1 //===- MachOUniversalWriter.cpp - MachO universal binary writer---*- 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 // Defines the Slice class and writeUniversalBinary function for writing a MachO
10 // universal binary file.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "llvm/Object/MachOUniversalWriter.h"
15 #include "llvm/ADT/STLExtras.h"
16 #include "llvm/ADT/SmallVector.h"
17 #include "llvm/ADT/Triple.h"
18 #include "llvm/Object/Archive.h"
19 #include "llvm/Object/Binary.h"
20 #include "llvm/Object/IRObjectFile.h"
21 #include "llvm/Object/MachO.h"
22 #include "llvm/Object/MachOUniversal.h"
23 #include "llvm/Support/Casting.h"
24 #include "llvm/Support/ErrorHandling.h"
25 #include "llvm/Support/FileSystem.h"
26 #include "llvm/Support/MathExtras.h"
27 #include "llvm/Support/MemoryBufferRef.h"
28 #include "llvm/Support/SwapByteOrder.h"
29 #include "llvm/Support/raw_ostream.h"
30 
31 using namespace llvm;
32 using namespace object;
33 
34 // For compatibility with cctools lipo, a file's alignment is calculated as the
35 // minimum aligment of all segments. For object files, the file's alignment is
36 // the maximum alignment of its sections.
37 static uint32_t calculateFileAlignment(const MachOObjectFile &O) {
38   uint32_t P2CurrentAlignment;
39   uint32_t P2MinAlignment = MachOUniversalBinary::MaxSectionAlignment;
40   const bool Is64Bit = O.is64Bit();
41 
42   for (const auto &LC : O.load_commands()) {
43     if (LC.C.cmd != (Is64Bit ? MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT))
44       continue;
45     if (O.getHeader().filetype == MachO::MH_OBJECT) {
46       unsigned NumberOfSections =
47           (Is64Bit ? O.getSegment64LoadCommand(LC).nsects
48                    : O.getSegmentLoadCommand(LC).nsects);
49       P2CurrentAlignment = NumberOfSections ? 2 : P2MinAlignment;
50       for (unsigned SI = 0; SI < NumberOfSections; ++SI) {
51         P2CurrentAlignment = std::max(P2CurrentAlignment,
52                                       (Is64Bit ? O.getSection64(LC, SI).align
53                                                : O.getSection(LC, SI).align));
54       }
55     } else {
56       P2CurrentAlignment =
57           countTrailingZeros(Is64Bit ? O.getSegment64LoadCommand(LC).vmaddr
58                                      : O.getSegmentLoadCommand(LC).vmaddr);
59     }
60     P2MinAlignment = std::min(P2MinAlignment, P2CurrentAlignment);
61   }
62   // return a value >= 4 byte aligned, and less than MachO MaxSectionAlignment
63   return std::max(
64       static_cast<uint32_t>(2),
65       std::min(P2MinAlignment, static_cast<uint32_t>(
66                                    MachOUniversalBinary::MaxSectionAlignment)));
67 }
68 
69 static uint32_t calculateAlignment(const MachOObjectFile &ObjectFile) {
70   switch (ObjectFile.getHeader().cputype) {
71   case MachO::CPU_TYPE_I386:
72   case MachO::CPU_TYPE_X86_64:
73   case MachO::CPU_TYPE_POWERPC:
74   case MachO::CPU_TYPE_POWERPC64:
75     return 12; // log2 value of page size(4k) for x86 and PPC
76   case MachO::CPU_TYPE_ARM:
77   case MachO::CPU_TYPE_ARM64:
78   case MachO::CPU_TYPE_ARM64_32:
79     return 14; // log2 value of page size(16k) for Darwin ARM
80   default:
81     return calculateFileAlignment(ObjectFile);
82   }
83 }
84 
85 Slice::Slice(const Archive &A, uint32_t CPUType, uint32_t CPUSubType,
86              std::string ArchName, uint32_t Align)
87     : B(&A), CPUType(CPUType), CPUSubType(CPUSubType),
88       ArchName(std::move(ArchName)), P2Alignment(Align) {}
89 
90 Slice::Slice(const MachOObjectFile &O, uint32_t Align)
91     : B(&O), CPUType(O.getHeader().cputype),
92       CPUSubType(O.getHeader().cpusubtype),
