth_var_name_->hasInitializer())
768 return false;
769
770 Constant *omvn_initializer = _objc_meth_var_name_->getInitializer();
771
772 ConstantDataArray *omvn_initializer_array =
773 dyn_cast<ConstantDataArray>(omvn_initializer);
774
775 if (!omvn_initializer_array->isString())
776 return false;
777
778 std::string omvn_initializer_string =
779 std::string(omvn_initializer_array->getAsString());
780
781 LLDB_LOG(log, "Found Objective-C selector reference \"{0}\"",
782 omvn_initializer_string);
783
784 // Construct a call to sel_registerName
785
786 if (!m_sel_registerName) {
787 lldb::addr_t sel_registerName_addr;
788
789 bool missing_weak = false;
790 static lldb_private::ConstString g_sel_registerName_str("sel_registerName");
791 sel_registerName_addr = m_execution_unit.FindSymbol(g_sel_registerName_str,
792 missing_weak);
793 if (sel_registerName_addr == LLDB_INVALID_ADDRESS || missing_weak)
794 return false;
795
796 LLDB_LOG(log, "Found sel_registerName at {0}", sel_registerName_addr);
797
798 // Build the function type: struct objc_selector
799 // *sel_registerName(uint8_t*)
800
801 // The below code would be "more correct," but in actuality what's required
802 // is uint8_t*
803 // Type *sel_type = StructType::get(m_module->getContext());
804 // Type *sel_ptr_type = PointerType::getUnqual(sel_type);
805 Type *sel_ptr_type = Type::getInt8PtrTy(m_module->getContext());
806
807 Type *type_array[1];
808
809 type_array[0] = llvm::Type::getInt8PtrTy(m_module->getContext());
810
811 ArrayRef<Type *> srN_arg_types(type_array, 1);
812
813 llvm::FunctionType *srN_type =
814 FunctionType::get(sel_ptr_type, srN_arg_types, false);
815
816 // Build the constant containing the pointer to the function
817 PointerType *srN_ptr_ty = PointerType::getUnqual(srN_type);
818 Constant *srN_addr_int =
819 ConstantInt::get(m_intptr_ty, sel_registerName_addr, false);
820 m_sel_registerName = {srN_type,
821 ConstantExpr::getIntToPtr(srN_addr_int, srN_ptr_ty)};
822 }
823
824 Value *argument_array[1];
825
826 Constant *omvn_pointer = ConstantExpr::getBitCast(
827 _objc_meth_var_name_, Type::getInt8PtrTy(m_module->getContext()));
828
829 argument_array[0] = omvn_pointer;
830
831 ArrayRef<Value *> srN_arguments(argument_array, 1);
832
833 CallInst *srN_call = CallInst::Create(m_sel_registerName, srN_arguments,
834 "sel_registerName", selector_load);
835
836 // Replace the load with the call in all users
837
838 selector_load->replaceAllUsesWith(srN_call);
839
840 selector_load->eraseFromParent();
841
842 return true;
843 }
844
RewriteObjCSelectors(BasicBlock & basic_block)845 bool IRForTarget::RewriteObjCSelectors(BasicBlock &basic_block) {
846 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
847
848 InstrList selector_loads;
849
850 for (Instruction &inst : basic_block) {
851 if (LoadInst *load = dyn_cast<LoadInst>(&inst))
852 if (IsObjCSelectorRef(load->getPointerOperand()))
853 selector_loads.push_back(&inst);
854 }
855
856 for (Instruction *inst : selector_loads) {
857 if (!RewriteObjCSelector(inst)) {
858 m_error_stream.Printf("Internal error [IRForTarget]: Couldn't change a "
859 "static reference to an Objective-C selector to a "
860 "dynamic reference\n");
861
862 LLDB_LOG(log, "Couldn't rewrite a reference to an Objective-C selector");
863
864 return false;
865 }
866 }
867
868 return true;
869 }
870
IsObjCClassReference(Value * value)871 static bool IsObjCClassReference(Value *value) {
872 GlobalVariable *global_variable = dyn_cast<GlobalVariable>(value);
873
874 return !(!global_variable || !global_variable->hasName() ||
875 !global_variable->getName().startswith("OBJC_CLASS_REFERENCES_"));
876 }
877
878 // This function does not report errors; its callers are responsible.
RewriteObjCClassReference(Instruction * class_load)879 bool IRForTarget::RewriteObjCClassReference(Instruction *class_load) {
880 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
881
882 LoadInst *load = dyn_cast<LoadInst>(class_load);
883
884 if (!load)
885 return false;
886
887 // Unpack the class name from the reference. In LLVM IR, a reference to an
888 // Objective-C class gets represented as
889 //
890 // %tmp = load %struct._objc_class*,
891 // %struct._objc_class** @OBJC_CLASS_REFERENCES_, align 4
892 //
893 // @"OBJC_CLASS_REFERENCES_ is a bitcast of a character array called
894 // @OBJC_CLASS_NAME_. @OBJC_CLASS_NAME contains the string.
