blob: bc7c83ed7d8c444e5f5398dac7fe78809d65a109 [file] [log] [blame]
Andreas Gampe54fc26c2014-09-04 21:47:42 -07001/*
David Srbeckybc90fd02015-04-22 19:40:27 +01002 * Copyright (C) 2015 The Android Open Source Project
Andreas Gampe54fc26c2014-09-04 21:47:42 -07003 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#ifndef ART_COMPILER_ELF_BUILDER_H_
18#define ART_COMPILER_ELF_BUILDER_H_
19
David Srbeckybc90fd02015-04-22 19:40:27 +010020#include <vector>
21
Ian Rogersd582fa42014-11-05 23:46:43 -080022#include "arch/instruction_set.h"
Vladimir Marko80afd022015-05-19 18:08:00 +010023#include "base/bit_utils.h"
David Srbecky6d8c8f02015-10-26 10:57:09 +000024#include "base/casts.h"
David Srbeckybc90fd02015-04-22 19:40:27 +010025#include "base/unix_file/fd_file.h"
Andreas Gampe54fc26c2014-09-04 21:47:42 -070026#include "elf_utils.h"
David Srbecky6d8c8f02015-10-26 10:57:09 +000027#include "leb128.h"
Vladimir Marko131980f2015-12-03 18:29:23 +000028#include "linker/error_delaying_output_stream.h"
Vladimir Marko10c13562015-11-25 14:33:36 +000029#include "utils/array_ref.h"
Andreas Gampe54fc26c2014-09-04 21:47:42 -070030
31namespace art {
32
David Srbeckybc90fd02015-04-22 19:40:27 +010033// Writes ELF file.
David Srbecky6d8c8f02015-10-26 10:57:09 +000034//
35// The basic layout of the elf file:
36// Elf_Ehdr - The ELF header.
37// Elf_Phdr[] - Program headers for the linker.
38// .rodata - DEX files and oat metadata.
39// .text - Compiled code.
40// .bss - Zero-initialized writeable section.
41// .dynstr - Names for .dynsym.
42// .dynsym - A few oat-specific dynamic symbols.
43// .hash - Hash-table for .dynsym.
44// .dynamic - Tags which let the linker locate .dynsym.
45// .strtab - Names for .symtab.
46// .symtab - Debug symbols.
47// .eh_frame - Unwind information (CFI).
48// .eh_frame_hdr - Index of .eh_frame.
49// .debug_frame - Unwind information (CFI).
50// .debug_frame.oat_patches - Addresses for relocation.
51// .debug_info - Debug information.
52// .debug_info.oat_patches - Addresses for relocation.
53// .debug_abbrev - Decoding information for .debug_info.
54// .debug_str - Strings for .debug_info.
55// .debug_line - Line number tables.
56// .debug_line.oat_patches - Addresses for relocation.
57// .text.oat_patches - Addresses for relocation.
58// .shstrtab - Names of ELF sections.
59// Elf_Shdr[] - Section headers.
60//
61// Some section are optional (the debug sections in particular).
62//
63// We try write the section data directly into the file without much
64// in-memory buffering. This means we generally write sections based on the
65// dependency order (e.g. .dynamic points to .dynsym which points to .text).
66//
67// In the cases where we need to buffer, we write the larger section first
68// and buffer the smaller one (e.g. .strtab is bigger than .symtab).
69//
70// The debug sections are written last for easier stripping.
71//
David Srbecky533c2072015-04-22 12:20:22 +010072template <typename ElfTypes>
Andreas Gampe54fc26c2014-09-04 21:47:42 -070073class ElfBuilder FINAL {
74 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +000075 static constexpr size_t kMaxProgramHeaders = 16;
David Srbecky533c2072015-04-22 12:20:22 +010076 using Elf_Addr = typename ElfTypes::Addr;
David Srbeckybc90fd02015-04-22 19:40:27 +010077 using Elf_Off = typename ElfTypes::Off;
David Srbecky533c2072015-04-22 12:20:22 +010078 using Elf_Word = typename ElfTypes::Word;
79 using Elf_Sword = typename ElfTypes::Sword;
80 using Elf_Ehdr = typename ElfTypes::Ehdr;
81 using Elf_Shdr = typename ElfTypes::Shdr;
82 using Elf_Sym = typename ElfTypes::Sym;
83 using Elf_Phdr = typename ElfTypes::Phdr;
84 using Elf_Dyn = typename ElfTypes::Dyn;
85
David Srbeckybc90fd02015-04-22 19:40:27 +010086 // Base class of all sections.