93       ArchName(std::string(O.getArchTriple().getArchName())),
94       P2Alignment(Align) {}
95 
96 Slice::Slice(const IRObjectFile &IRO, uint32_t CPUType, uint32_t CPUSubType,
97              std::string ArchName, uint32_t Align)
98     : B(&IRO), CPUType(CPUType), CPUSubType(CPUSubType),
99       ArchName(std::move(ArchName)), P2Alignment(Align) {}
100 
101 Slice::Slice(const MachOObjectFile &O) : Slice(O, calculateAlignment(O)) {}
102 
103 using MachoCPUTy = std::pair<unsigned, unsigned>;
104 
105 static Expected<MachoCPUTy> getMachoCPUFromTriple(Triple TT) {
106   auto CPU = std::make_pair(MachO::getCPUType(TT), MachO::getCPUSubType(TT));
107   if (!CPU.first) {
108     return CPU.first.takeError();
109   }
110   if (!CPU.second) {
111     return CPU.second.takeError();
112   }
113   return std::make_pair(*CPU.first, *CPU.second);
114 }
115 
116 static Expected<MachoCPUTy> getMachoCPUFromTriple(StringRef TT) {
117   return getMachoCPUFromTriple(Triple{TT});
118 }
119 
120 Expected<Slice> Slice::create(const Archive &A, LLVMContext *LLVMCtx) {
121   Error Err = Error::success();
122   std::unique_ptr<MachOObjectFile> MFO = nullptr;
123   std::unique_ptr<IRObjectFile> IRFO = nullptr;
124   for (const Archive::Child &Child : A.children(Err)) {
125     Expected<std::unique_ptr<Binary>> ChildOrErr = Child.getAsBinary(LLVMCtx);
126     if (!ChildOrErr)
127       return createFileError(A.getFileName(), ChildOrErr.takeError());
128     Binary *Bin = ChildOrErr.get().get();
129     if (Bin->isMachOUniversalBinary())
130       return createStringError(std::errc::invalid_argument,
131                                ("archive member " + Bin->getFileName() +
132                                 " is a fat file (not allowed in an archive)")
133                                    .str()
134                                    .c_str());
135     if (Bin->isMachO()) {
136       MachOObjectFile *O = cast<MachOObjectFile>(Bin);
137       if (IRFO) {
138         return createStringError(
139             std::errc::invalid_argument,
140             "archive member %s is a MachO, while previous archive member "
141             "%s was an IR LLVM object",
142             O->getFileName().str().c_str(), IRFO->getFileName().str().c_str());
143       }
144       if (MFO &&
145           std::tie(MFO->getHeader().cputype, MFO->getHeader().cpusubtype) !=
146               std::tie(O->getHeader().cputype, O->getHeader().cpusubtype)) {
147         return createStringError(
148             std::errc::invalid_argument,
149             ("archive member " + O->getFileName() + " cputype (" +
150              Twine(O->getHeader().cputype) + ") and cpusubtype(" +
151              Twine(O->getHeader().cpusubtype) +
152              ") does not match previous archive members cputype (" +
153              Twine(MFO->getHeader().cputype) + ") and cpusubtype(" +
154              Twine(MFO->getHeader().cpusubtype) +
155              ") (all members must match) " + MFO->getFileName())
156                 .str()
157                 .c_str());
158       }
159       if (!MFO) {
160         ChildOrErr.get().release();
161         MFO.reset(O);
162       }
163     } else if (Bin->isIR()) {
164       IRObjectFile *O = cast<IRObjectFile>(Bin);
165       if (MFO) {
166         return createStringError(std::errc::invalid_argument,
167                                  "archive member '%s' is an LLVM IR object, "
168                                  "while previous archive member "
169                                  "'%s' was a MachO",
170                                  O->getFileName().str().c_str(),
171                                  MFO->getFileName().str().c_str());
172       }
173       if (IRFO) {