895
896 // Find the pointer's initializer (a ConstantExpr with opcode BitCast) and
897 // get the string from its target
898
899 GlobalVariable *_objc_class_references_ =
900 dyn_cast<GlobalVariable>(load->getPointerOperand());
901
902 if (!_objc_class_references_ ||
903 !_objc_class_references_->hasInitializer())
904 return false;
905
906 Constant *ocr_initializer = _objc_class_references_->getInitializer();
907
908 ConstantExpr *ocr_initializer_expr = dyn_cast<ConstantExpr>(ocr_initializer);
909
910 if (!ocr_initializer_expr ||
911 ocr_initializer_expr->getOpcode() != Instruction::BitCast)
912 return false;
913
914 Value *ocr_initializer_base = ocr_initializer_expr->getOperand(0);
915
916 if (!ocr_initializer_base)
917 return false;
918
919 // Find the string's initializer (a ConstantArray) and get the string from it
920
921 GlobalVariable *_objc_class_name_ =
922 dyn_cast<GlobalVariable>(ocr_initializer_base);
923
924 if (!_objc_class_name_ || !_objc_class_name_->hasInitializer())
925 return false;
926
927 Constant *ocn_initializer = _objc_class_name_->getInitializer();
928
929 ConstantDataArray *ocn_initializer_array =
930 dyn_cast<ConstantDataArray>(ocn_initializer);
931
932 if (!ocn_initializer_array->isString())
933 return false;
934
935 std::string ocn_initializer_string =
936 std::string(ocn_initializer_array->getAsString());
937
938 LLDB_LOG(log, "Found Objective-C class reference \"{0}\"",
939 ocn_initializer_string);
940
941 // Construct a call to objc_getClass
942
943 if (!m_objc_getClass) {
944 lldb::addr_t objc_getClass_addr;
945
946 bool missing_weak = false;
947 static lldb_private::ConstString g_objc_getClass_str("objc_getClass");
948 objc_getClass_addr = m_execution_unit.FindSymbol(g_objc_getClass_str,
949 missing_weak);
950 if (objc_getClass_addr == LLDB_INVALID_ADDRESS || missing_weak)
951 return false;
952
953 LLDB_LOG(log, "Found objc_getClass at {0}", objc_getClass_addr);
954
955 // Build the function type: %struct._objc_class *objc_getClass(i8*)
956
957 Type *class_type = load->getType();
958 Type *type_array[1];
959 type_array[0] = llvm::Type::getInt8PtrTy(m_module->getContext());
960
961 ArrayRef<Type *> ogC_arg_types(type_array, 1);
962
963 llvm::FunctionType *ogC_type =
964 FunctionType::get(class_type, ogC_arg_types, false);
965
966 // Build the constant containing the pointer to the function
967 PointerType *ogC_ptr_ty = PointerType::getUnqual(ogC_type);
968 Constant *ogC_addr_int =
969 ConstantInt::get(m_intptr_ty, objc_getClass_addr, false);
970 m_objc_getClass = {ogC_type,
971 ConstantExpr::getIntToPtr(ogC_addr_int, ogC_ptr_ty)};
972 }
973
974 Value *argument_array[1];
975
976 Constant *ocn_pointer = ConstantExpr::getBitCast(
977 _objc_class_name_, Type::getInt8PtrTy(m_module->getContext()));
978
979 argument_array[0] = ocn_pointer;
980
981 ArrayRef<Value *> ogC_arguments(argument_array, 1);
982
983 CallInst *ogC_call = CallInst::Create(m_objc_getClass, ogC_arguments,
984 "objc_getClass", class_load);
985
986 // Replace the load with the call in all users
987
988 class_load->replaceAllUsesWith(ogC_call);
989
990 class_load->eraseFromParent();
991
992 return true;
993 }
994
RewriteObjCClassReferences(BasicBlock & basic_block)995 bool IRForTarget::RewriteObjCClassReferences(BasicBlock &basic_block) {
996 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
997
998 InstrList class_loads;
999
1000 for (Instruction &inst : basic_block) {
1001 if (LoadInst *load = dyn_cast<LoadInst>(&inst))
1002 if (IsObjCClassReference(load->getPointerOperand()))
1003 class_loads.push_back(&inst);
1004 }
1005
1006 for (Instruction *inst : class_loads) {
1007 if (!RewriteObjCClassReference(inst)) {
1008 m_error_stream.Printf("Internal error [IRForTarget]: Couldn't change a "
1009 "static reference to an Objective-C class to a "
1010 "dynamic reference\n");
1011
1012 LLDB_LOG(log, "Couldn't rewrite a reference to an Objective-C class");
1013
1014 return false;
1015 }
1016 }
1017
1018 return true;
1019 }
1020
1021 // This function does not report errors; its callers are responsible.
RewritePersistentAlloc(llvm::Instruction * persistent_alloc)1022 bool IRForTarget::RewritePersistentAlloc(llvm::Instruction *persistent_alloc) {
1023 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1024
1025 AllocaInst *alloc = dyn_cast<AllocaInst>(persistent_alloc);
1026
1027 MDNode *alloc_md = alloc->getMetadata("clang.decl.ptr");
1028
1029 if (!alloc_md || !alloc_md->getNumOperands())
1030 return false;
1031
1032 ConstantInt *constant_int =
1033 mdconst::dyn_extract<ConstantInt>(alloc_md->getOperand(0));
1034
1035 if (!constant_int)
1036 return false;
1037
1038 // We attempt to register this as a new persistent variable with the DeclMap.
1039
1040 uintptr_t ptr = constant_int->getZExtValue();
1041
1042 clang::VarDecl *decl = reinterpret_cast<clang::VarDecl *>(ptr);
1043
1044 lldb_private::TypeFromParser result_decl_type(
1045 m_decl_map->GetTypeSystem()->GetType(decl->getType()));
1046
1047 StringRef decl_name(decl->getName());
1048 lldb_private::ConstString persistent_variable_name(decl_name.data(),
1049 decl_name.size());
1050 if (!m_decl_map->AddPersistentVariable(decl, persistent_variable_name,
1051 result_decl_type, false, false))
1052 return false;
1053
1054 GlobalVariable *persistent_global = new GlobalVariable(
1055 (*m_module), alloc->getType(), false, /* not constant */
1056 GlobalValue::ExternalLinkage, nullptr, /* no initializer */
1057 alloc->getName().str());
1058
1059 // What we're going to do here is make believe this was a regular old
1060 // external variable. That means we need to make the metadata valid.