David Srbecky6d8c8f02015-10-26 10:57:09 +000087 class Section : public OutputStream {
David Srbecky0c5bbc12015-04-28 17:54:52 +010088 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +000089 Section(ElfBuilder<ElfTypes>* owner, const std::string& name,
90 Elf_Word type, Elf_Word flags, const Section* link,
91 Elf_Word info, Elf_Word align, Elf_Word entsize)
92 : OutputStream(name), owner_(owner), header_(),
93 section_index_(0), name_(name), link_(link),
94 started_(false), finished_(false), phdr_flags_(PF_R), phdr_type_(0) {
95 DCHECK_GE(align, 1u);
David Srbecky491a7fe2015-05-28 00:59:08 +010096 header_.sh_type = type;
97 header_.sh_flags = flags;
98 header_.sh_info = info;
99 header_.sh_addralign = align;
100 header_.sh_entsize = entsize;
David Srbecky0c5bbc12015-04-28 17:54:52 +0100101 }
David Srbecky0c5bbc12015-04-28 17:54:52 +0100102
David Srbecky6d8c8f02015-10-26 10:57:09 +0000103 // Start writing of this section.
104 void Start() {
105 CHECK(!started_);
106 CHECK(!finished_);
107 started_ = true;
108 auto& sections = owner_->sections_;
109 // Check that the previous section is complete.
110 CHECK(sections.empty() || sections.back()->finished_);
111 // The first ELF section index is 1. Index 0 is reserved for NULL.
112 section_index_ = sections.size() + 1;
David Srbecky579942f2016-01-28 20:01:28 +0000113 // Page-align if we switch between allocated and non-allocated sections,
114 // or if we change the type of allocation (e.g. executable vs non-executable).
115 if (!sections.empty()) {
116 if (header_.sh_flags != sections.back()->header_.sh_flags) {
117 header_.sh_addralign = kPageSize;
118 }
119 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000120 // Align file position.
121 if (header_.sh_type != SHT_NOBITS) {
David Srbecky579942f2016-01-28 20:01:28 +0000122 header_.sh_offset = owner_->AlignFileOffset(header_.sh_addralign);
123 } else {
124 header_.sh_offset = 0;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000125 }
126 // Align virtual memory address.
127 if ((header_.sh_flags & SHF_ALLOC) != 0) {
David Srbecky579942f2016-01-28 20:01:28 +0000128 header_.sh_addr = owner_->AlignVirtualAddress(header_.sh_addralign);
129 } else {
130 header_.sh_addr = 0;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000131 }
David Srbecky579942f2016-01-28 20:01:28 +0000132 // Push this section on the list of written sections.
133 sections.push_back(this);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100134 }
135
David Srbecky6d8c8f02015-10-26 10:57:09 +0000136 // Finish writing of this section.
137 void End() {
138 CHECK(started_);
139 CHECK(!finished_);
140 finished_ = true;
141 if (header_.sh_type == SHT_NOBITS) {
142 CHECK_GT(header_.sh_size, 0u);
143 } else {
144 // Use the current file position to determine section size.
Vladimir Marko131980f2015-12-03 18:29:23 +0000145 off_t file_offset = owner_->stream_.Seek(0, kSeekCurrent);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000146 CHECK_GE(file_offset, (off_t)header_.sh_offset);
147 header_.sh_size = file_offset - header_.sh_offset;
148 }
149 if ((header_.sh_flags & SHF_ALLOC) != 0) {
150 owner_->virtual_address_ += header_.sh_size;
151 }
152 }
153
David Srbecky5cc349f2015-12-18 15:04:48 +0000154 // Returns true if the section was written to disk.
155 // (Used to check whether we have .text when writing JIT debug info)
156 bool Exists() const {
157 return finished_;
158 }
159
David Srbecky6d8c8f02015-10-26 10:57:09 +0000160 // Get the location of this section in virtual memory.
161 Elf_Addr GetAddress() const {
162 CHECK(started_);
163 return header_.sh_addr;
164 }
165
166 // Returns the size of the content of this section.
167 Elf_Word GetSize() const {
David Srbeckyb851b492015-11-11 20:19:38 +0000168 if (finished_) {
169 return header_.sh_size;
170 } else {
171 CHECK(started_);
172 CHECK_NE(header_.sh_type, (Elf_Word)SHT_NOBITS);
Vladimir Marko131980f2015-12-03 18:29:23 +0000173 return owner_->stream_.Seek(0, kSeekCurrent) - header_.sh_offset;
David Srbeckyb851b492015-11-11 20:19:38 +0000174 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000175 }
176
David Srbecky5b1c2ca2016-01-25 17:32:41 +0000177 // Write this section as "NOBITS" section. (used for the .bss section)
178 // This means that the ELF file does not contain the initial data for this section
179 // and it will be zero-initialized when the ELF file is loaded in the running program.
180 void WriteNoBitsSection(Elf_Word size) {
181 DCHECK_NE(header_.sh_flags & SHF_ALLOC, 0u);
David Srbecky5b1c2ca2016-01-25 17:32:41 +0000182 header_.sh_type = SHT_NOBITS;
David Srbecky579942f2016-01-28 20:01:28 +0000183 Start();
David Srbecky6d8c8f02015-10-26 10:57:09 +0000184 header_.sh_size = size;
David Srbecky5b1c2ca2016-01-25 17:32:41 +0000185 End();
David Srbecky6d8c8f02015-10-26 10:57:09 +0000186 }
187
188 // This function always succeeds to simplify code.
