blob: caf6fcb643f49a45f50607a76d903345efeee2bc [file] [log] [blame]
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001//===- LinkerScript.cpp ---------------------------------------------------===//
2//
3// The LLVM Linker
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file contains the parser/evaluator of the linker script.
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000011//
12//===----------------------------------------------------------------------===//
13
Rui Ueyama717677a2016-02-11 21:17:59 +000014#include "LinkerScript.h"
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000015#include "Config.h"
16#include "Driver.h"
Rui Ueyama1ebc8ed2016-02-12 21:47:28 +000017#include "InputSection.h"
Rui Ueyama9381eb12016-12-18 14:06:06 +000018#include "Memory.h"
George Rimar652852c2016-04-16 10:10:32 +000019#include "OutputSections.h"
Rui Ueyama794366a2017-02-14 04:47:05 +000020#include "ScriptLexer.h"
Rui Ueyama93c9af42016-06-29 08:01:32 +000021#include "Strings.h"
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000022#include "SymbolTable.h"
Rui Ueyama55518e72016-10-28 20:57:25 +000023#include "Symbols.h"
George Rimar3fb5a6d2016-11-29 16:05:27 +000024#include "SyntheticSections.h"
Eugene Leviant467c4d52016-07-01 10:27:36 +000025#include "Target.h"
Eugene Leviantbbe38602016-07-19 09:25:43 +000026#include "Writer.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000027#include "llvm/ADT/STLExtras.h"
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +000028#include "llvm/ADT/SmallString.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000029#include "llvm/ADT/StringRef.h"
Rui Ueyama960504b2016-04-19 18:58:11 +000030#include "llvm/ADT/StringSwitch.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000031#include "llvm/Support/Casting.h"
George Rimar652852c2016-04-16 10:10:32 +000032#include "llvm/Support/ELF.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000033#include "llvm/Support/Endian.h"
34#include "llvm/Support/ErrorHandling.h"
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000035#include "llvm/Support/FileSystem.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000036#include "llvm/Support/MathExtras.h"
Rui Ueyamaf03f3cc2015-10-13 00:09:21 +000037#include "llvm/Support/Path.h"
Eugene Zelenko22886a22016-11-05 01:00:56 +000038#include <algorithm>
39#include <cassert>
40#include <cstddef>
41#include <cstdint>
42#include <iterator>
43#include <limits>
44#include <memory>
45#include <string>
46#include <tuple>
47#include <vector>
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000048
49using namespace llvm;
George Rimar652852c2016-04-16 10:10:32 +000050using namespace llvm::ELF;
Rui Ueyama1ebc8ed2016-02-12 21:47:28 +000051using namespace llvm::object;
George Rimare38cbab2016-09-26 19:22:50 +000052using namespace llvm::support::endian;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000053using namespace lld;
Rafael Espindolae0df00b2016-02-28 00:25:54 +000054using namespace lld::elf;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +000055
George Rimar884e7862016-09-08 08:19:13 +000056LinkerScriptBase *elf::ScriptBase;
Rui Ueyama07320e42016-04-20 20:13:41 +000057ScriptConfiguration *elf::ScriptConfig;
Rui Ueyama717677a2016-02-11 21:17:59 +000058
Meador Inge8f1f3c42017-01-09 18:36:57 +000059template <class ELFT> static SymbolBody *addRegular(SymbolAssignment *Cmd) {
Petr Hosek5e51f7d2017-02-21 22:32:51 +000060 Symbol *Sym;
Rafael Espindola3dabfc62016-10-31 13:14:53 +000061 uint8_t Visibility = Cmd->Hidden ? STV_HIDDEN : STV_DEFAULT;
Petr Hosek5e51f7d2017-02-21 22:32:51 +000062 std::tie(Sym, std::ignore) = Symtab<ELFT>::X->insert(
63 Cmd->Name, /*Type*/ 0, Visibility, /*CanOmitFromDynSym*/ false,
64 /*File*/ nullptr);
65 Sym->Binding = STB_GLOBAL;
Rui Ueyama80474a22017-02-28 19:29:55 +000066 replaceBody<DefinedRegular>(Sym, Cmd->Name, /*IsLocal=*/false, Visibility,
67 STT_NOTYPE, 0, 0, nullptr, nullptr);
Meador Inge8f1f3c42017-01-09 18:36:57 +000068 return Sym->body();
Eugene Leviantceabe802016-08-11 07:56:43 +000069}
70
Meador Inge8f1f3c42017-01-09 18:36:57 +000071template <class ELFT> static SymbolBody *addSynthetic(SymbolAssignment *Cmd) {
Petr Hosek5e51f7d2017-02-21 22:32:51 +000072 Symbol *Sym;
Meador Inge8f1f3c42017-01-09 18:36:57 +000073 uint8_t Visibility = Cmd->Hidden ? STV_HIDDEN : STV_DEFAULT;
Rafael Espindola24e6f362017-02-24 15:07:30 +000074 const OutputSection *Sec =
Eugene Leviantafaa9342016-11-16 09:49:39 +000075 ScriptConfig->HasSections ? nullptr : Cmd->Expression.Section();
Petr Hosek5e51f7d2017-02-21 22:32:51 +000076 std::tie(Sym, std::ignore) = Symtab<ELFT>::X->insert(
77 Cmd->Name, /*Type*/ 0, Visibility, /*CanOmitFromDynSym*/ false,
78 /*File*/ nullptr);
79 Sym->Binding = STB_GLOBAL;
Meador Inge8f1f3c42017-01-09 18:36:57 +000080 replaceBody<DefinedSynthetic>(Sym, Cmd->Name, 0, Sec);
81 return Sym->body();
Eugene Leviantceabe802016-08-11 07:56:43 +000082}
83
Rui Ueyama22375f22016-11-25 18:51:54 +000084static bool isUnderSysroot(StringRef Path) {
85 if (Config->Sysroot == "")
86 return false;
87 for (; !Path.empty(); Path = sys::path::parent_path(Path))
88 if (sys::fs::equivalent(Config->Sysroot, Path))
89 return true;
90 return false;
91}
92
George Rimar2ee2d2d2017-02-21 14:50:38 +000093template <class ELFT>
94void LinkerScript<ELFT>::setDot(Expr E, const Twine &Loc, bool InSec) {
Rafael Espindola679828f2017-02-17 16:26:13 +000095 uintX_t Val = E(Dot);
96 if (Val < Dot) {
97 if (InSec)
George Rimar2ee2d2d2017-02-21 14:50:38 +000098 error(Loc + ": unable to move location counter backward for: " +
99 CurOutSec->Name);
Rafael Espindola679828f2017-02-17 16:26:13 +0000100 else
George Rimar2ee2d2d2017-02-21 14:50:38 +0000101 error(Loc + ": unable to move location counter backward");
Rafael Espindola679828f2017-02-17 16:26:13 +0000102 }
103 Dot = Val;
104 // Update to location counter means update to section size.
105 if (InSec)
106 CurOutSec->Size = Dot - CurOutSec->Addr;
107}
108
George Rimarb2b70972017-02-07 10:23:28 +0000109// Sets value of a symbol. Two kinds of symbols are processed: synthetic
110// symbols, whose value is an offset from beginning of section and regular
111// symbols whose value is absolute.
112template <class ELFT>
Rafael Espindola4cd73522017-02-17 16:01:51 +0000113void LinkerScript<ELFT>::assignSymbol(SymbolAssignment *Cmd, bool InSec) {
114 if (Cmd->Name == ".") {
George Rimar2ee2d2d2017-02-21 14:50:38 +0000115 setDot(Cmd->Expression, Cmd->Location, InSec);
Rafael Espindola4cd73522017-02-17 16:01:51 +0000116 return;
117 }
118
George Rimarb2b70972017-02-07 10:23:28 +0000119 if (!Cmd->Sym)
Meador Inge8f1f3c42017-01-09 18:36:57 +0000120 return;
121
George Rimarb2b70972017-02-07 10:23:28 +0000122 if (auto *Body = dyn_cast<DefinedSynthetic>(Cmd->Sym)) {
123 Body->Section = Cmd->Expression.Section();
Rafael Espindolaea590d92017-02-08 15:19:03 +0000124 if (Body->Section) {
125 uint64_t VA = 0;
126 if (Body->Section->Flags & SHF_ALLOC)
127 VA = Body->Section->Addr;
128 Body->Value = Cmd->Expression(Dot) - VA;
129 }
George Rimarb2b70972017-02-07 10:23:28 +0000130 return;
Meador Inge8f1f3c42017-01-09 18:36:57 +0000131 }
George Rimarb2b70972017-02-07 10:23:28 +0000132
Rui Ueyama80474a22017-02-28 19:29:55 +0000133 cast<DefinedRegular>(Cmd->Sym)->Value = Cmd->Expression(Dot);
Eugene Leviantdb741e72016-09-07 07:08:43 +0000134}
Meador Inge8f1f3c42017-01-09 18:36:57 +0000135
Rafael Espindola4cd73522017-02-17 16:01:51 +0000136template <class ELFT>
137void LinkerScript<ELFT>::addSymbol(SymbolAssignment *Cmd) {
Rui Ueyama16024212016-08-11 23:22:52 +0000138 if (Cmd->Name == ".")
Meador Inge8f1f3c42017-01-09 18:36:57 +0000139 return;
140
141 // If a symbol was in PROVIDE(), we need to define it only when
142 // it is a referenced undefined symbol.
Rui Ueyama16024212016-08-11 23:22:52 +0000143 SymbolBody *B = Symtab<ELFT>::X->find(Cmd->Name);
Meador Inge8f1f3c42017-01-09 18:36:57 +0000144 if (Cmd->Provide && (!B || B->isDefined()))
145 return;
146
147 // Otherwise, create a new symbol if one does not exist or an
148 // undefined one does exist.
149 if (Cmd->Expression.IsAbsolute())
150 Cmd->Sym = addRegular<ELFT>(Cmd);
151 else
152 Cmd->Sym = addSynthetic<ELFT>(Cmd);
George Rimarb2b70972017-02-07 10:23:28 +0000153
154 // If there are sections, then let the value be assigned later in
155 // `assignAddresses`.
156 if (!ScriptConfig->HasSections)
Rafael Espindola4cd73522017-02-17 16:01:51 +0000157 assignSymbol(Cmd);
Eugene Leviantceabe802016-08-11 07:56:43 +0000158}
159
George Rimar076fe152016-07-21 06:43:01 +0000160bool SymbolAssignment::classof(const BaseCommand *C) {
161 return C->Kind == AssignmentKind;
162}
163
164bool OutputSectionCommand::classof(const BaseCommand *C) {
165 return C->Kind == OutputSectionKind;
166}
167
George Rimareea31142016-07-21 14:26:59 +0000168bool InputSectionDescription::classof(const BaseCommand *C) {
169 return C->Kind == InputSectionKind;
170}
171
George Rimareefa7582016-08-04 09:29:31 +0000172bool AssertCommand::classof(const BaseCommand *C) {
173 return C->Kind == AssertKind;
174}
175
George Rimare38cbab2016-09-26 19:22:50 +0000176bool BytesDataCommand::classof(const BaseCommand *C) {
177 return C->Kind == BytesDataKind;
178}
179
Eugene Zelenko22886a22016-11-05 01:00:56 +0000180template <class ELFT> LinkerScript<ELFT>::LinkerScript() = default;
181template <class ELFT> LinkerScript<ELFT>::~LinkerScript() = default;
Rui Ueyamaf34d0e02016-08-12 01:24:53 +0000182
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000183static StringRef basename(InputSectionBase *S) {
184 if (S->File)
185 return sys::path::filename(S->File->getName());
Rui Ueyamae0be2902016-11-21 02:10:12 +0000186 return "";
187}
188
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000189template <class ELFT> bool LinkerScript<ELFT>::shouldKeep(InputSectionBase *S) {
Rui Ueyamae0be2902016-11-21 02:10:12 +0000190 for (InputSectionDescription *ID : Opt.KeptSections)
191 if (ID->FilePat.match(basename(S)))
192 for (SectionPattern &P : ID->SectionPatterns)
193 if (P.SectionPat.match(S->Name))
194 return true;
George Rimareea31142016-07-21 14:26:59 +0000195 return false;
196}
197
Rafael Espindolac404d502017-02-23 02:32:18 +0000198static bool comparePriority(InputSectionBase *A, InputSectionBase *B) {
George Rimar575208c2016-09-15 19:15:12 +0000199 return getPriority(A->Name) < getPriority(B->Name);
200}
201
Rafael Espindolac404d502017-02-23 02:32:18 +0000202static bool compareName(InputSectionBase *A, InputSectionBase *B) {
Rafael Espindola042a3f22016-09-08 14:06:08 +0000203 return A->Name < B->Name;
Rui Ueyama742c3832016-08-04 22:27:00 +0000204}
George Rimar350ece42016-08-03 08:35:59 +0000205
Rafael Espindolac404d502017-02-23 02:32:18 +0000206static bool compareAlignment(InputSectionBase *A, InputSectionBase *B) {
Rui Ueyama742c3832016-08-04 22:27:00 +0000207 // ">" is not a mistake. Larger alignments are placed before smaller
208 // alignments in order to reduce the amount of padding necessary.
209 // This is compatible with GNU.
210 return A->Alignment > B->Alignment;
211}
George Rimar350ece42016-08-03 08:35:59 +0000212
Rafael Espindolac404d502017-02-23 02:32:18 +0000213static std::function<bool(InputSectionBase *, InputSectionBase *)>
George Rimarbe394db2016-09-16 20:21:55 +0000214getComparator(SortSectionPolicy K) {
215 switch (K) {
216 case SortSectionPolicy::Alignment:
217 return compareAlignment;
218 case SortSectionPolicy::Name:
Rafael Espindolac0028d32016-09-08 20:47:52 +0000219 return compareName;
George Rimarbe394db2016-09-16 20:21:55 +0000220 case SortSectionPolicy::Priority:
221 return comparePriority;
222 default:
223 llvm_unreachable("unknown sort policy");
224 }
Rui Ueyama742c3832016-08-04 22:27:00 +0000225}
George Rimar0702c4e2016-07-29 15:32:46 +0000226
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000227template <class ELFT>
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000228static bool matchConstraints(ArrayRef<InputSectionBase *> Sections,
George Rimar06ae6832016-08-12 09:07:57 +0000229 ConstraintKind Kind) {
George Rimar8f66df92016-08-12 20:38:20 +0000230 if (Kind == ConstraintKind::NoConstraint)
231 return true;
Rafael Espindolac404d502017-02-23 02:32:18 +0000232 bool IsRW = llvm::any_of(Sections, [=](InputSectionBase *Sec2) {
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000233 auto *Sec = static_cast<InputSectionBase *>(Sec2);
Rafael Espindola1854a8e2016-10-26 12:36:56 +0000234 return Sec->Flags & SHF_WRITE;
George Rimar06ae6832016-08-12 09:07:57 +0000235 });
Rafael Espindolae746e522016-09-21 18:33:44 +0000236 return (IsRW && Kind == ConstraintKind::ReadWrite) ||
237 (!IsRW && Kind == ConstraintKind::ReadOnly);
George Rimar06ae6832016-08-12 09:07:57 +0000238}
239
Rafael Espindolac404d502017-02-23 02:32:18 +0000240static void sortSections(InputSectionBase **Begin, InputSectionBase **End,
Rui Ueyamaee924702016-09-20 19:42:41 +0000241 SortSectionPolicy K) {
242 if (K != SortSectionPolicy::Default && K != SortSectionPolicy::None)
George Rimar07171f22016-09-21 15:56:44 +0000243 std::stable_sort(Begin, End, getComparator(K));
Rui Ueyamaee924702016-09-20 19:42:41 +0000244}
245
Rafael Espindolad3190792016-09-16 15:10:23 +0000246// Compute and remember which sections the InputSectionDescription matches.
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000247template <class ELFT>
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000248void LinkerScript<ELFT>::computeInputSections(InputSectionDescription *I) {
Rui Ueyama4dc07be2016-09-17 02:23:40 +0000249 // Collects all sections that satisfy constraints of I
250 // and attach them to I.
251 for (SectionPattern &Pat : I->SectionPatterns) {
George Rimar07171f22016-09-21 15:56:44 +0000252 size_t SizeBefore = I->Sections.size();
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000253
Rui Ueyama536a2672017-02-27 02:32:08 +0000254 for (InputSectionBase *S : InputSections) {
Rafael Espindola3773bca2017-02-17 19:37:30 +0000255 if (S->Assigned)
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000256 continue;
Rafael Espindola908a3d32017-02-16 14:36:09 +0000257 // For -emit-relocs we have to ignore entries like
258 // .rela.dyn : { *(.rela.data) }
259 // which are common because they are in the default bfd script.