1061
1062 NamedMDNode *named_metadata =
1063 m_module->getOrInsertNamedMetadata("clang.global.decl.ptrs");
1064
1065 llvm::Metadata *values[2];
1066 values[0] = ConstantAsMetadata::get(persistent_global);
1067 values[1] = ConstantAsMetadata::get(constant_int);
1068
1069 ArrayRef<llvm::Metadata *> value_ref(values, 2);
1070
1071 MDNode *persistent_global_md = MDNode::get(m_module->getContext(), value_ref);
1072 named_metadata->addOperand(persistent_global_md);
1073
1074 // Now, since the variable is a pointer variable, we will drop in a load of
1075 // that pointer variable.
1076
1077 LoadInst *persistent_load = new LoadInst(persistent_global->getValueType(),
1078 persistent_global, "", alloc);
1079
1080 LLDB_LOG(log, "Replacing \"{0}\" with \"{1}\"", PrintValue(alloc),
1081 PrintValue(persistent_load));
1082
1083 alloc->replaceAllUsesWith(persistent_load);
1084 alloc->eraseFromParent();
1085
1086 return true;
1087 }
1088
RewritePersistentAllocs(llvm::BasicBlock & basic_block)1089 bool IRForTarget::RewritePersistentAllocs(llvm::BasicBlock &basic_block) {
1090 if (!m_resolve_vars)
1091 return true;
1092
1093 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1094
1095 InstrList pvar_allocs;
1096
1097 for (Instruction &inst : basic_block) {
1098
1099 if (AllocaInst *alloc = dyn_cast<AllocaInst>(&inst)) {
1100 llvm::StringRef alloc_name = alloc->getName();
1101
1102 if (alloc_name.startswith("$") && !alloc_name.startswith("$__lldb")) {
1103 if (alloc_name.find_first_of("0123456789") == 1) {
1104 LLDB_LOG(log, "Rejecting a numeric persistent variable.");
1105
1106 m_error_stream.Printf("Error [IRForTarget]: Names starting with $0, "
1107 "$1, ... are reserved for use as result "
1108 "names\n");
1109
1110 return false;
1111 }
1112
1113 pvar_allocs.push_back(alloc);
1114 }
1115 }
1116 }
1117
1118 for (Instruction *inst : pvar_allocs) {
1119 if (!RewritePersistentAlloc(inst)) {
1120 m_error_stream.Printf("Internal error [IRForTarget]: Couldn't rewrite "
1121 "the creation of a persistent variable\n");
1122
1123 LLDB_LOG(log, "Couldn't rewrite the creation of a persistent variable");
1124
1125 return false;
1126 }
1127 }
1128
1129 return true;
1130 }
1131
1132 // This function does not report errors; its callers are responsible.
MaybeHandleVariable(Value * llvm_value_ptr)1133 bool IRForTarget::MaybeHandleVariable(Value *llvm_value_ptr) {
1134 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1135
1136 LLDB_LOG(log, "MaybeHandleVariable ({0})", PrintValue(llvm_value_ptr));
1137
1138 if (ConstantExpr *constant_expr = dyn_cast<ConstantExpr>(llvm_value_ptr)) {
1139 switch (constant_expr->getOpcode()) {
1140 default:
1141 break;
1142 case Instruction::GetElementPtr:
1143 case Instruction::BitCast:
1144 Value *s = constant_expr->getOperand(0);
1145 if (!MaybeHandleVariable(s))
1146 return false;
1147 }
1148 } else if (GlobalVariable *global_variable =
1149 dyn_cast<GlobalVariable>(llvm_value_ptr)) {
1150 if (!GlobalValue::isExternalLinkage(global_variable->getLinkage()))
1151 return true;
1152
1153 clang::NamedDecl *named_decl = DeclForGlobal(global_variable);
1154
1155 if (!named_decl) {
1156 if (IsObjCSelectorRef(llvm_value_ptr))
1157 return true;
1158
1159 if (!global_variable->hasExternalLinkage())
1160 return true;
1161
1162 LLDB_LOG(log, "Found global variable \"{0}\" without metadata",
1163 global_variable->getName());
1164
1165 return false;
1166 }
1167
1168 llvm::StringRef name(named_decl->getName());
1169
1170 clang::ValueDecl *value_decl = dyn_cast<clang::ValueDecl>(named_decl);
1171 if (value_decl == nullptr)
1172 return false;
1173
1174 lldb_private::CompilerType compiler_type =
1175 m_decl_map->GetTypeSystem()->GetType(value_decl->getType());
1176
1177 const Type *value_type = nullptr;
1178
1179 if (name.startswith("$")) {
1180 // The $__lldb_expr_result name indicates the return value has allocated
1181 // as a static variable. Per the comment at
1182 // ASTResultSynthesizer::SynthesizeBodyResult, accesses to this static
1183 // variable need to be redirected to the result of dereferencing a
1184 // pointer that is passed in as one of the arguments.
1185 //
1186 // Consequently, when reporting the size of the type, we report a pointer
1187 // type pointing to the type of $__lldb_expr_result, not the type itself.
1188 //
1189 // We also do this for any user-declared persistent variables.