174         Expected<MachoCPUTy> CPUO = getMachoCPUFromTriple(O->getTargetTriple());
175         Expected<MachoCPUTy> CPUFO =
176             getMachoCPUFromTriple(IRFO->getTargetTriple());
177         if (!CPUO)
178           return CPUO.takeError();
179         if (!CPUFO)
180           return CPUFO.takeError();
181         if (*CPUO != *CPUFO) {
182           return createStringError(
183               std::errc::invalid_argument,
184               ("archive member " + O->getFileName() + " cputype (" +
185                Twine(CPUO->first) + ") and cpusubtype(" + Twine(CPUO->second) +
186                ") does not match previous archive members cputype (" +
187                Twine(CPUFO->first) + ") and cpusubtype(" +
188                Twine(CPUFO->second) + ") (all members must match) " +
189                IRFO->getFileName())
190                   .str()
191                   .c_str());
192         }
193       } else {
194         ChildOrErr.get().release();
195         IRFO.reset(O);
196       }
197     } else
198       return createStringError(std::errc::invalid_argument,
199                                ("archive member " + Bin->getFileName() +
200                                 " is neither a MachO file or an LLVM IR file "
201                                 "(not allowed in an archive)")
202                                    .str()
203                                    .c_str());
204   }
205   if (Err)
206     return createFileError(A.getFileName(), std::move(Err));
207   if (!MFO && !IRFO)
208     return createStringError(
209         std::errc::invalid_argument,
210         ("empty archive with no architecture specification: " +
211          A.getFileName() + " (can't determine architecture for it)")
212             .str()
213             .c_str());
214 
215   if (MFO) {
216     Slice ArchiveSlice(*(MFO), MFO->is64Bit() ? 3 : 2);
217     ArchiveSlice.B = &A;
218     return ArchiveSlice;
219   }
220 
221   // For IR objects
222   Expected<Slice> ArchiveSliceOrErr = Slice::create(*IRFO, 0);
223   if (!ArchiveSliceOrErr)
224     return createFileError(A.getFileName(), ArchiveSliceOrErr.takeError());
225   auto &ArchiveSlice = ArchiveSliceOrErr.get();
226   ArchiveSlice.B = &A;
227   return std::move(ArchiveSlice);
228 }
229 
230 Expected<Slice> Slice::create(const IRObjectFile &IRO, uint32_t Align) {
231   Expected<MachoCPUTy> CPUOrErr = getMachoCPUFromTriple(IRO.getTargetTriple());
232   if (!CPUOrErr)
233     return CPUOrErr.takeError();
234   unsigned CPUType, CPUSubType;
235   std::tie(CPUType, CPUSubType) = CPUOrErr.get();
236   // We don't directly use the architecture name of the target triple T, as,
237   // for instance, thumb is treated as ARM by the MachOUniversal object.
238   std::string ArchName(
239       MachOObjectFile::getArchTriple(CPUType, CPUSubType).getArchName());
240   return Slice{IRO, CPUType, CPUSubType, std::move(ArchName), Align};
241 }
242 
243 static Expected<SmallVector<MachO::fat_arch, 2>>
244 buildFatArchList(ArrayRef<Slice> Slices) {
245   SmallVector<MachO::fat_arch, 2> FatArchList;
246   uint64_t Offset =
247       sizeof(MachO::fat_header) + Slices.size() * sizeof(MachO::fat_arch);
248 
249   for (const auto &S : Slices) {
250     Offset = alignTo(Offset, 1ull << S.getP2Alignment());
251     if (Offset > UINT32_MAX)
252       return createStringError(
253           std::errc::invalid_argument,
254           ("fat file too large to be created because the offset "
255            "field in struct fat_arch is only 32-bits and the offset " +
256            Twine(Offset) + " for " + S.getBinary()->getFileName() +
257            " for architecture " + S.getArchString() + "exceeds that.")
