1 //===- MetadataLoader.cpp - Internal BitcodeReader implementation ---------===//
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 "MetadataLoader.h"
10 #include "ValueList.h"
11 
12 #include "llvm/ADT/APInt.h"
13 #include "llvm/ADT/ArrayRef.h"
14 #include "llvm/ADT/BitmaskEnum.h"
15 #include "llvm/ADT/DenseMap.h"
16 #include "llvm/ADT/DenseSet.h"
17 #include "llvm/ADT/STLFunctionalExtras.h"
18 #include "llvm/ADT/SetVector.h"
19 #include "llvm/ADT/SmallString.h"
20 #include "llvm/ADT/SmallVector.h"
21 #include "llvm/ADT/Statistic.h"
22 #include "llvm/ADT/StringRef.h"
23 #include "llvm/ADT/Twine.h"
24 #include "llvm/ADT/ilist_iterator.h"
25 #include "llvm/BinaryFormat/Dwarf.h"
26 #include "llvm/Bitcode/BitcodeReader.h"
27 #include "llvm/Bitcode/LLVMBitCodes.h"
28 #include "llvm/Bitstream/BitstreamReader.h"
29 #include "llvm/IR/AutoUpgrade.h"
30 #include "llvm/IR/BasicBlock.h"
31 #include "llvm/IR/Constants.h"
32 #include "llvm/IR/DebugInfoMetadata.h"
33 #include "llvm/IR/Function.h"
34 #include "llvm/IR/GlobalObject.h"
35 #include "llvm/IR/GlobalVariable.h"
36 #include "llvm/IR/Instruction.h"
37 #include "llvm/IR/IntrinsicInst.h"
38 #include "llvm/IR/LLVMContext.h"
39 #include "llvm/IR/Metadata.h"
40 #include "llvm/IR/Module.h"
41 #include "llvm/IR/TrackingMDRef.h"
42 #include "llvm/IR/Type.h"
43 #include "llvm/Support/Casting.h"
44 #include "llvm/Support/CommandLine.h"
45 #include "llvm/Support/Compiler.h"
46 #include "llvm/Support/ErrorHandling.h"
47 #include "llvm/Support/type_traits.h"
48 
49 #include <algorithm>
50 #include <cassert>
51 #include <cstddef>
52 #include <cstdint>
53 #include <deque>
54 #include <iterator>
55 #include <limits>
56 #include <map>
57 #include <optional>
58 #include <string>
59 #include <tuple>
60 #include <type_traits>
61 #include <utility>
62 #include <vector>
63 namespace llvm {
64 class Argument;
65 }
66 
67 using namespace llvm;
68 
69 #define DEBUG_TYPE "bitcode-reader"
70 
71 STATISTIC(NumMDStringLoaded, "Number of MDStrings loaded");
72 STATISTIC(NumMDNodeTemporary, "Number of MDNode::Temporary created");
73 STATISTIC(NumMDRecordLoaded, "Number of Metadata records loaded");
74 
75 /// Flag whether we need to import full type definitions for ThinLTO.
76 /// Currently needed for Darwin and LLDB.
77 static cl::opt<bool> ImportFullTypeDefinitions(
78     "import-full-type-definitions", cl::init(false), cl::Hidden,
79     cl::desc("Import full type definitions for ThinLTO."));
80 
81 static cl::opt<bool> DisableLazyLoading(
82     "disable-ondemand-mds-loading", cl::init(false), cl::Hidden,
83     cl::desc("Force disable the lazy-loading on-demand of metadata when "
84              "loading bitcode for importing."));
85 
86 namespace {
87 
unrotateSign(uint64_t U)88 static int64_t unrotateSign(uint64_t U) { return (U & 1) ? ~(U >> 1) : U >> 1; }
89 
90 class BitcodeReaderMetadataList {
91   /// Array of metadata references.
92   ///
93   /// Don't use std::vector here.  Some versions of libc++ copy (instead of
94   /// move) on resize, and TrackingMDRef is very expensive to copy.
95   SmallVector<TrackingMDRef, 1> MetadataPtrs;
96 
97   /// The set of indices in MetadataPtrs above of forward references that were
98   /// generated.
99   SmallDenseSet<unsigned, 1> ForwardReference;
100 
101   /// The set of indices in MetadataPtrs above of Metadata that need to be
102   /// resolved.
103   SmallDenseSet<unsigned, 1> UnresolvedNodes;
104 
105   /// Structures for resolving old type refs.
106   struct {
107     SmallDenseMap<MDString *, TempMDTuple, 1> Unknown;
108     SmallDenseMap<MDString *, DICompositeType *, 1> Final;
109     SmallDenseMap<MDString *, DICompositeType *, 1> FwdDecls;
110     SmallVector<std::pair<TrackingMDRef, TempMDTuple>, 1> Arrays;
111   } OldTypeRefs;
112 
113   LLVMContext &Context;
114 
115   /// Maximum number of valid references. Forward references exceeding the
116   /// maximum must be invalid.
117   unsigned RefsUpperBound;
118 
119 public:
BitcodeReaderMetadataList(LLVMContext & C,size_t RefsUpperBound)120   BitcodeReaderMetadataList(LLVMContext &C, size_t RefsUpperBound)
121       : Context(C),
122         RefsUpperBound(std::min((size_t)std::numeric_limits<unsigned>::max(),
123                                 RefsUpperBound)) {}
124 
125   // vector compatibility methods
size() const126   unsigned size() const { return MetadataPtrs.size(); }
resize(unsigned N)127   void resize(unsigned N) { MetadataPtrs.resize(N); }
push_back(Metadata * MD)128   void push_back(Metadata *MD) { MetadataPtrs.emplace_back(MD); }
clear()129   void clear() { MetadataPtrs.clear(); }
back() const130   Metadata *back() const { return MetadataPtrs.back(); }
pop_back()131   void pop_back() { MetadataPtrs.pop_back(); }
empty() const132   bool empty() const { return MetadataPtrs.empty(); }
133 
operator [](unsigned i) const134   Metadata *operator[](unsigned i) const {
135     assert(i < MetadataPtrs.size());
136     return MetadataPtrs[i];
137   }
138 
lookup(unsigned I) const139   Metadata *lookup(unsigned I) const {
140     if (I < MetadataPtrs.size())
141       return MetadataPtrs[I];
142     return nullptr;
143   }
144 
shrinkTo(unsigned N)145   void shrinkTo(unsigned N) {
146     assert(N <= size() && "Invalid shrinkTo request!");
147     assert(ForwardReference.empty() && "Unexpected forward refs");
148     assert(UnresolvedNodes.empty() && "Unexpected unresolved node");
149     MetadataPtrs.resize(N);
150   }
151 
152   /// Return the given metadata, creating a replaceable forward reference if
153   /// necessary.
154   Metadata *getMetadataFwdRef(unsigned Idx);
155 
156   /// Return the given metadata only if it is fully resolved.
157   ///
158   /// Gives the same result as \a lookup(), unless \a MDNode::isResolved()
159   /// would give \c false.
160   Metadata *getMetadataIfResolved(unsigned Idx);
161 
162   MDNode *getMDNodeFwdRefOrNull(unsigned Idx);
163   void assignValue(Metadata *MD, unsigned Idx);
164   void tryToResolveCycles();
hasFwdRefs() const165   bool hasFwdRefs() const { return !ForwardReference.empty(); }
getNextFwdRef()166   int getNextFwdRef() {
167     assert(hasFwdRefs());
168     return *ForwardReference.begin();
169   }
170 
171   /// Upgrade a type that had an MDString reference.
172   void addTypeRef(MDString &UUID, DICompositeType &CT);
173 
174   /// Upgrade a type that had an MDString reference.
175   Metadata *upgradeTypeRef(Metadata *MaybeUUID);
176 
177   /// Upgrade a type ref array that may have MDString references.
178   Metadata *upgradeTypeRefArray(Metadata *MaybeTuple);
179 
180 private:
181   Metadata *resolveTypeRefArray(Metadata *MaybeTuple);
182 };
183 
assignValue(Metadata * MD,unsigned Idx)184 void BitcodeReaderMetadataList::assignValue(Metadata *MD, unsigned Idx) {
185   if (auto *MDN = dyn_cast<MDNode>(MD))
186     if (!MDN->isResolved())
187       UnresolvedNodes.insert(Idx);
188 
189   if (Idx == size()) {
190     push_back(MD);
191     return;
192   }
193 
194   if (Idx >= size())
195     resize(Idx + 1);
196 
197   TrackingMDRef &OldMD = MetadataPtrs[Idx];
198   if (!OldMD) {
199     OldMD.reset(MD);
200     return;
201   }
202 
203   // If there was a forward reference to this value, replace it.
204   TempMDTuple PrevMD(cast<MDTuple>(OldMD.get()));
205   PrevMD->replaceAllUsesWith(MD);
206   ForwardReference.erase(Idx);
207 }
208 
getMetadataFwdRef(unsigned Idx)209 Metadata *BitcodeReaderMetadataList::getMetadataFwdRef(unsigned Idx) {
210   // Bail out for a clearly invalid value.
211   if (Idx >= RefsUpperBound)
212     return nullptr;
213 
214   if (Idx >= size())
215     resize(Idx + 1);
216 
217   if (Metadata *MD = MetadataPtrs[Idx])
218     return MD;
219 
220   // Track forward refs to be resolved later.
221   ForwardReference.insert(Idx);
222 
223   // Create and return a placeholder, which will later be RAUW'd.
224   ++NumMDNodeTemporary;
225   Metadata *MD = MDNode::getTemporary(Context, std::nullopt).release();
226   MetadataPtrs[Idx].reset(MD);
227   return MD;
228 }
229 
getMetadataIfResolved(unsigned Idx)230 Metadata *BitcodeReaderMetadataList::getMetadataIfResolved(unsigned Idx) {
231   Metadata *MD = lookup(Idx);
232   if (auto *N = dyn_cast_or_null<MDNode>(MD))
233     if (!N->isResolved())
234       return nullptr;
235   return MD;
236 }
237 
getMDNodeFwdRefOrNull(unsigned Idx)238 MDNode *BitcodeReaderMetadataList::getMDNodeFwdRefOrNull(unsigned Idx) {
239   return dyn_cast_or_null<MDNode>(getMetadataFwdRef(Idx));
240 }
241 
tryToResolveCycles()242 void BitcodeReaderMetadataList::tryToResolveCycles() {
243   if (!ForwardReference.empty())
244     // Still forward references... can't resolve cycles.
245     return;
246 
247   // Give up on finding a full definition for any forward decls that remain.
248   for (const auto &Ref : OldTypeRefs.FwdDecls)
249     OldTypeRefs.Final.insert(Ref);
250   OldTypeRefs.FwdDecls.clear();
251 
252   // Upgrade from old type ref arrays.  In strange cases, this could add to
253   // OldTypeRefs.Unknown.
254   for (const auto &Array : OldTypeRefs.Arrays)
255     Array.second->replaceAllUsesWith(resolveTypeRefArray(Array.first.get()));
256   OldTypeRefs.Arrays.clear();
257 
258   // Replace old string-based type refs with the resolved node, if possible.
259   // If we haven't seen the node, leave it to the verifier to complain about
260   // the invalid string reference.
261   for (const auto &Ref : OldTypeRefs.Unknown) {
262     if (DICompositeType *CT = OldTypeRefs.Final.lookup(Ref.first))
263       Ref.second->replaceAllUsesWith(CT);
264     else
265       Ref.second->replaceAllUsesWith(Ref.first);
266   }
267   OldTypeRefs.Unknown.clear();
268 
269   if (UnresolvedNodes.empty())
270     // Nothing to do.
271     return;
272 
273   // Resolve any cycles.
274   for (unsigned I : UnresolvedNodes) {
275     auto &MD = MetadataPtrs[I];
276     auto *N = dyn_cast_or_null<MDNode>(MD);
277     if (!N)
278       continue;
279 
280     assert(!N->isTemporary() && "Unexpected forward reference");
281     N->resolveCycles();
282   }
283 
284   // Make sure we return early again until there's another unresolved ref.
285   UnresolvedNodes.clear();
286 }
287 
addTypeRef(MDString & UUID,DICompositeType & CT)288 void BitcodeReaderMetadataList::addTypeRef(MDString &UUID,
289                                            DICompositeType &CT) {
290   assert(CT.getRawIdentifier() == &UUID && "Mismatched UUID");
291   if (CT.isForwardDecl())
292     OldTypeRefs.FwdDecls.insert(std::make_pair(&UUID, &CT));
293   else
294     OldTypeRefs.Final.insert(std::make_pair(&UUID, &CT));
295 }
296 
upgradeTypeRef(Metadata * MaybeUUID)297 Metadata *BitcodeReaderMetadataList::upgradeTypeRef(Metadata *MaybeUUID) {
298   auto *UUID = dyn_cast_or_null<MDString>(MaybeUUID);
299   if (LLVM_LIKELY(!UUID))
300     return MaybeUUID;
301 
302   if (auto *CT = OldTypeRefs.Final.lookup(UUID))
303     return CT;
304 
305   auto &Ref = OldTypeRefs.Unknown[UUID];
306   if (!Ref)
307     Ref = MDNode::getTemporary(Context, std::nullopt);
308   return Ref.get();
309 }
310 
upgradeTypeRefArray(Metadata * MaybeTuple)311 Metadata *BitcodeReaderMetadataList::upgradeTypeRefArray(Metadata *MaybeTuple) {
312   auto *Tuple = dyn_cast_or_null<MDTuple>(MaybeTuple);
313   if (!Tuple || Tuple->isDistinct())
314     return MaybeTuple;
315 
316   // Look through the array immediately if possible.
317   if (!Tuple->isTemporary())
318     return resolveTypeRefArray(Tuple);
319 
320   // Create and return a placeholder to use for now.  Eventually
321   // resolveTypeRefArrays() will be resolve this forward reference.
322   OldTypeRefs.Arrays.emplace_back(
323       std::piecewise_construct, std::forward_as_tuple(Tuple),
324       std::forward_as_tuple(MDTuple::getTemporary(Context, std::nullopt)));
325   return OldTypeRefs.Arrays.back().second.get();
326 }
327 
resolveTypeRefArray(Metadata * MaybeTuple)328 Metadata *BitcodeReaderMetadataList::resolveTypeRefArray(Metadata *MaybeTuple) {
329   auto *Tuple = dyn_cast_or_null<MDTuple>(MaybeTuple);
330   if (!Tuple || Tuple->isDistinct())
331     return MaybeTuple;
332 
333   // Look through the DITypeRefArray, upgrading each DIType *.