260 if (S->Type == SHT_REL || S->Type == SHT_RELA)
261 continue;
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000262
Rui Ueyamae0be2902016-11-21 02:10:12 +0000263 StringRef Filename = basename(S);
264 if (!I->FilePat.match(Filename) || Pat.ExcludedFilePat.match(Filename))
265 continue;
266 if (!Pat.SectionPat.match(S->Name))
267 continue;
268 I->Sections.push_back(S);
269 S->Assigned = true;
George Rimar395281c2016-09-16 17:42:10 +0000270 }
Rafael Espindolad3190792016-09-16 15:10:23 +0000271
George Rimar07171f22016-09-21 15:56:44 +0000272 // Sort sections as instructed by SORT-family commands and --sort-section
273 // option. Because SORT-family commands can be nested at most two depth
274 // (e.g. SORT_BY_NAME(SORT_BY_ALIGNMENT(.text.*))) and because the command
275 // line option is respected even if a SORT command is given, the exact
276 // behavior we have here is a bit complicated. Here are the rules.
277 //
278 // 1. If two SORT commands are given, --sort-section is ignored.
279 // 2. If one SORT command is given, and if it is not SORT_NONE,
280 // --sort-section is handled as an inner SORT command.
281 // 3. If one SORT command is given, and if it is SORT_NONE, don't sort.
282 // 4. If no SORT command is given, sort according to --sort-section.
Rafael Espindolac404d502017-02-23 02:32:18 +0000283 InputSectionBase **Begin = I->Sections.data() + SizeBefore;
284 InputSectionBase **End = I->Sections.data() + I->Sections.size();
George Rimar07171f22016-09-21 15:56:44 +0000285 if (Pat.SortOuter != SortSectionPolicy::None) {
286 if (Pat.SortInner == SortSectionPolicy::Default)
287 sortSections(Begin, End, Config->SortSection);
288 else
289 sortSections(Begin, End, Pat.SortInner);
290 sortSections(Begin, End, Pat.SortOuter);
291 }
Rui Ueyamaee924702016-09-20 19:42:41 +0000292 }
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000293}
294
295template <class ELFT>
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000296void LinkerScript<ELFT>::discard(ArrayRef<InputSectionBase *> V) {
297 for (InputSectionBase *S : V) {
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000298 S->Live = false;
Rafael Espindolaecbfd872017-02-17 17:35:07 +0000299 if (S == In<ELFT>::ShStrTab)
300 error("discarding .shstrtab section is not allowed");
George Rimar647c1682017-02-17 19:34:05 +0000301 discard(S->DependentSections);
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000302 }
303}
304
George Rimar06ae6832016-08-12 09:07:57 +0000305template <class ELFT>
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000306std::vector<InputSectionBase *>
George Rimar06ae6832016-08-12 09:07:57 +0000307LinkerScript<ELFT>::createInputSectionList(OutputSectionCommand &OutCmd) {
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000308 std::vector<InputSectionBase *> Ret;
Rui Ueyamae7f912c2016-08-03 21:12:09 +0000309
George Rimar06ae6832016-08-12 09:07:57 +0000310 for (const std::unique_ptr<BaseCommand> &Base : OutCmd.Commands) {
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000311 auto *Cmd = dyn_cast<InputSectionDescription>(Base.get());
312 if (!Cmd)
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000313 continue;
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000314 computeInputSections(Cmd);
Rafael Espindolac404d502017-02-23 02:32:18 +0000315 for (InputSectionBase *S : Cmd->Sections)
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000316 Ret.push_back(static_cast<InputSectionBase *>(S));
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000317 }
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000318
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000319 return Ret;
320}
321
Petr Hoseke5d3ca52016-08-31 15:31:17 +0000322template <class ELFT>
Rui Ueyama02a036f2017-02-27 02:31:48 +0000323void LinkerScript<ELFT>::processCommands(OutputSectionFactory &Factory) {
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000324 for (unsigned I = 0; I < Opt.Commands.size(); ++I) {
325 auto Iter = Opt.Commands.begin() + I;
326 const std::unique_ptr<BaseCommand> &Base1 = *Iter;
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000327
328 // Handle symbol assignments outside of any output section.
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000329 if (auto *Cmd = dyn_cast<SymbolAssignment>(Base1.get())) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000330 addSymbol(Cmd);
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000331 continue;
332 }
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000333
Eugene Leviant20d03192016-09-16 15:30:47 +0000334 if (auto *Cmd = dyn_cast<AssertCommand>(Base1.get())) {
335 // If we don't have SECTIONS then output sections have already been
George Rimar194470cd2016-09-17 19:21:05 +0000336 // created by Writer<ELFT>. The LinkerScript<ELFT>::assignAddresses
Eugene Leviant20d03192016-09-16 15:30:47 +0000337 // will not be called, so ASSERT should be evaluated now.
338 if (!Opt.HasSections)
339 Cmd->Expression(0);
340 continue;
341 }
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000342
Eugene Leviantceabe802016-08-11 07:56:43 +0000343 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base1.get())) {
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000344 std::vector<InputSectionBase *> V = createInputSectionList(*Cmd);
Rafael Espindola7bd37872016-09-12 16:05:16 +0000345
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000346 // The output section name `/DISCARD/' is special.
347 // Any input section assigned to it is discarded.
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000348 if (Cmd->Name == "/DISCARD/") {
Rafael Espindola7bd37872016-09-12 16:05:16 +0000349 discard(V);
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000350 continue;
351 }
Eugene Leviantceabe802016-08-11 07:56:43 +0000352
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000353 // This is for ONLY_IF_RO and ONLY_IF_RW. An output section directive
354 // ".foo : ONLY_IF_R[OW] { ... }" is handled only if all member input
355 // sections satisfy a given constraint. If not, a directive is handled
356 // as if it wasn't present from the beginning.
357 //
358 // Because we'll iterate over Commands many more times, the easiest
359 // way to "make it as if it wasn't present" is to just remove it.
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000360 if (!matchConstraints<ELFT>(V, Cmd->Constraint)) {
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000361 for (InputSectionBase *S : V)
Rui Ueyamaf94efdd2016-11-20 23:15:52 +0000362 S->Assigned = false;
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000363 Opt.Commands.erase(Iter);
George Rimardfbbbc82016-09-17 09:50:10 +0000364 --I;
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000365 continue;
366 }
367
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000368 // A directive may contain symbol definitions like this:
369 // ".foo : { ...; bar = .; }". Handle them.
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000370 for (const std::unique_ptr<BaseCommand> &Base : Cmd->Commands)
371 if (auto *OutCmd = dyn_cast<SymbolAssignment>(Base.get()))
Rafael Espindola4cd73522017-02-17 16:01:51 +0000372 addSymbol(OutCmd);
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000373
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000374 // Handle subalign (e.g. ".foo : SUBALIGN(32) { ... }"). If subalign
375 // is given, input sections are aligned to that value, whether the
376 // given value is larger or smaller than the original section alignment.
377 if (Cmd->SubalignExpr) {
378 uint32_t Subalign = Cmd->SubalignExpr(0);
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000379 for (InputSectionBase *S : V)
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000380 S->Alignment = Subalign;
George Rimardb24d9c2016-08-19 15:18:23 +0000381 }
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000382
383 // Add input sections to an output section.
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000384 for (InputSectionBase *S : V)
Rui Ueyama02a036f2017-02-27 02:31:48 +0000385 Factory.addInputSec<ELFT>(S, Cmd->Name);
Eugene Leviantceabe802016-08-11 07:56:43 +0000386 }
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000387 }
Eugene Leviant20d03192016-09-16 15:30:47 +0000388}
Eugene Leviante63d81b2016-07-20 14:43:20 +0000389
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000390// Add sections that didn't match any sections command.
Eugene Leviant20d03192016-09-16 15:30:47 +0000391template <class ELFT>
Rui Ueyama02a036f2017-02-27 02:31:48 +0000392void LinkerScript<ELFT>::addOrphanSections(OutputSectionFactory &Factory) {
Rui Ueyama536a2672017-02-27 02:32:08 +0000393 for (InputSectionBase *S : InputSections)
Rafael Espindola8f9026b2016-11-08 18:23:02 +0000394 if (S->Live && !S->OutSec)
Rui Ueyama02a036f2017-02-27 02:31:48 +0000395 Factory.addInputSec<ELFT>(S, getOutputSectionName(S->Name));
Eugene Leviante63d81b2016-07-20 14:43:20 +0000396}
397
Rafael Espindola24e6f362017-02-24 15:07:30 +0000398template <class ELFT> static bool isTbss(OutputSection *Sec) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000399 return (Sec->Flags & SHF_TLS) && Sec->Type == SHT_NOBITS;
Rafael Espindolaa940e532016-09-22 12:35:44 +0000400}
401
Rafael Espindola774ea7d2017-02-23 16:49:07 +0000402template <class ELFT> void LinkerScript<ELFT>::output(InputSection *S) {
Rafael Espindolad3190792016-09-16 15:10:23 +0000403 if (!AlreadyOutputIS.insert(S).second)
404 return;
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000405 bool IsTbss = isTbss<ELFT>(CurOutSec);
Eugene Leviant20889c52016-08-31 08:13:33 +0000406
Rafael Espindolad3190792016-09-16 15:10:23 +0000407 uintX_t Pos = IsTbss ? Dot + ThreadBssOffset : Dot;
408 Pos = alignTo(Pos, S->Alignment);
Rafael Espindola04a2e342016-11-09 01:42:41 +0000409 S->OutSecOff = Pos - CurOutSec->Addr;
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000410 Pos += S->template getSize<ELFT>();
Rafael Espindolad3190792016-09-16 15:10:23 +0000411
412 // Update output section size after adding each section. This is so that
413 // SIZEOF works correctly in the case below:
414 // .foo { *(.aaa) a = SIZEOF(.foo); *(.bbb) }
Rafael Espindola04a2e342016-11-09 01:42:41 +0000415 CurOutSec->Size = Pos - CurOutSec->Addr;
Rafael Espindolad3190792016-09-16 15:10:23 +0000416
Meador Ingeb8897442017-01-24 02:34:00 +0000417 // If there is a memory region associated with this input section, then
418 // place the section in that region and update the region index.
419 if (CurMemRegion) {
420 CurMemRegion->Offset += CurOutSec->Size;
421 uint64_t CurSize = CurMemRegion->Offset - CurMemRegion->Origin;
422 if (CurSize > CurMemRegion->Length) {
423 uint64_t OverflowAmt = CurSize - CurMemRegion->Length;
424 error("section '" + CurOutSec->Name + "' will not fit in region '" +
425 CurMemRegion->Name + "': overflowed by " + Twine(OverflowAmt) +
426 " bytes");
427 }
428 }
429
Rafael Espindola7252ae52016-09-22 12:00:08 +0000430 if (IsTbss)
431 ThreadBssOffset = Pos - Dot;
432 else
Rafael Espindolad3190792016-09-16 15:10:23 +0000433 Dot = Pos;
434}
435
436template <class ELFT> void LinkerScript<ELFT>::flush() {
Rafael Espindola65499b92016-09-23 20:10:47 +0000437 if (!CurOutSec || !AlreadyOutputOS.insert(CurOutSec).second)
438 return;
Rafael Espindola24e6f362017-02-24 15:07:30 +0000439 for (InputSection *I : CurOutSec->Sections)
440 output(I);
Eugene Leviant20889c52016-08-31 08:13:33 +0000441}
442
Rafael Espindola24e6f362017-02-24 15:07:30 +0000443template <class ELFT> void LinkerScript<ELFT>::switchTo(OutputSection *Sec) {
Rafael Espindolad3190792016-09-16 15:10:23 +0000444 if (CurOutSec == Sec)
445 return;
446 if (AlreadyOutputOS.count(Sec))
447 return;
448
449 flush();
450 CurOutSec = Sec;
451
Rafael Espindola04a2e342016-11-09 01:42:41 +0000452 Dot = alignTo(Dot, CurOutSec->Addralign);
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000453 CurOutSec->Addr = isTbss<ELFT>(CurOutSec) ? Dot + ThreadBssOffset : Dot;
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000454
455 // If neither AT nor AT> is specified for an allocatable section, the linker
456 // will set the LMA such that the difference between VMA and LMA for the
457 // section is the same as the preceding output section in the same region
458 // https://sourceware.org/binutils/docs-2.20/ld/Output-Section-LMA.html
George Rimar21467872017-02-23 07:57:55 +0000459 if (LMAOffset)
Rafael Espindola29c1afb2017-02-24 14:34:12 +0000460 CurOutSec->LMAOffset = LMAOffset();
Rafael Espindolad3190792016-09-16 15:10:23 +0000461}
462
463template <class ELFT> void LinkerScript<ELFT>::process(BaseCommand &Base) {
George Rimare38cbab2016-09-26 19:22:50 +0000464 // This handles the assignments to symbol or to a location counter (.)
Rafael Espindolad3190792016-09-16 15:10:23 +0000465 if (auto *AssignCmd = dyn_cast<SymbolAssignment>(&Base)) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000466 assignSymbol(AssignCmd, true);
Eugene Leviantceabe802016-08-11 07:56:43 +0000467 return;
Rui Ueyama2de509c2016-08-12 00:55:08 +0000468 }
George Rimare38cbab2016-09-26 19:22:50 +0000469
470 // Handle BYTE(), SHORT(), LONG(), or QUAD().
471 if (auto *DataCmd = dyn_cast<BytesDataCommand>(&Base)) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000472 DataCmd->Offset = Dot - CurOutSec->Addr;
George Rimare38cbab2016-09-26 19:22:50 +0000473 Dot += DataCmd->Size;
Rafael Espindola04a2e342016-11-09 01:42:41 +0000474 CurOutSec->Size = Dot - CurOutSec->Addr;
George Rimare38cbab2016-09-26 19:22:50 +0000475 return;
476 }
477
Meador Ingeb2d99d62016-11-22 18:01:50 +0000478 if (auto *AssertCmd = dyn_cast<AssertCommand>(&Base)) {
479 AssertCmd->Expression(Dot);
480 return;
481 }
482
George Rimare38cbab2016-09-26 19:22:50 +0000483 // It handles single input section description command,
484 // calculates and assigns the offsets for each section and also
485 // updates the output section size.
Rafael Espindolad3190792016-09-16 15:10:23 +0000486 auto &ICmd = cast<InputSectionDescription>(Base);
Rafael Espindolac404d502017-02-23 02:32:18 +0000487 for (InputSectionBase *ID : ICmd.Sections) {
George Rimar3fb5a6d2016-11-29 16:05:27 +0000488 // We tentatively added all synthetic sections at the beginning and removed
489 // empty ones afterwards (because there is no way to know whether they were
490 // going be empty or not other than actually running linker scripts.)
491 // We need to ignore remains of empty sections.
Rui Ueyama9320cb02017-02-27 02:56:02 +0000492 if (auto *Sec = dyn_cast<SyntheticSection>(ID))
George Rimar3fb5a6d2016-11-29 16:05:27 +0000493 if (Sec->empty())
494 continue;
495
Rafael Espindolab4c9b812017-02-23 02:28:28 +0000496 auto *IB = static_cast<InputSectionBase *>(ID);
George Rimar78ef6452017-02-21 15:46:43 +0000497 if (!IB->Live)
498 continue;
Rafael Espindolad3190792016-09-16 15:10:23 +0000499 switchTo(IB->OutSec);
Rafael Espindola774ea7d2017-02-23 16:49:07 +0000500 if (auto *I = dyn_cast<InputSection>(IB))
Rafael Espindolad3190792016-09-16 15:10:23 +0000501 output(I);
Rafael Espindola65499b92016-09-23 20:10:47 +0000502 else
503 flush();
Eugene Leviantceabe802016-08-11 07:56:43 +0000504 }
505}
506
George Rimar8f66df92016-08-12 20:38:20 +0000507template <class ELFT>
Rafael Espindola24e6f362017-02-24 15:07:30 +0000508static OutputSection *
509findSection(StringRef Name, const std::vector<OutputSection *> &Sections) {
Rafael Espindola2b074552017-02-03 22:27:05 +0000510 auto End = Sections.end();
Rafael Espindola24e6f362017-02-24 15:07:30 +0000511 auto HasName = [=](OutputSection *Sec) { return Sec->Name == Name; };
Rafael Espindola2b074552017-02-03 22:27:05 +0000512 auto I = std::find_if(Sections.begin(), End, HasName);
Rafael Espindola24e6f362017-02-24 15:07:30 +0000513 std::vector<OutputSection *> Ret;
Rafael Espindola2b074552017-02-03 22:27:05 +0000514 if (I == End)
515 return nullptr;
516 assert(std::find_if(I + 1, End, HasName) == End);
517 return *I;
George Rimar8f66df92016-08-12 20:38:20 +0000518}
519
Meador Ingeb8897442017-01-24 02:34:00 +0000520// This function searches for a memory region to place the given output
521// section in. If found, a pointer to the appropriate memory region is
522// returned. Otherwise, a nullptr is returned.