189 // Use builder's Good() to check the actual status.
190 bool WriteFully(const void* buffer, size_t byte_count) OVERRIDE {
191 CHECK(started_);
192 CHECK(!finished_);
Vladimir Marko131980f2015-12-03 18:29:23 +0000193 return owner_->stream_.WriteFully(buffer, byte_count);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000194 }
195
196 // This function always succeeds to simplify code.
197 // Use builder's Good() to check the actual status.
198 off_t Seek(off_t offset, Whence whence) OVERRIDE {
199 // Forward the seek as-is and trust the caller to use it reasonably.
Vladimir Marko131980f2015-12-03 18:29:23 +0000200 return owner_->stream_.Seek(offset, whence);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100201 }
202
Vladimir Marko10c13562015-11-25 14:33:36 +0000203 // This function flushes the output and returns whether it succeeded.
204 // If there was a previous failure, this does nothing and returns false, i.e. failed.
205 bool Flush() OVERRIDE {
Vladimir Marko131980f2015-12-03 18:29:23 +0000206 return owner_->stream_.Flush();
Vladimir Marko10c13562015-11-25 14:33:36 +0000207 }
208
David Srbecky0c5bbc12015-04-28 17:54:52 +0100209 Elf_Word GetSectionIndex() const {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000210 DCHECK(started_);
David Srbeckybc90fd02015-04-22 19:40:27 +0100211 DCHECK_NE(section_index_, 0u);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100212 return section_index_;
213 }
214
David Srbecky0c5bbc12015-04-28 17:54:52 +0100215 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000216 ElfBuilder<ElfTypes>* owner_;
David Srbecky491a7fe2015-05-28 00:59:08 +0100217 Elf_Shdr header_;
David Srbecky0c5bbc12015-04-28 17:54:52 +0100218 Elf_Word section_index_;
219 const std::string name_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100220 const Section* const link_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000221 bool started_;
222 bool finished_;
223 Elf_Word phdr_flags_;
224 Elf_Word phdr_type_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100225
David Srbecky6d8c8f02015-10-26 10:57:09 +0000226 friend class ElfBuilder;
David Srbecky4d247f72015-11-09 11:56:52 +0000227
228 DISALLOW_COPY_AND_ASSIGN(Section);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100229 };
230
David Srbeckybc90fd02015-04-22 19:40:27 +0100231 // Writer of .dynstr .strtab and .shstrtab sections.
David Srbecky6d8c8f02015-10-26 10:57:09 +0000232 class StringSection FINAL : public Section {
David Srbeckybc90fd02015-04-22 19:40:27 +0100233 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000234 StringSection(ElfBuilder<ElfTypes>* owner, const std::string& name,
235 Elf_Word flags, Elf_Word align)
236 : Section(owner, name, SHT_STRTAB, flags, nullptr, 0, align, 0),
237 current_offset_(0) {
David Srbecky0c5bbc12015-04-28 17:54:52 +0100238 }
David Srbecky0c5bbc12015-04-28 17:54:52 +0100239
David Srbecky6d8c8f02015-10-26 10:57:09 +0000240 Elf_Word Write(const std::string& name) {
241 if (current_offset_ == 0) {
242 DCHECK(name.empty());
243 }
244 Elf_Word offset = current_offset_;
245 this->WriteFully(name.c_str(), name.length() + 1);
246 current_offset_ += name.length() + 1;
David Srbeckybc90fd02015-04-22 19:40:27 +0100247 return offset;
248 }
249
David Srbeckybc90fd02015-04-22 19:40:27 +0100250 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000251 Elf_Word current_offset_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100252 };
253
David Srbeckybc90fd02015-04-22 19:40:27 +0100254 // Writer of .dynsym and .symtab sections.
David Srbecky6d8c8f02015-10-26 10:57:09 +0000255 class SymbolSection FINAL : public Section {
David Srbecky0c5bbc12015-04-28 17:54:52 +0100256 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000257 SymbolSection(ElfBuilder<ElfTypes>* owner, const std::string& name,
258 Elf_Word type, Elf_Word flags, StringSection* strtab)
259 : Section(owner, name, type, flags, strtab, 0,
260 sizeof(Elf_Off), sizeof(Elf_Sym)) {
261 }
262
263 // Buffer symbol for this section. It will be written later.
David Srbecky5cc349f2015-12-18 15:04:48 +0000264 // If the symbol's section is null, it will be considered absolute (SHN_ABS).
