1 //===- Archive.h - ar archive file format -----------------------*- C++ -*-===//
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 // This file declares the ar archive file format class.
10 //
11 //===----------------------------------------------------------------------===//
12 
13 #ifndef LLVM_OBJECT_ARCHIVE_H
14 #define LLVM_OBJECT_ARCHIVE_H
15 
16 #include "llvm/ADT/Optional.h"
17 #include "llvm/ADT/StringRef.h"
18 #include "llvm/ADT/fallible_iterator.h"
19 #include "llvm/ADT/iterator_range.h"
20 #include "llvm/Object/Binary.h"
21 #include "llvm/Support/Chrono.h"
22 #include "llvm/Support/Error.h"
23 #include "llvm/Support/FileSystem.h"
24 #include "llvm/Support/MemoryBuffer.h"
25 #include <algorithm>
26 #include <cassert>
27 #include <cstdint>
28 #include <memory>
29 #include <string>
30 #include <vector>
31 
32 namespace llvm {
33 namespace object {
34 
35 const char ArchiveMagic[] = "!<arch>\n";
36 const char ThinArchiveMagic[] = "!<thin>\n";
37 const char BigArchiveMagic[] = "<bigaf>\n";
38 
39 class Archive;
40 
41 class AbstractArchiveMemberHeader {
42 protected:
43   AbstractArchiveMemberHeader(const Archive *Parent) : Parent(Parent){};
44 
45 public:
46   friend class Archive;
47   virtual std::unique_ptr<AbstractArchiveMemberHeader> clone() const = 0;
48   virtual ~AbstractArchiveMemberHeader() = default;
49 
50   /// Get the name without looking up long names.
51   virtual Expected<StringRef> getRawName() const = 0;
52   virtual StringRef getRawAccessMode() const = 0;
53   virtual StringRef getRawLastModified() const = 0;
54   virtual StringRef getRawUID() const = 0;
55   virtual StringRef getRawGID() const = 0;
56 
57   /// Get the name looking up long names.
58   virtual Expected<StringRef> getName(uint64_t Size) const = 0;
59   virtual Expected<uint64_t> getSize() const = 0;
60   virtual uint64_t getOffset() const = 0;
61 
62   /// Get next file member location.
63   virtual Expected<const char *> getNextChildLoc() const = 0;
64   virtual Expected<bool> isThin() const = 0;
65 
66   Expected<sys::fs::perms> getAccessMode() const;
67   Expected<sys::TimePoint<std::chrono::seconds>> getLastModified() const;
68   Expected<unsigned> getUID() const;
69   Expected<unsigned> getGID() const;
70 
71   /// Returns the size in bytes of the format-defined member header of the
72   /// concrete archive type.
73   virtual uint64_t getSizeOf() const = 0;
74 
75   const Archive *Parent;
76 };
77 
78 template <typename T>
79 class CommonArchiveMemberHeader : public AbstractArchiveMemberHeader {
80 public:
81   CommonArchiveMemberHeader(const Archive *Parent, const T *RawHeaderPtr)
82       : AbstractArchiveMemberHeader(Parent), ArMemHdr(RawHeaderPtr){};
83   StringRef getRawAccessMode() const override;
84   StringRef getRawLastModified() const override;
85   StringRef getRawUID() const override;
86   StringRef getRawGID() const override;
87 
88   uint64_t getOffset() const override;
89   uint64_t getSizeOf() const override { return sizeof(T); }
90 
91   T const *ArMemHdr;
92 };
93 
94 struct UnixArMemHdrType {
95   char Name[16];
96   char LastModified[12];
97   char UID[6];
98   char GID[6];
99   char AccessMode[8];
100   char Size[10]; ///< Size of data, not including header or padding.
101   char Terminator[2];
102 };
103 
104 class ArchiveMemberHeader : public CommonArchiveMemberHeader<UnixArMemHdrType> {
105 public:
106   ArchiveMemberHeader(const Archive *Parent, const char *RawHeaderPtr,
107                       uint64_t Size, Error *Err);
108 
109   std::unique_ptr<AbstractArchiveMemberHeader> clone() const override {
110     return std::make_unique<ArchiveMemberHeader>(*this);
111   }
112 
113   Expected<StringRef> getRawName() const override;
114 
115   Expected<StringRef> getName(uint64_t Size) const override;
116   Expected<uint64_t> getSize() const override;
117   Expected<const char *> getNextChildLoc() const override;
118   Expected<bool> isThin() const override;
119 };
120 
121 // File Member Header
122 struct BigArMemHdrType {
123   char Size[20];       // File member size in decimal
124   char NextOffset[20]; // Next member offset in decimal
125   char PrevOffset[20]; // Previous member offset in decimal
126   char LastModified[12];
127   char UID[12];
128   char GID[12];
129   char AccessMode[12];
130   char NameLen[4]; // File member name length in decimal
131   union {
132     char Name[2]; // Start of member name
133     char Terminator[2];
134   };
135 };
136 
137 // Define file member header of AIX big archive.
