blob: a7461a5525e7d8168bb8d54e89b3bdb6637681af [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
Vladimir Marko10c13562015-11-25 14:33:36 +0000103 ~Section() OVERRIDE {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000104 if (started_) {
105 CHECK(finished_);
106 }
David Srbecky0c5bbc12015-04-28 17:54:52 +0100107 }
108
David Srbecky6d8c8f02015-10-26 10:57:09 +0000109 // Start writing of this section.
110 void Start() {
111 CHECK(!started_);
112 CHECK(!finished_);
113 started_ = true;
114 auto& sections = owner_->sections_;
115 // Check that the previous section is complete.
116 CHECK(sections.empty() || sections.back()->finished_);
117 // The first ELF section index is 1. Index 0 is reserved for NULL.
118 section_index_ = sections.size() + 1;
119 // Push this section on the list of written sections.
120 sections.push_back(this);
121 // Align file position.
122 if (header_.sh_type != SHT_NOBITS) {
Vladimir Marko131980f2015-12-03 18:29:23 +0000123 header_.sh_offset = RoundUp(owner_->stream_.Seek(0, kSeekCurrent), header_.sh_addralign);
124 owner_->stream_.Seek(header_.sh_offset, kSeekSet);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000125 }
126 // Align virtual memory address.
127 if ((header_.sh_flags & SHF_ALLOC) != 0) {
128 header_.sh_addr = RoundUp(owner_->virtual_address_, header_.sh_addralign);
129 owner_->virtual_address_ = header_.sh_addr;
130 }
David Srbecky0c5bbc12015-04-28 17:54:52 +0100131 }
132
David Srbecky6d8c8f02015-10-26 10:57:09 +0000133 // Finish writing of this section.
134 void End() {
135 CHECK(started_);
136 CHECK(!finished_);
137 finished_ = true;
138 if (header_.sh_type == SHT_NOBITS) {
139 CHECK_GT(header_.sh_size, 0u);
140 } else {
141 // Use the current file position to determine section size.
Vladimir Marko131980f2015-12-03 18:29:23 +0000142 off_t file_offset = owner_->stream_.Seek(0, kSeekCurrent);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000143 CHECK_GE(file_offset, (off_t)header_.sh_offset);
144 header_.sh_size = file_offset - header_.sh_offset;
145 }
146 if ((header_.sh_flags & SHF_ALLOC) != 0) {
147 owner_->virtual_address_ += header_.sh_size;
148 }
149 }
150
David Srbecky5cc349f2015-12-18 15:04:48 +0000151 // Returns true if the section was written to disk.
152 // (Used to check whether we have .text when writing JIT debug info)
153 bool Exists() const {
154 return finished_;
155 }
156
David Srbecky6d8c8f02015-10-26 10:57:09 +0000157 // Get the location of this section in virtual memory.
158 Elf_Addr GetAddress() const {
159 CHECK(started_);
160 return header_.sh_addr;
161 }
162
163 // Returns the size of the content of this section.
164 Elf_Word GetSize() const {
David Srbeckyb851b492015-11-11 20:19:38 +0000165 if (finished_) {
166 return header_.sh_size;
167 } else {
168 CHECK(started_);
169 CHECK_NE(header_.sh_type, (Elf_Word)SHT_NOBITS);
Vladimir Marko131980f2015-12-03 18:29:23 +0000170 return owner_->stream_.Seek(0, kSeekCurrent) - header_.sh_offset;
David Srbeckyb851b492015-11-11 20:19:38 +0000171 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000172 }
173
174 // Set desired allocation size for .bss section.
175 void SetSize(Elf_Word size) {
176 CHECK_EQ(header_.sh_type, (Elf_Word)SHT_NOBITS);
177 header_.sh_size = size;
178 }
179
180 // This function always succeeds to simplify code.
181 // Use builder's Good() to check the actual status.