265 // (we use this in JIT to reference code which is stored outside the debug ELF file)
David Srbecky6d8c8f02015-10-26 10:57:09 +0000266 void Add(Elf_Word name, const Section* section,
267 Elf_Addr addr, bool is_relative, Elf_Word size,
268 uint8_t binding, uint8_t type, uint8_t other = 0) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000269 Elf_Sym sym = Elf_Sym();
270 sym.st_name = name;
271 sym.st_value = addr + (is_relative ? section->GetAddress() : 0);
272 sym.st_size = size;
273 sym.st_other = other;
David Srbecky5cc349f2015-12-18 15:04:48 +0000274 sym.st_shndx = (section != nullptr ? section->GetSectionIndex()
275 : static_cast<Elf_Word>(SHN_ABS));
David Srbecky6d8c8f02015-10-26 10:57:09 +0000276 sym.st_info = (binding << 4) + (type & 0xf);
277 symbols_.push_back(sym);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100278 }
279
David Srbecky6d8c8f02015-10-26 10:57:09 +0000280 void Write() {
281 // The symbol table always has to start with NULL symbol.
282 Elf_Sym null_symbol = Elf_Sym();
283 this->WriteFully(&null_symbol, sizeof(null_symbol));
284 this->WriteFully(symbols_.data(), symbols_.size() * sizeof(symbols_[0]));
285 symbols_.clear();
286 symbols_.shrink_to_fit();
David Srbeckybc90fd02015-04-22 19:40:27 +0100287 }
288
289 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000290 std::vector<Elf_Sym> symbols_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100291 };
292
David Srbecky6d8c8f02015-10-26 10:57:09 +0000293 ElfBuilder(InstructionSet isa, OutputStream* output)
Vladimir Marko131980f2015-12-03 18:29:23 +0000294 : isa_(isa),
295 stream_(output),
296 rodata_(this, ".rodata", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
297 text_(this, ".text", SHT_PROGBITS, SHF_ALLOC | SHF_EXECINSTR, nullptr, 0, kPageSize, 0),
298 bss_(this, ".bss", SHT_NOBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
299 dynstr_(this, ".dynstr", SHF_ALLOC, kPageSize),
300 dynsym_(this, ".dynsym", SHT_DYNSYM, SHF_ALLOC, &dynstr_),
301 hash_(this, ".hash", SHT_HASH, SHF_ALLOC, &dynsym_, 0, sizeof(Elf_Word), sizeof(Elf_Word)),
302 dynamic_(this, ".dynamic", SHT_DYNAMIC, SHF_ALLOC, &dynstr_, 0, kPageSize, sizeof(Elf_Dyn)),
303 eh_frame_(this, ".eh_frame", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
304 eh_frame_hdr_(this, ".eh_frame_hdr", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, 4, 0),
David Srbecky579942f2016-01-28 20:01:28 +0000305 strtab_(this, ".strtab", 0, 1),
Vladimir Marko131980f2015-12-03 18:29:23 +0000306 symtab_(this, ".symtab", SHT_SYMTAB, 0, &strtab_),
307 debug_frame_(this, ".debug_frame", SHT_PROGBITS, 0, nullptr, 0, sizeof(Elf_Addr), 0),
308 debug_info_(this, ".debug_info", SHT_PROGBITS, 0, nullptr, 0, 1, 0),
309 debug_line_(this, ".debug_line", SHT_PROGBITS, 0, nullptr, 0, 1, 0),
310 shstrtab_(this, ".shstrtab", 0, 1),
David Srbecky579942f2016-01-28 20:01:28 +0000311 started_(false),
Vladimir Marko131980f2015-12-03 18:29:23 +0000312 virtual_address_(0) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000313 text_.phdr_flags_ = PF_R | PF_X;
314 bss_.phdr_flags_ = PF_R | PF_W;
315 dynamic_.phdr_flags_ = PF_R | PF_W;
316 dynamic_.phdr_type_ = PT_DYNAMIC;
317 eh_frame_hdr_.phdr_type_ = PT_GNU_EH_FRAME;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700318 }
319 ~ElfBuilder() {}
320
David Srbecky6d8c8f02015-10-26 10:57:09 +0000321 InstructionSet GetIsa() { return isa_; }
322 Section* GetRoData() { return &rodata_; }
323 Section* GetText() { return &text_; }
324 Section* GetBss() { return &bss_; }
325 StringSection* GetStrTab() { return &strtab_; }
326 SymbolSection* GetSymTab() { return &symtab_; }
327 Section* GetEhFrame() { return &eh_frame_; }
328 Section* GetEhFrameHdr() { return &eh_frame_hdr_; }
329 Section* GetDebugFrame() { return &debug_frame_; }
David Srbeckyb851b492015-11-11 20:19:38 +0000330 Section* GetDebugInfo() { return &debug_info_; }
331 Section* GetDebugLine() { return &debug_line_; }
Ian Rogers0279ebb2014-10-08 17:27:48 -0700332
David Srbecky6d8c8f02015-10-26 10:57:09 +0000333 // Encode patch locations as LEB128 list of deltas between consecutive addresses.
334 // (exposed publicly for tests)
Vladimir Marko10c13562015-11-25 14:33:36 +0000335 static void EncodeOatPatches(const ArrayRef<const uintptr_t>& locations,
David Srbecky6d8c8f02015-10-26 10:57:09 +0000336 std::vector<uint8_t>* buffer) {
337 buffer->reserve(buffer->size() + locations.size() * 2); // guess 2 bytes per ULEB128.