523template <class ELFT>
524MemoryRegion *LinkerScript<ELFT>::findMemoryRegion(OutputSectionCommand *Cmd,
Rafael Espindola24e6f362017-02-24 15:07:30 +0000525 OutputSection *Sec) {
Meador Ingeb8897442017-01-24 02:34:00 +0000526 // If a memory region name was specified in the output section command,
527 // then try to find that region first.
528 if (!Cmd->MemoryRegionName.empty()) {
529 auto It = Opt.MemoryRegions.find(Cmd->MemoryRegionName);
530 if (It != Opt.MemoryRegions.end())
531 return &It->second;
532 error("memory region '" + Cmd->MemoryRegionName + "' not declared");
533 return nullptr;
534 }
535
536 // The memory region name is empty, thus a suitable region must be
537 // searched for in the region map. If the region map is empty, just
538 // return. Note that this check doesn't happen at the very beginning
539 // so that uses of undeclared regions can be caught.
540 if (!Opt.MemoryRegions.size())
541 return nullptr;
542
543 // See if a region can be found by matching section flags.
544 for (auto &MRI : Opt.MemoryRegions) {
545 MemoryRegion &MR = MRI.second;
Rui Ueyama8a8a9532017-01-26 02:58:59 +0000546 if ((MR.Flags & Sec->Flags) != 0 && (MR.NegFlags & Sec->Flags) == 0)
Meador Ingeb8897442017-01-24 02:34:00 +0000547 return &MR;
548 }
549
550 // Otherwise, no suitable region was found.
551 if (Sec->Flags & SHF_ALLOC)
552 error("no memory region specified for section '" + Sec->Name + "'");
553 return nullptr;
554}
555
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000556// This function assigns offsets to input sections and an output section
557// for a single sections command (e.g. ".text { *(.text); }").
Rafael Espindolad3190792016-09-16 15:10:23 +0000558template <class ELFT>
559void LinkerScript<ELFT>::assignOffsets(OutputSectionCommand *Cmd) {
George Rimar21467872017-02-23 07:57:55 +0000560 if (Cmd->LMAExpr) {
561 uintX_t D = Dot;
562 LMAOffset = [=] { return Cmd->LMAExpr(D) - D; };
563 }
Rafael Espindola24e6f362017-02-24 15:07:30 +0000564 OutputSection *Sec = findSection<ELFT>(Cmd->Name, *OutputSections);
Rafael Espindola2b074552017-02-03 22:27:05 +0000565 if (!Sec)
Rafael Espindolad3190792016-09-16 15:10:23 +0000566 return;
Meador Ingeb8897442017-01-24 02:34:00 +0000567
Rafael Espindola679828f2017-02-17 16:26:13 +0000568 if (Cmd->AddrExpr && Sec->Flags & SHF_ALLOC)
George Rimar2ee2d2d2017-02-21 14:50:38 +0000569 setDot(Cmd->AddrExpr, Cmd->Location);
Rafael Espindola679828f2017-02-17 16:26:13 +0000570
Petr Hosek165088a2017-02-07 23:42:31 +0000571 // Handle align (e.g. ".foo : ALIGN(16) { ... }").
572 if (Cmd->AlignExpr)
573 Sec->updateAlignment(Cmd->AlignExpr(0));
574
Meador Ingeb8897442017-01-24 02:34:00 +0000575 // Try and find an appropriate memory region to assign offsets in.
576 CurMemRegion = findMemoryRegion(Cmd, Sec);
577 if (CurMemRegion)
578 Dot = CurMemRegion->Offset;
579 switchTo(Sec);
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000580
Rafael Espindolad3190792016-09-16 15:10:23 +0000581 // Find the last section output location. We will output orphan sections
582 // there so that end symbols point to the correct location.
583 auto E = std::find_if(Cmd->Commands.rbegin(), Cmd->Commands.rend(),
584 [](const std::unique_ptr<BaseCommand> &Cmd) {
585 return !isa<SymbolAssignment>(*Cmd);
586 })
587 .base();
588 for (auto I = Cmd->Commands.begin(); I != E; ++I)
589 process(**I);
Rafael Espindola65499b92016-09-23 20:10:47 +0000590 flush();
George Rimarb31dd372016-09-19 13:27:31 +0000591 std::for_each(E, Cmd->Commands.end(),
592 [this](std::unique_ptr<BaseCommand> &B) { process(*B.get()); });
Rafael Espindolad3190792016-09-16 15:10:23 +0000593}
594
Rafael Espindola07fe6122016-11-14 14:23:35 +0000595template <class ELFT> void LinkerScript<ELFT>::removeEmptyCommands() {
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000596 // It is common practice to use very generic linker scripts. So for any
597 // given run some of the output sections in the script will be empty.
598 // We could create corresponding empty output sections, but that would
599 // clutter the output.
600 // We instead remove trivially empty sections. The bfd linker seems even
601 // more aggressive at removing them.
602 auto Pos = std::remove_if(
603 Opt.Commands.begin(), Opt.Commands.end(),
604 [&](const std::unique_ptr<BaseCommand> &Base) {
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000605 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
Rafael Espindola2b074552017-02-03 22:27:05 +0000606 return !findSection<ELFT>(Cmd->Name, *OutputSections);
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000607 return false;
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000608 });
609 Opt.Commands.erase(Pos, Opt.Commands.end());
Rafael Espindola07fe6122016-11-14 14:23:35 +0000610}
611
Rafael Espindola6a537372016-11-14 14:33:49 +0000612static bool isAllSectionDescription(const OutputSectionCommand &Cmd) {
613 for (const std::unique_ptr<BaseCommand> &I : Cmd.Commands)
614 if (!isa<InputSectionDescription>(*I))
615 return false;
616 return true;
617}
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000618
Rafael Espindola6a537372016-11-14 14:33:49 +0000619template <class ELFT> void LinkerScript<ELFT>::adjustSectionsBeforeSorting() {
Rafael Espindola9546fff2016-09-22 14:40:50 +0000620 // If the output section contains only symbol assignments, create a
621 // corresponding output section. The bfd linker seems to only create them if
622 // '.' is assigned to, but creating these section should not have any bad
623 // consequeces and gives us a section to put the symbol in.
624 uintX_t Flags = SHF_ALLOC;
Rafael Espindolaf93b8c22016-11-26 06:55:35 +0000625 uint32_t Type = SHT_NOBITS;
Rafael Espindola9546fff2016-09-22 14:40:50 +0000626 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
627 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
628 if (!Cmd)
629 continue;
Rafael Espindola24e6f362017-02-24 15:07:30 +0000630 if (OutputSection *Sec = findSection<ELFT>(Cmd->Name, *OutputSections)) {
Rafael Espindola2b074552017-02-03 22:27:05 +0000631 Flags = Sec->Flags;
632 Type = Sec->Type;
Rafael Espindola9546fff2016-09-22 14:40:50 +0000633 continue;
634 }
635
Rafael Espindola6a537372016-11-14 14:33:49 +0000636 if (isAllSectionDescription(*Cmd))
637 continue;
638
Rafael Espindola24e6f362017-02-24 15:07:30 +0000639 auto *OutSec = make<OutputSection>(Cmd->Name, Type, Flags);
Rafael Espindola9546fff2016-09-22 14:40:50 +0000640 OutputSections->push_back(OutSec);
641 }
Rafael Espindolaf7a17442016-11-14 15:39:38 +0000642}
643
644template <class ELFT> void LinkerScript<ELFT>::adjustSectionsAfterSorting() {
645 placeOrphanSections();
646
647 // If output section command doesn't specify any segments,
648 // and we haven't previously assigned any section to segment,
649 // then we simply assign section to the very first load segment.
650 // Below is an example of such linker script:
651 // PHDRS { seg PT_LOAD; }
652 // SECTIONS { .aaa : { *(.aaa) } }
653 std::vector<StringRef> DefPhdrs;
654 auto FirstPtLoad =
655 std::find_if(Opt.PhdrsCommands.begin(), Opt.PhdrsCommands.end(),
656 [](const PhdrsCommand &Cmd) { return Cmd.Type == PT_LOAD; });
657 if (FirstPtLoad != Opt.PhdrsCommands.end())
658 DefPhdrs.push_back(FirstPtLoad->Name);
659
660 // Walk the commands and propagate the program headers to commands that don't
661 // explicitly specify them.
662 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
663 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
664 if (!Cmd)
665 continue;
666 if (Cmd->Phdrs.empty())
667 Cmd->Phdrs = DefPhdrs;
668 else
669 DefPhdrs = Cmd->Phdrs;
670 }
Rafael Espindola6a537372016-11-14 14:33:49 +0000671
672 removeEmptyCommands();
Rafael Espindola9546fff2016-09-22 14:40:50 +0000673}
674
Rafael Espindola15c57952016-09-22 18:05:49 +0000675// When placing orphan sections, we want to place them after symbol assignments
676// so that an orphan after
677// begin_foo = .;
678// foo : { *(foo) }
679// end_foo = .;
680// doesn't break the intended meaning of the begin/end symbols.
681// We don't want to go over sections since Writer<ELFT>::sortSections is the
682// one in charge of deciding the order of the sections.
683// We don't want to go over alignments, since doing so in
684// rx_sec : { *(rx_sec) }
685// . = ALIGN(0x1000);
686// /* The RW PT_LOAD starts here*/
687// rw_sec : { *(rw_sec) }
688// would mean that the RW PT_LOAD would become unaligned.
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000689static bool shouldSkip(const BaseCommand &Cmd) {
Rafael Espindola15c57952016-09-22 18:05:49 +0000690 if (isa<OutputSectionCommand>(Cmd))
691 return false;
692 const auto *Assign = dyn_cast<SymbolAssignment>(&Cmd);
693 if (!Assign)
694 return true;
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000695 return Assign->Name != ".";
Rafael Espindola15c57952016-09-22 18:05:49 +0000696}
697
Rui Ueyama6697ec22017-02-02 23:26:12 +0000698// Orphan sections are sections present in the input files which are
699// not explicitly placed into the output file by the linker script.
700//
701// When the control reaches this function, Opt.Commands contains
702// output section commands for non-orphan sections only. This function
703// adds new elements for orphan sections to Opt.Commands so that all
704// sections are explicitly handled by Opt.Commands.
705//
706// Writer<ELFT>::sortSections has already sorted output sections.
707// What we need to do is to scan OutputSections vector and
708// Opt.Commands in parallel to find orphan sections. If there is an
709// output section that doesn't have a corresponding entry in
710// Opt.Commands, we will insert a new entry to Opt.Commands.
711//
712// There is some ambiguity as to where exactly a new entry should be
713// inserted, because Opt.Commands contains not only output section
714// commands but other types of commands such as symbol assignment
715// expressions. There's no correct answer here due to the lack of the
716// formal specification of the linker script. We use heuristics to
717// determine whether a new output command should be added before or
718// after another commands. For the details, look at shouldSkip
719// function.
George Rimar93c64022016-12-15 15:38:09 +0000720template <class ELFT> void LinkerScript<ELFT>::placeOrphanSections() {
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000721 // The OutputSections are already in the correct order.
722 // This loops creates or moves commands as needed so that they are in the
723 // correct order.
724 int CmdIndex = 0;
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000725
726 // As a horrible special case, skip the first . assignment if it is before any
727 // section. We do this because it is common to set a load address by starting
728 // the script with ". = 0xabcd" and the expectation is that every section is
729 // after that.
730 auto FirstSectionOrDotAssignment =
731 std::find_if(Opt.Commands.begin(), Opt.Commands.end(),
732 [](const std::unique_ptr<BaseCommand> &Cmd) {
733 if (isa<OutputSectionCommand>(*Cmd))
734 return true;
735 const auto *Assign = dyn_cast<SymbolAssignment>(Cmd.get());
736 if (!Assign)
737 return false;
738 return Assign->Name == ".";
739 });
740 if (FirstSectionOrDotAssignment != Opt.Commands.end()) {
741 CmdIndex = FirstSectionOrDotAssignment - Opt.Commands.begin();
742 if (isa<SymbolAssignment>(**FirstSectionOrDotAssignment))
743 ++CmdIndex;
744 }
745
Rafael Espindola24e6f362017-02-24 15:07:30 +0000746 for (OutputSection *Sec : *OutputSections) {
Rafael Espindola40849412017-02-24 14:28:00 +0000747 StringRef Name = Sec->Name;
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000748
749 // Find the last spot where we can insert a command and still get the
Rafael Espindola15c57952016-09-22 18:05:49 +0000750 // correct result.
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000751 auto CmdIter = Opt.Commands.begin() + CmdIndex;
752 auto E = Opt.Commands.end();
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000753 while (CmdIter != E && shouldSkip(**CmdIter)) {
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000754 ++CmdIter;
755 ++CmdIndex;
756 }
757
758 auto Pos =
759 std::find_if(CmdIter, E, [&](const std::unique_ptr<BaseCommand> &Base) {
760 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
761 return Cmd && Cmd->Name == Name;
762 });
763 if (Pos == E) {
764 Opt.Commands.insert(CmdIter,
765 llvm::make_unique<OutputSectionCommand>(Name));
Rafael Espindola15c57952016-09-22 18:05:49 +0000766 ++CmdIndex;
767 continue;
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000768 }
Rafael Espindola15c57952016-09-22 18:05:49 +0000769
770 // Continue from where we found it.
771 CmdIndex = (Pos - Opt.Commands.begin()) + 1;
George Rimar652852c2016-04-16 10:10:32 +0000772 }
Rafael Espindola337f9032016-11-14 14:13:32 +0000773}
774
775template <class ELFT>
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000776void LinkerScript<ELFT>::assignAddresses(std::vector<PhdrEntry> &Phdrs) {
Rui Ueyama7c18c282016-04-18 21:00:40 +0000777 // Assign addresses as instructed by linker script SECTIONS sub-commands.
Rafael Espindolabe607332016-09-30 00:16:11 +0000778 Dot = 0;
George Rimar652852c2016-04-16 10:10:32 +0000779
Rafael Espindola06f47432017-02-06 22:21:46 +0000780 // A symbol can be assigned before any section is mentioned in the linker
781 // script. In an DSO, the symbol values are addresses, so the only important
782 // section values are:
783 // * SHN_UNDEF
784 // * SHN_ABS
785 // * Any value meaning a regular section.
786 // To handle that, create a dummy aether section that fills the void before
787 // the linker scripts switches to another section. It has an index of one
788 // which will map to whatever the first actual section is.
Rafael Espindola24e6f362017-02-24 15:07:30 +0000789 auto *Aether = make<OutputSection>("", 0, SHF_ALLOC);
Rafael Espindola06f47432017-02-06 22:21:46 +0000790 Aether->SectionIndex = 1;
791 switchTo(Aether);
792
George Rimar076fe152016-07-21 06:43:01 +0000793 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
794 if (auto *Cmd = dyn_cast<SymbolAssignment>(Base.get())) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000795 assignSymbol(Cmd);
George Rimar652852c2016-04-16 10:10:32 +0000796 continue;
797 }
798
George Rimareefa7582016-08-04 09:29:31 +0000799 if (auto *Cmd = dyn_cast<AssertCommand>(Base.get())) {
800 Cmd->Expression(Dot);
801 continue;
802 }
803
George Rimar076fe152016-07-21 06:43:01 +0000804 auto *Cmd = cast<OutputSectionCommand>(Base.get());
Rafael Espindolad3190792016-09-16 15:10:23 +0000805 assignOffsets(Cmd);
George Rimar652852c2016-04-16 10:10:32 +0000806 }
Rui Ueyama52c4e172016-07-01 10:42:25 +0000807
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000808 uintX_t MinVA = std::numeric_limits<uintX_t>::max();
Rafael Espindola24e6f362017-02-24 15:07:30 +0000809 for (OutputSection *Sec : *OutputSections) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000810 if (Sec->Flags & SHF_ALLOC)
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000811 MinVA = std::min<uint64_t>(MinVA, Sec->Addr);
Rafael Espindolaea590d92017-02-08 15:19:03 +0000812 else
813 Sec->Addr = 0;
814 }
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000815
Rafael Espindola8c495e22017-01-20 20:41:18 +0000816 allocateHeaders<ELFT>(Phdrs, *OutputSections, MinVA);
George Rimar652852c2016-04-16 10:10:32 +0000817}
818
Rui Ueyama464daad2016-08-22 04:55:20 +0000819// Creates program headers as instructed by PHDRS linker script command.