182 bool WriteFully(const void* buffer, size_t byte_count) OVERRIDE {
183 CHECK(started_);
184 CHECK(!finished_);
Vladimir Marko131980f2015-12-03 18:29:23 +0000185 return owner_->stream_.WriteFully(buffer, byte_count);
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 off_t Seek(off_t offset, Whence whence) OVERRIDE {
191 // Forward the seek as-is and trust the caller to use it reasonably.
Vladimir Marko131980f2015-12-03 18:29:23 +0000192 return owner_->stream_.Seek(offset, whence);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100193 }
194
Vladimir Marko10c13562015-11-25 14:33:36 +0000195 // This function flushes the output and returns whether it succeeded.
196 // If there was a previous failure, this does nothing and returns false, i.e. failed.
197 bool Flush() OVERRIDE {
Vladimir Marko131980f2015-12-03 18:29:23 +0000198 return owner_->stream_.Flush();
Vladimir Marko10c13562015-11-25 14:33:36 +0000199 }
200
David Srbecky0c5bbc12015-04-28 17:54:52 +0100201 Elf_Word GetSectionIndex() const {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000202 DCHECK(started_);
David Srbeckybc90fd02015-04-22 19:40:27 +0100203 DCHECK_NE(section_index_, 0u);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100204 return section_index_;
205 }
206
David Srbecky0c5bbc12015-04-28 17:54:52 +0100207 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000208 ElfBuilder<ElfTypes>* owner_;
David Srbecky491a7fe2015-05-28 00:59:08 +0100209 Elf_Shdr header_;
David Srbecky0c5bbc12015-04-28 17:54:52 +0100210 Elf_Word section_index_;
211 const std::string name_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100212 const Section* const link_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000213 bool started_;
214 bool finished_;
215 Elf_Word phdr_flags_;
216 Elf_Word phdr_type_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100217
David Srbecky6d8c8f02015-10-26 10:57:09 +0000218 friend class ElfBuilder;
David Srbecky4d247f72015-11-09 11:56:52 +0000219
220 DISALLOW_COPY_AND_ASSIGN(Section);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100221 };
222
David Srbeckybc90fd02015-04-22 19:40:27 +0100223 // Writer of .dynstr .strtab and .shstrtab sections.
David Srbecky6d8c8f02015-10-26 10:57:09 +0000224 class StringSection FINAL : public Section {
David Srbeckybc90fd02015-04-22 19:40:27 +0100225 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000226 StringSection(ElfBuilder<ElfTypes>* owner, const std::string& name,
227 Elf_Word flags, Elf_Word align)
228 : Section(owner, name, SHT_STRTAB, flags, nullptr, 0, align, 0),
229 current_offset_(0) {
David Srbecky0c5bbc12015-04-28 17:54:52 +0100230 }
David Srbecky0c5bbc12015-04-28 17:54:52 +0100231
David Srbecky6d8c8f02015-10-26 10:57:09 +0000232 Elf_Word Write(const std::string& name) {
233 if (current_offset_ == 0) {
234 DCHECK(name.empty());
235 }
236 Elf_Word offset = current_offset_;
237 this->WriteFully(name.c_str(), name.length() + 1);
238 current_offset_ += name.length() + 1;
David Srbeckybc90fd02015-04-22 19:40:27 +0100239 return offset;
240 }
241
David Srbeckybc90fd02015-04-22 19:40:27 +0100242 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000243 Elf_Word current_offset_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100244 };
245
David Srbeckybc90fd02015-04-22 19:40:27 +0100246 // Writer of .dynsym and .symtab sections.
David Srbecky6d8c8f02015-10-26 10:57:09 +0000247 class SymbolSection FINAL : public Section {
David Srbecky0c5bbc12015-04-28 17:54:52 +0100248 public:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000249 SymbolSection(ElfBuilder<ElfTypes>* owner, const std::string& name,
250 Elf_Word type, Elf_Word flags, StringSection* strtab)
251 : Section(owner, name, type, flags, strtab, 0,
252 sizeof(Elf_Off), sizeof(Elf_Sym)) {
253 }
254
255 // Buffer symbol for this section. It will be written later.
David Srbecky5cc349f2015-12-18 15:04:48 +0000256 // If the symbol's section is null, it will be considered absolute (SHN_ABS).