338 uintptr_t address = 0; // relative to start of section.
339 for (uintptr_t location : locations) {
340 DCHECK_GE(location, address) << "Patch locations are not in sorted order";
341 EncodeUnsignedLeb128(buffer, dchecked_integral_cast<uint32_t>(location - address));
342 address = location;
343 }
344 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700345
Vladimir Marko10c13562015-11-25 14:33:36 +0000346 void WritePatches(const char* name, const ArrayRef<const uintptr_t>& patch_locations) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000347 std::vector<uint8_t> buffer;
Vladimir Marko10c13562015-11-25 14:33:36 +0000348 EncodeOatPatches(patch_locations, &buffer);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000349 std::unique_ptr<Section> s(new Section(this, name, SHT_OAT_PATCH, 0, nullptr, 0, 1, 0));
350 s->Start();
351 s->WriteFully(buffer.data(), buffer.size());
352 s->End();
353 other_sections_.push_back(std::move(s));
354 }
David Srbecky527c9c72015-04-17 21:14:10 +0100355
David Srbecky6d8c8f02015-10-26 10:57:09 +0000356 void WriteSection(const char* name, const std::vector<uint8_t>* buffer) {
357 std::unique_ptr<Section> s(new Section(this, name, SHT_PROGBITS, 0, nullptr, 0, 1, 0));
358 s->Start();
359 s->WriteFully(buffer->data(), buffer->size());
360 s->End();
361 other_sections_.push_back(std::move(s));
362 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700363
David Srbecky5b1c2ca2016-01-25 17:32:41 +0000364 // Set where the next section will be allocated in the virtual address space.
365 void SetVirtualAddress(Elf_Addr address) {
366 DCHECK_GE(address, virtual_address_);
367 virtual_address_ = address;
368 }
369
David Srbecky579942f2016-01-28 20:01:28 +0000370 // Reserve space for ELF header and program headers.
371 // We do not know the number of headers until later, so
372 // it is easiest to just reserve a fixed amount of space.
373 // Program headers are required for loading by the linker.
374 // It is possible to omit them for ELF files used for debugging.
375 void Start(bool write_program_headers = true) {
376 int size = sizeof(Elf_Ehdr);
377 if (write_program_headers) {
378 size += sizeof(Elf_Phdr) * kMaxProgramHeaders;
379 }
Vladimir Marko131980f2015-12-03 18:29:23 +0000380 stream_.Seek(size, kSeekSet);
David Srbecky579942f2016-01-28 20:01:28 +0000381 started_ = true;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000382 virtual_address_ += size;
David Srbecky579942f2016-01-28 20:01:28 +0000383 write_program_headers_ = write_program_headers;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000384 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700385
David Srbecky6d8c8f02015-10-26 10:57:09 +0000386 void End() {
David Srbecky579942f2016-01-28 20:01:28 +0000387 DCHECK(started_);
388
David Srbecky6d8c8f02015-10-26 10:57:09 +0000389 // Write section names and finish the section headers.
390 shstrtab_.Start();
391 shstrtab_.Write("");
392 for (auto* section : sections_) {
393 section->header_.sh_name = shstrtab_.Write(section->name_);
394 if (section->link_ != nullptr) {
395 section->header_.sh_link = section->link_->GetSectionIndex();
David Srbeckyb0a962c2015-04-28 19:43:56 +0100396 }
David Srbecky527c9c72015-04-17 21:14:10 +0100397 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000398 shstrtab_.End();
David Srbecky527c9c72015-04-17 21:14:10 +0100399
David Srbecky6d8c8f02015-10-26 10:57:09 +0000400 // Write section headers at the end of the ELF file.
401 std::vector<Elf_Shdr> shdrs;
402 shdrs.reserve(1u + sections_.size());
403 shdrs.push_back(Elf_Shdr()); // NULL at index 0.
404 for (auto* section : sections_) {
405 shdrs.push_back(section->header_);
406 }
407 Elf_Off section_headers_offset;
David Srbecky579942f2016-01-28 20:01:28 +0000408 section_headers_offset = AlignFileOffset(sizeof(Elf_Off));
Vladimir Marko131980f2015-12-03 18:29:23 +0000409 stream_.WriteFully(shdrs.data(), shdrs.size() * sizeof(shdrs[0]));
410
411 // Flush everything else before writing the program headers. This should prevent
412 // the OS from reordering writes, so that we don't end up with valid headers
413 // and partially written data if we suddenly lose power, for example.
414 stream_.Flush();
David Srbecky6d8c8f02015-10-26 10:57:09 +0000415
David Srbecky579942f2016-01-28 20:01:28 +0000416 // The main ELF header.
David Srbeckybc90fd02015-04-22 19:40:27 +0100417 Elf_Ehdr elf_header = MakeElfHeader(isa_);
David Srbeckybc90fd02015-04-22 19:40:27 +0100418 elf_header.e_shoff = section_headers_offset;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000419 elf_header.e_shnum = shdrs.size();
David Srbeckybc90fd02015-04-22 19:40:27 +0100420 elf_header.e_shstrndx = shstrtab_.GetSectionIndex();
David Srbecky579942f2016-01-28 20:01:28 +0000421
422 // Program headers (i.e. mmap instructions).