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000820template <class ELFT> std::vector<PhdrEntry> LinkerScript<ELFT>::createPhdrs() {
821 std::vector<PhdrEntry> Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +0000822
Rui Ueyama464daad2016-08-22 04:55:20 +0000823 // Process PHDRS and FILEHDR keywords because they are not
824 // real output sections and cannot be added in the following loop.
Eugene Leviantbbe38602016-07-19 09:25:43 +0000825 for (const PhdrsCommand &Cmd : Opt.PhdrsCommands) {
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000826 Ret.emplace_back(Cmd.Type, Cmd.Flags == UINT_MAX ? PF_R : Cmd.Flags);
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000827 PhdrEntry &Phdr = Ret.back();
Eugene Leviantbbe38602016-07-19 09:25:43 +0000828
829 if (Cmd.HasFilehdr)
Rui Ueyama9d1bacb12017-02-27 02:31:26 +0000830 Phdr.add(Out::ElfHeader);
Eugene Leviantbbe38602016-07-19 09:25:43 +0000831 if (Cmd.HasPhdrs)
Rui Ueyama9d1bacb12017-02-27 02:31:26 +0000832 Phdr.add(Out::ProgramHeaders);
Eugene Leviant56b21c82016-09-09 09:46:16 +0000833
834 if (Cmd.LMAExpr) {
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000835 Phdr.p_paddr = Cmd.LMAExpr(0);
Eugene Leviant56b21c82016-09-09 09:46:16 +0000836 Phdr.HasLMA = true;
837 }
Eugene Leviantbbe38602016-07-19 09:25:43 +0000838 }
839
Rui Ueyama464daad2016-08-22 04:55:20 +0000840 // Add output sections to program headers.
Rafael Espindola24e6f362017-02-24 15:07:30 +0000841 for (OutputSection *Sec : *OutputSections) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000842 if (!(Sec->Flags & SHF_ALLOC))
Eugene Leviantbbe38602016-07-19 09:25:43 +0000843 break;
844
Eugene Leviantce30b1c2016-10-19 15:04:49 +0000845 // Assign headers specified by linker script
Rafael Espindola40849412017-02-24 14:28:00 +0000846 for (size_t Id : getPhdrIndices(Sec->Name)) {
Eugene Leviantce30b1c2016-10-19 15:04:49 +0000847 Ret[Id].add(Sec);
848 if (Opt.PhdrsCommands[Id].Flags == UINT_MAX)
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000849 Ret[Id].p_flags |= Sec->getPhdrFlags();
Eugene Leviantbbe38602016-07-19 09:25:43 +0000850 }
Eugene Leviantbbe38602016-07-19 09:25:43 +0000851 }
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000852 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +0000853}
854
Eugene Leviantf9bc3bd2016-08-16 06:40:58 +0000855template <class ELFT> bool LinkerScript<ELFT>::ignoreInterpSection() {
856 // Ignore .interp section in case we have PHDRS specification
857 // and PT_INTERP isn't listed.
858 return !Opt.PhdrsCommands.empty() &&
859 llvm::find_if(Opt.PhdrsCommands, [](const PhdrsCommand &Cmd) {
860 return Cmd.Type == PT_INTERP;
861 }) == Opt.PhdrsCommands.end();
862}
863
George Rimar93c64022016-12-15 15:38:09 +0000864template <class ELFT> uint32_t LinkerScript<ELFT>::getFiller(StringRef Name) {
George Rimarf6c3cce2016-07-21 07:48:54 +0000865 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands)
866 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
867 if (Cmd->Name == Name)
868 return Cmd->Filler;
Rui Ueyama16068ae2016-11-19 18:05:56 +0000869 return 0;
George Rimare2ee72b2016-02-26 14:48:31 +0000870}
871
George Rimare38cbab2016-09-26 19:22:50 +0000872template <class ELFT>
873static void writeInt(uint8_t *Buf, uint64_t Data, uint64_t Size) {
874 const endianness E = ELFT::TargetEndianness;
875
876 switch (Size) {
877 case 1:
878 *Buf = (uint8_t)Data;
879 break;
880 case 2:
881 write16<E>(Buf, Data);
882 break;
883 case 4:
884 write32<E>(Buf, Data);
885 break;
886 case 8:
887 write64<E>(Buf, Data);
888 break;
889 default:
890 llvm_unreachable("unsupported Size argument");
891 }
892}
893
894template <class ELFT>
895void LinkerScript<ELFT>::writeDataBytes(StringRef Name, uint8_t *Buf) {
896 int I = getSectionIndex(Name);
897 if (I == INT_MAX)
898 return;
899
Rui Ueyama6e68c5e2016-11-19 18:05:58 +0000900 auto *Cmd = dyn_cast<OutputSectionCommand>(Opt.Commands[I].get());
901 for (const std::unique_ptr<BaseCommand> &Base : Cmd->Commands)
902 if (auto *Data = dyn_cast<BytesDataCommand>(Base.get()))
Meador Inge95c7d8d2016-12-08 23:21:30 +0000903 writeInt<ELFT>(Buf + Data->Offset, Data->Expression(0), Data->Size);
George Rimare38cbab2016-09-26 19:22:50 +0000904}
905
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000906template <class ELFT> bool LinkerScript<ELFT>::hasLMA(StringRef Name) {
George Rimar8ceadb32016-08-17 07:44:19 +0000907 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands)
908 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000909 if (Cmd->LMAExpr && Cmd->Name == Name)
910 return true;
911 return false;
George Rimar8ceadb32016-08-17 07:44:19 +0000912}
913
Rui Ueyamac3e2a4b2016-04-21 20:30:00 +0000914// Returns the index of the given section name in linker script
915// SECTIONS commands. Sections are laid out as the same order as they
916// were in the script. If a given name did not appear in the script,
917// it returns INT_MAX, so that it will be laid out at end of file.
George Rimar076fe152016-07-21 06:43:01 +0000918template <class ELFT> int LinkerScript<ELFT>::getSectionIndex(StringRef Name) {
Rui Ueyama6e68c5e2016-11-19 18:05:58 +0000919 for (int I = 0, E = Opt.Commands.size(); I != E; ++I)
920 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Opt.Commands[I].get()))
Rui Ueyamaf510fa62016-07-26 00:21:15 +0000921 if (Cmd->Name == Name)
922 return I;
Rui Ueyamaf510fa62016-07-26 00:21:15 +0000923 return INT_MAX;
George Rimar71b26e92016-04-21 10:22:02 +0000924}
925
Eugene Leviantbbe38602016-07-19 09:25:43 +0000926template <class ELFT> bool LinkerScript<ELFT>::hasPhdrsCommands() {
927 return !Opt.PhdrsCommands.empty();
928}
929
George Rimar9e694502016-07-29 16:18:47 +0000930template <class ELFT>
Rafael Espindola24e6f362017-02-24 15:07:30 +0000931const OutputSection *LinkerScript<ELFT>::getOutputSection(const Twine &Loc,
932 StringRef Name) {
933 static OutputSection FakeSec("", 0, 0);
George Rimar96659df2016-08-30 09:54:01 +0000934
Rafael Espindola24e6f362017-02-24 15:07:30 +0000935 for (OutputSection *Sec : *OutputSections)
Rafael Espindola40849412017-02-24 14:28:00 +0000936 if (Sec->Name == Name)
Eugene Leviantafaa9342016-11-16 09:49:39 +0000937 return Sec;
Eugene Levianted30ce72016-11-28 09:58:04 +0000938
939 error(Loc + ": undefined section " + Name);
Eugene Leviantafaa9342016-11-16 09:49:39 +0000940 return &FakeSec;
Eugene Leviant36fac7f2016-09-08 09:08:30 +0000941}
942
Rui Ueyamaedf75e72016-11-17 20:27:10 +0000943// This function is essentially the same as getOutputSection(Name)->Size,
944// but it won't print out an error message if a given section is not found.
945//
946// Linker script does not create an output section if its content is empty.
947// We want to allow SIZEOF(.foo) where .foo is a section which happened to
948// be empty. That is why this function is different from getOutputSection().
949template <class ELFT>
950uint64_t LinkerScript<ELFT>::getOutputSectionSize(StringRef Name) {
Rafael Espindola24e6f362017-02-24 15:07:30 +0000951 for (OutputSection *Sec : *OutputSections)
Rafael Espindola40849412017-02-24 14:28:00 +0000952 if (Sec->Name == Name)
Rui Ueyamaedf75e72016-11-17 20:27:10 +0000953 return Sec->Size;
954 return 0;
955}
956
George Rimar884e7862016-09-08 08:19:13 +0000957template <class ELFT> uint64_t LinkerScript<ELFT>::getHeaderSize() {
Rafael Espindola0d4b6d52016-09-22 16:47:21 +0000958 return elf::getHeaderSize<ELFT>();
George Rimare32a3592016-08-10 07:59:34 +0000959}
960
Eugene Leviantf6aeed32016-12-22 13:13:12 +0000961template <class ELFT>
962uint64_t LinkerScript<ELFT>::getSymbolValue(const Twine &Loc, StringRef S) {
George Rimar884e7862016-09-08 08:19:13 +0000963 if (SymbolBody *B = Symtab<ELFT>::X->find(S))
964 return B->getVA<ELFT>();
Eugene Leviantf6aeed32016-12-22 13:13:12 +0000965 error(Loc + ": symbol not found: " + S);
George Rimar884e7862016-09-08 08:19:13 +0000966 return 0;
967}
968
George Rimarf34f45f2016-09-23 13:17:23 +0000969template <class ELFT> bool LinkerScript<ELFT>::isDefined(StringRef S) {
970 return Symtab<ELFT>::X->find(S) != nullptr;
971}
972
Rafael Espindola2f831dc2016-10-31 19:56:37 +0000973template <class ELFT> bool LinkerScript<ELFT>::isAbsolute(StringRef S) {
974 SymbolBody *Sym = Symtab<ELFT>::X->find(S);
Rui Ueyama80474a22017-02-28 19:29:55 +0000975 auto *DR = dyn_cast_or_null<DefinedRegular>(Sym);
Rafael Espindola2f831dc2016-10-31 19:56:37 +0000976 return DR && !DR->Section;
977}
978
Eugene Leviantafaa9342016-11-16 09:49:39 +0000979// Gets section symbol belongs to. Symbol "." doesn't belong to any
980// specific section but isn't absolute at the same time, so we try
981// to find suitable section for it as well.
982template <class ELFT>
Rafael Espindola24e6f362017-02-24 15:07:30 +0000983const OutputSection *LinkerScript<ELFT>::getSymbolSection(StringRef S) {
Rafael Espindola06f47432017-02-06 22:21:46 +0000984 if (SymbolBody *Sym = Symtab<ELFT>::X->find(S))
Rui Ueyama968db482017-02-28 04:02:42 +0000985 return Sym->getOutputSection<ELFT>();
Rafael Espindola06f47432017-02-06 22:21:46 +0000986 return CurOutSec;
Eugene Leviantafaa9342016-11-16 09:49:39 +0000987}
988
Eugene Leviantbbe38602016-07-19 09:25:43 +0000989// Returns indices of ELF headers containing specific section, identified
990// by Name. Each index is a zero based number of ELF header listed within
991// PHDRS {} script block.
992template <class ELFT>
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000993std::vector<size_t> LinkerScript<ELFT>::getPhdrIndices(StringRef SectionName) {
George Rimar076fe152016-07-21 06:43:01 +0000994 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
995 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000996 if (!Cmd || Cmd->Name != SectionName)
George Rimar31d842f2016-07-20 16:43:03 +0000997 continue;
998
Rui Ueyama29c5a2a2016-07-26 00:27:36 +0000999 std::vector<size_t> Ret;
1000 for (StringRef PhdrName : Cmd->Phdrs)
Eugene Leviant2a942c42016-12-05 16:38:32 +00001001 Ret.push_back(getPhdrIndex(Cmd->Location, PhdrName));
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001002 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +00001003 }
George Rimar31d842f2016-07-20 16:43:03 +00001004 return {};
Eugene Leviantbbe38602016-07-19 09:25:43 +00001005}
1006
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001007template <class ELFT>
Eugene Leviant2a942c42016-12-05 16:38:32 +00001008size_t LinkerScript<ELFT>::getPhdrIndex(const Twine &Loc, StringRef PhdrName) {
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001009 size_t I = 0;
1010 for (PhdrsCommand &Cmd : Opt.PhdrsCommands) {
1011 if (Cmd.Name == PhdrName)
1012 return I;
1013 ++I;
1014 }
Eugene Leviant2a942c42016-12-05 16:38:32 +00001015 error(Loc + ": section header '" + PhdrName + "' is not listed in PHDRS");
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001016 return 0;
1017}
1018
Rui Ueyama794366a2017-02-14 04:47:05 +00001019class elf::ScriptParser final : public ScriptLexer {
George Rimarc3794e52016-02-24 09:21:47 +00001020 typedef void (ScriptParser::*Handler)();
1021
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001022public:
Rui Ueyama22375f22016-11-25 18:51:54 +00001023 ScriptParser(MemoryBufferRef MB)
Rui Ueyama794366a2017-02-14 04:47:05 +00001024 : ScriptLexer(MB),
Rui Ueyama22375f22016-11-25 18:51:54 +00001025 IsUnderSysroot(isUnderSysroot(MB.getBufferIdentifier())) {}
George Rimarf23b2322016-02-19 10:45:45 +00001026
George Rimar20b65982016-08-31 09:08:26 +00001027 void readLinkerScript();
1028 void readVersionScript();
Rafael Espindolad0ebd842016-12-08 17:54:26 +00001029 void readDynamicList();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001030
1031private:
Rui Ueyama52a15092015-10-11 03:28:42 +00001032 void addFile(StringRef Path);
1033
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001034 void readAsNeeded();
Denis Protivensky90c50992015-10-08 06:48:38 +00001035 void readEntry();
George Rimar83f406c2015-10-19 17:35:12 +00001036 void readExtern();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001037 void readGroup();
Rui Ueyama31aa1f82015-10-11 01:31:55 +00001038 void readInclude();
Meador Ingeb8897442017-01-24 02:34:00 +00001039 void readMemory();
Rui Ueyamaee592822015-10-07 00:25:09 +00001040 void readOutput();
Davide Italiano9159ce92015-10-12 21:50:08 +00001041 void readOutputArch();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001042 void readOutputFormat();
Eugene Leviantbbe38602016-07-19 09:25:43 +00001043 void readPhdrs();
Davide Italiano68a39a62015-10-08 17:51:41 +00001044 void readSearchDir();
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001045 void readSections();
Rui Ueyama95769b42016-08-31 20:03:54 +00001046 void readVersion();
1047 void readVersionScriptCommand();
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001048
Rui Ueyama113cdec2016-07-24 23:05:57 +00001049 SymbolAssignment *readAssignment(StringRef Name);
George Rimare38cbab2016-09-26 19:22:50 +00001050 BytesDataCommand *readBytesDataCommand(StringRef Tok);
Rui Ueyama16068ae2016-11-19 18:05:56 +00001051 uint32_t readFill();
Rui Ueyama10416562016-08-04 02:03:27 +00001052 OutputSectionCommand *readOutputSectionDescription(StringRef OutSec);
Rui Ueyama16068ae2016-11-19 18:05:56 +00001053 uint32_t readOutputSectionFiller(StringRef Tok);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001054 std::vector<StringRef> readOutputSectionPhdrs();
George Rimara2496cb2016-08-30 09:46:59 +00001055 InputSectionDescription *readInputSectionDescription(StringRef Tok);
Eugene Leviantdb688452016-11-03 10:54:58 +00001056 StringMatcher readFilePatterns();
George Rimar07171f22016-09-21 15:56:44 +00001057 std::vector<SectionPattern> readInputSectionsList();
George Rimara2496cb2016-08-30 09:46:59 +00001058 InputSectionDescription *readInputSectionRules(StringRef FilePattern);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001059 unsigned readPhdrType();
George Rimarbe394db2016-09-16 20:21:55 +00001060 SortSectionPolicy readSortKind();
Petr Hoseka35e39c2016-08-16 01:11:16 +00001061 SymbolAssignment *readProvideHidden(bool Provide, bool Hidden);
Rafael Espindolac96da112016-11-01 11:30:45 +00001062 SymbolAssignment *readProvideOrAssignment(StringRef Tok);
George Rimar03fc0102016-07-28 07:18:23 +00001063 void readSort();
George Rimareefa7582016-08-04 09:29:31 +00001064 Expr readAssert();
Rui Ueyama708019c2016-07-24 18:19:40 +00001065
Rui Ueyama24e626c2017-01-26 02:58:19 +00001066 uint64_t readMemoryAssignment(StringRef, StringRef, StringRef);
1067 std::pair<uint32_t, uint32_t> readMemoryAttributes();
1068
Rui Ueyama708019c2016-07-24 18:19:40 +00001069 Expr readExpr();
1070 Expr readExpr1(Expr Lhs, int MinPrec);
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001071 StringRef readParenLiteral();
Rui Ueyama708019c2016-07-24 18:19:40 +00001072 Expr readPrimary();
1073 Expr readTernary(Expr Cond);
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001074 Expr readParenExpr();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001075
George Rimar20b65982016-08-31 09:08:26 +00001076 // For parsing version script.