257 // (we use this in JIT to reference code which is stored outside the debug ELF file)
David Srbecky6d8c8f02015-10-26 10:57:09 +0000258 void Add(Elf_Word name, const Section* section,
259 Elf_Addr addr, bool is_relative, Elf_Word size,
260 uint8_t binding, uint8_t type, uint8_t other = 0) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000261 Elf_Sym sym = Elf_Sym();
262 sym.st_name = name;
263 sym.st_value = addr + (is_relative ? section->GetAddress() : 0);
264 sym.st_size = size;
265 sym.st_other = other;
David Srbecky5cc349f2015-12-18 15:04:48 +0000266 sym.st_shndx = (section != nullptr ? section->GetSectionIndex()
267 : static_cast<Elf_Word>(SHN_ABS));
David Srbecky6d8c8f02015-10-26 10:57:09 +0000268 sym.st_info = (binding << 4) + (type & 0xf);
269 symbols_.push_back(sym);
David Srbecky0c5bbc12015-04-28 17:54:52 +0100270 }
271
David Srbecky6d8c8f02015-10-26 10:57:09 +0000272 void Write() {
273 // The symbol table always has to start with NULL symbol.
274 Elf_Sym null_symbol = Elf_Sym();
275 this->WriteFully(&null_symbol, sizeof(null_symbol));
276 this->WriteFully(symbols_.data(), symbols_.size() * sizeof(symbols_[0]));
277 symbols_.clear();
278 symbols_.shrink_to_fit();
David Srbeckybc90fd02015-04-22 19:40:27 +0100279 }
280
281 private:
David Srbecky6d8c8f02015-10-26 10:57:09 +0000282 std::vector<Elf_Sym> symbols_;
David Srbeckybc90fd02015-04-22 19:40:27 +0100283 };
284
David Srbecky6d8c8f02015-10-26 10:57:09 +0000285 ElfBuilder(InstructionSet isa, OutputStream* output)
Vladimir Marko131980f2015-12-03 18:29:23 +0000286 : isa_(isa),
287 stream_(output),
288 rodata_(this, ".rodata", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
289 text_(this, ".text", SHT_PROGBITS, SHF_ALLOC | SHF_EXECINSTR, nullptr, 0, kPageSize, 0),
290 bss_(this, ".bss", SHT_NOBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
291 dynstr_(this, ".dynstr", SHF_ALLOC, kPageSize),
292 dynsym_(this, ".dynsym", SHT_DYNSYM, SHF_ALLOC, &dynstr_),
293 hash_(this, ".hash", SHT_HASH, SHF_ALLOC, &dynsym_, 0, sizeof(Elf_Word), sizeof(Elf_Word)),
294 dynamic_(this, ".dynamic", SHT_DYNAMIC, SHF_ALLOC, &dynstr_, 0, kPageSize, sizeof(Elf_Dyn)),
295 eh_frame_(this, ".eh_frame", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, kPageSize, 0),
296 eh_frame_hdr_(this, ".eh_frame_hdr", SHT_PROGBITS, SHF_ALLOC, nullptr, 0, 4, 0),
297 strtab_(this, ".strtab", 0, kPageSize),
298 symtab_(this, ".symtab", SHT_SYMTAB, 0, &strtab_),
299 debug_frame_(this, ".debug_frame", SHT_PROGBITS, 0, nullptr, 0, sizeof(Elf_Addr), 0),
300 debug_info_(this, ".debug_info", SHT_PROGBITS, 0, nullptr, 0, 1, 0),
301 debug_line_(this, ".debug_line", SHT_PROGBITS, 0, nullptr, 0, 1, 0),
302 shstrtab_(this, ".shstrtab", 0, 1),
303 virtual_address_(0) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000304 text_.phdr_flags_ = PF_R | PF_X;
305 bss_.phdr_flags_ = PF_R | PF_W;
306 dynamic_.phdr_flags_ = PF_R | PF_W;
307 dynamic_.phdr_type_ = PT_DYNAMIC;
308 eh_frame_hdr_.phdr_type_ = PT_GNU_EH_FRAME;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700309 }
310 ~ElfBuilder() {}
311
David Srbecky6d8c8f02015-10-26 10:57:09 +0000312 InstructionSet GetIsa() { return isa_; }
313 Section* GetRoData() { return &rodata_; }
314 Section* GetText() { return &text_; }
315 Section* GetBss() { return &bss_; }
316 StringSection* GetStrTab() { return &strtab_; }
317 SymbolSection* GetSymTab() { return &symtab_; }
318 Section* GetEhFrame() { return &eh_frame_; }
319 Section* GetEhFrameHdr() { return &eh_frame_hdr_; }
320 Section* GetDebugFrame() { return &debug_frame_; }
David Srbeckyb851b492015-11-11 20:19:38 +0000321 Section* GetDebugInfo() { return &debug_info_; }
322 Section* GetDebugLine() { return &debug_line_; }
Ian Rogers0279ebb2014-10-08 17:27:48 -0700323
David Srbecky6d8c8f02015-10-26 10:57:09 +0000324 // Encode patch locations as LEB128 list of deltas between consecutive addresses.