138 class BigArchiveMemberHeader
139     : public CommonArchiveMemberHeader<BigArMemHdrType> {
140 
141 public:
142   BigArchiveMemberHeader(Archive const *Parent, const char *RawHeaderPtr,
143                          uint64_t Size, Error *Err);
144   std::unique_ptr<AbstractArchiveMemberHeader> clone() const override {
145     return std::make_unique<BigArchiveMemberHeader>(*this);
146   }
147 
148   Expected<StringRef> getRawName() const override;
149   Expected<uint64_t> getRawNameSize() const;
150 
151   Expected<StringRef> getName(uint64_t Size) const override;
152   Expected<uint64_t> getSize() const override;
153   Expected<const char *> getNextChildLoc() const override;
154   Expected<uint64_t> getNextOffset() const;
155   Expected<bool> isThin() const override { return false; }
156 };
157 
158 class Archive : public Binary {
159   virtual void anchor();
160 
161 public:
162   class Child {
163     friend Archive;
164     friend AbstractArchiveMemberHeader;
165 
166     const Archive *Parent;
167     std::unique_ptr<AbstractArchiveMemberHeader> Header;
168     /// Includes header but not padding byte.
169     StringRef Data;
170     /// Offset from Data to the start of the file.
171     uint16_t StartOfFile;
172 
173     Expected<bool> isThinMember() const;
174 
175   public:
176     Child(const Archive *Parent, const char *Start, Error *Err);
177     Child(const Archive *Parent, StringRef Data, uint16_t StartOfFile);
178 
179     Child(const Child &C)
180         : Parent(C.Parent), Data(C.Data), StartOfFile(C.StartOfFile) {
181       if (C.Header)
182         Header = C.Header->clone();
183     }
184 
185     Child(Child &&C) {
186       Parent = std::move(C.Parent);
187       Header = std::move(C.Header);
188       Data = C.Data;
189       StartOfFile = C.StartOfFile;
190     }
191 
192     Child &operator=(Child &&C) noexcept {
193       if (&C == this)
194         return *this;
195 
196       Parent = std::move(C.Parent);
197       Header = std::move(C.Header);
198       Data = C.Data;
199       StartOfFile = C.StartOfFile;
200 
201       return *this;
202     }
203 
204     Child &operator=(const Child &C) {
205       if (&C == this)
206         return *this;
207 
208       Parent = C.Parent;
209       if (C.Header)
210         Header = C.Header->clone();
211       Data = C.Data;
212       StartOfFile = C.StartOfFile;
213 
214       return *this;
215     }
216 
217     bool operator==(const Child &other) const {
218       assert(!Parent || !other.Parent || Parent == other.Parent);
219       return Data.begin() == other.Data.begin();
220     }
221 
222     const Archive *getParent() const { return Parent; }
223     Expected<Child> getNext() const;
224 
225     Expected<StringRef> getName() const;
226     Expected<std::string> getFullName() const;
227     Expected<StringRef> getRawName() const { return Header->getRawName(); }
228 
229     Expected<sys::TimePoint<std::chrono::seconds>> getLastModified() const {
230       return Header->getLastModified();
231     }
232 
233     StringRef getRawLastModified() const {
234       return Header->getRawLastModified();
235     }
236 
237     Expected<unsigned> getUID() const { return Header->getUID(); }
238     Expected<unsigned> getGID() const { return Header->getGID(); }
239 
240     Expected<sys::fs::perms> getAccessMode() const {
241       return Header->getAccessMode();
242     }
243 
244     /// \return the size of the archive member without the header or padding.
245     Expected<uint64_t> getSize() const;
246     /// \return the size in the archive header for this member.
247     Expected<uint64_t> getRawSize() const;
248 
249     Expected<StringRef> getBuffer() const;
250     uint64_t getChildOffset() const;
251     uint64_t getDataOffset() const { return getChildOffset() + StartOfFile; }
252 
253     Expected<MemoryBufferRef> getMemoryBufferRef() const;
254 
255     Expected<std::unique_ptr<Binary>>
256     getAsBinary(LLVMContext *Context = nullptr) const;
257   };
258 
259   class ChildFallibleIterator {
260     Child C;
261 
262   public:
263     ChildFallibleIterator() : C(Child(nullptr, nullptr, nullptr)) {}
264     ChildFallibleIterator(const Child &C) : C(C) {}
265 
266     const Child *operator->() const { return &C; }
267     const Child &operator*() const { return C; }
268 
269     bool operator==(const ChildFallibleIterator &other) const {
270       // Ignore errors here: If an error occurred during increment then getNext
271       // will have been set to child_end(), and the following comparison should
272       // do the right thing.
273       return C == other.C;
274     }
275 
276     bool operator!=(const ChildFallibleIterator &other) const {
277       return !(*this == other);
278     }
279 
280     Error inc() {
281       auto NextChild = C.getNext();
282       if (!NextChild)
283         return NextChild.takeError();
284       C = std::move(*NextChild);
285       return Error::success();
286     }
287   };
288 
289   using child_iterator = fallible_iterator<ChildFallibleIterator>;
290 
291   class Symbol {
292     const Archive *Parent;
293     uint32_t SymbolIndex;
294     uint32_t StringIndex; // Extra index to the string.