258               .str()
259               .c_str());
260 
261     MachO::fat_arch FatArch;
262     FatArch.cputype = S.getCPUType();
263     FatArch.cpusubtype = S.getCPUSubType();
264     FatArch.offset = Offset;
265     FatArch.size = S.getBinary()->getMemoryBufferRef().getBufferSize();
266     FatArch.align = S.getP2Alignment();
267     Offset += FatArch.size;
268     FatArchList.push_back(FatArch);
269   }
270   return FatArchList;
271 }
272 
273 Error object::writeUniversalBinaryToStream(ArrayRef<Slice> Slices,
274                                            raw_ostream &Out) {
275   MachO::fat_header FatHeader;
276   FatHeader.magic = MachO::FAT_MAGIC;
277   FatHeader.nfat_arch = Slices.size();
278 
279   Expected<SmallVector<MachO::fat_arch, 2>> FatArchListOrErr =
280       buildFatArchList(Slices);
281   if (!FatArchListOrErr)
282     return FatArchListOrErr.takeError();
283   SmallVector<MachO::fat_arch, 2> FatArchList = *FatArchListOrErr;
284 
285   if (sys::IsLittleEndianHost)
286     MachO::swapStruct(FatHeader);
287   Out.write(reinterpret_cast<const char *>(&FatHeader),
288             sizeof(MachO::fat_header));
289 
290   if (sys::IsLittleEndianHost)
291     for (MachO::fat_arch &FA : FatArchList)
292       MachO::swapStruct(FA);
293   Out.write(reinterpret_cast<const char *>(FatArchList.data()),
294             sizeof(MachO::fat_arch) * FatArchList.size());
295 
296   if (sys::IsLittleEndianHost)
297     for (MachO::fat_arch &FA : FatArchList)
298       MachO::swapStruct(FA);
299 
300   size_t Offset =
301       sizeof(MachO::fat_header) + sizeof(MachO::fat_arch) * FatArchList.size();
302   for (size_t Index = 0, Size = Slices.size(); Index < Size; ++Index) {
303     MemoryBufferRef BufferRef = Slices[Index].getBinary()->getMemoryBufferRef();
304     assert((Offset <= FatArchList[Index].offset) && "Incorrect slice offset");
305     Out.write_zeros(FatArchList[Index].offset - Offset);
306     Out.write(BufferRef.getBufferStart(), BufferRef.getBufferSize());
307     Offset = FatArchList[Index].offset + BufferRef.getBufferSize();
308   }
309 
310   Out.flush();
311   return Error::success();
312 }
313 
314 Error object::writeUniversalBinary(ArrayRef<Slice> Slices,
315                                    StringRef OutputFileName) {
316   const bool IsExecutable = any_of(Slices, [](Slice S) {
317     return sys::fs::can_execute(S.getBinary()->getFileName());
318   });
319   unsigned Mode = sys::fs::all_read | sys::fs::all_write;
320   if (IsExecutable)
321     Mode |= sys::fs::all_exe;
322   Expected<sys::fs::TempFile> Temp = sys::fs::TempFile::create(
323       OutputFileName + ".temp-universal-%%%%%%", Mode);
324   if (!Temp)
325     return Temp.takeError();
326   raw_fd_ostream Out(Temp->FD, false);
327   if (Error E = writeUniversalBinaryToStream(Slices, Out)) {
328     if (Error DiscardError = Temp->discard())
329       return joinErrors(std::move(E), std::move(DiscardError));
330     return E;
331   }
332   return Temp->keep(OutputFileName);
333 }
334