334   SmallVector<Metadata *, 32> Ops;
335   Ops.reserve(Tuple->getNumOperands());
336   for (Metadata *MD : Tuple->operands())
337     Ops.push_back(upgradeTypeRef(MD));
338 
339   return MDTuple::get(Context, Ops);
340 }
341 
342 namespace {
343 
344 class PlaceholderQueue {
345   // Placeholders would thrash around when moved, so store in a std::deque
346   // instead of some sort of vector.
347   std::deque<DistinctMDOperandPlaceholder> PHs;
348 
349 public:
~PlaceholderQueue()350   ~PlaceholderQueue() {
351     assert(empty() &&
352            "PlaceholderQueue hasn't been flushed before being destroyed");
353   }
empty() const354   bool empty() const { return PHs.empty(); }
355   DistinctMDOperandPlaceholder &getPlaceholderOp(unsigned ID);
356   void flush(BitcodeReaderMetadataList &MetadataList);
357 
358   /// Return the list of temporaries nodes in the queue, these need to be
359   /// loaded before we can flush the queue.
getTemporaries(BitcodeReaderMetadataList & MetadataList,DenseSet<unsigned> & Temporaries)360   void getTemporaries(BitcodeReaderMetadataList &MetadataList,
361                       DenseSet<unsigned> &Temporaries) {
362     for (auto &PH : PHs) {
363       auto ID = PH.getID();
364       auto *MD = MetadataList.lookup(ID);
365       if (!MD) {
366         Temporaries.insert(ID);
367         continue;
368       }
369       auto *N = dyn_cast_or_null<MDNode>(MD);
370       if (N && N->isTemporary())
371         Temporaries.insert(ID);
372     }
373   }
374 };
375 
376 } // end anonymous namespace
377 
getPlaceholderOp(unsigned ID)378 DistinctMDOperandPlaceholder &PlaceholderQueue::getPlaceholderOp(unsigned ID) {
379   PHs.emplace_back(ID);
380   return PHs.back();
381 }
382 
flush(BitcodeReaderMetadataList & MetadataList)383 void PlaceholderQueue::flush(BitcodeReaderMetadataList &MetadataList) {
384   while (!PHs.empty()) {
385     auto *MD = MetadataList.lookup(PHs.front().getID());
386     assert(MD && "Flushing placeholder on unassigned MD");
387 #ifndef NDEBUG
388     if (auto *MDN = dyn_cast<MDNode>(MD))
389       assert(MDN->isResolved() &&
390              "Flushing Placeholder while cycles aren't resolved");
391 #endif
392     PHs.front().replaceUseWith(MD);
393     PHs.pop_front();
394   }
395 }
396 
397 } // anonymous namespace
398 
error(const Twine & Message)399 static Error error(const Twine &Message) {
400   return make_error<StringError>(
401       Message, make_error_code(BitcodeError::CorruptedBitcode));
402 }
403 
404 class MetadataLoader::MetadataLoaderImpl {
405   BitcodeReaderMetadataList MetadataList;
406   BitcodeReaderValueList &ValueList;
407   BitstreamCursor &Stream;
408   LLVMContext &Context;
409   Module &TheModule;
410   MetadataLoaderCallbacks Callbacks;
411 
412   /// Cursor associated with the lazy-loading of Metadata. This is the easy way
413   /// to keep around the right "context" (Abbrev list) to be able to jump in
414   /// the middle of the metadata block and load any record.
415   BitstreamCursor IndexCursor;
416 
417   /// Index that keeps track of MDString values.
418   std::vector<StringRef> MDStringRef;
419 
420   /// On-demand loading of a single MDString. Requires the index above to be
421   /// populated.
422   MDString *lazyLoadOneMDString(unsigned Idx);
423 
424   /// Index that keeps track of where to find a metadata record in the stream.
425   std::vector<uint64_t> GlobalMetadataBitPosIndex;
426 
427   /// Cursor position of the start of the global decl attachments, to enable
428   /// loading using the index built for lazy loading, instead of forward
429   /// references.
430   uint64_t GlobalDeclAttachmentPos = 0;
431 
432 #ifndef NDEBUG
433   /// Baisic correctness check that we end up parsing all of the global decl
434   /// attachments.
435   unsigned NumGlobalDeclAttachSkipped = 0;
436   unsigned NumGlobalDeclAttachParsed = 0;
437 #endif
438 
439   /// Load the global decl attachments, using the index built for lazy loading.
440   Expected<bool> loadGlobalDeclAttachments();
441 
442   /// Populate the index above to enable lazily loading of metadata, and load
443   /// the named metadata as well as the transitively referenced global
444   /// Metadata.
445   Expected<bool> lazyLoadModuleMetadataBlock();
446 
447   /// On-demand loading of a single metadata. Requires the index above to be
448   /// populated.
449   void lazyLoadOneMetadata(unsigned Idx, PlaceholderQueue &Placeholders);
450 
451   // Keep mapping of seens pair of old-style CU <-> SP, and update pointers to
452   // point from SP to CU after a block is completly parsed.
453   std::vector<std::pair<DICompileUnit *, Metadata *>> CUSubprograms;
454 
455   /// Functions that need to be matched with subprograms when upgrading old
456   /// metadata.
457   SmallDenseMap<Function *, DISubprogram *, 16> FunctionsWithSPs;
458 
459   // Map the bitcode's custom MDKind ID to the Module's MDKind ID.
460   DenseMap<unsigned, unsigned> MDKindMap;
461 
462   bool StripTBAA = false;
463   bool HasSeenOldLoopTags = false;
464   bool NeedUpgradeToDIGlobalVariableExpression = false;
465   bool NeedDeclareExpressionUpgrade = false;
466 
467   /// Map DILocalScope to the enclosing DISubprogram, if any.
468   DenseMap<DILocalScope *, DISubprogram *> ParentSubprogram;
469 
470   /// True if metadata is being parsed for a module being ThinLTO imported.
471   bool IsImporting = false;
472 
473   Error parseOneMetadata(SmallVectorImpl<uint64_t> &Record, unsigned Code,
474                          PlaceholderQueue &Placeholders, StringRef Blob,
475                          unsigned &NextMetadataNo);
476   Error parseMetadataStrings(ArrayRef<uint64_t> Record, StringRef Blob,
477                              function_ref<void(StringRef)> CallBack);
478   Error parseGlobalObjectAttachment(GlobalObject &GO,
479                                     ArrayRef<uint64_t> Record);
480   Error parseMetadataKindRecord(SmallVectorImpl<uint64_t> &Record);
481 
482   void resolveForwardRefsAndPlaceholders(PlaceholderQueue &Placeholders);
483 
484   /// Upgrade old-style CU <-> SP pointers to point from SP to CU.
upgradeCUSubprograms()485   void upgradeCUSubprograms() {
486     for (auto CU_SP : CUSubprograms)
487       if (auto *SPs = dyn_cast_or_null<MDTuple>(CU_SP.second))
488         for (auto &Op : SPs->operands())
489           if (auto *SP = dyn_cast_or_null<DISubprogram>(Op))
490             SP->replaceUnit(CU_SP.first);
491     CUSubprograms.clear();
492   }
493 
494   /// Upgrade old-style bare DIGlobalVariables to DIGlobalVariableExpressions.
upgradeCUVariables()495   void upgradeCUVariables() {
496     if (!NeedUpgradeToDIGlobalVariableExpression)
497       return;
498 
499     // Upgrade list of variables attached to the CUs.
500     if (NamedMDNode *CUNodes = TheModule.getNamedMetadata("llvm.dbg.cu"))
501       for (unsigned I = 0, E = CUNodes->getNumOperands(); I != E; ++I) {
502         auto *CU = cast<DICompileUnit>(CUNodes->getOperand(I));
503         if (auto *GVs = dyn_cast_or_null<MDTuple>(CU->getRawGlobalVariables()))
504           for (unsigned I = 0; I < GVs->getNumOperands(); I++)
505             if (auto *GV =
506                     dyn_cast_or_null<DIGlobalVariable>(GVs->getOperand(I))) {
507               auto *DGVE = DIGlobalVariableExpression::getDistinct(
508                   Context, GV, DIExpression::get(Context, {}));
509               GVs->replaceOperandWith(I, DGVE);
510             }
511       }
512 
513     // Upgrade variables attached to globals.
514     for (auto &GV : TheModule.globals()) {
515       SmallVector<MDNode *, 1> MDs;
516       GV.getMetadata(LLVMContext::MD_dbg, MDs);
517       GV.eraseMetadata(LLVMContext::MD_dbg);
518       for (auto *MD : MDs)
519         if (auto *DGV = dyn_cast<DIGlobalVariable>(MD)) {
520           auto *DGVE = DIGlobalVariableExpression::getDistinct(
521               Context, DGV, DIExpression::get(Context, {}));
522           GV.addMetadata(LLVMContext::MD_dbg, *DGVE);
523         } else
524           GV.addMetadata(LLVMContext::MD_dbg, *MD);
525     }
526   }
527 
findEnclosingSubprogram(DILocalScope * S)528   DISubprogram *findEnclosingSubprogram(DILocalScope *S) {
529     if (!S)
530       return nullptr;
531     if (auto *SP = ParentSubprogram[S]) {
532       return SP;
533     }
534 
535     DILocalScope *InitialScope = S;
536     DenseSet<DILocalScope *> Visited;
537     while (S && !isa<DISubprogram>(S)) {
538       S = dyn_cast_or_null<DILocalScope>(S->getScope());
539       if (Visited.contains(S))
540         break;
541       Visited.insert(S);
542     }
543     ParentSubprogram[InitialScope] = llvm::dyn_cast_or_null<DISubprogram>(S);
544 
545     return ParentSubprogram[InitialScope];
546   }
547 
548   /// Move local imports from DICompileUnit's 'imports' field to
549   /// DISubprogram's retainedNodes.
upgradeCULocals()550   void upgradeCULocals() {
551     if (NamedMDNode *CUNodes = TheModule.getNamedMetadata("llvm.dbg.cu")) {
552       for (unsigned I = 0, E = CUNodes->getNumOperands(); I != E; ++I) {
553         auto *CU = dyn_cast<DICompileUnit>(CUNodes->getOperand(I));
554         if (!CU)
555           continue;
556 
557         if (CU->getRawImportedEntities()) {
558           // Collect a set of imported entities to be moved.
559           SetVector<Metadata *> EntitiesToRemove;
560           for (Metadata *Op : CU->getImportedEntities()->operands()) {
561             auto *IE = cast<DIImportedEntity>(Op);
562             if (dyn_cast_or_null<DILocalScope>(IE->getScope())) {
563               EntitiesToRemove.insert(IE);
564             }
565           }
566 
567           if (!EntitiesToRemove.empty()) {
568             // Make a new list of CU's 'imports'.
569             SmallVector<Metadata *> NewImports;
570             for (Metadata *Op : CU->getImportedEntities()->operands()) {
571               if (!EntitiesToRemove.contains(cast<DIImportedEntity>(Op))) {
572                 NewImports.push_back(Op);
573               }
574             }
575 
576             // Find DISubprogram corresponding to each entity.
577             std::map<DISubprogram *, SmallVector<Metadata *>> SPToEntities;
578             for (auto *I : EntitiesToRemove) {
579               auto *Entity = cast<DIImportedEntity>(I);
580               if (auto *SP = findEnclosingSubprogram(
581                       cast<DILocalScope>(Entity->getScope()))) {
582                 SPToEntities[SP].push_back(Entity);
583               }
584             }
585 
586             // Update DISubprograms' retainedNodes.
587             for (auto I = SPToEntities.begin(); I != SPToEntities.end(); ++I) {
588               auto *SP = I->first;
589               auto RetainedNodes = SP->getRetainedNodes();
590               SmallVector<Metadata *> MDs(RetainedNodes.begin(),
591                                           RetainedNodes.end());
592               MDs.append(I->second);
593               SP->replaceRetainedNodes(MDNode::get(Context, MDs));
594             }
595 
596             // Remove entities with local scope from CU.
597             CU->replaceImportedEntities(MDTuple::get(Context, NewImports));
598           }
599         }
600       }
601     }
602 
603     ParentSubprogram.clear();
604   }
605 
606   /// Remove a leading DW_OP_deref from DIExpressions in a dbg.declare that
607   /// describes a function argument.
upgradeDeclareExpressions(Function & F)608   void upgradeDeclareExpressions(Function &F) {
609     if (!NeedDeclareExpressionUpgrade)
610       return;
611 
612     for (auto &BB : F)
613       for (auto &I : BB)
614         if (auto *DDI = dyn_cast<DbgDeclareInst>(&I))
615           if (auto *DIExpr = DDI->getExpression())
616             if (DIExpr->startsWithDeref() &&
617                 isa_and_nonnull<Argument>(DDI->getAddress())) {
618               SmallVector<uint64_t, 8> Ops;
619               Ops.append(std::next(DIExpr->elements_begin()),
620                          DIExpr->elements_end());
621               DDI->setExpression(DIExpression::get(Context, Ops));
622             }
623   }
624 
625   /// Upgrade the expression from previous versions.
upgradeDIExpression(uint64_t FromVersion,MutableArrayRef<uint64_t> & Expr,SmallVectorImpl<uint64_t> & Buffer)626   Error upgradeDIExpression(uint64_t FromVersion,
627                             MutableArrayRef<uint64_t> &Expr,
628                             SmallVectorImpl<uint64_t> &Buffer) {
629     auto N = Expr.size();
630     switch (FromVersion) {
631     default:
632       return error("Invalid record");
633     case 0:
634       if (N >= 3 && Expr[N - 3] == dwarf::DW_OP_bit_piece)
635         Expr[N - 3] = dwarf::DW_OP_LLVM_fragment;
636       [[fallthrough]];
637     case 1:
638       // Move DW_OP_deref to the end.
639       if (N && Expr[0] == dwarf::DW_OP_deref) {
640         auto End = Expr.end();
641         if (Expr.size() >= 3 &&
642             *std::prev(End, 3) == dwarf::DW_OP_LLVM_fragment)
643           End = std::prev(End, 3);
644         std::move(std::next(Expr.begin()), End, Expr.begin());
645         *std::prev(End) = dwarf::DW_OP_deref;
646       }
647       NeedDeclareExpressionUpgrade = true;
648       [[fallthrough]];
649     case 2: {
650       // Change DW_OP_plus to DW_OP_plus_uconst.