325 // (exposed publicly for tests)
Vladimir Marko10c13562015-11-25 14:33:36 +0000326 static void EncodeOatPatches(const ArrayRef<const uintptr_t>& locations,
David Srbecky6d8c8f02015-10-26 10:57:09 +0000327 std::vector<uint8_t>* buffer) {
328 buffer->reserve(buffer->size() + locations.size() * 2); // guess 2 bytes per ULEB128.
329 uintptr_t address = 0; // relative to start of section.
330 for (uintptr_t location : locations) {
331 DCHECK_GE(location, address) << "Patch locations are not in sorted order";
332 EncodeUnsignedLeb128(buffer, dchecked_integral_cast<uint32_t>(location - address));
333 address = location;
334 }
335 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700336
Vladimir Marko10c13562015-11-25 14:33:36 +0000337 void WritePatches(const char* name, const ArrayRef<const uintptr_t>& patch_locations) {
David Srbecky6d8c8f02015-10-26 10:57:09 +0000338 std::vector<uint8_t> buffer;
Vladimir Marko10c13562015-11-25 14:33:36 +0000339 EncodeOatPatches(patch_locations, &buffer);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000340 std::unique_ptr<Section> s(new Section(this, name, SHT_OAT_PATCH, 0, nullptr, 0, 1, 0));
341 s->Start();
342 s->WriteFully(buffer.data(), buffer.size());
343 s->End();
344 other_sections_.push_back(std::move(s));
345 }
David Srbecky527c9c72015-04-17 21:14:10 +0100346
David Srbecky6d8c8f02015-10-26 10:57:09 +0000347 void WriteSection(const char* name, const std::vector<uint8_t>* buffer) {
348 std::unique_ptr<Section> s(new Section(this, name, SHT_PROGBITS, 0, nullptr, 0, 1, 0));
349 s->Start();
350 s->WriteFully(buffer->data(), buffer->size());
351 s->End();
352 other_sections_.push_back(std::move(s));
353 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700354
David Srbecky6d8c8f02015-10-26 10:57:09 +0000355 void Start() {
356 // Reserve space for ELF header and program headers.
357 // We do not know the number of headers until later, so
358 // it is easiest to just reserve a fixed amount of space.
359 int size = sizeof(Elf_Ehdr) + sizeof(Elf_Phdr) * kMaxProgramHeaders;
Vladimir Marko131980f2015-12-03 18:29:23 +0000360 stream_.Seek(size, kSeekSet);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000361 virtual_address_ += size;
362 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700363
David Srbecky6d8c8f02015-10-26 10:57:09 +0000364 void End() {
365 // Write section names and finish the section headers.