1190 compiler_type = compiler_type.GetPointerType();
1191 value_type = PointerType::get(global_variable->getType(), 0);
1192 } else {
1193 value_type = global_variable->getType();
1194 }
1195
1196 auto *target = m_execution_unit.GetTarget().get();
1197 std::optional<uint64_t> value_size = compiler_type.GetByteSize(target);
1198 if (!value_size)
1199 return false;
1200 std::optional<size_t> opt_alignment = compiler_type.GetTypeBitAlign(target);
1201 if (!opt_alignment)
1202 return false;
1203 lldb::offset_t value_alignment = (*opt_alignment + 7ull) / 8ull;
1204
1205 LLDB_LOG(log,
1206 "Type of \"{0}\" is [clang \"{1}\", llvm \"{2}\"] [size {3}, "
1207 "align {4}]",
1208 name,
1209 lldb_private::ClangUtil::GetQualType(compiler_type).getAsString(),
1210 PrintType(value_type), *value_size, value_alignment);
1211
1212 if (named_decl)
1213 m_decl_map->AddValueToStruct(named_decl, lldb_private::ConstString(name),
1214 llvm_value_ptr, *value_size,
1215 value_alignment);
1216 } else if (isa<llvm::Function>(llvm_value_ptr)) {
1217 LLDB_LOG(log, "Function pointers aren't handled right now");
1218
1219 return false;
1220 }
1221
1222 return true;
1223 }
1224
1225 // This function does not report errors; its callers are responsible.
HandleSymbol(Value * symbol)1226 bool IRForTarget::HandleSymbol(Value *symbol) {
1227 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1228
1229 lldb_private::ConstString name(symbol->getName().str().c_str());
1230
1231 lldb::addr_t symbol_addr =
1232 m_decl_map->GetSymbolAddress(name, lldb::eSymbolTypeAny);
1233
1234 if (symbol_addr == LLDB_INVALID_ADDRESS) {
1235 LLDB_LOG(log, "Symbol \"{0}\" had no address", name);
1236
1237 return false;
1238 }
1239
1240 LLDB_LOG(log, "Found \"{0}\" at {1}", name, symbol_addr);
1241
1242 Type *symbol_type = symbol->getType();
1243
1244 Constant *symbol_addr_int = ConstantInt::get(m_intptr_ty, symbol_addr, false);
1245
1246 Value *symbol_addr_ptr =
1247 ConstantExpr::getIntToPtr(symbol_addr_int, symbol_type);
1248
1249 LLDB_LOG(log, "Replacing {0} with {1}", PrintValue(symbol),
1250 PrintValue(symbol_addr_ptr));
1251
1252 symbol->replaceAllUsesWith(symbol_addr_ptr);
1253
1254 return true;
1255 }
1256
MaybeHandleCallArguments(CallInst * Old)1257 bool IRForTarget::MaybeHandleCallArguments(CallInst *Old) {
1258 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1259
1260 LLDB_LOG(log, "MaybeHandleCallArguments({0})", PrintValue(Old));
1261
1262 for (unsigned op_index = 0, num_ops = Old->arg_size();
1263 op_index < num_ops; ++op_index)
1264 // conservatively believe that this is a store
1265 if (!MaybeHandleVariable(Old->getArgOperand(op_index))) {
1266 m_error_stream.Printf("Internal error [IRForTarget]: Couldn't rewrite "
1267 "one of the arguments of a function call.\n");
1268
1269 return false;
1270 }
1271
1272 return true;
1273 }
1274
HandleObjCClass(Value * classlist_reference)1275 bool IRForTarget::HandleObjCClass(Value *classlist_reference) {
1276 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1277
1278 GlobalVariable *global_variable =
1279 dyn_cast<GlobalVariable>(classlist_reference);
1280
1281 if (!global_variable)
1282 return false;
1283
1284 Constant *initializer = global_variable->getInitializer();
1285
1286 if (!initializer)
1287 return false;
1288
1289 if (!initializer->hasName())
1290 return false;
1291
1292 StringRef name(initializer->getName());
1293 lldb_private::ConstString name_cstr(name.str().c_str());
1294 lldb::addr_t class_ptr =
1295 m_decl_map->GetSymbolAddress(name_cstr, lldb::eSymbolTypeObjCClass);
1296
1297 LLDB_LOG(log, "Found reference to Objective-C class {0} ({1})", name,
1298 (unsigned long long)class_ptr);
1299
1300 if (class_ptr == LLDB_INVALID_ADDRESS)
1301 return false;
1302
1303 if (global_variable->use_empty())
1304 return false;
1305
1306 SmallVector<LoadInst *, 2> load_instructions;
1307
1308 for (llvm::User *u : global_variable->users()) {
1309 if (LoadInst *load_instruction = dyn_cast<LoadInst>(u))
1310 load_instructions.push_back(load_instruction);
1311 }
1312
1313 if (load_instructions.empty())
1314 return false;
1315
1316 Constant *class_addr = ConstantInt::get(m_intptr_ty, (uint64_t)class_ptr);
1317
1318 for (LoadInst *load_instruction : load_instructions) {
1319 Constant *class_bitcast =
1320 ConstantExpr::getIntToPtr(class_addr, load_instruction->getType());
1321
1322 load_instruction->replaceAllUsesWith(class_bitcast);
1323
1324 load_instruction->eraseFromParent();
1325 }
1326
1327 return true;
1328 }
1329
RemoveCXAAtExit(BasicBlock & basic_block)1330 bool IRForTarget::RemoveCXAAtExit(BasicBlock &basic_block) {
1331 std::vector<CallInst *> calls_to_remove;
1332
1333 for (Instruction &inst : basic_block) {
1334 CallInst *call = dyn_cast<CallInst>(&inst);
1335
1336 // MaybeHandleCallArguments handles error reporting; we are silent here
1337 if (!call)
1338 continue;
