blob: 7c9328d845a6c2455387034e57d32602bf175efc [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;
Meador Inge8f1f3c42017-01-09 18:36:57 +000066 replaceBody<DefinedRegular<ELFT>>(Sym, Cmd->Name, /*IsLocal=*/false,
67 Visibility, STT_NOTYPE, 0, 0, nullptr,
68 nullptr);
69 return Sym->body();
Eugene Leviantceabe802016-08-11 07:56:43 +000070}
71
Meador Inge8f1f3c42017-01-09 18:36:57 +000072template <class ELFT> static SymbolBody *addSynthetic(SymbolAssignment *Cmd) {
Petr Hosek5e51f7d2017-02-21 22:32:51 +000073 Symbol *Sym;
Meador Inge8f1f3c42017-01-09 18:36:57 +000074 uint8_t Visibility = Cmd->Hidden ? STV_HIDDEN : STV_DEFAULT;
Eugene Leviantafaa9342016-11-16 09:49:39 +000075 const OutputSectionBase *Sec =
76 ScriptConfig->HasSections ? nullptr : Cmd->Expression.Section();
Petr Hosek5e51f7d2017-02-21 22:32:51 +000077 std::tie(Sym, std::ignore) = Symtab<ELFT>::X->insert(
78 Cmd->Name, /*Type*/ 0, Visibility, /*CanOmitFromDynSym*/ false,
79 /*File*/ nullptr);
80 Sym->Binding = STB_GLOBAL;
Meador Inge8f1f3c42017-01-09 18:36:57 +000081 replaceBody<DefinedSynthetic>(Sym, Cmd->Name, 0, Sec);
82 return Sym->body();
Eugene Leviantceabe802016-08-11 07:56:43 +000083}
84
Rui Ueyama22375f22016-11-25 18:51:54 +000085static bool isUnderSysroot(StringRef Path) {
86 if (Config->Sysroot == "")
87 return false;
88 for (; !Path.empty(); Path = sys::path::parent_path(Path))
89 if (sys::fs::equivalent(Config->Sysroot, Path))
90 return true;
91 return false;
92}
93
George Rimar2ee2d2d2017-02-21 14:50:38 +000094template <class ELFT>
95void LinkerScript<ELFT>::setDot(Expr E, const Twine &Loc, bool InSec) {
Rafael Espindola679828f2017-02-17 16:26:13 +000096 uintX_t Val = E(Dot);
97 if (Val < Dot) {
98 if (InSec)
George Rimar2ee2d2d2017-02-21 14:50:38 +000099 error(Loc + ": unable to move location counter backward for: " +
100 CurOutSec->Name);
Rafael Espindola679828f2017-02-17 16:26:13 +0000101 else
George Rimar2ee2d2d2017-02-21 14:50:38 +0000102 error(Loc + ": unable to move location counter backward");
Rafael Espindola679828f2017-02-17 16:26:13 +0000103 }
104 Dot = Val;
105 // Update to location counter means update to section size.
106 if (InSec)
107 CurOutSec->Size = Dot - CurOutSec->Addr;
108}
109
George Rimarb2b70972017-02-07 10:23:28 +0000110// Sets value of a symbol. Two kinds of symbols are processed: synthetic
111// symbols, whose value is an offset from beginning of section and regular
112// symbols whose value is absolute.
113template <class ELFT>
Rafael Espindola4cd73522017-02-17 16:01:51 +0000114void LinkerScript<ELFT>::assignSymbol(SymbolAssignment *Cmd, bool InSec) {
115 if (Cmd->Name == ".") {
George Rimar2ee2d2d2017-02-21 14:50:38 +0000116 setDot(Cmd->Expression, Cmd->Location, InSec);
Rafael Espindola4cd73522017-02-17 16:01:51 +0000117 return;
118 }
119
George Rimarb2b70972017-02-07 10:23:28 +0000120 if (!Cmd->Sym)
Meador Inge8f1f3c42017-01-09 18:36:57 +0000121 return;
122
George Rimarb2b70972017-02-07 10:23:28 +0000123 if (auto *Body = dyn_cast<DefinedSynthetic>(Cmd->Sym)) {
124 Body->Section = Cmd->Expression.Section();
Rafael Espindolaea590d92017-02-08 15:19:03 +0000125 if (Body->Section) {
126 uint64_t VA = 0;
127 if (Body->Section->Flags & SHF_ALLOC)
128 VA = Body->Section->Addr;
129 Body->Value = Cmd->Expression(Dot) - VA;
130 }
George Rimarb2b70972017-02-07 10:23:28 +0000131 return;
Meador Inge8f1f3c42017-01-09 18:36:57 +0000132 }
George Rimarb2b70972017-02-07 10:23:28 +0000133
134 cast<DefinedRegular<ELFT>>(Cmd->Sym)->Value = Cmd->Expression(Dot);
Eugene Leviantdb741e72016-09-07 07:08:43 +0000135}
Meador Inge8f1f3c42017-01-09 18:36:57 +0000136
Rafael Espindola4cd73522017-02-17 16:01:51 +0000137template <class ELFT>
138void LinkerScript<ELFT>::addSymbol(SymbolAssignment *Cmd) {
Rui Ueyama16024212016-08-11 23:22:52 +0000139 if (Cmd->Name == ".")
Meador Inge8f1f3c42017-01-09 18:36:57 +0000140 return;
141
142 // If a symbol was in PROVIDE(), we need to define it only when
143 // it is a referenced undefined symbol.
Rui Ueyama16024212016-08-11 23:22:52 +0000144 SymbolBody *B = Symtab<ELFT>::X->find(Cmd->Name);
Meador Inge8f1f3c42017-01-09 18:36:57 +0000145 if (Cmd->Provide && (!B || B->isDefined()))
146 return;
147
148 // Otherwise, create a new symbol if one does not exist or an
149 // undefined one does exist.
150 if (Cmd->Expression.IsAbsolute())
151 Cmd->Sym = addRegular<ELFT>(Cmd);
152 else
153 Cmd->Sym = addSynthetic<ELFT>(Cmd);
George Rimarb2b70972017-02-07 10:23:28 +0000154
155 // If there are sections, then let the value be assigned later in
156 // `assignAddresses`.