366 shstrtab_.Start();
367 shstrtab_.Write("");
368 for (auto* section : sections_) {
369 section->header_.sh_name = shstrtab_.Write(section->name_);
370 if (section->link_ != nullptr) {
371 section->header_.sh_link = section->link_->GetSectionIndex();
David Srbeckyb0a962c2015-04-28 19:43:56 +0100372 }
David Srbecky527c9c72015-04-17 21:14:10 +0100373 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000374 shstrtab_.End();
David Srbecky527c9c72015-04-17 21:14:10 +0100375
David Srbecky6d8c8f02015-10-26 10:57:09 +0000376 // Write section headers at the end of the ELF file.
377 std::vector<Elf_Shdr> shdrs;
378 shdrs.reserve(1u + sections_.size());
379 shdrs.push_back(Elf_Shdr()); // NULL at index 0.
380 for (auto* section : sections_) {
381 shdrs.push_back(section->header_);
382 }
383 Elf_Off section_headers_offset;
Vladimir Marko131980f2015-12-03 18:29:23 +0000384 section_headers_offset = RoundUp(stream_.Seek(0, kSeekCurrent), sizeof(Elf_Off));
385 stream_.Seek(section_headers_offset, kSeekSet);
386 stream_.WriteFully(shdrs.data(), shdrs.size() * sizeof(shdrs[0]));
387
388 // Flush everything else before writing the program headers. This should prevent
389 // the OS from reordering writes, so that we don't end up with valid headers
390 // and partially written data if we suddenly lose power, for example.
391 stream_.Flush();
David Srbecky6d8c8f02015-10-26 10:57:09 +0000392
393 // Write the initial file headers.
394 std::vector<Elf_Phdr> phdrs = MakeProgramHeaders();
David Srbeckybc90fd02015-04-22 19:40:27 +0100395 Elf_Ehdr elf_header = MakeElfHeader(isa_);
David Srbecky6d8c8f02015-10-26 10:57:09 +0000396 elf_header.e_phoff = sizeof(Elf_Ehdr);
David Srbeckybc90fd02015-04-22 19:40:27 +0100397 elf_header.e_shoff = section_headers_offset;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000398 elf_header.e_phnum = phdrs.size();
399 elf_header.e_shnum = shdrs.size();
David Srbeckybc90fd02015-04-22 19:40:27 +0100400 elf_header.e_shstrndx = shstrtab_.GetSectionIndex();
Vladimir Marko131980f2015-12-03 18:29:23 +0000401 stream_.Seek(0, kSeekSet);
402 stream_.WriteFully(&elf_header, sizeof(elf_header));
403 stream_.WriteFully(phdrs.data(), phdrs.size() * sizeof(phdrs[0]));
404 stream_.Flush();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700405 }
406
David Srbecky6d8c8f02015-10-26 10:57:09 +0000407 // The running program does not have access to section headers
408 // and the loader is not supposed to use them either.
409 // The dynamic sections therefore replicates some of the layout
410 // information like the address and size of .rodata and .text.
411 // It also contains other metadata like the SONAME.
412 // The .dynamic section is found using the PT_DYNAMIC program header.
413 void WriteDynamicSection(const std::string& elf_file_path) {
414 std::string soname(elf_file_path);
415 size_t directory_separator_pos = soname.rfind('/');
416 if (directory_separator_pos != std::string::npos) {
417 soname = soname.substr(directory_separator_pos + 1);
418 }
419
420 dynstr_.Start();
421 dynstr_.Write(""); // dynstr should start with empty string.
422 dynsym_.Add(dynstr_.Write("oatdata"), &rodata_, 0, true,
423 rodata_.GetSize(), STB_GLOBAL, STT_OBJECT);
424 if (text_.GetSize() != 0u) {
425 dynsym_.Add(dynstr_.Write("oatexec"), &text_, 0, true,
426 text_.GetSize(), STB_GLOBAL, STT_OBJECT);
427 dynsym_.Add(dynstr_.Write("oatlastword"), &text_, text_.GetSize() - 4,
428 true, 4, STB_GLOBAL, STT_OBJECT);
429 } else if (rodata_.GetSize() != 0) {
430 // rodata_ can be size 0 for dwarf_test.