423 std::vector<Elf_Phdr> phdrs;
424 if (write_program_headers_) {
425 phdrs = MakeProgramHeaders();
426 CHECK_LE(phdrs.size(), kMaxProgramHeaders);
427 elf_header.e_phoff = sizeof(Elf_Ehdr);
428 elf_header.e_phnum = phdrs.size();
429 }
430
Vladimir Marko131980f2015-12-03 18:29:23 +0000431 stream_.Seek(0, kSeekSet);
432 stream_.WriteFully(&elf_header, sizeof(elf_header));
433 stream_.WriteFully(phdrs.data(), phdrs.size() * sizeof(phdrs[0]));
434 stream_.Flush();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700435 }
436
David Srbecky6d8c8f02015-10-26 10:57:09 +0000437 // The running program does not have access to section headers
438 // and the loader is not supposed to use them either.
439 // The dynamic sections therefore replicates some of the layout
440 // information like the address and size of .rodata and .text.
441 // It also contains other metadata like the SONAME.
442 // The .dynamic section is found using the PT_DYNAMIC program header.
443 void WriteDynamicSection(const std::string& elf_file_path) {
444 std::string soname(elf_file_path);
445 size_t directory_separator_pos = soname.rfind('/');
446 if (directory_separator_pos != std::string::npos) {
447 soname = soname.substr(directory_separator_pos + 1);
448 }
449
450 dynstr_.Start();
451 dynstr_.Write(""); // dynstr should start with empty string.
452 dynsym_.Add(dynstr_.Write("oatdata"), &rodata_, 0, true,
453 rodata_.GetSize(), STB_GLOBAL, STT_OBJECT);
454 if (text_.GetSize() != 0u) {
455 dynsym_.Add(dynstr_.Write("oatexec"), &text_, 0, true,
456 text_.GetSize(), STB_GLOBAL, STT_OBJECT);
457 dynsym_.Add(dynstr_.Write("oatlastword"), &text_, text_.GetSize() - 4,
458 true, 4, STB_GLOBAL, STT_OBJECT);
459 } else if (rodata_.GetSize() != 0) {
460 // rodata_ can be size 0 for dwarf_test.
461 dynsym_.Add(dynstr_.Write("oatlastword"), &rodata_, rodata_.GetSize() - 4,
462 true, 4, STB_GLOBAL, STT_OBJECT);
463 }
464 if (bss_.finished_) {
465 dynsym_.Add(dynstr_.Write("oatbss"), &bss_,
466 0, true, bss_.GetSize(), STB_GLOBAL, STT_OBJECT);
467 dynsym_.Add(dynstr_.Write("oatbsslastword"), &bss_,
468 bss_.GetSize() - 4, true, 4, STB_GLOBAL, STT_OBJECT);
469 }
470 Elf_Word soname_offset = dynstr_.Write(soname);
471 dynstr_.End();
472
473 dynsym_.Start();
474 dynsym_.Write();
475 dynsym_.End();
476
477 // We do not really need a hash-table since there is so few entries.
478 // However, the hash-table is the only way the linker can actually
479 // determine the number of symbols in .dynsym so it is required.
480 hash_.Start();
481 int count = dynsym_.GetSize() / sizeof(Elf_Sym); // Includes NULL.
482 std::vector<Elf_Word> hash;
483 hash.push_back(1); // Number of buckets.
484 hash.push_back(count); // Number of chains.
485 // Buckets. Having just one makes it linear search.
486 hash.push_back(1); // Point to first non-NULL symbol.
487 // Chains. This creates linked list of symbols.
488 hash.push_back(0); // Dummy entry for the NULL symbol.
489 for (int i = 1; i < count - 1; i++) {
490 hash.push_back(i + 1); // Each symbol points to the next one.
491 }
492 hash.push_back(0); // Last symbol terminates the chain.
493 hash_.WriteFully(hash.data(), hash.size() * sizeof(hash[0]));
494 hash_.End();
495
496 dynamic_.Start();
497 Elf_Dyn dyns[] = {
498 { DT_HASH, { hash_.GetAddress() } },
499 { DT_STRTAB, { dynstr_.GetAddress() } },
500 { DT_SYMTAB, { dynsym_.GetAddress() } },
501 { DT_SYMENT, { sizeof(Elf_Sym) } },
502 { DT_STRSZ, { dynstr_.GetSize() } },
503 { DT_SONAME, { soname_offset } },
504 { DT_NULL, { 0 } },
505 };
506 dynamic_.WriteFully(&dyns, sizeof(dyns));
507 dynamic_.End();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700508 }
509
David Srbecky6d8c8f02015-10-26 10:57:09 +0000510 // Returns true if all writes and seeks on the output stream succeeded.