Rui Ueyama12450b22016-11-18 06:30:09 +00001077 std::vector<SymbolVersion> readVersionExtern();
1078 void readAnonymousDeclaration();
Rui Ueyama95769b42016-08-31 20:03:54 +00001079 void readVersionDeclaration(StringRef VerStr);
Rui Ueyama12450b22016-11-18 06:30:09 +00001080 std::vector<SymbolVersion> readSymbols();
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001081 void readLocals();
George Rimar20b65982016-08-31 09:08:26 +00001082
Rui Ueyama07320e42016-04-20 20:13:41 +00001083 ScriptConfiguration &Opt = *ScriptConfig;
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001084 bool IsUnderSysroot;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001085};
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001086
Rafael Espindolad0ebd842016-12-08 17:54:26 +00001087void ScriptParser::readDynamicList() {
1088 expect("{");
1089 readAnonymousDeclaration();
1090 if (!atEOF())
1091 setError("EOF expected, but got " + next());
1092}
1093
George Rimar20b65982016-08-31 09:08:26 +00001094void ScriptParser::readVersionScript() {
Rui Ueyama95769b42016-08-31 20:03:54 +00001095 readVersionScriptCommand();
1096 if (!atEOF())
1097 setError("EOF expected, but got " + next());
1098}
1099
1100void ScriptParser::readVersionScriptCommand() {
Rui Ueyama83043f22016-10-17 16:01:53 +00001101 if (consume("{")) {
Rui Ueyama12450b22016-11-18 06:30:09 +00001102 readAnonymousDeclaration();
George Rimar20b65982016-08-31 09:08:26 +00001103 return;
1104 }
1105
Rui Ueyama95769b42016-08-31 20:03:54 +00001106 while (!atEOF() && !Error && peek() != "}") {
George Rimar20b65982016-08-31 09:08:26 +00001107 StringRef VerStr = next();
1108 if (VerStr == "{") {
Rui Ueyama95769b42016-08-31 20:03:54 +00001109 setError("anonymous version definition is used in "
1110 "combination with other version definitions");
George Rimar20b65982016-08-31 09:08:26 +00001111 return;
1112 }
1113 expect("{");
Rui Ueyama95769b42016-08-31 20:03:54 +00001114 readVersionDeclaration(VerStr);
George Rimar20b65982016-08-31 09:08:26 +00001115 }
1116}
1117
Rui Ueyama95769b42016-08-31 20:03:54 +00001118void ScriptParser::readVersion() {
1119 expect("{");
1120 readVersionScriptCommand();
1121 expect("}");
1122}
1123
George Rimar20b65982016-08-31 09:08:26 +00001124void ScriptParser::readLinkerScript() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001125 while (!atEOF()) {
1126 StringRef Tok = next();
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001127 if (Tok == ";")
1128 continue;
1129
Eugene Leviant20d03192016-09-16 15:30:47 +00001130 if (Tok == "ASSERT") {
1131 Opt.Commands.emplace_back(new AssertCommand(readAssert()));
1132 } else if (Tok == "ENTRY") {
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001133 readEntry();
1134 } else if (Tok == "EXTERN") {
1135 readExtern();
1136 } else if (Tok == "GROUP" || Tok == "INPUT") {
1137 readGroup();
1138 } else if (Tok == "INCLUDE") {
1139 readInclude();
Meador Ingeb8897442017-01-24 02:34:00 +00001140 } else if (Tok == "MEMORY") {
1141 readMemory();
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001142 } else if (Tok == "OUTPUT") {
1143 readOutput();
1144 } else if (Tok == "OUTPUT_ARCH") {
1145 readOutputArch();
1146 } else if (Tok == "OUTPUT_FORMAT") {
1147 readOutputFormat();
1148 } else if (Tok == "PHDRS") {
1149 readPhdrs();
1150 } else if (Tok == "SEARCH_DIR") {
1151 readSearchDir();
1152 } else if (Tok == "SECTIONS") {
1153 readSections();
1154 } else if (Tok == "VERSION") {
1155 readVersion();
Rafael Espindolac96da112016-11-01 11:30:45 +00001156 } else if (SymbolAssignment *Cmd = readProvideOrAssignment(Tok)) {
Eugene Leviant20d03192016-09-16 15:30:47 +00001157 Opt.Commands.emplace_back(Cmd);
Petr Hoseke5d3ca52016-08-31 15:31:17 +00001158 } else {
George Rimar57610422016-03-11 14:43:02 +00001159 setError("unknown directive: " + Tok);
Petr Hoseke5d3ca52016-08-31 15:31:17 +00001160 }
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001161 }
1162}
1163
Rui Ueyama717677a2016-02-11 21:17:59 +00001164void ScriptParser::addFile(StringRef S) {
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001165 if (IsUnderSysroot && S.startswith("/")) {
Justin Bogner5af16872016-10-17 06:08:48 +00001166 SmallString<128> PathData;
1167 StringRef Path = (Config->Sysroot + S).toStringRef(PathData);
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001168 if (sys::fs::exists(Path)) {
Justin Bogner5af16872016-10-17 06:08:48 +00001169 Driver->addFile(Saver.save(Path));
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001170 return;
1171 }
1172 }
1173
Rui Ueyamaf03f3cc2015-10-13 00:09:21 +00001174 if (sys::path::is_absolute(S)) {
Rui Ueyama52a15092015-10-11 03:28:42 +00001175 Driver->addFile(S);
1176 } else if (S.startswith("=")) {
1177 if (Config->Sysroot.empty())
1178 Driver->addFile(S.substr(1));
1179 else
1180 Driver->addFile(Saver.save(Config->Sysroot + "/" + S.substr(1)));
1181 } else if (S.startswith("-l")) {
Rui Ueyama21eecb42016-02-02 21:13:09 +00001182 Driver->addLibrary(S.substr(2));
Simon Atanasyana1b8fc32015-11-26 20:23:46 +00001183 } else if (sys::fs::exists(S)) {
1184 Driver->addFile(S);
Rui Ueyama52a15092015-10-11 03:28:42 +00001185 } else {
Rui Ueyama061f9282016-11-19 19:23:58 +00001186 if (Optional<std::string> Path = findFromSearchPaths(S))
1187 Driver->addFile(Saver.save(*Path));
Rui Ueyama025d59b2016-02-02 20:27:59 +00001188 else
Rui Ueyama061f9282016-11-19 19:23:58 +00001189 setError("unable to find " + S);
Rui Ueyama52a15092015-10-11 03:28:42 +00001190 }
1191}
1192
Rui Ueyama717677a2016-02-11 21:17:59 +00001193void ScriptParser::readAsNeeded() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001194 expect("(");
Rui Ueyama35da9b62015-10-11 20:59:12 +00001195 bool Orig = Config->AsNeeded;
1196 Config->AsNeeded = true;
Rui Ueyama83043f22016-10-17 16:01:53 +00001197 while (!Error && !consume(")"))
George Rimarcd574a52016-09-09 14:35:36 +00001198 addFile(unquote(next()));
Rui Ueyama35da9b62015-10-11 20:59:12 +00001199 Config->AsNeeded = Orig;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001200}
1201
Rui Ueyama717677a2016-02-11 21:17:59 +00001202void ScriptParser::readEntry() {
Denis Protivensky90c50992015-10-08 06:48:38 +00001203 // -e <symbol> takes predecence over ENTRY(<symbol>).
1204 expect("(");
1205 StringRef Tok = next();
1206 if (Config->Entry.empty())
1207 Config->Entry = Tok;
1208 expect(")");
1209}
1210
Rui Ueyama717677a2016-02-11 21:17:59 +00001211void ScriptParser::readExtern() {
George Rimar83f406c2015-10-19 17:35:12 +00001212 expect("(");
Rui Ueyama83043f22016-10-17 16:01:53 +00001213 while (!Error && !consume(")"))
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001214 Config->Undefined.push_back(next());
George Rimar83f406c2015-10-19 17:35:12 +00001215}
1216
Rui Ueyama717677a2016-02-11 21:17:59 +00001217void ScriptParser::readGroup() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001218 expect("(");
Rui Ueyama83043f22016-10-17 16:01:53 +00001219 while (!Error && !consume(")")) {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001220 StringRef Tok = next();
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001221 if (Tok == "AS_NEEDED")
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001222 readAsNeeded();
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001223 else
George Rimarcd574a52016-09-09 14:35:36 +00001224 addFile(unquote(Tok));
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001225 }
1226}
1227
Rui Ueyama717677a2016-02-11 21:17:59 +00001228void ScriptParser::readInclude() {
George Rimard4500652016-12-21 09:42:25 +00001229 StringRef Tok = unquote(next());
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001230
George Rimard4500652016-12-21 09:42:25 +00001231 // https://sourceware.org/binutils/docs/ld/File-Commands.html:
1232 // The file will be searched for in the current directory, and in any
1233 // directory specified with the -L option.
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001234 if (sys::fs::exists(Tok)) {
1235 if (Optional<MemoryBufferRef> MB = readFile(Tok))
1236 tokenize(*MB);
Rui Ueyama025d59b2016-02-02 20:27:59 +00001237 return;
1238 }
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001239 if (Optional<std::string> Path = findFromSearchPaths(Tok)) {
1240 if (Optional<MemoryBufferRef> MB = readFile(*Path))
1241 tokenize(*MB);
1242 return;
1243 }
1244 setError("cannot open " + Tok);
Rui Ueyama31aa1f82015-10-11 01:31:55 +00001245}
1246
Rui Ueyama717677a2016-02-11 21:17:59 +00001247void ScriptParser::readOutput() {
Rui Ueyamaee592822015-10-07 00:25:09 +00001248 // -o <file> takes predecence over OUTPUT(<file>).
1249 expect("(");
1250 StringRef Tok = next();
1251 if (Config->OutputFile.empty())
George Rimarcd574a52016-09-09 14:35:36 +00001252 Config->OutputFile = unquote(Tok);
Rui Ueyamaee592822015-10-07 00:25:09 +00001253 expect(")");
1254}
1255
Rui Ueyama717677a2016-02-11 21:17:59 +00001256void ScriptParser::readOutputArch() {
George Rimar4e01c3e2017-02-08 09:59:06 +00001257 // OUTPUT_ARCH is ignored for now.
Davide Italiano9159ce92015-10-12 21:50:08 +00001258 expect("(");
George Rimar4e01c3e2017-02-08 09:59:06 +00001259 while (!Error && !consume(")"))
1260 skip();
Davide Italiano9159ce92015-10-12 21:50:08 +00001261}
1262
Rui Ueyama717677a2016-02-11 21:17:59 +00001263void ScriptParser::readOutputFormat() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001264 // Error checking only for now.
1265 expect("(");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001266 skip();
Davide Italiano6836c612015-10-12 21:08:41 +00001267 StringRef Tok = next();
1268 if (Tok == ")")
George Rimar6c55f0e2016-09-08 08:20:30 +00001269 return;
Rui Ueyama025d59b2016-02-02 20:27:59 +00001270 if (Tok != ",") {
George Rimar57610422016-03-11 14:43:02 +00001271 setError("unexpected token: " + Tok);
Rui Ueyama025d59b2016-02-02 20:27:59 +00001272 return;
1273 }
Justin Bogner5424e7c2016-10-17 06:21:13 +00001274 skip();
Davide Italiano6836c612015-10-12 21:08:41 +00001275 expect(",");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001276 skip();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001277 expect(")");
1278}
1279
Eugene Leviantbbe38602016-07-19 09:25:43 +00001280void ScriptParser::readPhdrs() {
1281 expect("{");
Rui Ueyama83043f22016-10-17 16:01:53 +00001282 while (!Error && !consume("}")) {
Eugene Leviantbbe38602016-07-19 09:25:43 +00001283 StringRef Tok = next();
Eugene Leviant56b21c82016-09-09 09:46:16 +00001284 Opt.PhdrsCommands.push_back(
1285 {Tok, PT_NULL, false, false, UINT_MAX, nullptr});
Eugene Leviantbbe38602016-07-19 09:25:43 +00001286 PhdrsCommand &PhdrCmd = Opt.PhdrsCommands.back();
1287
1288 PhdrCmd.Type = readPhdrType();
1289 do {
1290 Tok = next();
1291 if (Tok == ";")
1292 break;
1293 if (Tok == "FILEHDR")
1294 PhdrCmd.HasFilehdr = true;
1295 else if (Tok == "PHDRS")
1296 PhdrCmd.HasPhdrs = true;
Eugene Leviant56b21c82016-09-09 09:46:16 +00001297 else if (Tok == "AT")
1298 PhdrCmd.LMAExpr = readParenExpr();
Eugene Leviant865bf862016-07-21 10:43:25 +00001299 else if (Tok == "FLAGS") {
1300 expect("(");
Rafael Espindolaeb685cd2016-08-02 22:14:57 +00001301 // Passing 0 for the value of dot is a bit of a hack. It means that
1302 // we accept expressions like ".|1".
1303 PhdrCmd.Flags = readExpr()(0);
Eugene Leviant865bf862016-07-21 10:43:25 +00001304 expect(")");
1305 } else
Eugene Leviantbbe38602016-07-19 09:25:43 +00001306 setError("unexpected header attribute: " + Tok);
1307 } while (!Error);
1308 }
1309}
1310
Rui Ueyama717677a2016-02-11 21:17:59 +00001311void ScriptParser::readSearchDir() {
Davide Italiano68a39a62015-10-08 17:51:41 +00001312 expect("(");
Rui Ueyama86c5fb82016-09-08 23:26:54 +00001313 StringRef Tok = next();
Rui Ueyama6c7ad132016-09-02 19:20:33 +00001314 if (!Config->Nostdlib)
George Rimarcd574a52016-09-09 14:35:36 +00001315 Config->SearchPaths.push_back(unquote(Tok));
Davide Italiano68a39a62015-10-08 17:51:41 +00001316 expect(")");
1317}
1318
Rui Ueyama717677a2016-02-11 21:17:59 +00001319void ScriptParser::readSections() {
Eugene Leviante05336ff2016-09-14 08:32:36 +00001320 Opt.HasSections = true;
George Rimar18a30962016-11-28 10:11:10 +00001321 // -no-rosegment is used to avoid placing read only non-executable sections in
1322 // their own segment. We do the same if SECTIONS command is present in linker
1323 // script. See comment for computeFlags().