295 
296   public:
297     Symbol(const Archive *p, uint32_t symi, uint32_t stri)
298         : Parent(p), SymbolIndex(symi), StringIndex(stri) {}
299 
300     bool operator==(const Symbol &other) const {
301       return (Parent == other.Parent) && (SymbolIndex == other.SymbolIndex);
302     }
303 
304     StringRef getName() const;
305     Expected<Child> getMember() const;
306     Symbol getNext() const;
307   };
308 
309   class symbol_iterator {
310     Symbol symbol;
311 
312   public:
313     symbol_iterator(const Symbol &s) : symbol(s) {}
314 
315     const Symbol *operator->() const { return &symbol; }
316     const Symbol &operator*() const { return symbol; }
317 
318     bool operator==(const symbol_iterator &other) const {
319       return symbol == other.symbol;
320     }
321 
322     bool operator!=(const symbol_iterator &other) const {
323       return !(*this == other);
324     }
325 
326     symbol_iterator &operator++() { // Preincrement
327       symbol = symbol.getNext();
328       return *this;
329     }
330   };
331 
332   Archive(MemoryBufferRef Source, Error &Err);
333   static Expected<std::unique_ptr<Archive>> create(MemoryBufferRef Source);
334 
335   /// Size field is 10 decimal digits long
336   static const uint64_t MaxMemberSize = 9999999999;
337 
338   enum Kind { K_GNU, K_GNU64, K_BSD, K_DARWIN, K_DARWIN64, K_COFF, K_AIXBIG };
339 
340   Kind kind() const { return (Kind)Format; }
341   bool isThin() const { return IsThin; }
342 
343   child_iterator child_begin(Error &Err, bool SkipInternal = true) const;
344   child_iterator child_end() const;
345   iterator_range<child_iterator> children(Error &Err,
346                                           bool SkipInternal = true) const {
347     return make_range(child_begin(Err, SkipInternal), child_end());
348   }
349 
350   symbol_iterator symbol_begin() const;
351   symbol_iterator symbol_end() const;
352   iterator_range<symbol_iterator> symbols() const {
353     return make_range(symbol_begin(), symbol_end());
354   }
355 
356   static bool classof(Binary const *v) { return v->isArchive(); }
357 
358   // check if a symbol is in the archive
359   Expected<Optional<Child>> findSym(StringRef name) const;
360 
361   bool isEmpty() const;
362   bool hasSymbolTable() const;
363   StringRef getSymbolTable() const { return SymbolTable; }
364   StringRef getStringTable() const { return StringTable; }
365   uint32_t getNumberOfSymbols() const;
366   virtual uint64_t getFirstChildOffset() const { return getArchiveMagicLen(); }
367 
368   std::vector<std::unique_ptr<MemoryBuffer>> takeThinBuffers() {
369     return std::move(ThinBuffers);
370   }
371 
372   std::unique_ptr<AbstractArchiveMemberHeader>
373   createArchiveMemberHeader(const char *RawHeaderPtr, uint64_t Size,
374                             Error *Err) const;
375 
376 protected:
377   uint64_t getArchiveMagicLen() const;
378   void setFirstRegular(const Child &C);
379 
380 private:
381   StringRef SymbolTable;
382   StringRef StringTable;
383 
384   StringRef FirstRegularData;
385   uint16_t FirstRegularStartOfFile = -1;
386 
387   unsigned Format : 3;
388   unsigned IsThin : 1;
389   mutable std::vector<std::unique_ptr<MemoryBuffer>> ThinBuffers;
390 };
391 
392 class BigArchive : public Archive {
393   /// Fixed-Length Header.
394   struct FixLenHdr {
395     char Magic[sizeof(BigArchiveMagic) - 1]; ///< Big archive magic string.
396     char MemOffset[20];                      ///< Offset to member table.
397     char GlobSymOffset[20];                  ///< Offset to global symbol table.
398     char
399         GlobSym64Offset[20]; ///< Offset global symbol table for 64-bit objects.
400     char FirstChildOffset[20]; ///< Offset to first archive member.
401     char LastChildOffset[20];  ///< Offset to last archive member.
402     char FreeOffset[20];       ///< Offset to first mem on free list.
403   };
404 
405   const FixLenHdr *ArFixLenHdr;
406   uint64_t FirstChildOffset = 0;
407   uint64_t LastChildOffset = 0;
408 
409 public:
410   BigArchive(MemoryBufferRef Source, Error &Err);
411   uint64_t getFirstChildOffset() const override { return FirstChildOffset; }
412   uint64_t getLastChildOffset() const { return LastChildOffset; }
413 };
414 
415 } // end namespace object
416 } // end namespace llvm
417 
418 #endif // LLVM_OBJECT_ARCHIVE_H
419