431 dynsym_.Add(dynstr_.Write("oatlastword"), &rodata_, rodata_.GetSize() - 4,
432 true, 4, STB_GLOBAL, STT_OBJECT);
433 }
434 if (bss_.finished_) {
435 dynsym_.Add(dynstr_.Write("oatbss"), &bss_,
436 0, true, bss_.GetSize(), STB_GLOBAL, STT_OBJECT);
437 dynsym_.Add(dynstr_.Write("oatbsslastword"), &bss_,
438 bss_.GetSize() - 4, true, 4, STB_GLOBAL, STT_OBJECT);
439 }
440 Elf_Word soname_offset = dynstr_.Write(soname);
441 dynstr_.End();
442
443 dynsym_.Start();
444 dynsym_.Write();
445 dynsym_.End();
446
447 // We do not really need a hash-table since there is so few entries.
448 // However, the hash-table is the only way the linker can actually
449 // determine the number of symbols in .dynsym so it is required.
450 hash_.Start();
451 int count = dynsym_.GetSize() / sizeof(Elf_Sym); // Includes NULL.
452 std::vector<Elf_Word> hash;
453 hash.push_back(1); // Number of buckets.
454 hash.push_back(count); // Number of chains.
455 // Buckets. Having just one makes it linear search.
456 hash.push_back(1); // Point to first non-NULL symbol.
457 // Chains. This creates linked list of symbols.
458 hash.push_back(0); // Dummy entry for the NULL symbol.
459 for (int i = 1; i < count - 1; i++) {
460 hash.push_back(i + 1); // Each symbol points to the next one.
461 }
462 hash.push_back(0); // Last symbol terminates the chain.
463 hash_.WriteFully(hash.data(), hash.size() * sizeof(hash[0]));
464 hash_.End();
465
466 dynamic_.Start();
467 Elf_Dyn dyns[] = {
468 { DT_HASH, { hash_.GetAddress() } },
469 { DT_STRTAB, { dynstr_.GetAddress() } },
470 { DT_SYMTAB, { dynsym_.GetAddress() } },
471 { DT_SYMENT, { sizeof(Elf_Sym) } },
472 { DT_STRSZ, { dynstr_.GetSize() } },
473 { DT_SONAME, { soname_offset } },
474 { DT_NULL, { 0 } },
475 };
476 dynamic_.WriteFully(&dyns, sizeof(dyns));
477 dynamic_.End();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700478 }
479
David Srbecky6d8c8f02015-10-26 10:57:09 +0000480 // Returns true if all writes and seeks on the output stream succeeded.
481 bool Good() {
Vladimir Marko131980f2015-12-03 18:29:23 +0000482 return stream_.Good();
483 }
484
485 // Returns the builder's internal stream.
486 OutputStream* GetStream() {
487 return &stream_;
David Srbecky527c9c72015-04-17 21:14:10 +0100488 }
489
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700490 private:
David Srbeckybc90fd02015-04-22 19:40:27 +0100491 static Elf_Ehdr MakeElfHeader(InstructionSet isa) {
492 Elf_Ehdr elf_header = Elf_Ehdr();
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700493 switch (isa) {
494 case kArm:
495 // Fall through.