1324 Config->SingleRoRx = true;
1325
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001326 expect("{");
Rui Ueyama83043f22016-10-17 16:01:53 +00001327 while (!Error && !consume("}")) {
Rui Ueyama113cdec2016-07-24 23:05:57 +00001328 StringRef Tok = next();
Rafael Espindolac96da112016-11-01 11:30:45 +00001329 BaseCommand *Cmd = readProvideOrAssignment(Tok);
Eugene Leviantceabe802016-08-11 07:56:43 +00001330 if (!Cmd) {
1331 if (Tok == "ASSERT")
1332 Cmd = new AssertCommand(readAssert());
1333 else
1334 Cmd = readOutputSectionDescription(Tok);
Rui Ueyama708019c2016-07-24 18:19:40 +00001335 }
Rui Ueyama10416562016-08-04 02:03:27 +00001336 Opt.Commands.emplace_back(Cmd);
George Rimar652852c2016-04-16 10:10:32 +00001337 }
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001338}
1339
Rui Ueyama708019c2016-07-24 18:19:40 +00001340static int precedence(StringRef Op) {
1341 return StringSwitch<int>(Op)
Rui Ueyama0120e3f2016-09-23 18:06:51 +00001342 .Cases("*", "/", 5)
1343 .Cases("+", "-", 4)
1344 .Cases("<<", ">>", 3)
Rui Ueyama9c4ac5f2016-09-23 22:22:34 +00001345 .Cases("<", "<=", ">", ">=", "==", "!=", 2)
Rui Ueyama0120e3f2016-09-23 18:06:51 +00001346 .Cases("&", "|", 1)
Rui Ueyama708019c2016-07-24 18:19:40 +00001347 .Default(-1);
1348}
1349
Eugene Leviantdb688452016-11-03 10:54:58 +00001350StringMatcher ScriptParser::readFilePatterns() {
Rui Ueyama10416562016-08-04 02:03:27 +00001351 std::vector<StringRef> V;
Rui Ueyama83043f22016-10-17 16:01:53 +00001352 while (!Error && !consume(")"))
Rui Ueyama10416562016-08-04 02:03:27 +00001353 V.push_back(next());
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001354 return StringMatcher(V);
George Rimar0702c4e2016-07-29 15:32:46 +00001355}
1356
George Rimarbe394db2016-09-16 20:21:55 +00001357SortSectionPolicy ScriptParser::readSortKind() {
Rui Ueyama83043f22016-10-17 16:01:53 +00001358 if (consume("SORT") || consume("SORT_BY_NAME"))
George Rimarbe394db2016-09-16 20:21:55 +00001359 return SortSectionPolicy::Name;
Rui Ueyama83043f22016-10-17 16:01:53 +00001360 if (consume("SORT_BY_ALIGNMENT"))
George Rimarbe394db2016-09-16 20:21:55 +00001361 return SortSectionPolicy::Alignment;
Rui Ueyama83043f22016-10-17 16:01:53 +00001362 if (consume("SORT_BY_INIT_PRIORITY"))
George Rimarbe394db2016-09-16 20:21:55 +00001363 return SortSectionPolicy::Priority;
Rui Ueyama83043f22016-10-17 16:01:53 +00001364 if (consume("SORT_NONE"))
Rui Ueyamab2a0abd2016-09-16 21:14:55 +00001365 return SortSectionPolicy::None;
1366 return SortSectionPolicy::Default;
George Rimarbe394db2016-09-16 20:21:55 +00001367}
1368
George Rimar395281c2016-09-16 17:42:10 +00001369// Method reads a list of sequence of excluded files and section globs given in
1370// a following form: ((EXCLUDE_FILE(file_pattern+))? section_pattern+)+
1371// Example: *(.foo.1 EXCLUDE_FILE (*a.o) .foo.2 EXCLUDE_FILE (*b.o) .foo.3)
George Rimaraf03be12016-09-17 19:17:25 +00001372// The semantics of that is next:
1373// * Include .foo.1 from every file.
1374// * Include .foo.2 from every file but a.o
1375// * Include .foo.3 from every file but b.o
George Rimar07171f22016-09-21 15:56:44 +00001376std::vector<SectionPattern> ScriptParser::readInputSectionsList() {
1377 std::vector<SectionPattern> Ret;
George Rimar601e9892016-09-21 08:53:21 +00001378 while (!Error && peek() != ")") {
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001379 StringMatcher ExcludeFilePat;
Rui Ueyama83043f22016-10-17 16:01:53 +00001380 if (consume("EXCLUDE_FILE")) {
George Rimar395281c2016-09-16 17:42:10 +00001381 expect("(");
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001382 ExcludeFilePat = readFilePatterns();
George Rimar395281c2016-09-16 17:42:10 +00001383 }
1384
George Rimar601e9892016-09-21 08:53:21 +00001385 std::vector<StringRef> V;
1386 while (!Error && peek() != ")" && peek() != "EXCLUDE_FILE")
1387 V.push_back(next());
1388
1389 if (!V.empty())
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001390 Ret.push_back({std::move(ExcludeFilePat), StringMatcher(V)});
George Rimar601e9892016-09-21 08:53:21 +00001391 else
1392 setError("section pattern is expected");
George Rimar395281c2016-09-16 17:42:10 +00001393 }
George Rimar07171f22016-09-21 15:56:44 +00001394 return Ret;
George Rimar395281c2016-09-16 17:42:10 +00001395}
1396
Rui Ueyamaf8f6f1e2016-11-18 07:03:56 +00001397// Reads contents of "SECTIONS" directive. That directive contains a
1398// list of glob patterns for input sections. The grammar is as follows.
1399//
1400// <patterns> ::= <section-list>
1401// | <sort> "(" <section-list> ")"
1402// | <sort> "(" <sort> "(" <section-list> ")" ")"
1403//
1404// <sort> ::= "SORT" | "SORT_BY_NAME" | "SORT_BY_ALIGNMENT"
1405// | "SORT_BY_INIT_PRIORITY" | "SORT_NONE"
1406//
1407// <section-list> is parsed by readInputSectionsList().
George Rimara2496cb2016-08-30 09:46:59 +00001408InputSectionDescription *
1409ScriptParser::readInputSectionRules(StringRef FilePattern) {
George Rimarc91930a2016-09-02 21:17:20 +00001410 auto *Cmd = new InputSectionDescription(FilePattern);
Davide Italiano0ed42b02016-07-25 21:47:13 +00001411 expect("(");
Rui Ueyamaf373dd72016-11-24 01:43:21 +00001412 while (!Error && !consume(")")) {
George Rimar07171f22016-09-21 15:56:44 +00001413 SortSectionPolicy Outer = readSortKind();
1414 SortSectionPolicy Inner = SortSectionPolicy::Default;
1415 std::vector<SectionPattern> V;
1416 if (Outer != SortSectionPolicy::Default) {
George Rimar350ece42016-08-03 08:35:59 +00001417 expect("(");
George Rimar07171f22016-09-21 15:56:44 +00001418 Inner = readSortKind();
1419 if (Inner != SortSectionPolicy::Default) {
1420 expect("(");
1421 V = readInputSectionsList();
1422 expect(")");
1423 } else {
1424 V = readInputSectionsList();
1425 }
George Rimar350ece42016-08-03 08:35:59 +00001426 expect(")");
1427 } else {
George Rimar07171f22016-09-21 15:56:44 +00001428 V = readInputSectionsList();
George Rimar350ece42016-08-03 08:35:59 +00001429 }
George Rimar0702c4e2016-07-29 15:32:46 +00001430
George Rimar07171f22016-09-21 15:56:44 +00001431 for (SectionPattern &Pat : V) {
1432 Pat.SortInner = Inner;
1433 Pat.SortOuter = Outer;
1434 }
1435
1436 std::move(V.begin(), V.end(), std::back_inserter(Cmd->SectionPatterns));
1437 }
Rui Ueyama10416562016-08-04 02:03:27 +00001438 return Cmd;
Davide Italianoe7282792016-07-27 01:44:01 +00001439}
1440
George Rimara2496cb2016-08-30 09:46:59 +00001441InputSectionDescription *
1442ScriptParser::readInputSectionDescription(StringRef Tok) {
George Rimar06598002016-07-28 21:51:30 +00001443 // Input section wildcard can be surrounded by KEEP.
1444 // https://sourceware.org/binutils/docs/ld/Input-Section-Keep.html#Input-Section-Keep
George Rimara2496cb2016-08-30 09:46:59 +00001445 if (Tok == "KEEP") {
George Rimar06598002016-07-28 21:51:30 +00001446 expect("(");
George Rimara2496cb2016-08-30 09:46:59 +00001447 StringRef FilePattern = next();
1448 InputSectionDescription *Cmd = readInputSectionRules(FilePattern);
George Rimar06598002016-07-28 21:51:30 +00001449 expect(")");
Eugene Leviantcf43f172016-10-05 09:36:59 +00001450 Opt.KeptSections.push_back(Cmd);
Rui Ueyama10416562016-08-04 02:03:27 +00001451 return Cmd;
George Rimar06598002016-07-28 21:51:30 +00001452 }
George Rimara2496cb2016-08-30 09:46:59 +00001453 return readInputSectionRules(Tok);
Davide Italiano0ed42b02016-07-25 21:47:13 +00001454}
1455
George Rimar03fc0102016-07-28 07:18:23 +00001456void ScriptParser::readSort() {
1457 expect("(");
1458 expect("CONSTRUCTORS");
1459 expect(")");
1460}
1461
George Rimareefa7582016-08-04 09:29:31 +00001462Expr ScriptParser::readAssert() {
1463 expect("(");
1464 Expr E = readExpr();
1465 expect(",");
George Rimarcd574a52016-09-09 14:35:36 +00001466 StringRef Msg = unquote(next());
George Rimareefa7582016-08-04 09:29:31 +00001467 expect(")");
1468 return [=](uint64_t Dot) {
George Rimar60f1fe82017-02-21 07:33:38 +00001469 if (!E(Dot))
George Rimareefa7582016-08-04 09:29:31 +00001470 error(Msg);
George Rimar60f1fe82017-02-21 07:33:38 +00001471 return Dot;
George Rimareefa7582016-08-04 09:29:31 +00001472 };
1473}
1474
Rui Ueyama25150e82016-09-06 17:46:43 +00001475// Reads a FILL(expr) command. We handle the FILL command as an
1476// alias for =fillexp section attribute, which is different from
1477// what GNU linkers do.
1478// https://sourceware.org/binutils/docs/ld/Output-Section-Data.html
Rui Ueyama16068ae2016-11-19 18:05:56 +00001479uint32_t ScriptParser::readFill() {
George Rimarff1f29e2016-09-06 13:51:57 +00001480 expect("(");
Rui Ueyama16068ae2016-11-19 18:05:56 +00001481 uint32_t V = readOutputSectionFiller(next());
George Rimarff1f29e2016-09-06 13:51:57 +00001482 expect(")");
1483 expect(";");
1484 return V;
1485}
1486
Rui Ueyama10416562016-08-04 02:03:27 +00001487OutputSectionCommand *
1488ScriptParser::readOutputSectionDescription(StringRef OutSec) {
George Rimar076fe152016-07-21 06:43:01 +00001489 OutputSectionCommand *Cmd = new OutputSectionCommand(OutSec);
Eugene Leviant2a942c42016-12-05 16:38:32 +00001490 Cmd->Location = getCurrentLocation();
George Rimar58e5c4d2016-07-25 08:29:46 +00001491
1492 // Read an address expression.
1493 // https://sourceware.org/binutils/docs/ld/Output-Section-Address.html#Output-Section-Address
1494 if (peek() != ":")
1495 Cmd->AddrExpr = readExpr();
1496
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001497 expect(":");
Davide Italiano246f6812016-07-22 03:36:24 +00001498
Rui Ueyama83043f22016-10-17 16:01:53 +00001499 if (consume("AT"))
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001500 Cmd->LMAExpr = readParenExpr();
Rui Ueyama83043f22016-10-17 16:01:53 +00001501 if (consume("ALIGN"))
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001502 Cmd->AlignExpr = readParenExpr();
Rui Ueyama83043f22016-10-17 16:01:53 +00001503 if (consume("SUBALIGN"))
George Rimardb24d9c2016-08-19 15:18:23 +00001504 Cmd->SubalignExpr = readParenExpr();
George Rimar630c6172016-07-26 18:06:29 +00001505
Davide Italiano246f6812016-07-22 03:36:24 +00001506 // Parse constraints.
Rui Ueyama83043f22016-10-17 16:01:53 +00001507 if (consume("ONLY_IF_RO"))
Rui Ueyamaefc40662016-07-25 22:00:10 +00001508 Cmd->Constraint = ConstraintKind::ReadOnly;
Rui Ueyama83043f22016-10-17 16:01:53 +00001509 if (consume("ONLY_IF_RW"))
Rui Ueyamaefc40662016-07-25 22:00:10 +00001510 Cmd->Constraint = ConstraintKind::ReadWrite;
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001511 expect("{");
Rui Ueyama8ec77e62016-04-21 22:00:51 +00001512
Rui Ueyama83043f22016-10-17 16:01:53 +00001513 while (!Error && !consume("}")) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001514 StringRef Tok = next();
George Rimar2fe07922017-01-31 08:50:11 +00001515 if (Tok == ";") {
George Rimar69750752017-02-01 09:14:22 +00001516 // Empty commands are allowed. Do nothing here.
George Rimar2fe07922017-01-31 08:50:11 +00001517 } else if (SymbolAssignment *Assignment = readProvideOrAssignment(Tok)) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001518 Cmd->Commands.emplace_back(Assignment);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001519 } else if (BytesDataCommand *Data = readBytesDataCommand(Tok)) {
George Rimare38cbab2016-09-26 19:22:50 +00001520 Cmd->Commands.emplace_back(Data);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001521 } else if (Tok == "ASSERT") {
1522 Cmd->Commands.emplace_back(new AssertCommand(readAssert()));
1523 expect(";");
George Rimar8e2eca22017-01-23 09:36:19 +00001524 } else if (Tok == "CONSTRUCTORS") {
1525 // CONSTRUCTORS is a keyword to make the linker recognize C++ ctors/dtors
1526 // by name. This is for very old file formats such as ECOFF/XCOFF.
1527 // For ELF, we should ignore.
Meador Ingeb2d99d62016-11-22 18:01:50 +00001528 } else if (Tok == "FILL") {
George Rimarff1f29e2016-09-06 13:51:57 +00001529 Cmd->Filler = readFill();
Meador Ingeb2d99d62016-11-22 18:01:50 +00001530 } else if (Tok == "SORT") {
George Rimar03fc0102016-07-28 07:18:23 +00001531 readSort();
Meador Ingeb2d99d62016-11-22 18:01:50 +00001532 } else if (peek() == "(") {
George Rimara2496cb2016-08-30 09:46:59 +00001533 Cmd->Commands.emplace_back(readInputSectionDescription(Tok));
Meador Ingeb2d99d62016-11-22 18:01:50 +00001534 } else {
Eugene Leviantceabe802016-08-11 07:56:43 +00001535 setError("unknown command " + Tok);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001536 }
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001537 }
Meador Ingeb8897442017-01-24 02:34:00 +00001538
1539 if (consume(">"))
1540 Cmd->MemoryRegionName = next();
1541
George Rimar076fe152016-07-21 06:43:01 +00001542 Cmd->Phdrs = readOutputSectionPhdrs();
George Rimar4ebc5622016-09-23 13:29:20 +00001543
Rui Ueyama83043f22016-10-17 16:01:53 +00001544 if (consume("="))
George Rimar4ebc5622016-09-23 13:29:20 +00001545 Cmd->Filler = readOutputSectionFiller(next());
1546 else if (peek().startswith("="))
George Rimarff1f29e2016-09-06 13:51:57 +00001547 Cmd->Filler = readOutputSectionFiller(next().drop_front());
George Rimar4ebc5622016-09-23 13:29:20 +00001548
George Rimar7185a1a2017-01-17 15:32:12 +00001549 // Consume optional comma following output section command.
1550 consume(",");
1551
Rui Ueyama10416562016-08-04 02:03:27 +00001552 return Cmd;
Rui Ueyamaf71caa22016-07-29 06:14:07 +00001553}
Rui Ueyama8ec77e62016-04-21 22:00:51 +00001554
Rui Ueyama2c8f1f02016-08-29 22:01:21 +00001555// Read "=<number>" where <number> is an octal/decimal/hexadecimal number.
1556// https://sourceware.org/binutils/docs/ld/Output-Section-Fill.html
1557//
1558// ld.gold is not fully compatible with ld.bfd. ld.bfd handles
1559// hexstrings as blobs of arbitrary sizes, while ld.gold handles them
1560// as 32-bit big-endian values. We will do the same as ld.gold does
1561// because it's simpler than what ld.bfd does.
Rui Ueyama16068ae2016-11-19 18:05:56 +00001562uint32_t ScriptParser::readOutputSectionFiller(StringRef Tok) {
Rui Ueyama965827d2016-08-03 23:25:15 +00001563 uint32_t V;
Rui Ueyama16068ae2016-11-19 18:05:56 +00001564 if (!Tok.getAsInteger(0, V))
1565 return V;
1566 setError("invalid filler expression: " + Tok);
1567 return 0;
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001568}
1569
Petr Hoseka35e39c2016-08-16 01:11:16 +00001570SymbolAssignment *ScriptParser::readProvideHidden(bool Provide, bool Hidden) {
Eugene Levianta31c91b2016-07-22 07:38:40 +00001571 expect("(");
Rui Ueyama174e0a12016-07-29 00:29:25 +00001572 SymbolAssignment *Cmd = readAssignment(next());
Petr Hoseka35e39c2016-08-16 01:11:16 +00001573 Cmd->Provide = Provide;
Rui Ueyama174e0a12016-07-29 00:29:25 +00001574 Cmd->Hidden = Hidden;
Eugene Levianta31c91b2016-07-22 07:38:40 +00001575 expect(")");
1576 expect(";");
Rui Ueyama10416562016-08-04 02:03:27 +00001577 return Cmd;
Eugene Levianteda81a12016-07-12 06:39:48 +00001578}
1579
Rafael Espindolac96da112016-11-01 11:30:45 +00001580SymbolAssignment *ScriptParser::readProvideOrAssignment(StringRef Tok) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001581 SymbolAssignment *Cmd = nullptr;
1582 if (peek() == "=" || peek() == "+=") {
1583 Cmd = readAssignment(Tok);
1584 expect(";");
1585 } else if (Tok == "PROVIDE") {
Petr Hoseka35e39c2016-08-16 01:11:16 +00001586 Cmd = readProvideHidden(true, false);
1587 } else if (Tok == "HIDDEN") {
1588 Cmd = readProvideHidden(false, true);
Eugene Leviantceabe802016-08-11 07:56:43 +00001589 } else if (Tok == "PROVIDE_HIDDEN") {
Petr Hoseka35e39c2016-08-16 01:11:16 +00001590 Cmd = readProvideHidden(true, true);
Eugene Leviantceabe802016-08-11 07:56:43 +00001591 }
1592 return Cmd;
1593}
1594
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001595static uint64_t getSymbolValue(const Twine &Loc, StringRef S, uint64_t Dot) {
George Rimar30835ea2016-07-28 21:08:56 +00001596 if (S == ".")