1339
1340 bool remove = false;
1341
1342 llvm::Function *func = call->getCalledFunction();
1343
1344 if (func && func->getName() == "__cxa_atexit")
1345 remove = true;
1346
1347 llvm::Value *val = call->getCalledOperand();
1348
1349 if (val && val->getName() == "__cxa_atexit")
1350 remove = true;
1351
1352 if (remove)
1353 calls_to_remove.push_back(call);
1354 }
1355
1356 for (CallInst *ci : calls_to_remove)
1357 ci->eraseFromParent();
1358
1359 return true;
1360 }
1361
ResolveCalls(BasicBlock & basic_block)1362 bool IRForTarget::ResolveCalls(BasicBlock &basic_block) {
1363 // Prepare the current basic block for execution in the remote process
1364
1365 for (Instruction &inst : basic_block) {
1366 CallInst *call = dyn_cast<CallInst>(&inst);
1367
1368 // MaybeHandleCallArguments handles error reporting; we are silent here
1369 if (call && !MaybeHandleCallArguments(call))
1370 return false;
1371 }
1372
1373 return true;
1374 }
1375
ResolveExternals(Function & llvm_function)1376 bool IRForTarget::ResolveExternals(Function &llvm_function) {
1377 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1378
1379 for (GlobalVariable &global_var : m_module->globals()) {
1380 llvm::StringRef global_name = global_var.getName();
1381
1382 LLDB_LOG(log, "Examining {0}, DeclForGlobalValue returns {1}", global_name,
1383 static_cast<void *>(DeclForGlobal(&global_var)));
1384
1385 if (global_name.startswith("OBJC_IVAR")) {
1386 if (!HandleSymbol(&global_var)) {
1387 m_error_stream.Format("Error [IRForTarget]: Couldn't find Objective-C "
1388 "indirect ivar symbol {0}\n",
1389 global_name);
1390
1391 return false;
1392 }
1393 } else if (global_name.contains("OBJC_CLASSLIST_REFERENCES_$")) {
1394 if (!HandleObjCClass(&global_var)) {
1395 m_error_stream.Printf("Error [IRForTarget]: Couldn't resolve the class "
1396 "for an Objective-C static method call\n");
1397
1398 return false;
1399 }
1400 } else if (global_name.contains("OBJC_CLASSLIST_SUP_REFS_$")) {
1401 if (!HandleObjCClass(&global_var)) {
1402 m_error_stream.Printf("Error [IRForTarget]: Couldn't resolve the class "
1403 "for an Objective-C static method call\n");
1404
1405 return false;
1406 }
1407 } else if (DeclForGlobal(&global_var)) {
1408 if (!MaybeHandleVariable(&global_var)) {
1409 m_error_stream.Format("Internal error [IRForTarget]: Couldn't rewrite "
1410 "external variable {0}\n",
1411 global_name);
1412
1413 return false;
1414 }
1415 }
1416 }
1417
1418 return true;
1419 }
1420
isGuardVariableRef(Value * V)1421 static bool isGuardVariableRef(Value *V) {
1422 Constant *Old = dyn_cast<Constant>(V);
1423
1424 if (!Old)
1425 return false;
1426
1427 if (auto CE = dyn_cast<ConstantExpr>(V)) {
1428 if (CE->getOpcode() != Instruction::BitCast)
1429 return false;
1430
1431 Old = CE->getOperand(0);
1432 }
1433
1434 GlobalVariable *GV = dyn_cast<GlobalVariable>(Old);
1435
1436 if (!GV || !GV->hasName() || !isGuardVariableSymbol(GV->getName()))
1437 return false;
1438
1439 return true;
1440 }
1441
TurnGuardLoadIntoZero(llvm::Instruction * guard_load)1442 void IRForTarget::TurnGuardLoadIntoZero(llvm::Instruction *guard_load) {
1443 Constant *zero(Constant::getNullValue(guard_load->getType()));
1444 guard_load->replaceAllUsesWith(zero);
1445 guard_load->eraseFromParent();
1446 }
1447
ExciseGuardStore(Instruction * guard_store)1448 static void ExciseGuardStore(Instruction *guard_store) {
1449 guard_store->eraseFromParent();
1450 }
1451
RemoveGuards(BasicBlock & basic_block)1452 bool IRForTarget::RemoveGuards(BasicBlock &basic_block) {
1453 // Eliminate any reference to guard variables found.
1454
1455 InstrList guard_loads;
1456 InstrList guard_stores;
1457
1458 for (Instruction &inst : basic_block) {
1459
1460 if (LoadInst *load = dyn_cast<LoadInst>(&inst))
1461 if (isGuardVariableRef(load->getPointerOperand()))
1462 guard_loads.push_back(&inst);
1463
1464 if (StoreInst *store = dyn_cast<StoreInst>(&inst))
1465 if (isGuardVariableRef(store->getPointerOperand()))
1466 guard_stores.push_back(&inst);
1467 }
1468
1469 for (Instruction *inst : guard_loads)
1470 TurnGuardLoadIntoZero(inst);
1471
1472 for (Instruction *inst : guard_stores)
1473 ExciseGuardStore(inst);
1474
1475 return true;
1476 }
1477
1478 // This function does not report errors; its callers are responsible.
UnfoldConstant(Constant * old_constant,llvm::Function * llvm_function,FunctionValueCache & value_maker,FunctionValueCache & entry_instruction_finder,lldb_private::Stream & error_stream)1479 bool IRForTarget::UnfoldConstant(Constant *old_constant,
1480 llvm::Function *llvm_function,
1481 FunctionValueCache &value_maker,
1482 FunctionValueCache &entry_instruction_finder,
1483 lldb_private::Stream &error_stream) {
1484 SmallVector<User *, 16> users;
1485
1486 // We do this because the use list might change, invalidating our iterator.
1487 // Much better to keep a work list ourselves.