651       // Change DW_OP_minus to DW_OP_uconst, DW_OP_minus
652       auto SubExpr = ArrayRef<uint64_t>(Expr);
653       while (!SubExpr.empty()) {
654         // Skip past other operators with their operands
655         // for this version of the IR, obtained from
656         // from historic DIExpression::ExprOperand::getSize().
657         size_t HistoricSize;
658         switch (SubExpr.front()) {
659         default:
660           HistoricSize = 1;
661           break;
662         case dwarf::DW_OP_constu:
663         case dwarf::DW_OP_minus:
664         case dwarf::DW_OP_plus:
665           HistoricSize = 2;
666           break;
667         case dwarf::DW_OP_LLVM_fragment:
668           HistoricSize = 3;
669           break;
670         }
671 
672         // If the expression is malformed, make sure we don't
673         // copy more elements than we should.
674         HistoricSize = std::min(SubExpr.size(), HistoricSize);
675         ArrayRef<uint64_t> Args = SubExpr.slice(1, HistoricSize - 1);
676 
677         switch (SubExpr.front()) {
678         case dwarf::DW_OP_plus:
679           Buffer.push_back(dwarf::DW_OP_plus_uconst);
680           Buffer.append(Args.begin(), Args.end());
681           break;
682         case dwarf::DW_OP_minus:
683           Buffer.push_back(dwarf::DW_OP_constu);
684           Buffer.append(Args.begin(), Args.end());
685           Buffer.push_back(dwarf::DW_OP_minus);
686           break;
687         default:
688           Buffer.push_back(*SubExpr.begin());
689           Buffer.append(Args.begin(), Args.end());
690           break;
691         }
692 
693         // Continue with remaining elements.
694         SubExpr = SubExpr.slice(HistoricSize);
695       }
696       Expr = MutableArrayRef<uint64_t>(Buffer);
697       [[fallthrough]];
698     }
699     case 3:
700       // Up-to-date!
701       break;
702     }
703 
704     return Error::success();
705   }
706 
upgradeDebugInfo(bool ModuleLevel)707   void upgradeDebugInfo(bool ModuleLevel) {
708     upgradeCUSubprograms();
709     upgradeCUVariables();
710     if (ModuleLevel)
711       upgradeCULocals();
712   }
713 
714   void callMDTypeCallback(Metadata **Val, unsigned TypeID);
715 
716 public:
MetadataLoaderImpl(BitstreamCursor & Stream,Module & TheModule,BitcodeReaderValueList & ValueList,MetadataLoaderCallbacks Callbacks,bool IsImporting)717   MetadataLoaderImpl(BitstreamCursor &Stream, Module &TheModule,
718                      BitcodeReaderValueList &ValueList,
719                      MetadataLoaderCallbacks Callbacks, bool IsImporting)
720       : MetadataList(TheModule.getContext(), Stream.SizeInBytes()),
721         ValueList(ValueList), Stream(Stream), Context(TheModule.getContext()),
722         TheModule(TheModule), Callbacks(std::move(Callbacks)),
723         IsImporting(IsImporting) {}
724 
725   Error parseMetadata(bool ModuleLevel);
726 
hasFwdRefs() const727   bool hasFwdRefs() const { return MetadataList.hasFwdRefs(); }
728 
getMetadataFwdRefOrLoad(unsigned ID)729   Metadata *getMetadataFwdRefOrLoad(unsigned ID) {
730     if (ID < MDStringRef.size())
731       return lazyLoadOneMDString(ID);
732     if (auto *MD = MetadataList.lookup(ID))
733       return MD;
734     // If lazy-loading is enabled, we try recursively to load the operand
735     // instead of creating a temporary.
736     if (ID < (MDStringRef.size() + GlobalMetadataBitPosIndex.size())) {
737       PlaceholderQueue Placeholders;
738       lazyLoadOneMetadata(ID, Placeholders);
739       resolveForwardRefsAndPlaceholders(Placeholders);
740       return MetadataList.lookup(ID);
741     }
742     return MetadataList.getMetadataFwdRef(ID);
743   }
744 
lookupSubprogramForFunction(Function * F)745   DISubprogram *lookupSubprogramForFunction(Function *F) {
746     return FunctionsWithSPs.lookup(F);
747   }
748 
hasSeenOldLoopTags() const749   bool hasSeenOldLoopTags() const { return HasSeenOldLoopTags; }
750 
751   Error parseMetadataAttachment(Function &F,
752                                 ArrayRef<Instruction *> InstructionList);
753 
754   Error parseMetadataKinds();
755 
setStripTBAA(bool Value)756   void setStripTBAA(bool Value) { StripTBAA = Value; }
isStrippingTBAA() const757   bool isStrippingTBAA() const { return StripTBAA; }
758 
size() const759   unsigned size() const { return MetadataList.size(); }
shrinkTo(unsigned N)760   void shrinkTo(unsigned N) { MetadataList.shrinkTo(N); }
upgradeDebugIntrinsics(Function & F)761   void upgradeDebugIntrinsics(Function &F) { upgradeDeclareExpressions(F); }
762 };
763 
764 Expected<bool>
lazyLoadModuleMetadataBlock()765 MetadataLoader::MetadataLoaderImpl::lazyLoadModuleMetadataBlock() {
766   IndexCursor = Stream;
767   SmallVector<uint64_t, 64> Record;
768   GlobalDeclAttachmentPos = 0;
769   // Get the abbrevs, and preload record positions to make them lazy-loadable.
770   while (true) {
771     uint64_t SavedPos = IndexCursor.GetCurrentBitNo();
772     BitstreamEntry Entry;
773     if (Error E =
774             IndexCursor
775                 .advanceSkippingSubblocks(BitstreamCursor::AF_DontPopBlockAtEnd)
776                 .moveInto(Entry))
777       return std::move(E);
778 
779     switch (Entry.Kind) {
780     case BitstreamEntry::SubBlock: // Handled for us already.
781     case BitstreamEntry::Error:
782       return error("Malformed block");
783     case BitstreamEntry::EndBlock: {
784       return true;
785     }
786     case BitstreamEntry::Record: {
787       // The interesting case.
788       ++NumMDRecordLoaded;
789       uint64_t CurrentPos = IndexCursor.GetCurrentBitNo();
790       unsigned Code;
791       if (Error E = IndexCursor.skipRecord(Entry.ID).moveInto(Code))
792         return std::move(E);
793       switch (Code) {
794       case bitc::METADATA_STRINGS: {
795         // Rewind and parse the strings.
796         if (Error Err = IndexCursor.JumpToBit(CurrentPos))
797           return std::move(Err);
798         StringRef Blob;
799         Record.clear();
800         if (Expected<unsigned> MaybeRecord =
801                 IndexCursor.readRecord(Entry.ID, Record, &Blob))
802           ;
803         else
804           return MaybeRecord.takeError();
805         unsigned NumStrings = Record[0];
806         MDStringRef.reserve(NumStrings);
807         auto IndexNextMDString = [&](StringRef Str) {
808           MDStringRef.push_back(Str);
809         };
810         if (auto Err = parseMetadataStrings(Record, Blob, IndexNextMDString))
811           return std::move(Err);
812         break;
813       }
814       case bitc::METADATA_INDEX_OFFSET: {
815         // This is the offset to the index, when we see this we skip all the
816         // records and load only an index to these.
817         if (Error Err = IndexCursor.JumpToBit(CurrentPos))
818           return std::move(Err);
819         Record.clear();
820         if (Expected<unsigned> MaybeRecord =
821                 IndexCursor.readRecord(Entry.ID, Record))
822           ;
823         else
824           return MaybeRecord.takeError();
825         if (Record.size() != 2)
826           return error("Invalid record");
827         auto Offset = Record[0] + (Record[1] << 32);
828         auto BeginPos = IndexCursor.GetCurrentBitNo();
829         if (Error Err = IndexCursor.JumpToBit(BeginPos + Offset))
830           return std::move(Err);
831         Expected<BitstreamEntry> MaybeEntry =
832             IndexCursor.advanceSkippingSubblocks(
833                 BitstreamCursor::AF_DontPopBlockAtEnd);
834         if (!MaybeEntry)
835           return MaybeEntry.takeError();
836         Entry = MaybeEntry.get();
837         assert(Entry.Kind == BitstreamEntry::Record &&
838                "Corrupted bitcode: Expected `Record` when trying to find the "
839                "Metadata index");
840         Record.clear();
841         if (Expected<unsigned> MaybeCode =
842                 IndexCursor.readRecord(Entry.ID, Record))
843           assert(MaybeCode.get() == bitc::METADATA_INDEX &&
844                  "Corrupted bitcode: Expected `METADATA_INDEX` when trying to "
845                  "find the Metadata index");
846         else
847           return MaybeCode.takeError();
848         // Delta unpack
849         auto CurrentValue = BeginPos;
850         GlobalMetadataBitPosIndex.reserve(Record.size());
851         for (auto &Elt : Record) {
852           CurrentValue += Elt;
853           GlobalMetadataBitPosIndex.push_back(CurrentValue);
854         }
855         break;
856       }
857       case bitc::METADATA_INDEX:
858         // We don't expect to get there, the Index is loaded when we encounter
859         // the offset.
860         return error("Corrupted Metadata block");
861       case bitc::METADATA_NAME: {
862         // Named metadata need to be materialized now and aren't deferred.
863         if (Error Err = IndexCursor.JumpToBit(CurrentPos))
864           return std::move(Err);
865         Record.clear();
866 
867         unsigned Code;
868         if (Expected<unsigned> MaybeCode =
869                 IndexCursor.readRecord(Entry.ID, Record)) {
870           Code = MaybeCode.get();
871           assert(Code == bitc::METADATA_NAME);
872         } else
873           return MaybeCode.takeError();
874 
875         // Read name of the named metadata.
876         SmallString<8> Name(Record.begin(), Record.end());
877         if (Expected<unsigned> MaybeCode = IndexCursor.ReadCode())
878           Code = MaybeCode.get();
879         else
880           return MaybeCode.takeError();
881 
882         // Named Metadata comes in two parts, we expect the name to be followed
883         // by the node
884         Record.clear();
885         if (Expected<unsigned> MaybeNextBitCode =
886                 IndexCursor.readRecord(Code, Record))
887           assert(MaybeNextBitCode.get() == bitc::METADATA_NAMED_NODE);
888         else
889           return MaybeNextBitCode.takeError();
890 
891         // Read named metadata elements.
892         unsigned Size = Record.size();
893         NamedMDNode *NMD = TheModule.getOrInsertNamedMetadata(Name);
894         for (unsigned i = 0; i != Size; ++i) {
895           // FIXME: We could use a placeholder here, however NamedMDNode are
896           // taking MDNode as operand and not using the Metadata infrastructure.
897           // It is acknowledged by 'TODO: Inherit from Metadata' in the
898           // NamedMDNode class definition.
899           MDNode *MD = MetadataList.getMDNodeFwdRefOrNull(Record[i]);
900           assert(MD && "Invalid metadata: expect fwd ref to MDNode");
901           NMD->addOperand(MD);
902         }
903         break;
904       }
905       case bitc::METADATA_GLOBAL_DECL_ATTACHMENT: {
906         if (!GlobalDeclAttachmentPos)
907           GlobalDeclAttachmentPos = SavedPos;
908 #ifndef NDEBUG
909         NumGlobalDeclAttachSkipped++;
910 #endif
911         break;
912       }
913       case bitc::METADATA_KIND:
914       case bitc::METADATA_STRING_OLD:
915       case bitc::METADATA_OLD_FN_NODE:
916       case bitc::METADATA_OLD_NODE:
917       case bitc::METADATA_VALUE:
918       case bitc::METADATA_DISTINCT_NODE:
919       case bitc::METADATA_NODE:
920       case bitc::METADATA_LOCATION:
921       case bitc::METADATA_GENERIC_DEBUG:
922       case bitc::METADATA_SUBRANGE:
923       case bitc::METADATA_ENUMERATOR:
924       case bitc::METADATA_BASIC_TYPE:
925       case bitc::METADATA_STRING_TYPE:
926       case bitc::METADATA_DERIVED_TYPE:
927       case bitc::METADATA_COMPOSITE_TYPE:
928       case bitc::METADATA_SUBROUTINE_TYPE:
929       case bitc::METADATA_MODULE:
930       case bitc::METADATA_FILE:
931       case bitc::METADATA_COMPILE_UNIT:
932       case bitc::METADATA_SUBPROGRAM:
933       case bitc::METADATA_LEXICAL_BLOCK:
934       case bitc::METADATA_LEXICAL_BLOCK_FILE:
935       case bitc::METADATA_NAMESPACE:
936       case bitc::METADATA_COMMON_BLOCK:
937       case bitc::METADATA_MACRO:
938       case bitc::METADATA_MACRO_FILE:
939       case bitc::METADATA_TEMPLATE_TYPE:
940       case bitc::METADATA_TEMPLATE_VALUE:
941       case bitc::METADATA_GLOBAL_VAR:
942       case bitc::METADATA_LOCAL_VAR:
943       case bitc::METADATA_ASSIGN_ID:
944       case bitc::METADATA_LABEL:
945       case bitc::METADATA_EXPRESSION:
946       case bitc::METADATA_OBJC_PROPERTY:
947       case bitc::METADATA_IMPORTED_ENTITY:
948       case bitc::METADATA_GLOBAL_VAR_EXPR:
949       case bitc::METADATA_GENERIC_SUBRANGE:
950         // We don't expect to see any of these, if we see one, give up on
951         // lazy-loading and fallback.
952         MDStringRef.clear();
953         GlobalMetadataBitPosIndex.clear();
954         return false;
955       }
956       break;
957     }
958     }
959   }
960 }
961 
962 // Load the global decl attachments after building the lazy loading index.
963 // We don't load them "lazily" - all global decl attachments must be
964 // parsed since they aren't materialized on demand. However, by delaying
965 // their parsing until after the index is created, we can use the index
966 // instead of creating temporaries.
loadGlobalDeclAttachments()967 Expected<bool> MetadataLoader::MetadataLoaderImpl::loadGlobalDeclAttachments() {
968   // Nothing to do if we didn't find any of these metadata records.