1597 return Dot;
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001598 return ScriptBase->getSymbolValue(Loc, S);
George Rimare32a3592016-08-10 07:59:34 +00001599}
1600
Rafael Espindola2f831dc2016-10-31 19:56:37 +00001601static bool isAbsolute(StringRef S) {
1602 if (S == ".")
1603 return false;
1604 return ScriptBase->isAbsolute(S);
1605}
1606
George Rimar30835ea2016-07-28 21:08:56 +00001607SymbolAssignment *ScriptParser::readAssignment(StringRef Name) {
1608 StringRef Op = next();
Eugene Leviantdb741e72016-09-07 07:08:43 +00001609 Expr E;
George Rimar30835ea2016-07-28 21:08:56 +00001610 assert(Op == "=" || Op == "+=");
Rui Ueyama83043f22016-10-17 16:01:53 +00001611 if (consume("ABSOLUTE")) {
Rui Ueyama731a66a2017-02-15 19:58:17 +00001612 E = readExpr();
Rui Ueyama009d1742016-11-18 06:49:09 +00001613 E.IsAbsolute = [] { return true; };
Eugene Leviantdb741e72016-09-07 07:08:43 +00001614 } else {
1615 E = readExpr();
1616 }
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001617 if (Op == "+=") {
1618 std::string Loc = getCurrentLocation();
1619 E = [=](uint64_t Dot) {
1620 return getSymbolValue(Loc, Name, Dot) + E(Dot);
1621 };
1622 }
George Rimar2ee2d2d2017-02-21 14:50:38 +00001623 return new SymbolAssignment(Name, E, getCurrentLocation());
George Rimar30835ea2016-07-28 21:08:56 +00001624}
1625
1626// This is an operator-precedence parser to parse a linker
1627// script expression.
Rui Ueyama731a66a2017-02-15 19:58:17 +00001628Expr ScriptParser::readExpr() {
1629 // Our lexer is context-aware. Set the in-expression bit so that
1630 // they apply different tokenization rules.
1631 bool Orig = InExpr;
1632 InExpr = true;
1633 Expr E = readExpr1(readPrimary(), 0);
1634 InExpr = Orig;
1635 return E;
1636}
George Rimar30835ea2016-07-28 21:08:56 +00001637
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001638static Expr combine(StringRef Op, Expr L, Expr R) {
George Rimarcc4d3e52017-02-01 09:01:16 +00001639 auto IsAbs = [=] { return L.IsAbsolute() && R.IsAbsolute(); };
1640 auto GetOutSec = [=] {
Rafael Espindola24e6f362017-02-24 15:07:30 +00001641 const OutputSection *S = L.Section();
George Rimarcc4d3e52017-02-01 09:01:16 +00001642 return S ? S : R.Section();
1643 };
1644
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001645 if (Op == "*")
1646 return [=](uint64_t Dot) { return L(Dot) * R(Dot); };
1647 if (Op == "/") {
1648 return [=](uint64_t Dot) -> uint64_t {
1649 uint64_t RHS = R(Dot);
1650 if (RHS == 0) {
1651 error("division by zero");
1652 return 0;
1653 }
1654 return L(Dot) / RHS;
1655 };
1656 }
1657 if (Op == "+")
George Rimarcc4d3e52017-02-01 09:01:16 +00001658 return {[=](uint64_t Dot) { return L(Dot) + R(Dot); }, IsAbs, GetOutSec};
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001659 if (Op == "-")
George Rimarcc4d3e52017-02-01 09:01:16 +00001660 return {[=](uint64_t Dot) { return L(Dot) - R(Dot); }, IsAbs, GetOutSec};
George Rimarc8ccd1f2016-09-23 13:13:55 +00001661 if (Op == "<<")
1662 return [=](uint64_t Dot) { return L(Dot) << R(Dot); };
1663 if (Op == ">>")
1664 return [=](uint64_t Dot) { return L(Dot) >> R(Dot); };
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001665 if (Op == "<")
1666 return [=](uint64_t Dot) { return L(Dot) < R(Dot); };
1667 if (Op == ">")
1668 return [=](uint64_t Dot) { return L(Dot) > R(Dot); };
1669 if (Op == ">=")
1670 return [=](uint64_t Dot) { return L(Dot) >= R(Dot); };
1671 if (Op == "<=")
1672 return [=](uint64_t Dot) { return L(Dot) <= R(Dot); };
1673 if (Op == "==")
1674 return [=](uint64_t Dot) { return L(Dot) == R(Dot); };
1675 if (Op == "!=")
1676 return [=](uint64_t Dot) { return L(Dot) != R(Dot); };
1677 if (Op == "&")
1678 return [=](uint64_t Dot) { return L(Dot) & R(Dot); };
Rafael Espindolacc3dd622016-08-22 21:33:35 +00001679 if (Op == "|")
1680 return [=](uint64_t Dot) { return L(Dot) | R(Dot); };
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001681 llvm_unreachable("invalid operator");
1682}
1683
Rui Ueyama708019c2016-07-24 18:19:40 +00001684// This is a part of the operator-precedence parser. This function
1685// assumes that the remaining token stream starts with an operator.
1686Expr ScriptParser::readExpr1(Expr Lhs, int MinPrec) {
1687 while (!atEOF() && !Error) {
1688 // Read an operator and an expression.
Rui Ueyama46247b82016-11-18 06:49:07 +00001689 if (consume("?"))
Rui Ueyama708019c2016-07-24 18:19:40 +00001690 return readTernary(Lhs);
Rui Ueyama46247b82016-11-18 06:49:07 +00001691 StringRef Op1 = peek();
Rui Ueyama708019c2016-07-24 18:19:40 +00001692 if (precedence(Op1) < MinPrec)
Eugene Levianteda81a12016-07-12 06:39:48 +00001693 break;
Justin Bogner5424e7c2016-10-17 06:21:13 +00001694 skip();
Rui Ueyama708019c2016-07-24 18:19:40 +00001695 Expr Rhs = readPrimary();
1696
1697 // Evaluate the remaining part of the expression first if the
1698 // next operator has greater precedence than the previous one.
1699 // For example, if we have read "+" and "3", and if the next
1700 // operator is "*", then we'll evaluate 3 * ... part first.
1701 while (!atEOF()) {
1702 StringRef Op2 = peek();
1703 if (precedence(Op2) <= precedence(Op1))
1704 break;
1705 Rhs = readExpr1(Rhs, precedence(Op2));
1706 }
1707
1708 Lhs = combine(Op1, Lhs, Rhs);
Eugene Levianteda81a12016-07-12 06:39:48 +00001709 }
Rui Ueyama708019c2016-07-24 18:19:40 +00001710 return Lhs;
1711}
1712
1713uint64_t static getConstant(StringRef S) {
Michael J. Spencere2cc07b2016-08-17 02:10:51 +00001714 if (S == "COMMONPAGESIZE")
Rui Ueyama708019c2016-07-24 18:19:40 +00001715 return Target->PageSize;
Michael J. Spencere2cc07b2016-08-17 02:10:51 +00001716 if (S == "MAXPAGESIZE")
Petr Hosek997f8832016-09-28 15:20:47 +00001717 return Config->MaxPageSize;
Rui Ueyama708019c2016-07-24 18:19:40 +00001718 error("unknown constant: " + S);
1719 return 0;
1720}
1721
Rui Ueyama626e0b02016-09-02 18:19:00 +00001722// Parses Tok as an integer. Returns true if successful.
1723// It recognizes hexadecimal (prefixed with "0x" or suffixed with "H")
1724// and decimal numbers. Decimal numbers may have "K" (kilo) or
1725// "M" (mega) prefixes.
George Rimar9f2f7ad2016-09-02 16:01:42 +00001726static bool readInteger(StringRef Tok, uint64_t &Result) {
Rui Ueyama46247b82016-11-18 06:49:07 +00001727 // Negative number
Simon Atanasyaneaeafb22016-09-02 21:54:35 +00001728 if (Tok.startswith("-")) {
1729 if (!readInteger(Tok.substr(1), Result))
1730 return false;
1731 Result = -Result;
1732 return true;
1733 }
Rui Ueyama46247b82016-11-18 06:49:07 +00001734
1735 // Hexadecimal
George Rimar9f2f7ad2016-09-02 16:01:42 +00001736 if (Tok.startswith_lower("0x"))
1737 return !Tok.substr(2).getAsInteger(16, Result);
1738 if (Tok.endswith_lower("H"))
1739 return !Tok.drop_back().getAsInteger(16, Result);
1740
Rui Ueyama46247b82016-11-18 06:49:07 +00001741 // Decimal
George Rimar9f2f7ad2016-09-02 16:01:42 +00001742 int Suffix = 1;
1743 if (Tok.endswith_lower("K")) {
1744 Suffix = 1024;
1745 Tok = Tok.drop_back();
1746 } else if (Tok.endswith_lower("M")) {
1747 Suffix = 1024 * 1024;
1748 Tok = Tok.drop_back();
1749 }
1750 if (Tok.getAsInteger(10, Result))
1751 return false;
1752 Result *= Suffix;
1753 return true;
1754}
1755
George Rimare38cbab2016-09-26 19:22:50 +00001756BytesDataCommand *ScriptParser::readBytesDataCommand(StringRef Tok) {
1757 int Size = StringSwitch<unsigned>(Tok)
1758 .Case("BYTE", 1)
1759 .Case("SHORT", 2)
1760 .Case("LONG", 4)
1761 .Case("QUAD", 8)
1762 .Default(-1);
1763 if (Size == -1)
1764 return nullptr;
1765
Meador Inge95c7d8d2016-12-08 23:21:30 +00001766 return new BytesDataCommand(readParenExpr(), Size);
George Rimare38cbab2016-09-26 19:22:50 +00001767}
1768
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001769StringRef ScriptParser::readParenLiteral() {
1770 expect("(");
1771 StringRef Tok = next();
1772 expect(")");
1773 return Tok;
1774}
1775
Rui Ueyama708019c2016-07-24 18:19:40 +00001776Expr ScriptParser::readPrimary() {
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001777 if (peek() == "(")
1778 return readParenExpr();
Rui Ueyama708019c2016-07-24 18:19:40 +00001779
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001780 StringRef Tok = next();
Rui Ueyamab5f1c3e2016-12-01 04:36:49 +00001781 std::string Location = getCurrentLocation();
Rui Ueyama708019c2016-07-24 18:19:40 +00001782
Simon Atanasyaneaeafb22016-09-02 21:54:35 +00001783 if (Tok == "~") {
1784 Expr E = readPrimary();
1785 return [=](uint64_t Dot) { return ~E(Dot); };
1786 }
1787 if (Tok == "-") {
1788 Expr E = readPrimary();
1789 return [=](uint64_t Dot) { return -E(Dot); };
1790 }
1791
Rui Ueyama708019c2016-07-24 18:19:40 +00001792 // Built-in functions are parsed here.
1793 // https://sourceware.org/binutils/docs/ld/Builtin-Functions.html.
George Rimar96659df2016-08-30 09:54:01 +00001794 if (Tok == "ADDR") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001795 StringRef Name = readParenLiteral();
Eugene Levianted30ce72016-11-28 09:58:04 +00001796 return {[=](uint64_t Dot) {
1797 return ScriptBase->getOutputSection(Location, Name)->Addr;
1798 },
1799 [=] { return false; },
1800 [=] { return ScriptBase->getOutputSection(Location, Name); }};
George Rimar96659df2016-08-30 09:54:01 +00001801 }
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001802 if (Tok == "LOADADDR") {
1803 StringRef Name = readParenLiteral();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001804 return [=](uint64_t Dot) {
Eugene Levianted30ce72016-11-28 09:58:04 +00001805 return ScriptBase->getOutputSection(Location, Name)->getLMA();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001806 };
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001807 }
George Rimareefa7582016-08-04 09:29:31 +00001808 if (Tok == "ASSERT")
1809 return readAssert();
Rui Ueyama708019c2016-07-24 18:19:40 +00001810 if (Tok == "ALIGN") {
Rui Ueyama5d804dc2016-12-16 18:19:35 +00001811 expect("(");
1812 Expr E = readExpr();
1813 if (consume(",")) {
1814 Expr E2 = readExpr();
1815 expect(")");
1816 return [=](uint64_t Dot) { return alignTo(E(Dot), E2(Dot)); };
1817 }
1818 expect(")");
Rui Ueyama708019c2016-07-24 18:19:40 +00001819 return [=](uint64_t Dot) { return alignTo(Dot, E(Dot)); };
1820 }
1821 if (Tok == "CONSTANT") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001822 StringRef Name = readParenLiteral();
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001823 return [=](uint64_t Dot) { return getConstant(Name); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001824 }
George Rimarf34f45f2016-09-23 13:17:23 +00001825 if (Tok == "DEFINED") {
Rui Ueyama0ee25a62016-11-17 03:52:14 +00001826 StringRef Name = readParenLiteral();
1827 return [=](uint64_t Dot) { return ScriptBase->isDefined(Name) ? 1 : 0; };
George Rimarf34f45f2016-09-23 13:17:23 +00001828 }
Rafael Espindola54c145c2016-07-28 18:16:24 +00001829 if (Tok == "SEGMENT_START") {
1830 expect("(");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001831 skip();
Rafael Espindola54c145c2016-07-28 18:16:24 +00001832 expect(",");
George Rimar8c658bf2016-09-17 18:14:56 +00001833 Expr E = readExpr();
Rafael Espindola54c145c2016-07-28 18:16:24 +00001834 expect(")");
George Rimar8c658bf2016-09-17 18:14:56 +00001835 return [=](uint64_t Dot) { return E(Dot); };
Rafael Espindola54c145c2016-07-28 18:16:24 +00001836 }
Rui Ueyama708019c2016-07-24 18:19:40 +00001837 if (Tok == "DATA_SEGMENT_ALIGN") {
1838 expect("(");
1839 Expr E = readExpr();
1840 expect(",");
1841 readExpr();
1842 expect(")");
Rui Ueyamaf7791bb2016-07-26 19:34:10 +00001843 return [=](uint64_t Dot) { return alignTo(Dot, E(Dot)); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001844 }
1845 if (Tok == "DATA_SEGMENT_END") {
1846 expect("(");
1847 expect(".");
1848 expect(")");
1849 return [](uint64_t Dot) { return Dot; };
1850 }
George Rimar276b4e62016-07-26 17:58:44 +00001851 // GNU linkers implements more complicated logic to handle
1852 // DATA_SEGMENT_RELRO_END. We instead ignore the arguments and just align to
1853 // the next page boundary for simplicity.
1854 if (Tok == "DATA_SEGMENT_RELRO_END") {
1855 expect("(");
Rafael Espindola97bdc722016-09-14 19:14:01 +00001856 readExpr();
George Rimar276b4e62016-07-26 17:58:44 +00001857 expect(",");
1858 readExpr();
1859 expect(")");
1860 return [](uint64_t Dot) { return alignTo(Dot, Target->PageSize); };
1861 }
George Rimar9e694502016-07-29 16:18:47 +00001862 if (Tok == "SIZEOF") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001863 StringRef Name = readParenLiteral();
Rui Ueyamaedf75e72016-11-17 20:27:10 +00001864 return [=](uint64_t Dot) { return ScriptBase->getOutputSectionSize(Name); };
George Rimar9e694502016-07-29 16:18:47 +00001865 }
Eugene Leviant36fac7f2016-09-08 09:08:30 +00001866 if (Tok == "ALIGNOF") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001867 StringRef Name = readParenLiteral();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001868 return [=](uint64_t Dot) {
Eugene Levianted30ce72016-11-28 09:58:04 +00001869 return ScriptBase->getOutputSection(Location, Name)->Addralign;
Eugene Leviantafaa9342016-11-16 09:49:39 +00001870 };
Eugene Leviant36fac7f2016-09-08 09:08:30 +00001871 }
George Rimare32a3592016-08-10 07:59:34 +00001872 if (Tok == "SIZEOF_HEADERS")
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001873 return [=](uint64_t Dot) { return ScriptBase->getHeaderSize(); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001874
George Rimar9f2f7ad2016-09-02 16:01:42 +00001875 // Tok is a literal number.