1488 for (llvm::User *u : old_constant->users())
1489 users.push_back(u);
1490
1491 for (size_t i = 0; i < users.size(); ++i) {
1492 User *user = users[i];
1493
1494 if (Constant *constant = dyn_cast<Constant>(user)) {
1495 // synthesize a new non-constant equivalent of the constant
1496
1497 if (ConstantExpr *constant_expr = dyn_cast<ConstantExpr>(constant)) {
1498 switch (constant_expr->getOpcode()) {
1499 default:
1500 error_stream.Printf("error [IRForTarget internal]: Unhandled "
1501 "constant expression type: \"%s\"",
1502 PrintValue(constant_expr).c_str());
1503 return false;
1504 case Instruction::BitCast: {
1505 FunctionValueCache bit_cast_maker(
1506 [&value_maker, &entry_instruction_finder, old_constant,
1507 constant_expr](llvm::Function *function) -> llvm::Value * {
1508 // UnaryExpr
1509 // OperandList[0] is value
1510
1511 if (constant_expr->getOperand(0) != old_constant)
1512 return constant_expr;
1513
1514 return new BitCastInst(
1515 value_maker.GetValue(function), constant_expr->getType(),
1516 "", llvm::cast<Instruction>(
1517 entry_instruction_finder.GetValue(function)));
1518 });
1519
1520 if (!UnfoldConstant(constant_expr, llvm_function, bit_cast_maker,
1521 entry_instruction_finder, error_stream))
1522 return false;
1523 } break;
1524 case Instruction::GetElementPtr: {
1525 // GetElementPtrConstantExpr
1526 // OperandList[0] is base
1527 // OperandList[1]... are indices
1528
1529 FunctionValueCache get_element_pointer_maker(
1530 [&value_maker, &entry_instruction_finder, old_constant,
1531 constant_expr](llvm::Function *function) -> llvm::Value * {
1532 auto *gep = cast<llvm::GEPOperator>(constant_expr);
1533 Value *ptr = gep->getPointerOperand();
1534
1535 if (ptr == old_constant)
1536 ptr = value_maker.GetValue(function);
1537
1538 std::vector<Value *> index_vector;
1539 for (Value *operand : gep->indices()) {
1540 if (operand == old_constant)
1541 operand = value_maker.GetValue(function);
1542
1543 index_vector.push_back(operand);
1544 }
1545
1546 ArrayRef<Value *> indices(index_vector);
1547
1548 return GetElementPtrInst::Create(
1549 gep->getSourceElementType(), ptr, indices, "",
1550 llvm::cast<Instruction>(
1551 entry_instruction_finder.GetValue(function)));
1552 });
1553
1554 if (!UnfoldConstant(constant_expr, llvm_function,
1555 get_element_pointer_maker,
1556 entry_instruction_finder, error_stream))
1557 return false;
1558 } break;
1559 }
1560 } else {
1561 error_stream.Printf(
1562 "error [IRForTarget internal]: Unhandled constant type: \"%s\"",
1563 PrintValue(constant).c_str());
1564 return false;
1565 }
1566 } else {
1567 if (Instruction *inst = llvm::dyn_cast<Instruction>(user)) {
1568 if (llvm_function && inst->getParent()->getParent() != llvm_function) {
1569 error_stream.PutCString("error: Capturing non-local variables in "
1570 "expressions is unsupported.\n");
1571 return false;
1572 }
1573 inst->replaceUsesOfWith(
1574 old_constant, value_maker.GetValue(inst->getParent()->getParent()));
1575 } else {
1576 error_stream.Printf(
1577 "error [IRForTarget internal]: Unhandled non-constant type: \"%s\"",
1578 PrintValue(user).c_str());
1579 return false;
1580 }
1581 }
1582 }
1583
1584 if (!isa<GlobalValue>(old_constant)) {
1585 old_constant->destroyConstant();
1586 }
1587
1588 return true;
1589 }
1590
ReplaceVariables(Function & llvm_function)1591 bool IRForTarget::ReplaceVariables(Function &llvm_function) {
1592 if (!m_resolve_vars)
1593 return true;
1594
1595 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1596
1597 m_decl_map->DoStructLayout();
1598
1599 LLDB_LOG(log, "Element arrangement:");
1600
1601 uint32_t num_elements;
1602 uint32_t element_index;
1603
1604 size_t size;
1605 lldb::offset_t alignment;
1606
1607 if (!m_decl_map->GetStructInfo(num_elements, size, alignment))
1608 return false;
1609
1610 Function::arg_iterator iter(llvm_function.arg_begin());
1611
1612 if (iter == llvm_function.arg_end()) {
1613 m_error_stream.Printf("Internal error [IRForTarget]: Wrapper takes no "
1614 "arguments (should take at least a struct pointer)");
1615
1616 return false;
1617 }
1618
1619 Argument *argument = &*iter;
1620
1621 if (argument->getName().equals("this")) {
1622 ++iter;
1623
1624 if (iter == llvm_function.arg_end()) {
1625 m_error_stream.Printf("Internal error [IRForTarget]: Wrapper takes only "
1626 "'this' argument (should take a struct pointer "
1627 "too)");
1628
1629 return false;
1630 }
1631
1632 argument = &*iter;
1633 } else if (argument->getName().equals("self")) {
1634 ++iter;
1635
1636 if (iter == llvm_function.arg_end()) {
1637 m_error_stream.Printf("Internal error [IRForTarget]: Wrapper takes only "
1638 "'self' argument (should take '_cmd' and a struct "
1639 "pointer too)");
1640
1641 return false;