157 if (!ScriptConfig->HasSections)
Rafael Espindola4cd73522017-02-17 16:01:51 +0000158 assignSymbol(Cmd);
Eugene Leviantceabe802016-08-11 07:56:43 +0000159}
160
George Rimar076fe152016-07-21 06:43:01 +0000161bool SymbolAssignment::classof(const BaseCommand *C) {
162 return C->Kind == AssignmentKind;
163}
164
165bool OutputSectionCommand::classof(const BaseCommand *C) {
166 return C->Kind == OutputSectionKind;
167}
168
George Rimareea31142016-07-21 14:26:59 +0000169bool InputSectionDescription::classof(const BaseCommand *C) {
170 return C->Kind == InputSectionKind;
171}
172
George Rimareefa7582016-08-04 09:29:31 +0000173bool AssertCommand::classof(const BaseCommand *C) {
174 return C->Kind == AssertKind;
175}
176
George Rimare38cbab2016-09-26 19:22:50 +0000177bool BytesDataCommand::classof(const BaseCommand *C) {
178 return C->Kind == BytesDataKind;
179}
180
Eugene Zelenko22886a22016-11-05 01:00:56 +0000181template <class ELFT> LinkerScript<ELFT>::LinkerScript() = default;
182template <class ELFT> LinkerScript<ELFT>::~LinkerScript() = default;
Rui Ueyamaf34d0e02016-08-12 01:24:53 +0000183
Rui Ueyamae0be2902016-11-21 02:10:12 +0000184template <class ELFT> static StringRef basename(InputSectionBase<ELFT> *S) {
185 if (S->getFile())
186 return sys::path::filename(S->getFile()->getName());
187 return "";
188}
189
Rui Ueyama07320e42016-04-20 20:13:41 +0000190template <class ELFT>
191bool LinkerScript<ELFT>::shouldKeep(InputSectionBase<ELFT> *S) {
Rui Ueyamae0be2902016-11-21 02:10:12 +0000192 for (InputSectionDescription *ID : Opt.KeptSections)
193 if (ID->FilePat.match(basename(S)))
194 for (SectionPattern &P : ID->SectionPatterns)
195 if (P.SectionPat.match(S->Name))
196 return true;
George Rimareea31142016-07-21 14:26:59 +0000197 return false;
198}
199
George Rimar575208c2016-09-15 19:15:12 +0000200static bool comparePriority(InputSectionData *A, InputSectionData *B) {
201 return getPriority(A->Name) < getPriority(B->Name);
202}
203
Rafael Espindolac0028d32016-09-08 20:47:52 +0000204static bool compareName(InputSectionData *A, InputSectionData *B) {
Rafael Espindola042a3f22016-09-08 14:06:08 +0000205 return A->Name < B->Name;
Rui Ueyama742c3832016-08-04 22:27:00 +0000206}
George Rimar350ece42016-08-03 08:35:59 +0000207
Rafael Espindolac0028d32016-09-08 20:47:52 +0000208static bool compareAlignment(InputSectionData *A, InputSectionData *B) {
Rui Ueyama742c3832016-08-04 22:27:00 +0000209 // ">" is not a mistake. Larger alignments are placed before smaller
210 // alignments in order to reduce the amount of padding necessary.
211 // This is compatible with GNU.
212 return A->Alignment > B->Alignment;
213}
George Rimar350ece42016-08-03 08:35:59 +0000214
Rafael Espindolac0028d32016-09-08 20:47:52 +0000215static std::function<bool(InputSectionData *, InputSectionData *)>
George Rimarbe394db2016-09-16 20:21:55 +0000216getComparator(SortSectionPolicy K) {
217 switch (K) {
218 case SortSectionPolicy::Alignment:
219 return compareAlignment;
220 case SortSectionPolicy::Name:
Rafael Espindolac0028d32016-09-08 20:47:52 +0000221 return compareName;
George Rimarbe394db2016-09-16 20:21:55 +0000222 case SortSectionPolicy::Priority:
223 return comparePriority;
224 default:
225 llvm_unreachable("unknown sort policy");
226 }
Rui Ueyama742c3832016-08-04 22:27:00 +0000227}
George Rimar0702c4e2016-07-29 15:32:46 +0000228
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000229template <class ELFT>
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000230static bool matchConstraints(ArrayRef<InputSectionBase<ELFT> *> Sections,
George Rimar06ae6832016-08-12 09:07:57 +0000231 ConstraintKind Kind) {
George Rimar8f66df92016-08-12 20:38:20 +0000232 if (Kind == ConstraintKind::NoConstraint)
233 return true;
Rafael Espindolae746e522016-09-21 18:33:44 +0000234 bool IsRW = llvm::any_of(Sections, [=](InputSectionData *Sec2) {
Rafael Espindolad3190792016-09-16 15:10:23 +0000235 auto *Sec = static_cast<InputSectionBase<ELFT> *>(Sec2);
Rafael Espindola1854a8e2016-10-26 12:36:56 +0000236 return Sec->Flags & SHF_WRITE;
George Rimar06ae6832016-08-12 09:07:57 +0000237 });
Rafael Espindolae746e522016-09-21 18:33:44 +0000238 return (IsRW && Kind == ConstraintKind::ReadWrite) ||
239 (!IsRW && Kind == ConstraintKind::ReadOnly);
George Rimar06ae6832016-08-12 09:07:57 +0000240}
241
George Rimar07171f22016-09-21 15:56:44 +0000242static void sortSections(InputSectionData **Begin, InputSectionData **End,
Rui Ueyamaee924702016-09-20 19:42:41 +0000243 SortSectionPolicy K) {
244 if (K != SortSectionPolicy::Default && K != SortSectionPolicy::None)
George Rimar07171f22016-09-21 15:56:44 +0000245 std::stable_sort(Begin, End, getComparator(K));
Rui Ueyamaee924702016-09-20 19:42:41 +0000246}
247
Rafael Espindolad3190792016-09-16 15:10:23 +0000248// Compute and remember which sections the InputSectionDescription matches.
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000249template <class ELFT>
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000250void LinkerScript<ELFT>::computeInputSections(InputSectionDescription *I) {
Rui Ueyama4dc07be2016-09-17 02:23:40 +0000251 // Collects all sections that satisfy constraints of I
252 // and attach them to I.
253 for (SectionPattern &Pat : I->SectionPatterns) {
George Rimar07171f22016-09-21 15:56:44 +0000254 size_t SizeBefore = I->Sections.size();
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000255
256 for (InputSectionBase<ELFT> *S : Symtab<ELFT>::X->Sections) {
Rafael Espindola3773bca2017-02-17 19:37:30 +0000257 if (S->Assigned)
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000258 continue;
Rafael Espindola908a3d32017-02-16 14:36:09 +0000259 // For -emit-relocs we have to ignore entries like
260 // .rela.dyn : { *(.rela.data) }
261 // which are common because they are in the default bfd script.