969   if (!GlobalDeclAttachmentPos)
970     return true;
971   // Use a temporary cursor so that we don't mess up the main Stream cursor or
972   // the lazy loading IndexCursor (which holds the necessary abbrev ids).
973   BitstreamCursor TempCursor = Stream;
974   SmallVector<uint64_t, 64> Record;
975   // Jump to the position before the first global decl attachment, so we can
976   // scan for the first BitstreamEntry record.
977   if (Error Err = TempCursor.JumpToBit(GlobalDeclAttachmentPos))
978     return std::move(Err);
979   while (true) {
980     BitstreamEntry Entry;
981     if (Error E =
982             TempCursor
983                 .advanceSkippingSubblocks(BitstreamCursor::AF_DontPopBlockAtEnd)
984                 .moveInto(Entry))
985       return std::move(E);
986 
987     switch (Entry.Kind) {
988     case BitstreamEntry::SubBlock: // Handled for us already.
989     case BitstreamEntry::Error:
990       return error("Malformed block");
991     case BitstreamEntry::EndBlock:
992       // Check that we parsed them all.
993       assert(NumGlobalDeclAttachSkipped == NumGlobalDeclAttachParsed);
994       return true;
995     case BitstreamEntry::Record:
996       break;
997     }
998     uint64_t CurrentPos = TempCursor.GetCurrentBitNo();
999     Expected<unsigned> MaybeCode = TempCursor.skipRecord(Entry.ID);
1000     if (!MaybeCode)
1001       return MaybeCode.takeError();
1002     if (MaybeCode.get() != bitc::METADATA_GLOBAL_DECL_ATTACHMENT) {
1003       // Anything other than a global decl attachment signals the end of
1004       // these records. Check that we parsed them all.
1005       assert(NumGlobalDeclAttachSkipped == NumGlobalDeclAttachParsed);
1006       return true;
1007     }
1008 #ifndef NDEBUG
1009     NumGlobalDeclAttachParsed++;
1010 #endif
1011     // FIXME: we need to do this early because we don't materialize global
1012     // value explicitly.
1013     if (Error Err = TempCursor.JumpToBit(CurrentPos))
1014       return std::move(Err);
1015     Record.clear();
1016     if (Expected<unsigned> MaybeRecord =
1017             TempCursor.readRecord(Entry.ID, Record))
1018       ;
1019     else
1020       return MaybeRecord.takeError();
1021     if (Record.size() % 2 == 0)
1022       return error("Invalid record");
1023     unsigned ValueID = Record[0];
1024     if (ValueID >= ValueList.size())
1025       return error("Invalid record");
1026     if (auto *GO = dyn_cast<GlobalObject>(ValueList[ValueID])) {
1027       // Need to save and restore the current position since
1028       // parseGlobalObjectAttachment will resolve all forward references which
1029       // would require parsing from locations stored in the index.
1030       CurrentPos = TempCursor.GetCurrentBitNo();
1031       if (Error Err = parseGlobalObjectAttachment(
1032               *GO, ArrayRef<uint64_t>(Record).slice(1)))
1033         return std::move(Err);
1034       if (Error Err = TempCursor.JumpToBit(CurrentPos))
1035         return std::move(Err);
1036     }
1037   }
1038 }
1039 
callMDTypeCallback(Metadata ** Val,unsigned TypeID)1040 void MetadataLoader::MetadataLoaderImpl::callMDTypeCallback(Metadata **Val,
1041                                                             unsigned TypeID) {
1042   if (Callbacks.MDType) {
1043     (*Callbacks.MDType)(Val, TypeID, Callbacks.GetTypeByID,
1044                         Callbacks.GetContainedTypeID);
1045   }
1046 }
1047 
1048 /// Parse a METADATA_BLOCK. If ModuleLevel is true then we are parsing
1049 /// module level metadata.
parseMetadata(bool ModuleLevel)1050 Error MetadataLoader::MetadataLoaderImpl::parseMetadata(bool ModuleLevel) {
1051   if (!ModuleLevel && MetadataList.hasFwdRefs())
1052     return error("Invalid metadata: fwd refs into function blocks");
1053 
1054   // Record the entry position so that we can jump back here and efficiently
1055   // skip the whole block in case we lazy-load.
1056   auto EntryPos = Stream.GetCurrentBitNo();
1057 
1058   if (Error Err = Stream.EnterSubBlock(bitc::METADATA_BLOCK_ID))
1059     return Err;
1060 
1061   SmallVector<uint64_t, 64> Record;
1062   PlaceholderQueue Placeholders;
1063 
1064   // We lazy-load module-level metadata: we build an index for each record, and
1065   // then load individual record as needed, starting with the named metadata.
1066   if (ModuleLevel && IsImporting && MetadataList.empty() &&
1067       !DisableLazyLoading) {
1068     auto SuccessOrErr = lazyLoadModuleMetadataBlock();
1069     if (!SuccessOrErr)
1070       return SuccessOrErr.takeError();
1071     if (SuccessOrErr.get()) {
1072       // An index was successfully created and we will be able to load metadata
1073       // on-demand.
1074       MetadataList.resize(MDStringRef.size() +
1075                           GlobalMetadataBitPosIndex.size());
1076 
1077       // Now that we have built the index, load the global decl attachments
1078       // that were deferred during that process. This avoids creating
1079       // temporaries.
1080       SuccessOrErr = loadGlobalDeclAttachments();
1081       if (!SuccessOrErr)
1082         return SuccessOrErr.takeError();
1083       assert(SuccessOrErr.get());
1084 
1085       // Reading the named metadata created forward references and/or
1086       // placeholders, that we flush here.
1087       resolveForwardRefsAndPlaceholders(Placeholders);
1088       upgradeDebugInfo(ModuleLevel);
1089       // Return at the beginning of the block, since it is easy to skip it
1090       // entirely from there.
1091       Stream.ReadBlockEnd(); // Pop the abbrev block context.
1092       if (Error Err = IndexCursor.JumpToBit(EntryPos))
1093         return Err;
1094       if (Error Err = Stream.SkipBlock()) {
1095         // FIXME this drops the error on the floor, which
1096         // ThinLTO/X86/debuginfo-cu-import.ll relies on.
1097         consumeError(std::move(Err));
1098         return Error::success();
1099       }
1100       return Error::success();
1101     }
1102     // Couldn't load an index, fallback to loading all the block "old-style".
1103   }
1104 
1105   unsigned NextMetadataNo = MetadataList.size();
1106 
1107   // Read all the records.
1108   while (true) {
1109     BitstreamEntry Entry;
1110     if (Error E = Stream.advanceSkippingSubblocks().moveInto(Entry))
1111       return E;
1112 
1113     switch (Entry.Kind) {
1114     case BitstreamEntry::SubBlock: // Handled for us already.
1115     case BitstreamEntry::Error:
1116       return error("Malformed block");
1117     case BitstreamEntry::EndBlock:
1118       resolveForwardRefsAndPlaceholders(Placeholders);
1119       upgradeDebugInfo(ModuleLevel);
1120       return Error::success();
1121     case BitstreamEntry::Record:
1122       // The interesting case.
1123       break;
1124     }
1125 
1126     // Read a record.
1127     Record.clear();
1128     StringRef Blob;
1129     ++NumMDRecordLoaded;
1130     if (Expected<unsigned> MaybeCode =
1131             Stream.readRecord(Entry.ID, Record, &Blob)) {
1132       if (Error Err = parseOneMetadata(Record, MaybeCode.get(), Placeholders,
1133                                        Blob, NextMetadataNo))
1134         return Err;
1135     } else
1136       return MaybeCode.takeError();
1137   }
1138 }
1139 
lazyLoadOneMDString(unsigned ID)1140 MDString *MetadataLoader::MetadataLoaderImpl::lazyLoadOneMDString(unsigned ID) {
1141   ++NumMDStringLoaded;
1142   if (Metadata *MD = MetadataList.lookup(ID))
1143     return cast<MDString>(MD);
1144   auto MDS = MDString::get(Context, MDStringRef[ID]);
1145   MetadataList.assignValue(MDS, ID);
1146   return MDS;
1147 }
1148 
lazyLoadOneMetadata(unsigned ID,PlaceholderQueue & Placeholders)1149 void MetadataLoader::MetadataLoaderImpl::lazyLoadOneMetadata(
1150     unsigned ID, PlaceholderQueue &Placeholders) {
1151   assert(ID < (MDStringRef.size()) + GlobalMetadataBitPosIndex.size());
1152   assert(ID >= MDStringRef.size() && "Unexpected lazy-loading of MDString");
1153   // Lookup first if the metadata hasn't already been loaded.
1154   if (auto *MD = MetadataList.lookup(ID)) {
1155     auto *N = cast<MDNode>(MD);
1156     if (!N->isTemporary())
1157       return;
1158   }
1159   SmallVector<uint64_t, 64> Record;
1160   StringRef Blob;
1161   if (Error Err = IndexCursor.JumpToBit(
1162           GlobalMetadataBitPosIndex[ID - MDStringRef.size()]))
1163     report_fatal_error("lazyLoadOneMetadata failed jumping: " +
1164                        Twine(toString(std::move(Err))));
1165   BitstreamEntry Entry;
1166   if (Error E = IndexCursor.advanceSkippingSubblocks().moveInto(Entry))
1167     // FIXME this drops the error on the floor.
1168     report_fatal_error("lazyLoadOneMetadata failed advanceSkippingSubblocks: " +
1169                        Twine(toString(std::move(E))));
1170   ++NumMDRecordLoaded;
1171   if (Expected<unsigned> MaybeCode =
1172           IndexCursor.readRecord(Entry.ID, Record, &Blob)) {
1173     if (Error Err =
1174             parseOneMetadata(Record, MaybeCode.get(), Placeholders, Blob, ID))
1175       report_fatal_error("Can't lazyload MD, parseOneMetadata: " +
1176                          Twine(toString(std::move(Err))));
1177   } else
1178     report_fatal_error("Can't lazyload MD: " +
1179                        Twine(toString(MaybeCode.takeError())));
1180 }
1181 
1182 /// Ensure that all forward-references and placeholders are resolved.
1183 /// Iteratively lazy-loading metadata on-demand if needed.
resolveForwardRefsAndPlaceholders(PlaceholderQueue & Placeholders)1184 void MetadataLoader::MetadataLoaderImpl::resolveForwardRefsAndPlaceholders(
1185     PlaceholderQueue &Placeholders) {
1186   DenseSet<unsigned> Temporaries;
1187   while (true) {
1188     // Populate Temporaries with the placeholders that haven't been loaded yet.
1189     Placeholders.getTemporaries(MetadataList, Temporaries);
1190 
1191     // If we don't have any temporary, or FwdReference, we're done!
1192     if (Temporaries.empty() && !MetadataList.hasFwdRefs())
1193       break;
1194 
1195     // First, load all the temporaries. This can add new placeholders or
1196     // forward references.
1197     for (auto ID : Temporaries)
1198       lazyLoadOneMetadata(ID, Placeholders);
1199     Temporaries.clear();
1200 
1201     // Second, load the forward-references. This can also add new placeholders
1202     // or forward references.
1203     while (MetadataList.hasFwdRefs())
1204       lazyLoadOneMetadata(MetadataList.getNextFwdRef(), Placeholders);
1205   }
1206   // At this point we don't have any forward reference remaining, or temporary
1207   // that haven't been loaded. We can safely drop RAUW support and mark cycles
1208   // as resolved.
1209   MetadataList.tryToResolveCycles();
1210 
1211   // Finally, everything is in place, we can replace the placeholders operands
1212   // with the final node they refer to.
1213   Placeholders.flush(MetadataList);
1214 }
1215 
getValueFwdRef(BitcodeReaderValueList & ValueList,unsigned Idx,Type * Ty,unsigned TyID)1216 static Value *getValueFwdRef(BitcodeReaderValueList &ValueList, unsigned Idx,
1217                              Type *Ty, unsigned TyID) {
1218   Value *V = ValueList.getValueFwdRef(Idx, Ty, TyID,
1219                                       /*ConstExprInsertBB*/ nullptr);
1220   if (V)
1221     return V;
1222 
1223   // This is a reference to a no longer supported constant expression.
1224   // Pretend that the constant was deleted, which will replace metadata
1225   // references with undef.
1226   // TODO: This is a rather indirect check. It would be more elegant to use
1227   // a separate ErrorInfo for constant materialization failure and thread
1228   // the error reporting through getValueFwdRef().
1229   if (Idx < ValueList.size() && ValueList[Idx] &&
1230       ValueList[Idx]->getType() == Ty)
1231     return UndefValue::get(Ty);
1232 
1233   return nullptr;
1234 }
1235 
parseOneMetadata(SmallVectorImpl<uint64_t> & Record,unsigned Code,PlaceholderQueue & Placeholders,StringRef Blob,unsigned & NextMetadataNo)1236 Error MetadataLoader::MetadataLoaderImpl::parseOneMetadata(
1237     SmallVectorImpl<uint64_t> &Record, unsigned Code,
1238     PlaceholderQueue &Placeholders, StringRef Blob, unsigned &NextMetadataNo) {
1239 
1240   bool IsDistinct = false;
1241   auto getMD = [&](unsigned ID) -> Metadata * {
1242     if (ID < MDStringRef.size())
1243       return lazyLoadOneMDString(ID);
1244     if (!IsDistinct) {
1245       if (auto *MD = MetadataList.lookup(ID))
1246         return MD;
1247       // If lazy-loading is enabled, we try recursively to load the operand
1248       // instead of creating a temporary.
1249       if (ID < (MDStringRef.size() + GlobalMetadataBitPosIndex.size())) {
1250         // Create a temporary for the node that is referencing the operand we
1251         // will lazy-load. It is needed before recursing in case there are
1252         // uniquing cycles.
1253         MetadataList.getMetadataFwdRef(NextMetadataNo);
1254         lazyLoadOneMetadata(ID, Placeholders);
1255         return MetadataList.lookup(ID);
1256       }
1257       // Return a temporary.
1258       return MetadataList.getMetadataFwdRef(ID);
1259     }
1260     if (auto *MD = MetadataList.getMetadataIfResolved(ID))
1261       return MD;
1262     return &Placeholders.getPlaceholderOp(ID);
1263   };
1264   auto getMDOrNull = [&](unsigned ID) -> Metadata * {
1265     if (ID)
1266       return getMD(ID - 1);
1267     return nullptr;
1268   };
1269   auto getMDOrNullWithoutPlaceholders = [&](unsigned ID) -> Metadata * {
1270     if (ID)
1271       return MetadataList.getMetadataFwdRef(ID - 1);
1272     return nullptr;
1273   };
1274   auto getMDString = [&](unsigned ID) -> MDString * {
1275     // This requires that the ID is not really a forward reference.  In
1276     // particular, the MDString must already have been resolved.