1642 }
1643
1644 if (!iter->getName().equals("_cmd")) {
1645 m_error_stream.Format("Internal error [IRForTarget]: Wrapper takes '{0}' "
1646 "after 'self' argument (should take '_cmd')",
1647 iter->getName());
1648
1649 return false;
1650 }
1651
1652 ++iter;
1653
1654 if (iter == llvm_function.arg_end()) {
1655 m_error_stream.Printf("Internal error [IRForTarget]: Wrapper takes only "
1656 "'self' and '_cmd' arguments (should take a struct "
1657 "pointer too)");
1658
1659 return false;
1660 }
1661
1662 argument = &*iter;
1663 }
1664
1665 if (!argument->getName().equals("$__lldb_arg")) {
1666 m_error_stream.Format("Internal error [IRForTarget]: Wrapper takes an "
1667 "argument named '{0}' instead of the struct pointer",
1668 argument->getName());
1669
1670 return false;
1671 }
1672
1673 LLDB_LOG(log, "Arg: \"{0}\"", PrintValue(argument));
1674
1675 BasicBlock &entry_block(llvm_function.getEntryBlock());
1676 Instruction *FirstEntryInstruction(entry_block.getFirstNonPHIOrDbg());
1677
1678 if (!FirstEntryInstruction) {
1679 m_error_stream.Printf("Internal error [IRForTarget]: Couldn't find the "
1680 "first instruction in the wrapper for use in "
1681 "rewriting");
1682
1683 return false;
1684 }
1685
1686 LLVMContext &context(m_module->getContext());
1687 IntegerType *offset_type(Type::getInt32Ty(context));
1688
1689 if (!offset_type) {
1690 m_error_stream.Printf(
1691 "Internal error [IRForTarget]: Couldn't produce an offset type");
1692
1693 return false;
1694 }
1695
1696 for (element_index = 0; element_index < num_elements; ++element_index) {
1697 const clang::NamedDecl *decl = nullptr;
1698 Value *value = nullptr;
1699 lldb::offset_t offset;
1700 lldb_private::ConstString name;
1701
1702 if (!m_decl_map->GetStructElement(decl, value, offset, name,
1703 element_index)) {
1704 m_error_stream.Printf(
1705 "Internal error [IRForTarget]: Structure information is incomplete");
1706
1707 return false;
1708 }
1709
1710 LLDB_LOG(log, " \"{0}\" (\"{1}\") placed at {2}", name,
1711 decl->getNameAsString(), offset);
1712
1713 if (value) {
1714 LLDB_LOG(log, " Replacing [{0}]", PrintValue(value));
1715
1716 FunctionValueCache body_result_maker(
1717 [this, name, offset_type, offset, argument,
1718 value](llvm::Function *function) -> llvm::Value * {
1719 // Per the comment at ASTResultSynthesizer::SynthesizeBodyResult,
1720 // in cases where the result variable is an rvalue, we have to
1721 // synthesize a dereference of the appropriate structure entry in
1722 // order to produce the static variable that the AST thinks it is
1723 // accessing.
1724
1725 llvm::Instruction *entry_instruction = llvm::cast<Instruction>(
1726 m_entry_instruction_finder.GetValue(function));
1727
1728 Type *int8Ty = Type::getInt8Ty(function->getContext());
1729 ConstantInt *offset_int(
1730 ConstantInt::get(offset_type, offset, true));
1731 GetElementPtrInst *get_element_ptr = GetElementPtrInst::Create(
1732 int8Ty, argument, offset_int, "", entry_instruction);
1733
1734 if (name == m_result_name && !m_result_is_pointer) {
1735 BitCastInst *bit_cast = new BitCastInst(
1736 get_element_ptr, value->getType()->getPointerTo(), "",
1737 entry_instruction);
1738
1739 LoadInst *load = new LoadInst(value->getType(), bit_cast, "",
1740 entry_instruction);
1741
1742 return load;
1743 } else {
1744 BitCastInst *bit_cast = new BitCastInst(
1745 get_element_ptr, value->getType(), "", entry_instruction);
1746
1747 return bit_cast;
1748 }
1749 });
1750
1751 if (Constant *constant = dyn_cast<Constant>(value)) {
1752 if (!UnfoldConstant(constant, &llvm_function, body_result_maker,
1753 m_entry_instruction_finder, m_error_stream)) {
1754 return false;
1755 }
1756 } else if (Instruction *instruction = dyn_cast<Instruction>(value)) {
1757 if (instruction->getParent()->getParent() != &llvm_function) {
1758 m_error_stream.PutCString("error: Capturing non-local variables in "
1759 "expressions is unsupported.\n");
1760 return false;
1761 }
1762 value->replaceAllUsesWith(
1763 body_result_maker.GetValue(instruction->getParent()->getParent()));
1764 } else {
1765 LLDB_LOG(log, "Unhandled non-constant type: \"{0}\"",
1766 PrintValue(value));
1767 return false;
1768 }
1769
1770 if (GlobalVariable *var = dyn_cast<GlobalVariable>(value))
1771 var->eraseFromParent();
1772 }
1773 }
1774
1775 LLDB_LOG(log, "Total structure [align {0}, size {1}]", (int64_t)alignment,
1776 (uint64_t)size);
1777
1778 return true;
1779 }
1780
runOnModule(Module & llvm_module)1781 bool IRForTarget::runOnModule(Module &llvm_module) {
1782 lldb_private::Log *log(GetLog(LLDBLog::Expressions));
1783
1784 m_module = &llvm_module;
1785 m_target_data = std::make_unique<DataLayout>(m_module);
1786 m_intptr_ty = llvm::Type::getIntNTy(m_module->getContext(),