1876 uint64_t V;
1877 if (readInteger(Tok, V))
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001878 return [=](uint64_t Dot) { return V; };
George Rimar9f2f7ad2016-09-02 16:01:42 +00001879
1880 // Tok is a symbol name.
1881 if (Tok != "." && !isValidCIdentifier(Tok))
1882 setError("malformed number: " + Tok);
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001883 return {[=](uint64_t Dot) { return getSymbolValue(Location, Tok, Dot); },
Rui Ueyama009d1742016-11-18 06:49:09 +00001884 [=] { return isAbsolute(Tok); },
1885 [=] { return ScriptBase->getSymbolSection(Tok); }};
Rui Ueyama708019c2016-07-24 18:19:40 +00001886}
1887
1888Expr ScriptParser::readTernary(Expr Cond) {
Rui Ueyama708019c2016-07-24 18:19:40 +00001889 Expr L = readExpr();
1890 expect(":");
1891 Expr R = readExpr();
1892 return [=](uint64_t Dot) { return Cond(Dot) ? L(Dot) : R(Dot); };
1893}
1894
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001895Expr ScriptParser::readParenExpr() {
1896 expect("(");
1897 Expr E = readExpr();
1898 expect(")");
1899 return E;
1900}
1901
Eugene Leviantbbe38602016-07-19 09:25:43 +00001902std::vector<StringRef> ScriptParser::readOutputSectionPhdrs() {
1903 std::vector<StringRef> Phdrs;
1904 while (!Error && peek().startswith(":")) {
1905 StringRef Tok = next();
George Rimarda841c12016-11-14 10:03:54 +00001906 Phdrs.push_back((Tok.size() == 1) ? next() : Tok.substr(1));
Eugene Leviantbbe38602016-07-19 09:25:43 +00001907 }
1908 return Phdrs;
1909}
1910
George Rimar95dd7182016-10-18 10:49:50 +00001911// Read a program header type name. The next token must be a
1912// name of a program header type or a constant (e.g. "0x3").
Eugene Leviantbbe38602016-07-19 09:25:43 +00001913unsigned ScriptParser::readPhdrType() {
Eugene Leviantbbe38602016-07-19 09:25:43 +00001914 StringRef Tok = next();
George Rimar95dd7182016-10-18 10:49:50 +00001915 uint64_t Val;
1916 if (readInteger(Tok, Val))
1917 return Val;
1918
Rui Ueyamab0f6c592016-07-20 19:36:38 +00001919 unsigned Ret = StringSwitch<unsigned>(Tok)
George Rimar6c55f0e2016-09-08 08:20:30 +00001920 .Case("PT_NULL", PT_NULL)
1921 .Case("PT_LOAD", PT_LOAD)
1922 .Case("PT_DYNAMIC", PT_DYNAMIC)
1923 .Case("PT_INTERP", PT_INTERP)
1924 .Case("PT_NOTE", PT_NOTE)
1925 .Case("PT_SHLIB", PT_SHLIB)
1926 .Case("PT_PHDR", PT_PHDR)
1927 .Case("PT_TLS", PT_TLS)
1928 .Case("PT_GNU_EH_FRAME", PT_GNU_EH_FRAME)
1929 .Case("PT_GNU_STACK", PT_GNU_STACK)
1930 .Case("PT_GNU_RELRO", PT_GNU_RELRO)
George Rimar270173f2016-10-14 13:02:22 +00001931 .Case("PT_OPENBSD_RANDOMIZE", PT_OPENBSD_RANDOMIZE)
George Rimarcc6e5672016-10-14 10:34:36 +00001932 .Case("PT_OPENBSD_WXNEEDED", PT_OPENBSD_WXNEEDED)
George Rimara2a32c22016-12-06 17:57:42 +00001933 .Case("PT_OPENBSD_BOOTDATA", PT_OPENBSD_BOOTDATA)
George Rimar6c55f0e2016-09-08 08:20:30 +00001934 .Default(-1);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001935
Rui Ueyamab0f6c592016-07-20 19:36:38 +00001936 if (Ret == (unsigned)-1) {
1937 setError("invalid program header type: " + Tok);
1938 return PT_NULL;
1939 }
1940 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +00001941}
1942
Rui Ueyama12450b22016-11-18 06:30:09 +00001943// Reads a list of symbols, e.g. "{ global: foo; bar; local: *; };".
1944void ScriptParser::readAnonymousDeclaration() {
1945 // Read global symbols first. "global:" is default, so if there's
1946 // no label, we assume global symbols.
Rafael Espindola45242682017-02-03 13:24:01 +00001947 if (peek() != "local") {
1948 if (consume("global"))
1949 expect(":");
Peter Collingbourne904c5ed2017-02-13 18:31:12 +00001950 for (SymbolVersion V : readSymbols())
1951 Config->VersionScriptGlobals.push_back(V);
Rafael Espindola45242682017-02-03 13:24:01 +00001952 }
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001953 readLocals();
Rui Ueyama12450b22016-11-18 06:30:09 +00001954 expect("}");
1955 expect(";");
1956}
1957
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001958void ScriptParser::readLocals() {
Rafael Espindola45242682017-02-03 13:24:01 +00001959 if (!consume("local"))
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001960 return;
Rafael Espindola45242682017-02-03 13:24:01 +00001961 expect(":");
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001962 std::vector<SymbolVersion> Locals = readSymbols();
1963 for (SymbolVersion V : Locals) {
1964 if (V.Name == "*") {
1965 Config->DefaultSymbolVersion = VER_NDX_LOCAL;
1966 continue;
1967 }
1968 Config->VersionScriptLocals.push_back(V);
1969 }
1970}
1971
Rui Ueyama12450b22016-11-18 06:30:09 +00001972// Reads a list of symbols, e.g. "VerStr { global: foo; bar; local: *; };".
Rui Ueyama95769b42016-08-31 20:03:54 +00001973void ScriptParser::readVersionDeclaration(StringRef VerStr) {
George Rimar20b65982016-08-31 09:08:26 +00001974 // Identifiers start at 2 because 0 and 1 are reserved
1975 // for VER_NDX_LOCAL and VER_NDX_GLOBAL constants.
Rui Ueyamada805c42016-11-17 03:39:21 +00001976 uint16_t VersionId = Config->VersionDefinitions.size() + 2;
George Rimar20b65982016-08-31 09:08:26 +00001977 Config->VersionDefinitions.push_back({VerStr, VersionId});
1978
Rui Ueyama12450b22016-11-18 06:30:09 +00001979 // Read global symbols.
Rafael Espindola45242682017-02-03 13:24:01 +00001980 if (peek() != "local") {
1981 if (consume("global"))
1982 expect(":");
Rui Ueyama12450b22016-11-18 06:30:09 +00001983 Config->VersionDefinitions.back().Globals = readSymbols();
Rafael Espindola45242682017-02-03 13:24:01 +00001984 }
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001985 readLocals();
George Rimar20b65982016-08-31 09:08:26 +00001986 expect("}");
1987
Rui Ueyama12450b22016-11-18 06:30:09 +00001988 // Each version may have a parent version. For example, "Ver2"
1989 // defined as "Ver2 { global: foo; local: *; } Ver1;" has "Ver1"
1990 // as a parent. This version hierarchy is, probably against your
1991 // instinct, purely for hint; the runtime doesn't care about it
1992 // at all. In LLD, we simply ignore it.
1993 if (peek() != ";")
Justin Bogner5424e7c2016-10-17 06:21:13 +00001994 skip();
George Rimar20b65982016-08-31 09:08:26 +00001995 expect(";");
1996}
1997
Rui Ueyama12450b22016-11-18 06:30:09 +00001998// Reads a list of symbols for a versions cript.
1999std::vector<SymbolVersion> ScriptParser::readSymbols() {
2000 std::vector<SymbolVersion> Ret;
George Rimare0fc2422016-11-16 17:59:10 +00002001 for (;;) {
Rafael Espindola1ef90d22016-12-09 16:44:05 +00002002 if (consume("extern")) {
Rui Ueyama12450b22016-11-18 06:30:09 +00002003 for (SymbolVersion V : readVersionExtern())
2004 Ret.push_back(V);
Rafael Espindola1ef90d22016-12-09 16:44:05 +00002005 continue;
2006 }
George Rimare0fc2422016-11-16 17:59:10 +00002007
Dmitry Mikulinf3965c02017-02-07 19:50:47 +00002008 if (peek() == "}" || (peek() == "local" && peek(1) == ":") || Error)
Rui Ueyama12450b22016-11-18 06:30:09 +00002009 break;
Rui Ueyama0ee25a62016-11-17 03:52:14 +00002010 StringRef Tok = next();
Rui Ueyama12450b22016-11-18 06:30:09 +00002011 Ret.push_back({unquote(Tok), false, hasWildcard(Tok)});
George Rimare0fc2422016-11-16 17:59:10 +00002012 expect(";");
2013 }
Rui Ueyama12450b22016-11-18 06:30:09 +00002014 return Ret;
George Rimare0fc2422016-11-16 17:59:10 +00002015}
2016
Rui Ueyama12450b22016-11-18 06:30:09 +00002017// Reads an "extern C++" directive, e.g.,
2018// "extern "C++" { ns::*; "f(int, double)"; };"
2019std::vector<SymbolVersion> ScriptParser::readVersionExtern() {
Rafael Espindola7e714152016-12-08 17:26:53 +00002020 StringRef Tok = next();
2021 bool IsCXX = Tok == "\"C++\"";
2022 if (!IsCXX && Tok != "\"C\"")
Rafael Espindolad0ebd842016-12-08 17:54:26 +00002023 setError("Unknown language");
George Rimar20b65982016-08-31 09:08:26 +00002024 expect("{");
2025
Rui Ueyama12450b22016-11-18 06:30:09 +00002026 std::vector<SymbolVersion> Ret;
Rui Ueyama0ee25a62016-11-17 03:52:14 +00002027 while (!Error && peek() != "}") {
2028 StringRef Tok = next();
2029 bool HasWildcard = !Tok.startswith("\"") && hasWildcard(Tok);
Rafael Espindola7e714152016-12-08 17:26:53 +00002030 Ret.push_back({unquote(Tok), IsCXX, HasWildcard});
George Rimar20b65982016-08-31 09:08:26 +00002031 expect(";");
2032 }
2033
2034 expect("}");
2035 expect(";");
Rui Ueyama12450b22016-11-18 06:30:09 +00002036 return Ret;
George Rimar20b65982016-08-31 09:08:26 +00002037}
2038
Rui Ueyama24e626c2017-01-26 02:58:19 +00002039uint64_t ScriptParser::readMemoryAssignment(
2040 StringRef S1, StringRef S2, StringRef S3) {
2041 if (!(consume(S1) || consume(S2) || consume(S3))) {
2042 setError("expected one of: " + S1 + ", " + S2 + ", or " + S3);
2043 return 0;
2044 }
2045 expect("=");
2046
2047 // TODO: Fully support constant expressions.
2048 uint64_t Val;
2049 if (!readInteger(next(), Val))
2050 setError("nonconstant expression for "+ S1);
2051 return Val;
2052}
2053
2054// Parse the MEMORY command as specified in:
2055// https://sourceware.org/binutils/docs/ld/MEMORY.html
2056//
2057// MEMORY { name [(attr)] : ORIGIN = origin, LENGTH = len ... }
Meador Ingeb8897442017-01-24 02:34:00 +00002058void ScriptParser::readMemory() {
2059 expect("{");
2060 while (!Error && !consume("}")) {
2061 StringRef Name = next();
Rui Ueyama24e626c2017-01-26 02:58:19 +00002062
Meador Ingeb8897442017-01-24 02:34:00 +00002063 uint32_t Flags = 0;
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002064 uint32_t NegFlags = 0;
Meador Ingeb8897442017-01-24 02:34:00 +00002065 if (consume("(")) {
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002066 std::tie(Flags, NegFlags) = readMemoryAttributes();
Meador Ingeb8897442017-01-24 02:34:00 +00002067 expect(")");
2068 }
2069 expect(":");
2070
Rui Ueyama24e626c2017-01-26 02:58:19 +00002071 uint64_t Origin = readMemoryAssignment("ORIGIN", "org", "o");
Meador Ingeb8897442017-01-24 02:34:00 +00002072 expect(",");
Rui Ueyama24e626c2017-01-26 02:58:19 +00002073 uint64_t Length = readMemoryAssignment("LENGTH", "len", "l");
Meador Ingeb8897442017-01-24 02:34:00 +00002074
Meador Ingeb8897442017-01-24 02:34:00 +00002075 // Add the memory region to the region map (if it doesn't already exist).
2076 auto It = Opt.MemoryRegions.find(Name);
2077 if (It != Opt.MemoryRegions.end())
2078 setError("region '" + Name + "' already defined");
2079 else
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002080 Opt.MemoryRegions[Name] = {Name, Origin, Length, Origin, Flags, NegFlags};
Meador Ingeb8897442017-01-24 02:34:00 +00002081 }
2082}
2083
2084// This function parses the attributes used to match against section
2085// flags when placing output sections in a memory region. These flags
2086// are only used when an explicit memory region name is not used.
2087std::pair<uint32_t, uint32_t> ScriptParser::readMemoryAttributes() {
2088 uint32_t Flags = 0;
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002089 uint32_t NegFlags = 0;
Meador Ingeb8897442017-01-24 02:34:00 +00002090 bool Invert = false;
Rui Ueyama481ac992017-01-26 02:58:39 +00002091
2092 for (char C : next().lower()) {
Meador Ingeb8897442017-01-24 02:34:00 +00002093 uint32_t Flag = 0;
2094 if (C == '!')
2095 Invert = !Invert;
Rui Ueyama481ac992017-01-26 02:58:39 +00002096 else if (C == 'w')
Meador Ingeb8897442017-01-24 02:34:00 +00002097 Flag = SHF_WRITE;
Rui Ueyama481ac992017-01-26 02:58:39 +00002098 else if (C == 'x')
Meador Ingeb8897442017-01-24 02:34:00 +00002099 Flag = SHF_EXECINSTR;
Rui Ueyama481ac992017-01-26 02:58:39 +00002100 else if (C == 'a')
Meador Ingeb8897442017-01-24 02:34:00 +00002101 Flag = SHF_ALLOC;
Rui Ueyama481ac992017-01-26 02:58:39 +00002102 else if (C != 'r')
Meador Ingeb8897442017-01-24 02:34:00 +00002103 setError("invalid memory region attribute");
Rui Ueyama481ac992017-01-26 02:58:39 +00002104
Meador Ingeb8897442017-01-24 02:34:00 +00002105 if (Invert)
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002106 NegFlags |= Flag;
Meador Ingeb8897442017-01-24 02:34:00 +00002107 else
2108 Flags |= Flag;
2109 }
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002110 return {Flags, NegFlags};
Meador Ingeb8897442017-01-24 02:34:00 +00002111}
2112
Rui Ueyama07320e42016-04-20 20:13:41 +00002113void elf::readLinkerScript(MemoryBufferRef MB) {
Rui Ueyama22375f22016-11-25 18:51:54 +00002114 ScriptParser(MB).readLinkerScript();
George Rimar20b65982016-08-31 09:08:26 +00002115}
2116
2117void elf::readVersionScript(MemoryBufferRef MB) {
Rui Ueyama22375f22016-11-25 18:51:54 +00002118 ScriptParser(MB).readVersionScript();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00002119}
Rui Ueyama1ebc8ed2016-02-12 21:47:28 +00002120
Rafael Espindolad0ebd842016-12-08 17:54:26 +00002121void elf::readDynamicList(MemoryBufferRef MB) {
2122 ScriptParser(MB).readDynamicList();
2123}
2124
Rui Ueyama07320e42016-04-20 20:13:41 +00002125template class elf::LinkerScript<ELF32LE>;
2126template class elf::LinkerScript<ELF32BE>;
2127template class elf::LinkerScript<ELF64LE>;
2128template class elf::LinkerScript<ELF64BE>;