262 if (S->Type == SHT_REL || S->Type == SHT_RELA)
263 continue;
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000264
Rui Ueyamae0be2902016-11-21 02:10:12 +0000265 StringRef Filename = basename(S);
266 if (!I->FilePat.match(Filename) || Pat.ExcludedFilePat.match(Filename))
267 continue;
268 if (!Pat.SectionPat.match(S->Name))
269 continue;
270 I->Sections.push_back(S);
271 S->Assigned = true;
George Rimar395281c2016-09-16 17:42:10 +0000272 }
Rafael Espindolad3190792016-09-16 15:10:23 +0000273
George Rimar07171f22016-09-21 15:56:44 +0000274 // Sort sections as instructed by SORT-family commands and --sort-section
275 // option. Because SORT-family commands can be nested at most two depth
276 // (e.g. SORT_BY_NAME(SORT_BY_ALIGNMENT(.text.*))) and because the command
277 // line option is respected even if a SORT command is given, the exact
278 // behavior we have here is a bit complicated. Here are the rules.
279 //
280 // 1. If two SORT commands are given, --sort-section is ignored.
281 // 2. If one SORT command is given, and if it is not SORT_NONE,
282 // --sort-section is handled as an inner SORT command.
283 // 3. If one SORT command is given, and if it is SORT_NONE, don't sort.
284 // 4. If no SORT command is given, sort according to --sort-section.
285 InputSectionData **Begin = I->Sections.data() + SizeBefore;
286 InputSectionData **End = I->Sections.data() + I->Sections.size();
287 if (Pat.SortOuter != SortSectionPolicy::None) {
288 if (Pat.SortInner == SortSectionPolicy::Default)
289 sortSections(Begin, End, Config->SortSection);
290 else
291 sortSections(Begin, End, Pat.SortInner);
292 sortSections(Begin, End, Pat.SortOuter);
293 }
Rui Ueyamaee924702016-09-20 19:42:41 +0000294 }
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000295}
296
297template <class ELFT>
298void LinkerScript<ELFT>::discard(ArrayRef<InputSectionBase<ELFT> *> V) {
299 for (InputSectionBase<ELFT> *S : V) {
300 S->Live = false;
Rafael Espindolaecbfd872017-02-17 17:35:07 +0000301 if (S == In<ELFT>::ShStrTab)
302 error("discarding .shstrtab section is not allowed");
George Rimar647c1682017-02-17 19:34:05 +0000303 discard(S->DependentSections);
Rafael Espindolabe94e1b2016-09-14 14:32:08 +0000304 }
305}
306
George Rimar06ae6832016-08-12 09:07:57 +0000307template <class ELFT>
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000308std::vector<InputSectionBase<ELFT> *>
George Rimar06ae6832016-08-12 09:07:57 +0000309LinkerScript<ELFT>::createInputSectionList(OutputSectionCommand &OutCmd) {
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000310 std::vector<InputSectionBase<ELFT> *> Ret;
Rui Ueyamae7f912c2016-08-03 21:12:09 +0000311
George Rimar06ae6832016-08-12 09:07:57 +0000312 for (const std::unique_ptr<BaseCommand> &Base : OutCmd.Commands) {
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000313 auto *Cmd = dyn_cast<InputSectionDescription>(Base.get());
314 if (!Cmd)
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000315 continue;
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000316 computeInputSections(Cmd);
Rafael Espindolad3190792016-09-16 15:10:23 +0000317 for (InputSectionData *S : Cmd->Sections)
318 Ret.push_back(static_cast<InputSectionBase<ELFT> *>(S));
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000319 }
Rafael Espindolae71a3f8a2016-09-16 20:34:02 +0000320
Rui Ueyama0b9ce6a2016-08-12 03:16:56 +0000321 return Ret;
322}
323
Petr Hoseke5d3ca52016-08-31 15:31:17 +0000324template <class ELFT>
Eugene Leviant20d03192016-09-16 15:30:47 +0000325void LinkerScript<ELFT>::processCommands(OutputSectionFactory<ELFT> &Factory) {
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000326 for (unsigned I = 0; I < Opt.Commands.size(); ++I) {
327 auto Iter = Opt.Commands.begin() + I;
328 const std::unique_ptr<BaseCommand> &Base1 = *Iter;
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000329
330 // Handle symbol assignments outside of any output section.
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000331 if (auto *Cmd = dyn_cast<SymbolAssignment>(Base1.get())) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000332 addSymbol(Cmd);
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000333 continue;
334 }
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000335
Eugene Leviant20d03192016-09-16 15:30:47 +0000336 if (auto *Cmd = dyn_cast<AssertCommand>(Base1.get())) {
337 // If we don't have SECTIONS then output sections have already been
George Rimar194470cd2016-09-17 19:21:05 +0000338 // created by Writer<ELFT>. The LinkerScript<ELFT>::assignAddresses
Eugene Leviant20d03192016-09-16 15:30:47 +0000339 // will not be called, so ASSERT should be evaluated now.
340 if (!Opt.HasSections)
341 Cmd->Expression(0);
342 continue;
343 }
Rui Ueyama2ab5f732016-08-12 03:33:04 +0000344
Eugene Leviantceabe802016-08-11 07:56:43 +0000345 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base1.get())) {
Rafael Espindola7bd37872016-09-12 16:05:16 +0000346 std::vector<InputSectionBase<ELFT> *> V = createInputSectionList(*Cmd);
347
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000348 // The output section name `/DISCARD/' is special.
349 // Any input section assigned to it is discarded.
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000350 if (Cmd->Name == "/DISCARD/") {
Rafael Espindola7bd37872016-09-12 16:05:16 +0000351 discard(V);
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000352 continue;
353 }
Eugene Leviantceabe802016-08-11 07:56:43 +0000354
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000355 // This is for ONLY_IF_RO and ONLY_IF_RW. An output section directive
356 // ".foo : ONLY_IF_R[OW] { ... }" is handled only if all member input
357 // sections satisfy a given constraint. If not, a directive is handled
358 // as if it wasn't present from the beginning.
359 //
360 // Because we'll iterate over Commands many more times, the easiest
361 // way to "make it as if it wasn't present" is to just remove it.
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000362 if (!matchConstraints<ELFT>(V, Cmd->Constraint)) {
363 for (InputSectionBase<ELFT> *S : V)
Rui Ueyamaf94efdd2016-11-20 23:15:52 +0000364 S->Assigned = false;
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000365 Opt.Commands.erase(Iter);
George Rimardfbbbc82016-09-17 09:50:10 +0000366 --I;
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000367 continue;
368 }
369
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000370 // A directive may contain symbol definitions like this:
371 // ".foo : { ...; bar = .; }". Handle them.
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000372 for (const std::unique_ptr<BaseCommand> &Base : Cmd->Commands)
373 if (auto *OutCmd = dyn_cast<SymbolAssignment>(Base.get()))
Rafael Espindola4cd73522017-02-17 16:01:51 +0000374 addSymbol(OutCmd);
Rafael Espindola7c3ff2e2016-09-16 21:05:36 +0000375
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000376 // Handle subalign (e.g. ".foo : SUBALIGN(32) { ... }"). If subalign
377 // is given, input sections are aligned to that value, whether the
378 // given value is larger or smaller than the original section alignment.