1787 m_target_data->getPointerSizeInBits());
1788
1789 if (log) {
1790 std::string s;
1791 raw_string_ostream oss(s);
1792
1793 m_module->print(oss, nullptr);
1794
1795 oss.flush();
1796
1797 LLDB_LOG(log, "Module as passed in to IRForTarget: \n\"{0}\"", s);
1798 }
1799
1800 Function *const main_function =
1801 m_func_name.IsEmpty() ? nullptr
1802 : m_module->getFunction(m_func_name.GetStringRef());
1803
1804 if (!m_func_name.IsEmpty() && !main_function) {
1805 LLDB_LOG(log, "Couldn't find \"{0}()\" in the module", m_func_name);
1806
1807 m_error_stream.Format("Internal error [IRForTarget]: Couldn't find wrapper "
1808 "'{0}' in the module",
1809 m_func_name);
1810
1811 return false;
1812 }
1813
1814 if (main_function) {
1815 if (!FixFunctionLinkage(*main_function)) {
1816 LLDB_LOG(log, "Couldn't fix the linkage for the function");
1817
1818 return false;
1819 }
1820 }
1821
1822 llvm::Type *int8_ty = Type::getInt8Ty(m_module->getContext());
1823
1824 m_reloc_placeholder = new llvm::GlobalVariable(
1825 (*m_module), int8_ty, false /* IsConstant */,
1826 GlobalVariable::InternalLinkage, Constant::getNullValue(int8_ty),
1827 "reloc_placeholder", nullptr /* InsertBefore */,
1828 GlobalVariable::NotThreadLocal /* ThreadLocal */, 0 /* AddressSpace */);
1829
1830 ////////////////////////////////////////////////////////////
1831 // Replace $__lldb_expr_result with a persistent variable
1832 //
1833
1834 if (main_function) {
1835 if (!CreateResultVariable(*main_function)) {
1836 LLDB_LOG(log, "CreateResultVariable() failed");
1837
1838 // CreateResultVariable() reports its own errors, so we don't do so here
1839
1840 return false;
1841 }
1842 }
1843
1844 if (log && log->GetVerbose()) {
1845 std::string s;
1846 raw_string_ostream oss(s);
1847
1848 m_module->print(oss, nullptr);
1849
1850 oss.flush();
1851
1852 LLDB_LOG(log, "Module after creating the result variable: \n\"{0}\"", s);
1853 }
1854
1855 for (llvm::Function &function : *m_module) {
1856 for (BasicBlock &bb : function) {
1857 if (!RemoveGuards(bb)) {
1858 LLDB_LOG(log, "RemoveGuards() failed");
1859
1860 // RemoveGuards() reports its own errors, so we don't do so here
1861
1862 return false;
1863 }
1864
1865 if (!RewritePersistentAllocs(bb)) {
1866 LLDB_LOG(log, "RewritePersistentAllocs() failed");
1867
1868 // RewritePersistentAllocs() reports its own errors, so we don't do so
1869 // here
1870
1871 return false;
1872 }
1873
1874 if (!RemoveCXAAtExit(bb)) {
1875 LLDB_LOG(log, "RemoveCXAAtExit() failed");
1876
1877 // RemoveCXAAtExit() reports its own errors, so we don't do so here
1878
1879 return false;
1880 }
1881 }
1882 }
1883
1884 ///////////////////////////////////////////////////////////////////////////////
1885 // Fix all Objective-C constant strings to use NSStringWithCString:encoding:
1886 //
1887
1888 if (!RewriteObjCConstStrings()) {
1889 LLDB_LOG(log, "RewriteObjCConstStrings() failed");
1890
1891 // RewriteObjCConstStrings() reports its own errors, so we don't do so here
1892
1893 return false;
1894 }
1895
1896 for (llvm::Function &function : *m_module) {
1897 for (llvm::BasicBlock &bb : function) {
1898 if (!RewriteObjCSelectors(bb)) {
1899 LLDB_LOG(log, "RewriteObjCSelectors() failed");
1900
1901 // RewriteObjCSelectors() reports its own errors, so we don't do so
1902 // here
1903
1904 return false;
1905 }
1906
1907 if (!RewriteObjCClassReferences(bb)) {
1908 LLDB_LOG(log, "RewriteObjCClassReferences() failed");
1909
1910 // RewriteObjCClasses() reports its own errors, so we don't do so here
1911
1912 return false;
1913 }
1914 }
1915 }
1916
1917 for (llvm::Function &function : *m_module) {
1918 for (BasicBlock &bb : function) {
1919 if (!ResolveCalls(bb)) {
1920 LLDB_LOG(log, "ResolveCalls() failed");
1921
1922 // ResolveCalls() reports its own errors, so we don't do so here
1923
1924 return false;
1925 }
1926 }
1927 }
1928
1929 ////////////////////////////////////////////////////////////////////////
1930 // Run function-level passes that only make sense on the main function
1931 //
1932
1933 if (main_function) {
1934 if (!ResolveExternals(*main_function)) {
1935 LLDB_LOG(log, "ResolveExternals() failed");
1936
1937 // ResolveExternals() reports its own errors, so we don't do so here
1938
1939 return false;
1940 }
1941
1942 if (!ReplaceVariables(*main_function)) {
1943 LLDB_LOG(log, "ReplaceVariables() failed");
1944
1945 // ReplaceVariables() reports its own errors, so we don't do so here
1946
1947 return false;
1948 }
1949 }
1950
1951 if (log && log->GetVerbose()) {
1952 std::string s;
1953 raw_string_ostream oss(s);
1954
1955 m_module->print(oss, nullptr);
1956
1957 oss.flush();
1958
1959 LLDB_LOG(log, "Module after preparing for execution: \n\"{0}\"", s);
1960 }
1961
1962 return true;
1963 }
1964