496 case kThumb2: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100497 elf_header.e_machine = EM_ARM;
498 elf_header.e_flags = EF_ARM_EABI_VER5;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700499 break;
500 }
501 case kArm64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100502 elf_header.e_machine = EM_AARCH64;
503 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700504 break;
505 }
506 case kX86: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100507 elf_header.e_machine = EM_386;
508 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700509 break;
510 }
511 case kX86_64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100512 elf_header.e_machine = EM_X86_64;
513 elf_header.e_flags = 0;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700514 break;
515 }
516 case kMips: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100517 elf_header.e_machine = EM_MIPS;
518 elf_header.e_flags = (EF_MIPS_NOREORDER |
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700519 EF_MIPS_PIC |
520 EF_MIPS_CPIC |
521 EF_MIPS_ABI_O32 |
522 EF_MIPS_ARCH_32R2);
523 break;
524 }
Andreas Gampe57b34292015-01-14 15:45:59 -0800525 case kMips64: {
David Srbeckybc90fd02015-04-22 19:40:27 +0100526 elf_header.e_machine = EM_MIPS;
527 elf_header.e_flags = (EF_MIPS_NOREORDER |
Andreas Gampe57b34292015-01-14 15:45:59 -0800528 EF_MIPS_PIC |
529 EF_MIPS_CPIC |
530 EF_MIPS_ARCH_64R6);
531 break;
532 }
David Srbeckybc90fd02015-04-22 19:40:27 +0100533 case kNone: {
534 LOG(FATAL) << "No instruction set";
David Srbecky6d8c8f02015-10-26 10:57:09 +0000535 break;
536 }
537 default: {
538 LOG(FATAL) << "Unknown instruction set " << isa;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700539 }
540 }
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700541
David Srbeckybc90fd02015-04-22 19:40:27 +0100542 elf_header.e_ident[EI_MAG0] = ELFMAG0;
543 elf_header.e_ident[EI_MAG1] = ELFMAG1;
544 elf_header.e_ident[EI_MAG2] = ELFMAG2;
545 elf_header.e_ident[EI_MAG3] = ELFMAG3;
546 elf_header.e_ident[EI_CLASS] = (sizeof(Elf_Addr) == sizeof(Elf32_Addr))
Tong Shen62d1ca32014-09-03 17:24:56 -0700547 ? ELFCLASS32 : ELFCLASS64;;
David Srbeckybc90fd02015-04-22 19:40:27 +0100548 elf_header.e_ident[EI_DATA] = ELFDATA2LSB;
549 elf_header.e_ident[EI_VERSION] = EV_CURRENT;
550 elf_header.e_ident[EI_OSABI] = ELFOSABI_LINUX;
551 elf_header.e_ident[EI_ABIVERSION] = 0;
552 elf_header.e_type = ET_DYN;
553 elf_header.e_version = 1;
554 elf_header.e_entry = 0;
555 elf_header.e_ehsize = sizeof(Elf_Ehdr);
556 elf_header.e_phentsize = sizeof(Elf_Phdr);
557 elf_header.e_shentsize = sizeof(Elf_Shdr);
558 elf_header.e_phoff = sizeof(Elf_Ehdr);
559 return elf_header;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700560 }
561
David Srbecky6d8c8f02015-10-26 10:57:09 +0000562 // Create program headers based on written sections.
563 std::vector<Elf_Phdr> MakeProgramHeaders() {
564 CHECK(!sections_.empty());
565 std::vector<Elf_Phdr> phdrs;
566 {
567 // The program headers must start with PT_PHDR which is used in
568 // loaded process to determine the number of program headers.
569 Elf_Phdr phdr = Elf_Phdr();
570 phdr.p_type = PT_PHDR;
571 phdr.p_flags = PF_R;
572 phdr.p_offset = phdr.p_vaddr = phdr.p_paddr = sizeof(Elf_Ehdr);
573 phdr.p_filesz = phdr.p_memsz = 0; // We need to fill this later.
574 phdr.p_align = sizeof(Elf_Off);
575 phdrs.push_back(phdr);
576 // Tell the linker to mmap the start of file to memory.
577 Elf_Phdr load = Elf_Phdr();
578 load.p_type = PT_LOAD;
579 load.p_flags = PF_R;
580 load.p_offset = load.p_vaddr = load.p_paddr = 0;
581 load.p_filesz = load.p_memsz = sections_[0]->header_.sh_offset;
582 load.p_align = kPageSize;
583 phdrs.push_back(load);
David Srbeckybc90fd02015-04-22 19:40:27 +0100584 }
David Srbecky6d8c8f02015-10-26 10:57:09 +0000585 // Create program headers for sections.
586 for (auto* section : sections_) {
587 const Elf_Shdr& shdr = section->header_;
588 if ((shdr.sh_flags & SHF_ALLOC) != 0 && shdr.sh_size != 0) {
589 // PT_LOAD tells the linker to mmap part of the file.
590 // The linker can only mmap page-aligned sections.