379 if (Cmd->SubalignExpr) {
380 uint32_t Subalign = Cmd->SubalignExpr(0);
381 for (InputSectionBase<ELFT> *S : V)
382 S->Alignment = Subalign;
George Rimardb24d9c2016-08-19 15:18:23 +0000383 }
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000384
385 // Add input sections to an output section.
386 for (InputSectionBase<ELFT> *S : V)
Rafael Espindola82902742017-02-16 17:32:26 +0000387 Factory.addInputSec(S, Cmd->Name);
Eugene Leviantceabe802016-08-11 07:56:43 +0000388 }
Rui Ueyama48c3f1c2016-08-12 00:27:23 +0000389 }
Eugene Leviant20d03192016-09-16 15:30:47 +0000390}
Eugene Leviante63d81b2016-07-20 14:43:20 +0000391
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000392// Add sections that didn't match any sections command.
Eugene Leviant20d03192016-09-16 15:30:47 +0000393template <class ELFT>
George Rimar93c64022016-12-15 15:38:09 +0000394void LinkerScript<ELFT>::addOrphanSections(
395 OutputSectionFactory<ELFT> &Factory) {
Rui Ueyama8c6a5aa2016-11-05 22:37:59 +0000396 for (InputSectionBase<ELFT> *S : Symtab<ELFT>::X->Sections)
Rafael Espindola8f9026b2016-11-08 18:23:02 +0000397 if (S->Live && !S->OutSec)
Rafael Espindola82902742017-02-16 17:32:26 +0000398 Factory.addInputSec(S, getOutputSectionName(S->Name));
Eugene Leviante63d81b2016-07-20 14:43:20 +0000399}
400
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000401template <class ELFT> static bool isTbss(OutputSectionBase *Sec) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000402 return (Sec->Flags & SHF_TLS) && Sec->Type == SHT_NOBITS;
Rafael Espindolaa940e532016-09-22 12:35:44 +0000403}
404
Rafael Espindolad3190792016-09-16 15:10:23 +0000405template <class ELFT> void LinkerScript<ELFT>::output(InputSection<ELFT> *S) {
406 if (!AlreadyOutputIS.insert(S).second)
407 return;
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000408 bool IsTbss = isTbss<ELFT>(CurOutSec);
Eugene Leviant20889c52016-08-31 08:13:33 +0000409
Rafael Espindolad3190792016-09-16 15:10:23 +0000410 uintX_t Pos = IsTbss ? Dot + ThreadBssOffset : Dot;
411 Pos = alignTo(Pos, S->Alignment);
Rafael Espindola04a2e342016-11-09 01:42:41 +0000412 S->OutSecOff = Pos - CurOutSec->Addr;
Rafael Espindolad3190792016-09-16 15:10:23 +0000413 Pos += S->getSize();
414
415 // Update output section size after adding each section. This is so that
416 // SIZEOF works correctly in the case below:
417 // .foo { *(.aaa) a = SIZEOF(.foo); *(.bbb) }
Rafael Espindola04a2e342016-11-09 01:42:41 +0000418 CurOutSec->Size = Pos - CurOutSec->Addr;
Rafael Espindolad3190792016-09-16 15:10:23 +0000419
Meador Ingeb8897442017-01-24 02:34:00 +0000420 // If there is a memory region associated with this input section, then
421 // place the section in that region and update the region index.
422 if (CurMemRegion) {
423 CurMemRegion->Offset += CurOutSec->Size;
424 uint64_t CurSize = CurMemRegion->Offset - CurMemRegion->Origin;
425 if (CurSize > CurMemRegion->Length) {
426 uint64_t OverflowAmt = CurSize - CurMemRegion->Length;
427 error("section '" + CurOutSec->Name + "' will not fit in region '" +
428 CurMemRegion->Name + "': overflowed by " + Twine(OverflowAmt) +
429 " bytes");
430 }
431 }
432
Rafael Espindola7252ae52016-09-22 12:00:08 +0000433 if (IsTbss)
434 ThreadBssOffset = Pos - Dot;
435 else
Rafael Espindolad3190792016-09-16 15:10:23 +0000436 Dot = Pos;
437}
438
439template <class ELFT> void LinkerScript<ELFT>::flush() {
Rafael Espindola65499b92016-09-23 20:10:47 +0000440 if (!CurOutSec || !AlreadyOutputOS.insert(CurOutSec).second)
441 return;
442 if (auto *OutSec = dyn_cast<OutputSection<ELFT>>(CurOutSec)) {
Rafael Espindolad3190792016-09-16 15:10:23 +0000443 for (InputSection<ELFT> *I : OutSec->Sections)
444 output(I);
Rafael Espindola65499b92016-09-23 20:10:47 +0000445 } else {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000446 Dot += CurOutSec->Size;
Eugene Leviant20889c52016-08-31 08:13:33 +0000447 }
448}
449
450template <class ELFT>
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000451void LinkerScript<ELFT>::switchTo(OutputSectionBase *Sec) {
Rafael Espindolad3190792016-09-16 15:10:23 +0000452 if (CurOutSec == Sec)
453 return;
454 if (AlreadyOutputOS.count(Sec))
455 return;
456
457 flush();
458 CurOutSec = Sec;
459
Rafael Espindola04a2e342016-11-09 01:42:41 +0000460 Dot = alignTo(Dot, CurOutSec->Addralign);
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000461 CurOutSec->Addr = isTbss<ELFT>(CurOutSec) ? Dot + ThreadBssOffset : Dot;
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000462
463 // If neither AT nor AT> is specified for an allocatable section, the linker
464 // will set the LMA such that the difference between VMA and LMA for the
465 // section is the same as the preceding output section in the same region
466 // https://sourceware.org/binutils/docs-2.20/ld/Output-Section-LMA.html
George Rimarae4761c2017-02-21 15:08:18 +0000467 Expr LMAExpr = CurLMA.first;
468 if (LMAExpr)
469 CurOutSec->setLMAOffset(LMAExpr(CurLMA.second) - CurLMA.second);
Rafael Espindolad3190792016-09-16 15:10:23 +0000470}
471
472template <class ELFT> void LinkerScript<ELFT>::process(BaseCommand &Base) {
George Rimare38cbab2016-09-26 19:22:50 +0000473 // This handles the assignments to symbol or to a location counter (.)
Rafael Espindolad3190792016-09-16 15:10:23 +0000474 if (auto *AssignCmd = dyn_cast<SymbolAssignment>(&Base)) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000475 assignSymbol(AssignCmd, true);
Eugene Leviantceabe802016-08-11 07:56:43 +0000476 return;
Rui Ueyama2de509c2016-08-12 00:55:08 +0000477 }
George Rimare38cbab2016-09-26 19:22:50 +0000478
479 // Handle BYTE(), SHORT(), LONG(), or QUAD().