511 bool Good() {
Vladimir Marko131980f2015-12-03 18:29:23 +0000512 return stream_.Good();
513 }
514
515 // Returns the builder's internal stream.
516 OutputStream* GetStream() {
517 return &stream_;
David Srbecky527c9c72015-04-17 21:14:10 +0100518 }
519
David Srbecky579942f2016-01-28 20:01:28 +0000520 off_t AlignFileOffset(size_t alignment) {
521 return stream_.Seek(RoundUp(stream_.Seek(0, kSeekCurrent), alignment), kSeekSet);
522 }
523
524 Elf_Addr AlignVirtualAddress(size_t alignment) {
525 return virtual_address_ = RoundUp(virtual_address_, alignment);
526 }
527
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700528 private:
David Srbeckybc90fd02015-04-22 19:40:27 +0100529 static Elf_Ehdr MakeElfHeader(InstructionSet isa) {
530 Elf_Ehdr elf_header = Elf_Ehdr();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700531 switch (isa) {
532 case kArm:
533 // Fall through.
534 case kThumb2: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100535 elf_header.e_machine = EM_ARM;
536 elf_header.e_flags = EF_ARM_EABI_VER5;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700537 break;
538 }
539 case kArm64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100540 elf_header.e_machine = EM_AARCH64;
541 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700542 break;
543 }
544 case kX86: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100545 elf_header.e_machine = EM_386;
546 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700547 break;
548 }
549 case kX86_64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100550 elf_header.e_machine = EM_X86_64;
551 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700552 break;
553 }
554 case kMips: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100555 elf_header.e_machine = EM_MIPS;
556 elf_header.e_flags = (EF_MIPS_NOREORDER |
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700557 EF_MIPS_PIC |
558 EF_MIPS_CPIC |
559 EF_MIPS_ABI_O32 |
560 EF_MIPS_ARCH_32R2);
561 break;
562 }
Andreas Gampe57b34292015-01-14 15:45:59 -0800563 case kMips64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100564 elf_header.e_machine = EM_MIPS;
565 elf_header.e_flags = (EF_MIPS_NOREORDER |
Andreas Gampe57b34292015-01-14 15:45:59 -0800566 EF_MIPS_PIC |
567 EF_MIPS_CPIC |
568 EF_MIPS_ARCH_64R6);
569 break;
570 }
David Srbeckybc90fd02015-04-22 19:40:27 +0100571 case kNone: {
572 LOG(FATAL) << "No instruction set";
David Srbecky6d8c8f02015-10-26 10:57:09 +0000573 break;
574 }
575 default: {
576 LOG(FATAL) << "Unknown instruction set " << isa;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700577 }
578 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700579
David Srbeckybc90fd02015-04-22 19:40:27 +0100580 elf_header.e_ident[EI_MAG0] = ELFMAG0;
581 elf_header.e_ident[EI_MAG1] = ELFMAG1;
582 elf_header.e_ident[EI_MAG2] = ELFMAG2;
583 elf_header.e_ident[EI_MAG3] = ELFMAG3;
584 elf_header.e_ident[EI_CLASS] = (sizeof(Elf_Addr) == sizeof(Elf32_Addr))
Tong Shen62d1ca32014-09-03 17:24:56 -0700585 ? ELFCLASS32 : ELFCLASS64;;
David Srbeckybc90fd02015-04-22 19:40:27 +0100586 elf_header.e_ident[EI_DATA] = ELFDATA2LSB;
587 elf_header.e_ident[EI_VERSION] = EV_CURRENT;
588 elf_header.e_ident[EI_OSABI] = ELFOSABI_LINUX;
589 elf_header.e_ident[EI_ABIVERSION] = 0;
590 elf_header.e_type = ET_DYN;
591 elf_header.e_version = 1;
592 elf_header.e_entry = 0;
593 elf_header.e_ehsize = sizeof(Elf_Ehdr);
594 elf_header.e_phentsize = sizeof(Elf_Phdr);
595 elf_header.e_shentsize = sizeof(Elf_Shdr);
596 elf_header.e_phoff = sizeof(Elf_Ehdr);
597 return elf_header;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700598 }
599
David Srbecky6d8c8f02015-10-26 10:57:09 +0000600 // Create program headers based on written sections.
601 std::vector<Elf_Phdr> MakeProgramHeaders() {
602 CHECK(!sections_.empty());
603 std::vector<Elf_Phdr> phdrs;
604 {
605 // The program headers must start with PT_PHDR which is used in
606 // loaded process to determine the number of program headers.
607 Elf_Phdr phdr = Elf_Phdr();
608 phdr.p_type = PT_PHDR;
609 phdr.p_flags = PF_R;
610 phdr.p_offset = phdr.p_vaddr = phdr.p_paddr = sizeof(Elf_Ehdr);
611 phdr.p_filesz = phdr.p_memsz = 0; // We need to fill this later.
612 phdr.p_align = sizeof(Elf_Off);
613 phdrs.push_back(phdr);
614 // Tell the linker to mmap the start of file to memory.