591 // Single PT_LOAD may contain several ELF sections.
592 Elf_Phdr& prev = phdrs.back();
593 Elf_Phdr load = Elf_Phdr();
594 load.p_type = PT_LOAD;
595 load.p_flags = section->phdr_flags_;
596 load.p_offset = shdr.sh_offset;
597 load.p_vaddr = load.p_paddr = shdr.sh_addr;
598 load.p_filesz = (shdr.sh_type != SHT_NOBITS ? shdr.sh_size : 0u);
599 load.p_memsz = shdr.sh_size;
600 load.p_align = shdr.sh_addralign;
601 if (prev.p_type == load.p_type &&
602 prev.p_flags == load.p_flags &&
603 prev.p_filesz == prev.p_memsz && // Do not merge .bss
604 load.p_filesz == load.p_memsz) { // Do not merge .bss
605 // Merge this PT_LOAD with the previous one.
606 Elf_Word size = shdr.sh_offset + shdr.sh_size - prev.p_offset;
607 prev.p_filesz = size;
608 prev.p_memsz = size;
609 } else {
610 // If we are adding new load, it must be aligned.
611 CHECK_EQ(shdr.sh_addralign, (Elf_Word)kPageSize);
612 phdrs.push_back(load);
613 }
614 }
615 }
616 for (auto* section : sections_) {
617 const Elf_Shdr& shdr = section->header_;
618 if ((shdr.sh_flags & SHF_ALLOC) != 0 && shdr.sh_size != 0) {
619 // Other PT_* types allow the program to locate interesting
620 // parts of memory at runtime. They must overlap with PT_LOAD.
621 if (section->phdr_type_ != 0) {
622 Elf_Phdr phdr = Elf_Phdr();
623 phdr.p_type = section->phdr_type_;
624 phdr.p_flags = section->phdr_flags_;
625 phdr.p_offset = shdr.sh_offset;
626 phdr.p_vaddr = phdr.p_paddr = shdr.sh_addr;
627 phdr.p_filesz = phdr.p_memsz = shdr.sh_size;
628 phdr.p_align = shdr.sh_addralign;
629 phdrs.push_back(phdr);
630 }
631 }
632 }
633 // Set the size of the initial PT_PHDR.
634 CHECK_EQ(phdrs[0].p_type, (Elf_Word)PT_PHDR);
635 phdrs[0].p_filesz = phdrs[0].p_memsz = phdrs.size() * sizeof(Elf_Phdr);
David Srbeckybc90fd02015-04-22 19:40:27 +0100636
David Srbecky6d8c8f02015-10-26 10:57:09 +0000637 return phdrs;
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700638 }
639
David Srbeckybc90fd02015-04-22 19:40:27 +0100640 InstructionSet isa_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000641
Vladimir Marko131980f2015-12-03 18:29:23 +0000642 ErrorDelayingOutputStream stream_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000643
644 Section rodata_;
645 Section text_;
646 Section bss_;
647 StringSection dynstr_;
648 SymbolSection dynsym_;
649 Section hash_;
650 Section dynamic_;
651 Section eh_frame_;
652 Section eh_frame_hdr_;
653 StringSection strtab_;
654 SymbolSection symtab_;
655 Section debug_frame_;
David Srbeckyb851b492015-11-11 20:19:38 +0000656 Section debug_info_;
657 Section debug_line_;
David Srbecky6d8c8f02015-10-26 10:57:09 +0000658 StringSection shstrtab_;
659 std::vector<std::unique_ptr<Section>> other_sections_;
660
661 // List of used section in the order in which they were written.
662 std::vector<Section*> sections_;
663
664 // Used for allocation of virtual address space.
665 Elf_Addr virtual_address_;
Ian Rogers0279ebb2014-10-08 17:27:48 -0700666
667 DISALLOW_COPY_AND_ASSIGN(ElfBuilder);
Andreas Gampe54fc26c2014-09-04 21:47:42 -0700668};
669
670} // namespace art
671
672#endif // ART_COMPILER_ELF_BUILDER_H_