1277     auto MDS = getMDOrNull(ID);
1278     return cast_or_null<MDString>(MDS);
1279   };
1280 
1281   // Support for old type refs.
1282   auto getDITypeRefOrNull = [&](unsigned ID) {
1283     return MetadataList.upgradeTypeRef(getMDOrNull(ID));
1284   };
1285 
1286 #define GET_OR_DISTINCT(CLASS, ARGS)                                           \
1287   (IsDistinct ? CLASS::getDistinct ARGS : CLASS::get ARGS)
1288 
1289   switch (Code) {
1290   default: // Default behavior: ignore.
1291     break;
1292   case bitc::METADATA_NAME: {
1293     // Read name of the named metadata.
1294     SmallString<8> Name(Record.begin(), Record.end());
1295     Record.clear();
1296     if (Error E = Stream.ReadCode().moveInto(Code))
1297       return E;
1298 
1299     ++NumMDRecordLoaded;
1300     if (Expected<unsigned> MaybeNextBitCode = Stream.readRecord(Code, Record)) {
1301       if (MaybeNextBitCode.get() != bitc::METADATA_NAMED_NODE)
1302         return error("METADATA_NAME not followed by METADATA_NAMED_NODE");
1303     } else
1304       return MaybeNextBitCode.takeError();
1305 
1306     // Read named metadata elements.
1307     unsigned Size = Record.size();
1308     NamedMDNode *NMD = TheModule.getOrInsertNamedMetadata(Name);
1309     for (unsigned i = 0; i != Size; ++i) {
1310       MDNode *MD = MetadataList.getMDNodeFwdRefOrNull(Record[i]);
1311       if (!MD)
1312         return error("Invalid named metadata: expect fwd ref to MDNode");
1313       NMD->addOperand(MD);
1314     }
1315     break;
1316   }
1317   case bitc::METADATA_OLD_FN_NODE: {
1318     // Deprecated, but still needed to read old bitcode files.
1319     // This is a LocalAsMetadata record, the only type of function-local
1320     // metadata.
1321     if (Record.size() % 2 == 1)
1322       return error("Invalid record");
1323 
1324     // If this isn't a LocalAsMetadata record, we're dropping it.  This used
1325     // to be legal, but there's no upgrade path.
1326     auto dropRecord = [&] {
1327       MetadataList.assignValue(MDNode::get(Context, std::nullopt),
1328                                NextMetadataNo);
1329       NextMetadataNo++;
1330     };
1331     if (Record.size() != 2) {
1332       dropRecord();
1333       break;
1334     }
1335 
1336     unsigned TyID = Record[0];
1337     Type *Ty = Callbacks.GetTypeByID(TyID);
1338     if (!Ty || Ty->isMetadataTy() || Ty->isVoidTy()) {
1339       dropRecord();
1340       break;
1341     }
1342 
1343     Value *V = ValueList.getValueFwdRef(Record[1], Ty, TyID,
1344                                         /*ConstExprInsertBB*/ nullptr);
1345     if (!V)
1346       return error("Invalid value reference from old fn metadata");
1347 
1348     MetadataList.assignValue(LocalAsMetadata::get(V), NextMetadataNo);
1349     NextMetadataNo++;
1350     break;
1351   }
1352   case bitc::METADATA_OLD_NODE: {
1353     // Deprecated, but still needed to read old bitcode files.
1354     if (Record.size() % 2 == 1)
1355       return error("Invalid record");
1356 
1357     unsigned Size = Record.size();
1358     SmallVector<Metadata *, 8> Elts;
1359     for (unsigned i = 0; i != Size; i += 2) {
1360       unsigned TyID = Record[i];
1361       Type *Ty = Callbacks.GetTypeByID(TyID);
1362       if (!Ty)
1363         return error("Invalid record");
1364       if (Ty->isMetadataTy())
1365         Elts.push_back(getMD(Record[i + 1]));
1366       else if (!Ty->isVoidTy()) {
1367         Value *V = getValueFwdRef(ValueList, Record[i + 1], Ty, TyID);
1368         if (!V)
1369           return error("Invalid value reference from old metadata");
1370         Metadata *MD = ValueAsMetadata::get(V);
1371         assert(isa<ConstantAsMetadata>(MD) &&
1372                "Expected non-function-local metadata");
1373         callMDTypeCallback(&MD, TyID);
1374         Elts.push_back(MD);
1375       } else
1376         Elts.push_back(nullptr);
1377     }
1378     MetadataList.assignValue(MDNode::get(Context, Elts), NextMetadataNo);
1379     NextMetadataNo++;
1380     break;
1381   }
1382   case bitc::METADATA_VALUE: {
1383     if (Record.size() != 2)
1384       return error("Invalid record");
1385 
1386     unsigned TyID = Record[0];
1387     Type *Ty = Callbacks.GetTypeByID(TyID);
1388     if (!Ty || Ty->isMetadataTy() || Ty->isVoidTy())
1389       return error("Invalid record");
1390 
1391     Value *V = getValueFwdRef(ValueList, Record[1], Ty, TyID);
1392     if (!V)
1393       return error("Invalid value reference from metadata");
1394 
1395     Metadata *MD = ValueAsMetadata::get(V);
1396     callMDTypeCallback(&MD, TyID);
1397     MetadataList.assignValue(MD, NextMetadataNo);
1398     NextMetadataNo++;
1399     break;
1400   }
1401   case bitc::METADATA_DISTINCT_NODE:
1402     IsDistinct = true;
1403     [[fallthrough]];
1404   case bitc::METADATA_NODE: {
1405     SmallVector<Metadata *, 8> Elts;
1406     Elts.reserve(Record.size());
1407     for (unsigned ID : Record)
1408       Elts.push_back(getMDOrNull(ID));
1409     MetadataList.assignValue(IsDistinct ? MDNode::getDistinct(Context, Elts)
1410                                         : MDNode::get(Context, Elts),
1411                              NextMetadataNo);
1412     NextMetadataNo++;
1413     break;
1414   }
1415   case bitc::METADATA_LOCATION: {
1416     if (Record.size() != 5 && Record.size() != 6)
1417       return error("Invalid record");
1418 
1419     IsDistinct = Record[0];
1420     unsigned Line = Record[1];
1421     unsigned Column = Record[2];
1422     Metadata *Scope = getMD(Record[3]);
1423     Metadata *InlinedAt = getMDOrNull(Record[4]);
1424     bool ImplicitCode = Record.size() == 6 && Record[5];
1425     MetadataList.assignValue(
1426         GET_OR_DISTINCT(DILocation, (Context, Line, Column, Scope, InlinedAt,
1427                                      ImplicitCode)),
1428         NextMetadataNo);
1429     NextMetadataNo++;
1430     break;
1431   }
1432   case bitc::METADATA_GENERIC_DEBUG: {
1433     if (Record.size() < 4)
1434       return error("Invalid record");
1435 
1436     IsDistinct = Record[0];
1437     unsigned Tag = Record[1];
1438     unsigned Version = Record[2];
1439 
1440     if (Tag >= 1u << 16 || Version != 0)
1441       return error("Invalid record");
1442 
1443     auto *Header = getMDString(Record[3]);
1444     SmallVector<Metadata *, 8> DwarfOps;
1445     for (unsigned I = 4, E = Record.size(); I != E; ++I)
1446       DwarfOps.push_back(getMDOrNull(Record[I]));
1447     MetadataList.assignValue(
1448         GET_OR_DISTINCT(GenericDINode, (Context, Tag, Header, DwarfOps)),
1449         NextMetadataNo);
1450     NextMetadataNo++;
1451     break;
1452   }
1453   case bitc::METADATA_SUBRANGE: {
1454     Metadata *Val = nullptr;
1455     // Operand 'count' is interpreted as:
1456     // - Signed integer (version 0)
1457     // - Metadata node  (version 1)
1458     // Operand 'lowerBound' is interpreted as:
1459     // - Signed integer (version 0 and 1)
1460     // - Metadata node  (version 2)
1461     // Operands 'upperBound' and 'stride' are interpreted as:
1462     // - Metadata node  (version 2)
1463     switch (Record[0] >> 1) {
1464     case 0:
1465       Val = GET_OR_DISTINCT(DISubrange,
1466                             (Context, Record[1], unrotateSign(Record[2])));
1467       break;
1468     case 1:
1469       Val = GET_OR_DISTINCT(DISubrange, (Context, getMDOrNull(Record[1]),
1470                                          unrotateSign(Record[2])));
1471       break;
1472     case 2:
1473       Val = GET_OR_DISTINCT(
1474           DISubrange, (Context, getMDOrNull(Record[1]), getMDOrNull(Record[2]),
1475                        getMDOrNull(Record[3]), getMDOrNull(Record[4])));
1476       break;
1477     default:
1478       return error("Invalid record: Unsupported version of DISubrange");
1479     }
1480 
1481     MetadataList.assignValue(Val, NextMetadataNo);
1482     IsDistinct = Record[0] & 1;
1483     NextMetadataNo++;
1484     break;
1485   }
1486   case bitc::METADATA_GENERIC_SUBRANGE: {
1487     Metadata *Val = nullptr;
1488     Val = GET_OR_DISTINCT(DIGenericSubrange,
1489                           (Context, getMDOrNull(Record[1]),
1490                            getMDOrNull(Record[2]), getMDOrNull(Record[3]),
1491                            getMDOrNull(Record[4])));
1492 
1493     MetadataList.assignValue(Val, NextMetadataNo);
1494     IsDistinct = Record[0] & 1;
1495     NextMetadataNo++;
1496     break;
1497   }
1498   case bitc::METADATA_ENUMERATOR: {
1499     if (Record.size() < 3)
1500       return error("Invalid record");
1501 
1502     IsDistinct = Record[0] & 1;
1503     bool IsUnsigned = Record[0] & 2;
1504     bool IsBigInt = Record[0] & 4;
1505     APInt Value;
1506 
1507     if (IsBigInt) {
1508       const uint64_t BitWidth = Record[1];
1509       const size_t NumWords = Record.size() - 3;
1510       Value = readWideAPInt(ArrayRef(&Record[3], NumWords), BitWidth);
1511     } else
1512       Value = APInt(64, unrotateSign(Record[1]), !IsUnsigned);
1513 
1514     MetadataList.assignValue(
1515         GET_OR_DISTINCT(DIEnumerator,
1516                         (Context, Value, IsUnsigned, getMDString(Record[2]))),
1517         NextMetadataNo);
1518     NextMetadataNo++;
1519     break;
1520   }
1521   case bitc::METADATA_BASIC_TYPE: {
1522     if (Record.size() < 6 || Record.size() > 7)
1523       return error("Invalid record");
1524 
1525     IsDistinct = Record[0];
1526     DINode::DIFlags Flags = (Record.size() > 6)
1527                                 ? static_cast<DINode::DIFlags>(Record[6])
1528                                 : DINode::FlagZero;
1529 
1530     MetadataList.assignValue(
1531         GET_OR_DISTINCT(DIBasicType,
1532                         (Context, Record[1], getMDString(Record[2]), Record[3],
1533                          Record[4], Record[5], Flags)),
1534         NextMetadataNo);
1535     NextMetadataNo++;
1536     break;
1537   }
1538   case bitc::METADATA_STRING_TYPE: {
1539     if (Record.size() > 9 || Record.size() < 8)
1540       return error("Invalid record");
1541 
1542     IsDistinct = Record[0];
1543     bool SizeIs8 = Record.size() == 8;
1544     // StringLocationExp (i.e. Record[5]) is added at a later time
1545     // than the other fields. The code here enables backward compatibility.
1546     Metadata *StringLocationExp = SizeIs8 ? nullptr : getMDOrNull(Record[5]);
1547     unsigned Offset = SizeIs8 ? 5 : 6;
1548     MetadataList.assignValue(
1549         GET_OR_DISTINCT(DIStringType,
1550                         (Context, Record[1], getMDString(Record[2]),
1551                          getMDOrNull(Record[3]), getMDOrNull(Record[4]),
1552                          StringLocationExp, Record[Offset], Record[Offset + 1],
1553                          Record[Offset + 2])),
1554         NextMetadataNo);
1555     NextMetadataNo++;
1556     break;
1557   }
1558   case bitc::METADATA_DERIVED_TYPE: {
1559     if (Record.size() < 12 || Record.size() > 14)
1560       return error("Invalid record");
1561 
1562     // DWARF address space is encoded as N->getDWARFAddressSpace() + 1. 0 means
1563     // that there is no DWARF address space associated with DIDerivedType.
1564     std::optional<unsigned> DWARFAddressSpace;
1565     if (Record.size() > 12 && Record[12])
1566       DWARFAddressSpace = Record[12] - 1;
1567 
1568     Metadata *Annotations = nullptr;
1569     if (Record.size() > 13 && Record[13])
1570       Annotations = getMDOrNull(Record[13]);
1571 
1572     IsDistinct = Record[0];
1573     DINode::DIFlags Flags = static_cast<DINode::DIFlags>(Record[10]);
1574     MetadataList.assignValue(
1575         GET_OR_DISTINCT(DIDerivedType,
1576                         (Context, Record[1], getMDString(Record[2]),
1577                          getMDOrNull(Record[3]), Record[4],
1578                          getDITypeRefOrNull(Record[5]),
1579                          getDITypeRefOrNull(Record[6]), Record[7], Record[8],
1580                          Record[9], DWARFAddressSpace, Flags,
1581                          getDITypeRefOrNull(Record[11]), Annotations)),
1582         NextMetadataNo);
1583     NextMetadataNo++;
1584     break;
1585   }
1586   case bitc::METADATA_COMPOSITE_TYPE: {
1587     if (Record.size() < 16 || Record.size() > 22)
1588       return error("Invalid record");
1589 
1590     // If we have a UUID and this is not a forward declaration, lookup the
1591     // mapping.