480 if (auto *DataCmd = dyn_cast<BytesDataCommand>(&Base)) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000481 DataCmd->Offset = Dot - CurOutSec->Addr;
George Rimare38cbab2016-09-26 19:22:50 +0000482 Dot += DataCmd->Size;
Rafael Espindola04a2e342016-11-09 01:42:41 +0000483 CurOutSec->Size = Dot - CurOutSec->Addr;
George Rimare38cbab2016-09-26 19:22:50 +0000484 return;
485 }
486
Meador Ingeb2d99d62016-11-22 18:01:50 +0000487 if (auto *AssertCmd = dyn_cast<AssertCommand>(&Base)) {
488 AssertCmd->Expression(Dot);
489 return;
490 }
491
George Rimare38cbab2016-09-26 19:22:50 +0000492 // It handles single input section description command,
493 // calculates and assigns the offsets for each section and also
494 // updates the output section size.
Rafael Espindolad3190792016-09-16 15:10:23 +0000495 auto &ICmd = cast<InputSectionDescription>(Base);
496 for (InputSectionData *ID : ICmd.Sections) {
George Rimar3fb5a6d2016-11-29 16:05:27 +0000497 // We tentatively added all synthetic sections at the beginning and removed
498 // empty ones afterwards (because there is no way to know whether they were
499 // going be empty or not other than actually running linker scripts.)
500 // We need to ignore remains of empty sections.
501 if (auto *Sec = dyn_cast<SyntheticSection<ELFT>>(ID))
502 if (Sec->empty())
503 continue;
504
Rafael Espindolad3190792016-09-16 15:10:23 +0000505 auto *IB = static_cast<InputSectionBase<ELFT> *>(ID);
George Rimar78ef6452017-02-21 15:46:43 +0000506 if (!IB->Live)
507 continue;
Rafael Espindolad3190792016-09-16 15:10:23 +0000508 switchTo(IB->OutSec);
509 if (auto *I = dyn_cast<InputSection<ELFT>>(IB))
510 output(I);
Rafael Espindola65499b92016-09-23 20:10:47 +0000511 else
512 flush();
Eugene Leviantceabe802016-08-11 07:56:43 +0000513 }
514}
515
George Rimar8f66df92016-08-12 20:38:20 +0000516template <class ELFT>
Rafael Espindola2b074552017-02-03 22:27:05 +0000517static OutputSectionBase *
518findSection(StringRef Name, const std::vector<OutputSectionBase *> &Sections) {
519 auto End = Sections.end();
520 auto HasName = [=](OutputSectionBase *Sec) { return Sec->getName() == Name; };
521 auto I = std::find_if(Sections.begin(), End, HasName);
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000522 std::vector<OutputSectionBase *> Ret;
Rafael Espindola2b074552017-02-03 22:27:05 +0000523 if (I == End)
524 return nullptr;
525 assert(std::find_if(I + 1, End, HasName) == End);
526 return *I;
George Rimar8f66df92016-08-12 20:38:20 +0000527}
528
Meador Ingeb8897442017-01-24 02:34:00 +0000529// This function searches for a memory region to place the given output
530// section in. If found, a pointer to the appropriate memory region is
531// returned. Otherwise, a nullptr is returned.
532template <class ELFT>
533MemoryRegion *LinkerScript<ELFT>::findMemoryRegion(OutputSectionCommand *Cmd,
534 OutputSectionBase *Sec) {
535 // If a memory region name was specified in the output section command,
536 // then try to find that region first.
537 if (!Cmd->MemoryRegionName.empty()) {
538 auto It = Opt.MemoryRegions.find(Cmd->MemoryRegionName);
539 if (It != Opt.MemoryRegions.end())
540 return &It->second;
541 error("memory region '" + Cmd->MemoryRegionName + "' not declared");
542 return nullptr;
543 }
544
545 // The memory region name is empty, thus a suitable region must be
546 // searched for in the region map. If the region map is empty, just
547 // return. Note that this check doesn't happen at the very beginning
548 // so that uses of undeclared regions can be caught.
549 if (!Opt.MemoryRegions.size())
550 return nullptr;
551
552 // See if a region can be found by matching section flags.
553 for (auto &MRI : Opt.MemoryRegions) {
554 MemoryRegion &MR = MRI.second;
Rui Ueyama8a8a9532017-01-26 02:58:59 +0000555 if ((MR.Flags & Sec->Flags) != 0 && (MR.NegFlags & Sec->Flags) == 0)
Meador Ingeb8897442017-01-24 02:34:00 +0000556 return &MR;
557 }
558
559 // Otherwise, no suitable region was found.
560 if (Sec->Flags & SHF_ALLOC)
561 error("no memory region specified for section '" + Sec->Name + "'");
562 return nullptr;
563}
564
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000565// This function assigns offsets to input sections and an output section
566// for a single sections command (e.g. ".text { *(.text); }").
Rafael Espindolad3190792016-09-16 15:10:23 +0000567template <class ELFT>
568void LinkerScript<ELFT>::assignOffsets(OutputSectionCommand *Cmd) {
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000569 if (Cmd->LMAExpr)
George Rimarae4761c2017-02-21 15:08:18 +0000570 CurLMA = {Cmd->LMAExpr, Dot};
Rafael Espindola2b074552017-02-03 22:27:05 +0000571 OutputSectionBase *Sec = findSection<ELFT>(Cmd->Name, *OutputSections);
572 if (!Sec)
Rafael Espindolad3190792016-09-16 15:10:23 +0000573 return;
Meador Ingeb8897442017-01-24 02:34:00 +0000574
Rafael Espindola679828f2017-02-17 16:26:13 +0000575 if (Cmd->AddrExpr && Sec->Flags & SHF_ALLOC)
George Rimar2ee2d2d2017-02-21 14:50:38 +0000576 setDot(Cmd->AddrExpr, Cmd->Location);
Rafael Espindola679828f2017-02-17 16:26:13 +0000577
Petr Hosek165088a2017-02-07 23:42:31 +0000578 // Handle align (e.g. ".foo : ALIGN(16) { ... }").
579 if (Cmd->AlignExpr)
580 Sec->updateAlignment(Cmd->AlignExpr(0));
581
Meador Ingeb8897442017-01-24 02:34:00 +0000582 // Try and find an appropriate memory region to assign offsets in.
583 CurMemRegion = findMemoryRegion(Cmd, Sec);
584 if (CurMemRegion)
585 Dot = CurMemRegion->Offset;
586 switchTo(Sec);
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000587
Rafael Espindolad3190792016-09-16 15:10:23 +0000588 // Find the last section output location. We will output orphan sections
589 // there so that end symbols point to the correct location.