615 Elf_Phdr load = Elf_Phdr();
616 load.p_type = PT_LOAD;
617 load.p_flags = PF_R;
618 load.p_offset = load.p_vaddr = load.p_paddr = 0;
619 load.p_filesz = load.p_memsz = sections_[0]->header_.sh_offset;
620 load.p_align = kPageSize;
621 phdrs.push_back(load);
David Srbeckybc90fd02015-04-22 19:40:27 +0100622 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000623 // Create program headers for sections.
624 for (auto* section : sections_) {
625 const Elf_Shdr& shdr = section->header_;
626 if ((shdr.sh_flags & SHF_ALLOC) != 0 && shdr.sh_size != 0) {
627 // PT_LOAD tells the linker to mmap part of the file.
628 // The linker can only mmap page-aligned sections.
629 // Single PT_LOAD may contain several ELF sections.
630 Elf_Phdr& prev = phdrs.back();
631 Elf_Phdr load = Elf_Phdr();
632 load.p_type = PT_LOAD;
633 load.p_flags = section->phdr_flags_;
634 load.p_offset = shdr.sh_offset;
635 load.p_vaddr = load.p_paddr = shdr.sh_addr;
636 load.p_filesz = (shdr.sh_type != SHT_NOBITS ? shdr.sh_size : 0u);
637 load.p_memsz = shdr.sh_size;
638 load.p_align = shdr.sh_addralign;
639 if (prev.p_type == load.p_type &&
640 prev.p_flags == load.p_flags &&
641 prev.p_filesz == prev.p_memsz && // Do not merge .bss
642 load.p_filesz == load.p_memsz) { // Do not merge .bss
643 // Merge this PT_LOAD with the previous one.
644 Elf_Word size = shdr.sh_offset + shdr.sh_size - prev.p_offset;
645 prev.p_filesz = size;
646 prev.p_memsz = size;
647 } else {
648 // If we are adding new load, it must be aligned.
649 CHECK_EQ(shdr.sh_addralign, (Elf_Word)kPageSize);
650 phdrs.push_back(load);
651 }
652 }
653 }
654 for (auto* section : sections_) {
655 const Elf_Shdr& shdr = section->header_;
656 if ((shdr.sh_flags & SHF_ALLOC) != 0 && shdr.sh_size != 0) {
657 // Other PT_* types allow the program to locate interesting
658 // parts of memory at runtime. They must overlap with PT_LOAD.
659 if (section->phdr_type_ != 0) {
660 Elf_Phdr phdr = Elf_Phdr();
661 phdr.p_type = section->phdr_type_;
662 phdr.p_flags = section->phdr_flags_;
663 phdr.p_offset = shdr.sh_offset;
664 phdr.p_vaddr = phdr.p_paddr = shdr.sh_addr;
665 phdr.p_filesz = phdr.p_memsz = shdr.sh_size;
666 phdr.p_align = shdr.sh_addralign;
667 phdrs.push_back(phdr);
668 }
669 }
670 }
671 // Set the size of the initial PT_PHDR.
672 CHECK_EQ(phdrs[0].p_type, (Elf_Word)PT_PHDR);
673 phdrs[0].p_filesz = phdrs[0].p_memsz = phdrs.size() * sizeof(Elf_Phdr);
David Srbeckybc90fd02015-04-22 19:40:27 +0100674
David Srbecky6d8c8f02015-10-26 10:57:09 +0000675 return phdrs;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700676 }
677
David Srbeckybc90fd02015-04-22 19:40:27 +0100678 InstructionSet isa_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000679
Vladimir Marko131980f2015-12-03 18:29:23 +0000680 ErrorDelayingOutputStream stream_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000681
682 Section rodata_;
683 Section text_;
684 Section bss_;
685 StringSection dynstr_;
686 SymbolSection dynsym_;
687 Section hash_;
688 Section dynamic_;
689 Section eh_frame_;
690 Section eh_frame_hdr_;
691 StringSection strtab_;
692 SymbolSection symtab_;
693 Section debug_frame_;
David Srbeckyb851b492015-11-11 20:19:38 +0000694 Section debug_info_;
695 Section debug_line_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000696 StringSection shstrtab_;
697 std::vector<std::unique_ptr<Section>> other_sections_;
698
699 // List of used section in the order in which they were written.
700 std::vector<Section*> sections_;
701
David Srbecky579942f2016-01-28 20:01:28 +0000702 bool started_;
703
David Srbecky6d8c8f02015-10-26 10:57:09 +0000704 // Used for allocation of virtual address space.
705 Elf_Addr virtual_address_;
Ian Rogers0279ebb2014-10-08 17:27:48 -0700706
David Srbecky579942f2016-01-28 20:01:28 +0000707 size_t write_program_headers_;
708
Ian Rogers0279ebb2014-10-08 17:27:48 -0700709 DISALLOW_COPY_AND_ASSIGN(ElfBuilder);
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700710};
711
712} // namespace art
713
714#endif // ART_COMPILER_ELF_BUILDER_H_