1592     IsDistinct = Record[0] & 0x1;
1593     bool IsNotUsedInTypeRef = Record[0] >= 2;
1594     unsigned Tag = Record[1];
1595     MDString *Name = getMDString(Record[2]);
1596     Metadata *File = getMDOrNull(Record[3]);
1597     unsigned Line = Record[4];
1598     Metadata *Scope = getDITypeRefOrNull(Record[5]);
1599     Metadata *BaseType = nullptr;
1600     uint64_t SizeInBits = Record[7];
1601     if (Record[8] > (uint64_t)std::numeric_limits<uint32_t>::max())
1602       return error("Alignment value is too large");
1603     uint32_t AlignInBits = Record[8];
1604     uint64_t OffsetInBits = 0;
1605     DINode::DIFlags Flags = static_cast<DINode::DIFlags>(Record[10]);
1606     Metadata *Elements = nullptr;
1607     unsigned RuntimeLang = Record[12];
1608     Metadata *VTableHolder = nullptr;
1609     Metadata *TemplateParams = nullptr;
1610     Metadata *Discriminator = nullptr;
1611     Metadata *DataLocation = nullptr;
1612     Metadata *Associated = nullptr;
1613     Metadata *Allocated = nullptr;
1614     Metadata *Rank = nullptr;
1615     Metadata *Annotations = nullptr;
1616     auto *Identifier = getMDString(Record[15]);
1617     // If this module is being parsed so that it can be ThinLTO imported
1618     // into another module, composite types only need to be imported as
1619     // type declarations (unless full type definitions are requested).
1620     // Create type declarations up front to save memory. This is only
1621     // done for types which have an Identifier, and are therefore
1622     // subject to the ODR.
1623     //
1624     // buildODRType handles the case where this is type ODRed with a
1625     // definition needed by the importing module, in which case the
1626     // existing definition is used.
1627     //
1628     // We always import full definitions for anonymous composite types,
1629     // as without a name, debuggers cannot easily resolve a declaration
1630     // to its definition.
1631     if (IsImporting && !ImportFullTypeDefinitions && Identifier && Name &&
1632         (Tag == dwarf::DW_TAG_enumeration_type ||
1633          Tag == dwarf::DW_TAG_class_type ||
1634          Tag == dwarf::DW_TAG_structure_type ||
1635          Tag == dwarf::DW_TAG_union_type)) {
1636       Flags = Flags | DINode::FlagFwdDecl;
1637       // This is a hack around preserving template parameters for simplified
1638       // template names - it should probably be replaced with a
1639       // DICompositeType flag specifying whether template parameters are
1640       // required on declarations of this type.
1641       StringRef NameStr = Name->getString();
1642       if (!NameStr.contains('<') || NameStr.starts_with("_STN|"))
1643         TemplateParams = getMDOrNull(Record[14]);
1644     } else {
1645       BaseType = getDITypeRefOrNull(Record[6]);
1646       OffsetInBits = Record[9];
1647       Elements = getMDOrNull(Record[11]);
1648       VTableHolder = getDITypeRefOrNull(Record[13]);
1649       TemplateParams = getMDOrNull(Record[14]);
1650       if (Record.size() > 16)
1651         Discriminator = getMDOrNull(Record[16]);
1652       if (Record.size() > 17)
1653         DataLocation = getMDOrNull(Record[17]);
1654       if (Record.size() > 19) {
1655         Associated = getMDOrNull(Record[18]);
1656         Allocated = getMDOrNull(Record[19]);
1657       }
1658       if (Record.size() > 20) {
1659         Rank = getMDOrNull(Record[20]);
1660       }
1661       if (Record.size() > 21) {
1662         Annotations = getMDOrNull(Record[21]);
1663       }
1664     }
1665     DICompositeType *CT = nullptr;
1666     if (Identifier)
1667       CT = DICompositeType::buildODRType(
1668           Context, *Identifier, Tag, Name, File, Line, Scope, BaseType,
1669           SizeInBits, AlignInBits, OffsetInBits, Flags, Elements, RuntimeLang,
1670           VTableHolder, TemplateParams, Discriminator, DataLocation, Associated,
1671           Allocated, Rank, Annotations);
1672 
1673     // Create a node if we didn't get a lazy ODR type.
1674     if (!CT)
1675       CT = GET_OR_DISTINCT(DICompositeType,
1676                            (Context, Tag, Name, File, Line, Scope, BaseType,
1677                             SizeInBits, AlignInBits, OffsetInBits, Flags,
1678                             Elements, RuntimeLang, VTableHolder, TemplateParams,
1679                             Identifier, Discriminator, DataLocation, Associated,
1680                             Allocated, Rank, Annotations));
1681     if (!IsNotUsedInTypeRef && Identifier)
1682       MetadataList.addTypeRef(*Identifier, *cast<DICompositeType>(CT));
1683 
1684     MetadataList.assignValue(CT, NextMetadataNo);
1685     NextMetadataNo++;
1686     break;
1687   }
1688   case bitc::METADATA_SUBROUTINE_TYPE: {
1689     if (Record.size() < 3 || Record.size() > 4)
1690       return error("Invalid record");
1691     bool IsOldTypeRefArray = Record[0] < 2;
1692     unsigned CC = (Record.size() > 3) ? Record[3] : 0;
1693 
1694     IsDistinct = Record[0] & 0x1;
1695     DINode::DIFlags Flags = static_cast<DINode::DIFlags>(Record[1]);
1696     Metadata *Types = getMDOrNull(Record[2]);
1697     if (LLVM_UNLIKELY(IsOldTypeRefArray))
1698       Types = MetadataList.upgradeTypeRefArray(Types);
1699 
1700     MetadataList.assignValue(
1701         GET_OR_DISTINCT(DISubroutineType, (Context, Flags, CC, Types)),
1702         NextMetadataNo);
1703     NextMetadataNo++;
1704     break;
1705   }
1706 
1707   case bitc::METADATA_MODULE: {
1708     if (Record.size() < 5 || Record.size() > 9)
1709       return error("Invalid record");
1710 
1711     unsigned Offset = Record.size() >= 8 ? 2 : 1;
1712     IsDistinct = Record[0];
1713     MetadataList.assignValue(
1714         GET_OR_DISTINCT(
1715             DIModule,
1716             (Context, Record.size() >= 8 ? getMDOrNull(Record[1]) : nullptr,
1717              getMDOrNull(Record[0 + Offset]), getMDString(Record[1 + Offset]),
1718              getMDString(Record[2 + Offset]), getMDString(Record[3 + Offset]),
1719              getMDString(Record[4 + Offset]),
1720              Record.size() <= 7 ? 0 : Record[7],
1721              Record.size() <= 8 ? false : Record[8])),
1722         NextMetadataNo);
1723     NextMetadataNo++;
1724     break;
1725   }
1726 
1727   case bitc::METADATA_FILE: {
1728     if (Record.size() != 3 && Record.size() != 5 && Record.size() != 6)
1729       return error("Invalid record");
1730 
1731     IsDistinct = Record[0];
1732     std::optional<DIFile::ChecksumInfo<MDString *>> Checksum;
1733     // The BitcodeWriter writes null bytes into Record[3:4] when the Checksum
1734     // is not present. This matches up with the old internal representation,
1735     // and the old encoding for CSK_None in the ChecksumKind. The new
1736     // representation reserves the value 0 in the ChecksumKind to continue to
1737     // encode None in a backwards-compatible way.
1738     if (Record.size() > 4 && Record[3] && Record[4])
1739       Checksum.emplace(static_cast<DIFile::ChecksumKind>(Record[3]),
1740                        getMDString(Record[4]));
1741     MetadataList.assignValue(
1742         GET_OR_DISTINCT(DIFile,
1743                         (Context, getMDString(Record[1]),
1744                          getMDString(Record[2]), Checksum,
1745                          Record.size() > 5 ? getMDString(Record[5]) : nullptr)),
1746         NextMetadataNo);
1747     NextMetadataNo++;
1748     break;
1749   }
1750   case bitc::METADATA_COMPILE_UNIT: {
1751     if (Record.size() < 14 || Record.size() > 22)
1752       return error("Invalid record");
1753 
1754     // Ignore Record[0], which indicates whether this compile unit is
1755     // distinct.  It's always distinct.
1756     IsDistinct = true;
1757     auto *CU = DICompileUnit::getDistinct(
1758         Context, Record[1], getMDOrNull(Record[2]), getMDString(Record[3]),
1759         Record[4], getMDString(Record[5]), Record[6], getMDString(Record[7]),
1760         Record[8], getMDOrNull(Record[9]), getMDOrNull(Record[10]),
1761         getMDOrNull(Record[12]), getMDOrNull(Record[13]),
1762         Record.size() <= 15 ? nullptr : getMDOrNull(Record[15]),
1763         Record.size() <= 14 ? 0 : Record[14],
1764         Record.size() <= 16 ? true : Record[16],
1765         Record.size() <= 17 ? false : Record[17],
1766         Record.size() <= 18 ? 0 : Record[18],
1767         Record.size() <= 19 ? false : Record[19],
1768         Record.size() <= 20 ? nullptr : getMDString(Record[20]),
1769         Record.size() <= 21 ? nullptr : getMDString(Record[21]));
1770 
1771     MetadataList.assignValue(CU, NextMetadataNo);
1772     NextMetadataNo++;
1773 
1774     // Move the Upgrade the list of subprograms.
1775     if (Metadata *SPs = getMDOrNullWithoutPlaceholders(Record[11]))
1776       CUSubprograms.push_back({CU, SPs});
1777     break;
1778   }
1779   case bitc::METADATA_SUBPROGRAM: {
1780     if (Record.size() < 18 || Record.size() > 21)
1781       return error("Invalid record");
1782 
1783     bool HasSPFlags = Record[0] & 4;
1784 
1785     DINode::DIFlags Flags;
1786     DISubprogram::DISPFlags SPFlags;
1787     if (!HasSPFlags)
1788       Flags = static_cast<DINode::DIFlags>(Record[11 + 2]);
1789     else {
1790       Flags = static_cast<DINode::DIFlags>(Record[11]);
1791       SPFlags = static_cast<DISubprogram::DISPFlags>(Record[9]);
1792     }
1793 
1794     // Support for old metadata when
1795     // subprogram specific flags are placed in DIFlags.
1796     const unsigned DIFlagMainSubprogram = 1 << 21;
1797     bool HasOldMainSubprogramFlag = Flags & DIFlagMainSubprogram;
1798     if (HasOldMainSubprogramFlag)
1799       // Remove old DIFlagMainSubprogram from DIFlags.
1800       // Note: This assumes that any future use of bit 21 defaults to it
1801       // being 0.
1802       Flags &= ~static_cast<DINode::DIFlags>(DIFlagMainSubprogram);
1803 
1804     if (HasOldMainSubprogramFlag && HasSPFlags)
1805       SPFlags |= DISubprogram::SPFlagMainSubprogram;
1806     else if (!HasSPFlags)
1807       SPFlags = DISubprogram::toSPFlags(
1808           /*IsLocalToUnit=*/Record[7], /*IsDefinition=*/Record[8],
1809           /*IsOptimized=*/Record[14], /*Virtuality=*/Record[11],
1810           /*IsMainSubprogram=*/HasOldMainSubprogramFlag);
1811 
1812     // All definitions should be distinct.
1813     IsDistinct = (Record[0] & 1) || (SPFlags & DISubprogram::SPFlagDefinition);
1814     // Version 1 has a Function as Record[15].
1815     // Version 2 has removed Record[15].
1816     // Version 3 has the Unit as Record[15].
1817     // Version 4 added thisAdjustment.
1818     // Version 5 repacked flags into DISPFlags, changing many element numbers.
1819     bool HasUnit = Record[0] & 2;
1820     if (!HasSPFlags && HasUnit && Record.size() < 19)
1821       return error("Invalid record");
1822     if (HasSPFlags && !HasUnit)
1823       return error("Invalid record");
1824     // Accommodate older formats.
1825     bool HasFn = false;
1826     bool HasThisAdj = true;
1827     bool HasThrownTypes = true;
1828     bool HasAnnotations = false;
1829     bool HasTargetFuncName = false;
1830     unsigned OffsetA = 0;
1831     unsigned OffsetB = 0;
1832     if (!HasSPFlags) {
1833       OffsetA = 2;
1834       OffsetB = 2;
1835       if (Record.size() >= 19) {
1836         HasFn = !HasUnit;
1837         OffsetB++;
1838       }
1839       HasThisAdj = Record.size() >= 20;
1840       HasThrownTypes = Record.size() >= 21;
1841     } else {
1842       HasAnnotations = Record.size() >= 19;
1843       HasTargetFuncName = Record.size() >= 20;
1844     }
1845     Metadata *CUorFn = getMDOrNull(Record[12 + OffsetB]);
1846     DISubprogram *SP = GET_OR_DISTINCT(
1847         DISubprogram,
1848         (Context,
1849          getDITypeRefOrNull(Record[1]),           // scope
1850          getMDString(Record[2]),                  // name
1851          getMDString(Record[3]),                  // linkageName
1852          getMDOrNull(Record[4]),                  // file
1853          Record[5],                               // line
1854          getMDOrNull(Record[6]),                  // type
1855          Record[7 + OffsetA],                     // scopeLine
1856          getDITypeRefOrNull(Record[8 + OffsetA]), // containingType
1857          Record[10 + OffsetA],                    // virtualIndex
1858          HasThisAdj ? Record[16 + OffsetB] : 0,   // thisAdjustment
1859          Flags,                                   // flags
1860          SPFlags,                                 // SPFlags
1861          HasUnit ? CUorFn : nullptr,              // unit
1862          getMDOrNull(Record[13 + OffsetB]),       // templateParams
1863          getMDOrNull(Record[14 + OffsetB]),       // declaration
1864          getMDOrNull(Record[15 + OffsetB]),       // retainedNodes
1865          HasThrownTypes ? getMDOrNull(Record[17 + OffsetB])
1866                         : nullptr, // thrownTypes
1867          HasAnnotations ? getMDOrNull(Record[18 + OffsetB])
1868                         : nullptr, // annotations
1869          HasTargetFuncName ? getMDString(Record[19 + OffsetB])
1870                            : nullptr // targetFuncName
1871          ));
1872     MetadataList.assignValue(SP, NextMetadataNo);
1873     NextMetadataNo++;
1874 
1875     // Upgrade sp->function mapping to function->sp mapping.
1876     if (HasFn) {
1877       if (auto *CMD = dyn_cast_or_null<ConstantAsMetadata>(CUorFn))
1878         if (auto *F = dyn_cast<Function>(CMD->getValue())) {
1879           if (F->isMaterializable())
1880             // Defer until materialized; unmaterialized functions may not have
1881             // metadata.