590 auto E = std::find_if(Cmd->Commands.rbegin(), Cmd->Commands.rend(),
591 [](const std::unique_ptr<BaseCommand> &Cmd) {
592 return !isa<SymbolAssignment>(*Cmd);
593 })
594 .base();
595 for (auto I = Cmd->Commands.begin(); I != E; ++I)
596 process(**I);
Rafael Espindola65499b92016-09-23 20:10:47 +0000597 flush();
George Rimarb31dd372016-09-19 13:27:31 +0000598 std::for_each(E, Cmd->Commands.end(),
599 [this](std::unique_ptr<BaseCommand> &B) { process(*B.get()); });
Rafael Espindolad3190792016-09-16 15:10:23 +0000600}
601
Rafael Espindola07fe6122016-11-14 14:23:35 +0000602template <class ELFT> void LinkerScript<ELFT>::removeEmptyCommands() {
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000603 // It is common practice to use very generic linker scripts. So for any
604 // given run some of the output sections in the script will be empty.
605 // We could create corresponding empty output sections, but that would
606 // clutter the output.
607 // We instead remove trivially empty sections. The bfd linker seems even
608 // more aggressive at removing them.
609 auto Pos = std::remove_if(
610 Opt.Commands.begin(), Opt.Commands.end(),
611 [&](const std::unique_ptr<BaseCommand> &Base) {
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000612 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
Rafael Espindola2b074552017-02-03 22:27:05 +0000613 return !findSection<ELFT>(Cmd->Name, *OutputSections);
Rui Ueyama0b1b6952016-11-21 02:11:05 +0000614 return false;
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000615 });
616 Opt.Commands.erase(Pos, Opt.Commands.end());
Rafael Espindola07fe6122016-11-14 14:23:35 +0000617}
618
Rafael Espindola6a537372016-11-14 14:33:49 +0000619static bool isAllSectionDescription(const OutputSectionCommand &Cmd) {
620 for (const std::unique_ptr<BaseCommand> &I : Cmd.Commands)
621 if (!isa<InputSectionDescription>(*I))
622 return false;
623 return true;
624}
Rafael Espindola6d38e4d2016-09-20 13:12:07 +0000625
Rafael Espindola6a537372016-11-14 14:33:49 +0000626template <class ELFT> void LinkerScript<ELFT>::adjustSectionsBeforeSorting() {
Rafael Espindola9546fff2016-09-22 14:40:50 +0000627 // If the output section contains only symbol assignments, create a
628 // corresponding output section. The bfd linker seems to only create them if
629 // '.' is assigned to, but creating these section should not have any bad
630 // consequeces and gives us a section to put the symbol in.
631 uintX_t Flags = SHF_ALLOC;
Rafael Espindolaf93b8c22016-11-26 06:55:35 +0000632 uint32_t Type = SHT_NOBITS;
Rafael Espindola9546fff2016-09-22 14:40:50 +0000633 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
634 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
635 if (!Cmd)
636 continue;
Rafael Espindola2b074552017-02-03 22:27:05 +0000637 if (OutputSectionBase *Sec =
638 findSection<ELFT>(Cmd->Name, *OutputSections)) {
639 Flags = Sec->Flags;
640 Type = Sec->Type;
Rafael Espindola9546fff2016-09-22 14:40:50 +0000641 continue;
642 }
643
Rafael Espindola6a537372016-11-14 14:33:49 +0000644 if (isAllSectionDescription(*Cmd))
645 continue;
646
Rui Ueyama95642b92016-11-01 23:09:07 +0000647 auto *OutSec = make<OutputSection<ELFT>>(Cmd->Name, Type, Flags);
Rafael Espindola9546fff2016-09-22 14:40:50 +0000648 OutputSections->push_back(OutSec);
649 }
Rafael Espindolaf7a17442016-11-14 15:39:38 +0000650}
651
652template <class ELFT> void LinkerScript<ELFT>::adjustSectionsAfterSorting() {
653 placeOrphanSections();
654
655 // If output section command doesn't specify any segments,
656 // and we haven't previously assigned any section to segment,
657 // then we simply assign section to the very first load segment.
658 // Below is an example of such linker script:
659 // PHDRS { seg PT_LOAD; }
660 // SECTIONS { .aaa : { *(.aaa) } }
661 std::vector<StringRef> DefPhdrs;
662 auto FirstPtLoad =
663 std::find_if(Opt.PhdrsCommands.begin(), Opt.PhdrsCommands.end(),
664 [](const PhdrsCommand &Cmd) { return Cmd.Type == PT_LOAD; });
665 if (FirstPtLoad != Opt.PhdrsCommands.end())
666 DefPhdrs.push_back(FirstPtLoad->Name);
667
668 // Walk the commands and propagate the program headers to commands that don't
669 // explicitly specify them.
670 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
671 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
672 if (!Cmd)
673 continue;
674 if (Cmd->Phdrs.empty())
675 Cmd->Phdrs = DefPhdrs;
676 else
677 DefPhdrs = Cmd->Phdrs;
678 }
Rafael Espindola6a537372016-11-14 14:33:49 +0000679
680 removeEmptyCommands();
Rafael Espindola9546fff2016-09-22 14:40:50 +0000681}
682
Rafael Espindola15c57952016-09-22 18:05:49 +0000683// When placing orphan sections, we want to place them after symbol assignments
684// so that an orphan after
685// begin_foo = .;
686// foo : { *(foo) }
687// end_foo = .;
688// doesn't break the intended meaning of the begin/end symbols.
689// We don't want to go over sections since Writer<ELFT>::sortSections is the
690// one in charge of deciding the order of the sections.
691// We don't want to go over alignments, since doing so in
692// rx_sec : { *(rx_sec) }
693// . = ALIGN(0x1000);
694// /* The RW PT_LOAD starts here*/
695// rw_sec : { *(rw_sec) }
696// would mean that the RW PT_LOAD would become unaligned.
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000697static bool shouldSkip(const BaseCommand &Cmd) {
Rafael Espindola15c57952016-09-22 18:05:49 +0000698 if (isa<OutputSectionCommand>(Cmd))
699 return false;
700 const auto *Assign = dyn_cast<SymbolAssignment>(&Cmd);
701 if (!Assign)
702 return true;
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000703 return Assign->Name != ".";
Rafael Espindola15c57952016-09-22 18:05:49 +0000704}
705
Rui Ueyama6697ec22017-02-02 23:26:12 +0000706// Orphan sections are sections present in the input files which are
707// not explicitly placed into the output file by the linker script.
708//
709// When the control reaches this function, Opt.Commands contains
710// output section commands for non-orphan sections only. This function
711// adds new elements for orphan sections to Opt.Commands so that all
712// sections are explicitly handled by Opt.Commands.
713//
714// Writer<ELFT>::sortSections has already sorted output sections.
715// What we need to do is to scan OutputSections vector and
716// Opt.Commands in parallel to find orphan sections. If there is an
717// output section that doesn't have a corresponding entry in
718// Opt.Commands, we will insert a new entry to Opt.Commands.
719//
720// There is some ambiguity as to where exactly a new entry should be
721// inserted, because Opt.Commands contains not only output section
722// commands but other types of commands such as symbol assignment
723// expressions. There's no correct answer here due to the lack of the
724// formal specification of the linker script. We use heuristics to
725// determine whether a new output command should be added before or
726// after another commands. For the details, look at shouldSkip
727// function.
George Rimar93c64022016-12-15 15:38:09 +0000728template <class ELFT> void LinkerScript<ELFT>::placeOrphanSections() {
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000729 // The OutputSections are already in the correct order.
730 // This loops creates or moves commands as needed so that they are in the
731 // correct order.
732 int CmdIndex = 0;
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000733
734 // As a horrible special case, skip the first . assignment if it is before any
735 // section. We do this because it is common to set a load address by starting
736 // the script with ". = 0xabcd" and the expectation is that every section is
737 // after that.
738 auto FirstSectionOrDotAssignment =
739 std::find_if(Opt.Commands.begin(), Opt.Commands.end(),
740 [](const std::unique_ptr<BaseCommand> &Cmd) {
741 if (isa<OutputSectionCommand>(*Cmd))
742 return true;
743 const auto *Assign = dyn_cast<SymbolAssignment>(Cmd.get());
744 if (!Assign)
745 return false;
746 return Assign->Name == ".";
747 });
748 if (FirstSectionOrDotAssignment != Opt.Commands.end()) {
749 CmdIndex = FirstSectionOrDotAssignment - Opt.Commands.begin();
750 if (isa<SymbolAssignment>(**FirstSectionOrDotAssignment))
751 ++CmdIndex;
752 }
753
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000754 for (OutputSectionBase *Sec : *OutputSections) {
George Rimar652852c2016-04-16 10:10:32 +0000755 StringRef Name = Sec->getName();
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000756
757 // Find the last spot where we can insert a command and still get the
Rafael Espindola15c57952016-09-22 18:05:49 +0000758 // correct result.
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000759 auto CmdIter = Opt.Commands.begin() + CmdIndex;
760 auto E = Opt.Commands.end();
Rafael Espindola5fcc99c2016-11-27 09:44:45 +0000761 while (CmdIter != E && shouldSkip(**CmdIter)) {
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000762 ++CmdIter;
763 ++CmdIndex;
764 }
765
766 auto Pos =
767 std::find_if(CmdIter, E, [&](const std::unique_ptr<BaseCommand> &Base) {
768 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
769 return Cmd && Cmd->Name == Name;
770 });
771 if (Pos == E) {
772 Opt.Commands.insert(CmdIter,
773 llvm::make_unique<OutputSectionCommand>(Name));
Rafael Espindola15c57952016-09-22 18:05:49 +0000774 ++CmdIndex;
775 continue;
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000776 }
Rafael Espindola15c57952016-09-22 18:05:49 +0000777
778 // Continue from where we found it.
779 CmdIndex = (Pos - Opt.Commands.begin()) + 1;
George Rimar652852c2016-04-16 10:10:32 +0000780 }
Rafael Espindola337f9032016-11-14 14:13:32 +0000781}
782
783template <class ELFT>
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000784void LinkerScript<ELFT>::assignAddresses(std::vector<PhdrEntry> &Phdrs) {
Rui Ueyama7c18c282016-04-18 21:00:40 +0000785 // Assign addresses as instructed by linker script SECTIONS sub-commands.
Rafael Espindolabe607332016-09-30 00:16:11 +0000786 Dot = 0;
George Rimar652852c2016-04-16 10:10:32 +0000787
Rafael Espindola06f47432017-02-06 22:21:46 +0000788 // A symbol can be assigned before any section is mentioned in the linker
789 // script. In an DSO, the symbol values are addresses, so the only important
790 // section values are:
791 // * SHN_UNDEF
792 // * SHN_ABS
793 // * Any value meaning a regular section.
794 // To handle that, create a dummy aether section that fills the void before
795 // the linker scripts switches to another section. It has an index of one
796 // which will map to whatever the first actual section is.
797 auto *Aether = make<OutputSectionBase>("", 0, SHF_ALLOC);
798 Aether->SectionIndex = 1;
799 switchTo(Aether);
800
George Rimar076fe152016-07-21 06:43:01 +0000801 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
802 if (auto *Cmd = dyn_cast<SymbolAssignment>(Base.get())) {
Rafael Espindola4cd73522017-02-17 16:01:51 +0000803 assignSymbol(Cmd);
George Rimar652852c2016-04-16 10:10:32 +0000804 continue;
805 }
806
George Rimareefa7582016-08-04 09:29:31 +0000807 if (auto *Cmd = dyn_cast<AssertCommand>(Base.get())) {
808 Cmd->Expression(Dot);
809 continue;
810 }
811
George Rimar076fe152016-07-21 06:43:01 +0000812 auto *Cmd = cast<OutputSectionCommand>(Base.get());
Rafael Espindolad3190792016-09-16 15:10:23 +0000813 assignOffsets(Cmd);
George Rimar652852c2016-04-16 10:10:32 +0000814 }
Rui Ueyama52c4e172016-07-01 10:42:25 +0000815
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000816 uintX_t MinVA = std::numeric_limits<uintX_t>::max();
Rafael Espindolaea590d92017-02-08 15:19:03 +0000817 for (OutputSectionBase *Sec : *OutputSections) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000818 if (Sec->Flags & SHF_ALLOC)
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000819 MinVA = std::min<uint64_t>(MinVA, Sec->Addr);
Rafael Espindolaea590d92017-02-08 15:19:03 +0000820 else
821 Sec->Addr = 0;
822 }
Rafael Espindolaaab6d5c2016-09-16 21:29:07 +0000823
Rafael Espindola8c495e22017-01-20 20:41:18 +0000824 allocateHeaders<ELFT>(Phdrs, *OutputSections, MinVA);
George Rimar652852c2016-04-16 10:10:32 +0000825}
826
Rui Ueyama464daad2016-08-22 04:55:20 +0000827// Creates program headers as instructed by PHDRS linker script command.
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000828template <class ELFT> std::vector<PhdrEntry> LinkerScript<ELFT>::createPhdrs() {
829 std::vector<PhdrEntry> Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +0000830
Rui Ueyama464daad2016-08-22 04:55:20 +0000831 // Process PHDRS and FILEHDR keywords because they are not
832 // real output sections and cannot be added in the following loop.