1882             FunctionsWithSPs[F] = SP;
1883           else if (!F->empty())
1884             F->setSubprogram(SP);
1885         }
1886     }
1887     break;
1888   }
1889   case bitc::METADATA_LEXICAL_BLOCK: {
1890     if (Record.size() != 5)
1891       return error("Invalid record");
1892 
1893     IsDistinct = Record[0];
1894     MetadataList.assignValue(
1895         GET_OR_DISTINCT(DILexicalBlock,
1896                         (Context, getMDOrNull(Record[1]),
1897                          getMDOrNull(Record[2]), Record[3], Record[4])),
1898         NextMetadataNo);
1899     NextMetadataNo++;
1900     break;
1901   }
1902   case bitc::METADATA_LEXICAL_BLOCK_FILE: {
1903     if (Record.size() != 4)
1904       return error("Invalid record");
1905 
1906     IsDistinct = Record[0];
1907     MetadataList.assignValue(
1908         GET_OR_DISTINCT(DILexicalBlockFile,
1909                         (Context, getMDOrNull(Record[1]),
1910                          getMDOrNull(Record[2]), Record[3])),
1911         NextMetadataNo);
1912     NextMetadataNo++;
1913     break;
1914   }
1915   case bitc::METADATA_COMMON_BLOCK: {
1916     IsDistinct = Record[0] & 1;
1917     MetadataList.assignValue(
1918         GET_OR_DISTINCT(DICommonBlock,
1919                         (Context, getMDOrNull(Record[1]),
1920                          getMDOrNull(Record[2]), getMDString(Record[3]),
1921                          getMDOrNull(Record[4]), Record[5])),
1922         NextMetadataNo);
1923     NextMetadataNo++;
1924     break;
1925   }
1926   case bitc::METADATA_NAMESPACE: {
1927     // Newer versions of DINamespace dropped file and line.
1928     MDString *Name;
1929     if (Record.size() == 3)
1930       Name = getMDString(Record[2]);
1931     else if (Record.size() == 5)
1932       Name = getMDString(Record[3]);
1933     else
1934       return error("Invalid record");
1935 
1936     IsDistinct = Record[0] & 1;
1937     bool ExportSymbols = Record[0] & 2;
1938     MetadataList.assignValue(
1939         GET_OR_DISTINCT(DINamespace,
1940                         (Context, getMDOrNull(Record[1]), Name, ExportSymbols)),
1941         NextMetadataNo);
1942     NextMetadataNo++;
1943     break;
1944   }
1945   case bitc::METADATA_MACRO: {
1946     if (Record.size() != 5)
1947       return error("Invalid record");
1948 
1949     IsDistinct = Record[0];
1950     MetadataList.assignValue(
1951         GET_OR_DISTINCT(DIMacro,
1952                         (Context, Record[1], Record[2], getMDString(Record[3]),
1953                          getMDString(Record[4]))),
1954         NextMetadataNo);
1955     NextMetadataNo++;
1956     break;
1957   }
1958   case bitc::METADATA_MACRO_FILE: {
1959     if (Record.size() != 5)
1960       return error("Invalid record");
1961 
1962     IsDistinct = Record[0];
1963     MetadataList.assignValue(
1964         GET_OR_DISTINCT(DIMacroFile,
1965                         (Context, Record[1], Record[2], getMDOrNull(Record[3]),
1966                          getMDOrNull(Record[4]))),
1967         NextMetadataNo);
1968     NextMetadataNo++;
1969     break;
1970   }
1971   case bitc::METADATA_TEMPLATE_TYPE: {
1972     if (Record.size() < 3 || Record.size() > 4)
1973       return error("Invalid record");
1974 
1975     IsDistinct = Record[0];
1976     MetadataList.assignValue(
1977         GET_OR_DISTINCT(DITemplateTypeParameter,
1978                         (Context, getMDString(Record[1]),
1979                          getDITypeRefOrNull(Record[2]),
1980                          (Record.size() == 4) ? getMDOrNull(Record[3])
1981                                               : getMDOrNull(false))),
1982         NextMetadataNo);
1983     NextMetadataNo++;
1984     break;
1985   }
1986   case bitc::METADATA_TEMPLATE_VALUE: {
1987     if (Record.size() < 5 || Record.size() > 6)
1988       return error("Invalid record");
1989 
1990     IsDistinct = Record[0];
1991 
1992     MetadataList.assignValue(
1993         GET_OR_DISTINCT(
1994             DITemplateValueParameter,
1995             (Context, Record[1], getMDString(Record[2]),
1996              getDITypeRefOrNull(Record[3]),
1997              (Record.size() == 6) ? getMDOrNull(Record[4]) : getMDOrNull(false),
1998              (Record.size() == 6) ? getMDOrNull(Record[5])
1999                                   : getMDOrNull(Record[4]))),
2000         NextMetadataNo);
2001     NextMetadataNo++;
2002     break;
2003   }
2004   case bitc::METADATA_GLOBAL_VAR: {
2005     if (Record.size() < 11 || Record.size() > 13)
2006       return error("Invalid record");
2007 
2008     IsDistinct = Record[0] & 1;
2009     unsigned Version = Record[0] >> 1;
2010 
2011     if (Version == 2) {
2012       Metadata *Annotations = nullptr;
2013       if (Record.size() > 12)
2014         Annotations = getMDOrNull(Record[12]);
2015 
2016       MetadataList.assignValue(
2017           GET_OR_DISTINCT(DIGlobalVariable,
2018                           (Context, getMDOrNull(Record[1]),
2019                            getMDString(Record[2]), getMDString(Record[3]),
2020                            getMDOrNull(Record[4]), Record[5],
2021                            getDITypeRefOrNull(Record[6]), Record[7], Record[8],
2022                            getMDOrNull(Record[9]), getMDOrNull(Record[10]),
2023                            Record[11], Annotations)),
2024           NextMetadataNo);
2025 
2026       NextMetadataNo++;
2027     } else if (Version == 1) {
2028       // No upgrade necessary. A null field will be introduced to indicate
2029       // that no parameter information is available.
2030       MetadataList.assignValue(
2031           GET_OR_DISTINCT(
2032               DIGlobalVariable,
2033               (Context, getMDOrNull(Record[1]), getMDString(Record[2]),
2034                getMDString(Record[3]), getMDOrNull(Record[4]), Record[5],
2035                getDITypeRefOrNull(Record[6]), Record[7], Record[8],
2036                getMDOrNull(Record[10]), nullptr, Record[11], nullptr)),
2037           NextMetadataNo);
2038 
2039       NextMetadataNo++;
2040     } else if (Version == 0) {
2041       // Upgrade old metadata, which stored a global variable reference or a
2042       // ConstantInt here.
2043       NeedUpgradeToDIGlobalVariableExpression = true;
2044       Metadata *Expr = getMDOrNull(Record[9]);
2045       uint32_t AlignInBits = 0;
2046       if (Record.size() > 11) {
2047         if (Record[11] > (uint64_t)std::numeric_limits<uint32_t>::max())
2048           return error("Alignment value is too large");
2049         AlignInBits = Record[11];
2050       }
2051       GlobalVariable *Attach = nullptr;
2052       if (auto *CMD = dyn_cast_or_null<ConstantAsMetadata>(Expr)) {
2053         if (auto *GV = dyn_cast<GlobalVariable>(CMD->getValue())) {
2054           Attach = GV;
2055           Expr = nullptr;
2056         } else if (auto *CI = dyn_cast<ConstantInt>(CMD->getValue())) {
2057           Expr = DIExpression::get(Context,
2058                                    {dwarf::DW_OP_constu, CI->getZExtValue(),
2059                                     dwarf::DW_OP_stack_value});
2060         } else {
2061           Expr = nullptr;
2062         }
2063       }
2064       DIGlobalVariable *DGV = GET_OR_DISTINCT(
2065           DIGlobalVariable,
2066           (Context, getMDOrNull(Record[1]), getMDString(Record[2]),
2067            getMDString(Record[3]), getMDOrNull(Record[4]), Record[5],
2068            getDITypeRefOrNull(Record[6]), Record[7], Record[8],
2069            getMDOrNull(Record[10]), nullptr, AlignInBits, nullptr));
2070 
2071       DIGlobalVariableExpression *DGVE = nullptr;
2072       if (Attach || Expr)
2073         DGVE = DIGlobalVariableExpression::getDistinct(
2074             Context, DGV, Expr ? Expr : DIExpression::get(Context, {}));
2075       if (Attach)
2076         Attach->addDebugInfo(DGVE);
2077 
2078       auto *MDNode = Expr ? cast<Metadata>(DGVE) : cast<Metadata>(DGV);
2079       MetadataList.assignValue(MDNode, NextMetadataNo);
2080       NextMetadataNo++;
2081     } else
2082       return error("Invalid record");
2083 
2084     break;
2085   }
2086   case bitc::METADATA_ASSIGN_ID: {
2087     if (Record.size() != 1)
2088       return error("Invalid DIAssignID record.");
2089 
2090     IsDistinct = Record[0] & 1;
2091     if (!IsDistinct)
2092       return error("Invalid DIAssignID record. Must be distinct");
2093 
2094     MetadataList.assignValue(DIAssignID::getDistinct(Context), NextMetadataNo);
2095     NextMetadataNo++;
2096     break;
2097   }
2098   case bitc::METADATA_LOCAL_VAR: {
2099     // 10th field is for the obseleted 'inlinedAt:' field.
2100     if (Record.size() < 8 || Record.size() > 10)
2101       return error("Invalid record");
2102 
2103     IsDistinct = Record[0] & 1;
2104     bool HasAlignment = Record[0] & 2;
2105     // 2nd field used to be an artificial tag, either DW_TAG_auto_variable or
2106     // DW_TAG_arg_variable, if we have alignment flag encoded it means, that
2107     // this is newer version of record which doesn't have artificial tag.
2108     bool HasTag = !HasAlignment && Record.size() > 8;
2109     DINode::DIFlags Flags = static_cast<DINode::DIFlags>(Record[7 + HasTag]);
2110     uint32_t AlignInBits = 0;
2111     Metadata *Annotations = nullptr;
2112     if (HasAlignment) {
2113       if (Record[8] > (uint64_t)std::numeric_limits<uint32_t>::max())
2114         return error("Alignment value is too large");
2115       AlignInBits = Record[8];
2116       if (Record.size() > 9)
2117         Annotations = getMDOrNull(Record[9]);
2118     }
2119 
2120     MetadataList.assignValue(
2121         GET_OR_DISTINCT(DILocalVariable,
2122                         (Context, getMDOrNull(Record[1 + HasTag]),
2123                          getMDString(Record[2 + HasTag]),
2124                          getMDOrNull(Record[3 + HasTag]), Record[4 + HasTag],
2125                          getDITypeRefOrNull(Record[5 + HasTag]),
2126                          Record[6 + HasTag], Flags, AlignInBits, Annotations)),
2127         NextMetadataNo);
2128     NextMetadataNo++;
2129     break;
2130   }
2131   case bitc::METADATA_LABEL: {
2132     if (Record.size() != 5)
2133       return error("Invalid record");
2134 
2135     IsDistinct = Record[0] & 1;
2136     MetadataList.assignValue(
2137         GET_OR_DISTINCT(DILabel, (Context, getMDOrNull(Record[1]),
2138                                   getMDString(Record[2]),
2139                                   getMDOrNull(Record[3]), Record[4])),
2140         NextMetadataNo);
2141     NextMetadataNo++;
2142     break;
2143   }
2144   case bitc::METADATA_EXPRESSION: {
2145     if (Record.size() < 1)
2146       return error("Invalid record");
2147 
2148     IsDistinct = Record[0] & 1;
2149     uint64_t Version = Record[0] >> 1;
2150     auto Elts = MutableArrayRef<uint64_t>(Record).slice(1);
2151 
2152     SmallVector<uint64_t, 6> Buffer;
2153     if (Error Err = upgradeDIExpression(Version, Elts, Buffer))
2154       return Err;
2155 
2156     MetadataList.assignValue(GET_OR_DISTINCT(DIExpression, (Context, Elts)),
2157                              NextMetadataNo);
2158     NextMetadataNo++;
2159     break;
2160   }
2161   case bitc::METADATA_GLOBAL_VAR_EXPR: {
2162     if (Record.size() != 3)
2163       return error("Invalid record");
2164 
2165     IsDistinct = Record[0];
2166     Metadata *Expr = getMDOrNull(Record[2]);
2167     if (!Expr)
2168       Expr = DIExpression::get(Context, {});
2169     MetadataList.assignValue(
2170         GET_OR_DISTINCT(DIGlobalVariableExpression,
2171                         (Context, getMDOrNull(Record[1]), Expr)),
2172         NextMetadataNo);
2173     NextMetadataNo++;
2174     break;
2175   }
2176   case bitc::METADATA_OBJC_PROPERTY: {
2177     if (Record.size() != 8)
2178       return error("Invalid record");
2179 
2180     IsDistinct = Record[0];
2181     MetadataList.assignValue(
2182         GET_OR_DISTINCT(DIObjCProperty,
2183                         (Context, getMDString(Record[1]),
2184                          getMDOrNull(Record[2]), Record[3],
2185                          getMDString(Record[4]), getMDString(Record[5]),
2186                          Record[6], getDITypeRefOrNull(Record[7]))),
2187         NextMetadataNo);
2188     NextMetadataNo++;
2189     break;
2190   }
2191   case bitc::METADATA_IMPORTED_ENTITY: {
2192     if (Record.size() < 6 || Record.size() > 8)
2193       return error("Invalid DIImportedEntity record");
2194 
2195     IsDistinct = Record[0];
2196     bool HasFile = (Record.size() >= 7);
2197     bool HasElements = (Record.size() >= 8);
2198     MetadataList.assignValue(
2199         GET_OR_DISTINCT(DIImportedEntity,
2200                         (Context, Record[1], getMDOrNull(Record[2]),
2201                          getDITypeRefOrNull(Record[3]),
2202                          HasFile ? getMDOrNull(Record[6]) : nullptr,
2203                          HasFile ? Record[4] : 0, getMDString(Record[5]),
2204                          HasElements ? getMDOrNull(Record[7]) : nullptr)),
2205         NextMetadataNo);
2206     NextMetadataNo++;
2207     break;
2208   }
2209   case bitc::METADATA_STRING_OLD: {
2210     std::string String(Record.begin(), Record.end());
2211 
2212     // Test for upgrading !llvm.loop.
2213     HasSeenOldLoopTags |= mayBeOldLoopAttachmentTag(String);
2214     ++NumMDStringLoaded;
2215     Metadata *MD = MDString::get(Context, String);
2216     MetadataList.assignValue(MD, NextMetadataNo);
2217     NextMetadataNo++;
2218     break;
2219   }
2220   case bitc::METADATA_STRINGS: {
2221     auto CreateNextMDString = [&](StringRef Str) {
2222       ++NumMDStringLoaded;
2223       MetadataList.assignValue(MDString::get(Context, Str), NextMetadataNo);
2224       NextMetadataNo++;
2225     };
2226     if (Error Err = parseMetadataStrings(Record, Blob, CreateNextMDString))
2227       return Err;
2228     break;
2229   }
2230   case bitc::METADATA_GLOBAL_DECL_ATTACHMENT: {
2231     if (Record.size() % 2 == 0)
2232       return error("Invalid record");
2233     unsigned ValueID = Record[0];
2234     if (ValueID >= ValueList.size())
2235       return error("Invalid record");
2236     if (auto *GO = dyn_cast<GlobalObject>(ValueList[ValueID]))
2237       if (Error Err = parseGlobalObjectAttachment(
2238               *GO, ArrayRef<uint64_t>(Record).slice(1)))
2239         return Err;
2240     break;
2241   }
2242   case bitc::METADATA_KIND: {
2243     // Support older bitcode files that had METADATA_KIND records in a
2244     // block with METADATA_BLOCK_ID.
2245     if (Error Err = parseMetadataKindRecord(Record))
2246       return Err;
2247     break;
2248   }
2249   case bitc::METADATA_ARG_LIST: {
2250     SmallVector<ValueAsMetadata *, 4> Elts;
2251     Elts.reserve(Record.size());
2252     for (uint64_t Elt : Record) {
2253       Metadata *MD = getMD(Elt);
2254       if (isa<MDNode>(MD) && cast<MDNode>(MD)->isTemporary())
2255         return error(
2256             "Invalid record: DIArgList should not contain forward refs");
2257       if (!isa<ValueAsMetadata>(MD))
2258         return error("Invalid record");
2259       Elts.push_back(cast<ValueAsMetadata>(MD));
2260     }
2261 
2262     MetadataList.assignValue(DIArgList::get(Context, Elts), NextMetadataNo);
2263     NextMetadataNo++;
2264     break;
2265   }
2266   }
2267   return Error::success();
2268 #undef GET_OR_DISTINCT
2269 }
2270 
parseMetadataStrings(ArrayRef<uint64_t> Record,StringRef Blob,function_ref<void (StringRef)> CallBack)2271 Error MetadataLoader::MetadataLoaderImpl::parseMetadataStrings(
2272     ArrayRef<uint64_t> Record, StringRef Blob,
2273     function_ref<void(StringRef)> CallBack) {
2274   // All the MDStrings in the block are emitted together in a single
2275   // record.  The strings are concatenated and stored in a blob along with
2276   // their sizes.
2277   if (Record.size() != 2)
2278     return error("Invalid record: metadata strings layout");
2279 
2280   unsigned NumStrings = Record[0];
2281   unsigned StringsOffset = Record[1];
2282   if (!NumStrings)
2283     return error("Invalid record: metadata strings with no strings");
2284   if (StringsOffset > Blob.size())
2285     return error("Invalid record: metadata strings corrupt offset");
2286 
2287   StringRef Lengths = Blob.slice(0, StringsOffset);
2288   SimpleBitstreamCursor R(Lengths);
2289 
2290   StringRef Strings = Blob.drop_front(StringsOffset);
2291   do {
2292     if (R.AtEndOfStream())
2293       return error("Invalid record: metadata strings bad length");
2294 
2295     uint32_t Size;
2296     if (Error E = R.ReadVBR(6).moveInto(Size))
2297       return E;
2298     if (Strings.size() < Size)
2299       return error("Invalid record: metadata strings truncated chars");
2300 
2301     CallBack(Strings.slice(0, Size));
2302     Strings = Strings.drop_front(Size);
2303   } while (--NumStrings);
2304 
2305   return Error::success();
2306 }
2307 
parseGlobalObjectAttachment(GlobalObject & GO,ArrayRef<uint64_t> Record)2308 Error MetadataLoader::MetadataLoaderImpl::parseGlobalObjectAttachment(
2309     GlobalObject &GO, ArrayRef<uint64_t> Record) {
2310   assert(Record.size() % 2 == 0);
2311   for (unsigned I = 0, E = Record.size(); I != E; I += 2) {
2312     auto K = MDKindMap.find(Record[I]);
2313     if (K == MDKindMap.end())
2314       return error("Invalid ID");
2315     MDNode *MD =
2316         dyn_cast_or_null<MDNode>(getMetadataFwdRefOrLoad(Record[I + 1]));
2317     if (!MD)
2318       return error("Invalid metadata attachment: expect fwd ref to MDNode");
2319     GO.addMetadata(K->second, *MD);
2320   }
2321   return Error::success();
2322 }
2323 
2324 /// Parse metadata attachments.
parseMetadataAttachment(Function & F,ArrayRef<Instruction * > InstructionList)2325 Error MetadataLoader::MetadataLoaderImpl::parseMetadataAttachment(
2326     Function &F, ArrayRef<Instruction *> InstructionList) {
2327   if (Error Err = Stream.EnterSubBlock(bitc::METADATA_ATTACHMENT_ID))
2328     return Err;
2329 
2330   SmallVector<uint64_t, 64> Record;
2331   PlaceholderQueue Placeholders;
2332 
2333   while (true) {
2334     BitstreamEntry Entry;
2335     if (Error E = Stream.advanceSkippingSubblocks().moveInto(Entry))
2336       return E;
2337 
2338     switch (Entry.Kind) {
2339     case BitstreamEntry::SubBlock: // Handled for us already.
2340     case BitstreamEntry::Error:
2341       return error("Malformed block");
2342     case BitstreamEntry::EndBlock:
2343       resolveForwardRefsAndPlaceholders(Placeholders);
2344       return Error::success();
2345     case BitstreamEntry::Record:
2346       // The interesting case.
2347       break;
2348     }
2349 
2350     // Read a metadata attachment record.
2351     Record.clear();
2352     ++NumMDRecordLoaded;
2353     Expected<unsigned> MaybeRecord = Stream.readRecord(Entry.ID, Record);
2354     if (!MaybeRecord)
2355       return MaybeRecord.takeError();
2356     switch (MaybeRecord.get()) {
2357     default: // Default behavior: ignore.
2358       break;
2359     case bitc::METADATA_ATTACHMENT: {
2360       unsigned RecordLength = Record.size();
2361       if (Record.empty())
2362         return error("Invalid record");
2363       if (RecordLength % 2 == 0) {
2364         // A function attachment.
2365         if (Error Err = parseGlobalObjectAttachment(F, Record))
2366           return Err;
2367         continue;
2368       }
2369 
2370       // An instruction attachment.
2371       Instruction *Inst = InstructionList[Record[0]];
2372       for (unsigned i = 1; i != RecordLength; i = i + 2) {
2373         unsigned Kind = Record[i];
2374         DenseMap<unsigned, unsigned>::iterator I = MDKindMap.find(Kind);
2375         if (I == MDKindMap.end())
2376           return error("Invalid ID");
2377         if (I->second == LLVMContext::MD_tbaa && StripTBAA)
2378           continue;
2379 
2380         auto Idx = Record[i + 1];
2381         if (Idx < (MDStringRef.size() + GlobalMetadataBitPosIndex.size()) &&
2382             !MetadataList.lookup(Idx)) {
2383           // Load the attachment if it is in the lazy-loadable range and hasn't
2384           // been loaded yet.
2385           lazyLoadOneMetadata(Idx, Placeholders);
2386           resolveForwardRefsAndPlaceholders(Placeholders);
2387         }
2388 
2389         Metadata *Node = MetadataList.getMetadataFwdRef(Idx);
2390         if (isa<LocalAsMetadata>(Node))
2391           // Drop the attachment.  This used to be legal, but there's no
2392           // upgrade path.
2393           break;
2394         MDNode *MD = dyn_cast_or_null<MDNode>(Node);
2395         if (!MD)
2396           return error("Invalid metadata attachment");
2397 
2398         if (HasSeenOldLoopTags && I->second == LLVMContext::MD_loop)
2399           MD = upgradeInstructionLoopAttachment(*MD);
2400 
2401         if (I->second == LLVMContext::MD_tbaa) {
2402           assert(!MD->isTemporary() && "should load MDs before attachments");
2403           MD = UpgradeTBAANode(*MD);
2404         }
2405         Inst->setMetadata(I->second, MD);
2406       }
2407       break;
2408     }
2409     }
2410   }
2411 }
2412 
2413 /// Parse a single METADATA_KIND record, inserting result in MDKindMap.
parseMetadataKindRecord(SmallVectorImpl<uint64_t> & Record)2414 Error MetadataLoader::MetadataLoaderImpl::parseMetadataKindRecord(
2415     SmallVectorImpl<uint64_t> &Record) {
2416   if (Record.size() < 2)
2417     return error("Invalid record");
2418 
2419   unsigned Kind = Record[0];
2420   SmallString<8> Name(Record.begin() + 1, Record.end());
2421 
2422   unsigned NewKind = TheModule.getMDKindID(Name.str());
2423   if (!MDKindMap.insert(std::make_pair(Kind, NewKind)).second)
2424     return error("Conflicting METADATA_KIND records");
2425   return Error::success();
2426 }
2427 
2428 /// Parse the metadata kinds out of the METADATA_KIND_BLOCK.
parseMetadataKinds()2429 Error MetadataLoader::MetadataLoaderImpl::parseMetadataKinds() {
2430   if (Error Err = Stream.EnterSubBlock(bitc::METADATA_KIND_BLOCK_ID))
2431     return Err;
2432 
2433   SmallVector<uint64_t, 64> Record;
2434 
2435   // Read all the records.
2436   while (true) {
2437     BitstreamEntry Entry;
2438     if (Error E = Stream.advanceSkippingSubblocks().moveInto(Entry))
2439       return E;
2440 
2441     switch (Entry.Kind) {
2442     case BitstreamEntry::SubBlock: // Handled for us already.
2443     case BitstreamEntry::Error:
2444       return error("Malformed block");
2445     case BitstreamEntry::EndBlock:
2446       return Error::success();
2447     case BitstreamEntry::Record:
2448       // The interesting case.
2449       break;
2450     }
2451 
2452     // Read a record.
2453     Record.clear();
2454     ++NumMDRecordLoaded;
2455     Expected<unsigned> MaybeCode = Stream.readRecord(Entry.ID, Record);
2456     if (!MaybeCode)
2457       return MaybeCode.takeError();
2458     switch (MaybeCode.get()) {
2459     default: // Default behavior: ignore.
2460       break;
2461     case bitc::METADATA_KIND: {
2462       if (Error Err = parseMetadataKindRecord(Record))
2463         return Err;
2464       break;
2465     }
2466     }
2467   }
2468 }
2469 
operator =(MetadataLoader && RHS)2470 MetadataLoader &MetadataLoader::operator=(MetadataLoader &&RHS) {
2471   Pimpl = std::move(RHS.Pimpl);
2472   return *this;
2473 }
MetadataLoader(MetadataLoader && RHS)2474 MetadataLoader::MetadataLoader(MetadataLoader &&RHS)
2475     : Pimpl(std::move(RHS.Pimpl)) {}
2476 
2477 MetadataLoader::~MetadataLoader() = default;
MetadataLoader(BitstreamCursor & Stream,Module & TheModule,BitcodeReaderValueList & ValueList,bool IsImporting,MetadataLoaderCallbacks Callbacks)2478 MetadataLoader::MetadataLoader(BitstreamCursor &Stream, Module &TheModule,
2479                                BitcodeReaderValueList &ValueList,
2480                                bool IsImporting,
2481                                MetadataLoaderCallbacks Callbacks)
2482     : Pimpl(std::make_unique<MetadataLoaderImpl>(
2483           Stream, TheModule, ValueList, std::move(Callbacks), IsImporting)) {}
2484 
parseMetadata(bool ModuleLevel)2485 Error MetadataLoader::parseMetadata(bool ModuleLevel) {
2486   return Pimpl->parseMetadata(ModuleLevel);
2487 }
2488 
hasFwdRefs() const2489 bool MetadataLoader::hasFwdRefs() const { return Pimpl->hasFwdRefs(); }
2490 
2491 /// Return the given metadata, creating a replaceable forward reference if
2492 /// necessary.
getMetadataFwdRefOrLoad(unsigned Idx)2493 Metadata *MetadataLoader::getMetadataFwdRefOrLoad(unsigned Idx) {
2494   return Pimpl->getMetadataFwdRefOrLoad(Idx);
2495 }
2496 
lookupSubprogramForFunction(Function * F)2497 DISubprogram *MetadataLoader::lookupSubprogramForFunction(Function *F) {
2498   return Pimpl->lookupSubprogramForFunction(F);
2499 }
2500 
parseMetadataAttachment(Function & F,ArrayRef<Instruction * > InstructionList)2501 Error MetadataLoader::parseMetadataAttachment(
2502     Function &F, ArrayRef<Instruction *> InstructionList) {
2503   return Pimpl->parseMetadataAttachment(F, InstructionList);
2504 }
2505 
parseMetadataKinds()2506 Error MetadataLoader::parseMetadataKinds() {
2507   return Pimpl->parseMetadataKinds();
2508 }
2509 
setStripTBAA(bool StripTBAA)2510 void MetadataLoader::setStripTBAA(bool StripTBAA) {
2511   return Pimpl->setStripTBAA(StripTBAA);
2512 }
2513 
isStrippingTBAA()2514 bool MetadataLoader::isStrippingTBAA() { return Pimpl->isStrippingTBAA(); }
2515 
size() const2516 unsigned MetadataLoader::size() const { return Pimpl->size(); }
shrinkTo(unsigned N)2517 void MetadataLoader::shrinkTo(unsigned N) { return Pimpl->shrinkTo(N); }
2518 
upgradeDebugIntrinsics(Function & F)2519 void MetadataLoader::upgradeDebugIntrinsics(Function &F) {
2520   return Pimpl->upgradeDebugIntrinsics(F);
2521 }
2522