Eugene Leviantbbe38602016-07-19 09:25:43 +0000833 for (const PhdrsCommand &Cmd : Opt.PhdrsCommands) {
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000834 Ret.emplace_back(Cmd.Type, Cmd.Flags == UINT_MAX ? PF_R : Cmd.Flags);
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000835 PhdrEntry &Phdr = Ret.back();
Eugene Leviantbbe38602016-07-19 09:25:43 +0000836
837 if (Cmd.HasFilehdr)
Rui Ueyamaadca2452016-07-23 14:18:48 +0000838 Phdr.add(Out<ELFT>::ElfHeader);
Eugene Leviantbbe38602016-07-19 09:25:43 +0000839 if (Cmd.HasPhdrs)
Rui Ueyamaadca2452016-07-23 14:18:48 +0000840 Phdr.add(Out<ELFT>::ProgramHeaders);
Eugene Leviant56b21c82016-09-09 09:46:16 +0000841
842 if (Cmd.LMAExpr) {
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000843 Phdr.p_paddr = Cmd.LMAExpr(0);
Eugene Leviant56b21c82016-09-09 09:46:16 +0000844 Phdr.HasLMA = true;
845 }
Eugene Leviantbbe38602016-07-19 09:25:43 +0000846 }
847
Rui Ueyama464daad2016-08-22 04:55:20 +0000848 // Add output sections to program headers.
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000849 for (OutputSectionBase *Sec : *OutputSections) {
Rafael Espindola04a2e342016-11-09 01:42:41 +0000850 if (!(Sec->Flags & SHF_ALLOC))
Eugene Leviantbbe38602016-07-19 09:25:43 +0000851 break;
852
Eugene Leviantce30b1c2016-10-19 15:04:49 +0000853 // Assign headers specified by linker script
Rafael Espindolaf7a17442016-11-14 15:39:38 +0000854 for (size_t Id : getPhdrIndices(Sec->getName())) {
Eugene Leviantce30b1c2016-10-19 15:04:49 +0000855 Ret[Id].add(Sec);
856 if (Opt.PhdrsCommands[Id].Flags == UINT_MAX)
Rafael Espindola17cb7c02016-12-19 17:01:01 +0000857 Ret[Id].p_flags |= Sec->getPhdrFlags();
Eugene Leviantbbe38602016-07-19 09:25:43 +0000858 }
Eugene Leviantbbe38602016-07-19 09:25:43 +0000859 }
Rui Ueyamaedebbdf2016-07-24 23:47:31 +0000860 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +0000861}
862
Eugene Leviantf9bc3bd2016-08-16 06:40:58 +0000863template <class ELFT> bool LinkerScript<ELFT>::ignoreInterpSection() {
864 // Ignore .interp section in case we have PHDRS specification
865 // and PT_INTERP isn't listed.
866 return !Opt.PhdrsCommands.empty() &&
867 llvm::find_if(Opt.PhdrsCommands, [](const PhdrsCommand &Cmd) {
868 return Cmd.Type == PT_INTERP;
869 }) == Opt.PhdrsCommands.end();
870}
871
George Rimar93c64022016-12-15 15:38:09 +0000872template <class ELFT> uint32_t LinkerScript<ELFT>::getFiller(StringRef Name) {
George Rimarf6c3cce2016-07-21 07:48:54 +0000873 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands)
874 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
875 if (Cmd->Name == Name)
876 return Cmd->Filler;
Rui Ueyama16068ae2016-11-19 18:05:56 +0000877 return 0;
George Rimare2ee72b2016-02-26 14:48:31 +0000878}
879
George Rimare38cbab2016-09-26 19:22:50 +0000880template <class ELFT>
881static void writeInt(uint8_t *Buf, uint64_t Data, uint64_t Size) {
882 const endianness E = ELFT::TargetEndianness;
883
884 switch (Size) {
885 case 1:
886 *Buf = (uint8_t)Data;
887 break;
888 case 2:
889 write16<E>(Buf, Data);
890 break;
891 case 4:
892 write32<E>(Buf, Data);
893 break;
894 case 8:
895 write64<E>(Buf, Data);
896 break;
897 default:
898 llvm_unreachable("unsupported Size argument");
899 }
900}
901
902template <class ELFT>
903void LinkerScript<ELFT>::writeDataBytes(StringRef Name, uint8_t *Buf) {
904 int I = getSectionIndex(Name);
905 if (I == INT_MAX)
906 return;
907
Rui Ueyama6e68c5e2016-11-19 18:05:58 +0000908 auto *Cmd = dyn_cast<OutputSectionCommand>(Opt.Commands[I].get());
909 for (const std::unique_ptr<BaseCommand> &Base : Cmd->Commands)
910 if (auto *Data = dyn_cast<BytesDataCommand>(Base.get()))
Meador Inge95c7d8d2016-12-08 23:21:30 +0000911 writeInt<ELFT>(Buf + Data->Offset, Data->Expression(0), Data->Size);
George Rimare38cbab2016-09-26 19:22:50 +0000912}
913
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000914template <class ELFT> bool LinkerScript<ELFT>::hasLMA(StringRef Name) {
George Rimar8ceadb32016-08-17 07:44:19 +0000915 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands)
916 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get()))
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000917 if (Cmd->LMAExpr && Cmd->Name == Name)
918 return true;
919 return false;
George Rimar8ceadb32016-08-17 07:44:19 +0000920}
921
Rui Ueyamac3e2a4b2016-04-21 20:30:00 +0000922// Returns the index of the given section name in linker script
923// SECTIONS commands. Sections are laid out as the same order as they
924// were in the script. If a given name did not appear in the script,
925// it returns INT_MAX, so that it will be laid out at end of file.
George Rimar076fe152016-07-21 06:43:01 +0000926template <class ELFT> int LinkerScript<ELFT>::getSectionIndex(StringRef Name) {
Rui Ueyama6e68c5e2016-11-19 18:05:58 +0000927 for (int I = 0, E = Opt.Commands.size(); I != E; ++I)
928 if (auto *Cmd = dyn_cast<OutputSectionCommand>(Opt.Commands[I].get()))
Rui Ueyamaf510fa62016-07-26 00:21:15 +0000929 if (Cmd->Name == Name)
930 return I;
Rui Ueyamaf510fa62016-07-26 00:21:15 +0000931 return INT_MAX;
George Rimar71b26e92016-04-21 10:22:02 +0000932}
933
Eugene Leviantbbe38602016-07-19 09:25:43 +0000934template <class ELFT> bool LinkerScript<ELFT>::hasPhdrsCommands() {
935 return !Opt.PhdrsCommands.empty();
936}
937
George Rimar9e694502016-07-29 16:18:47 +0000938template <class ELFT>
Eugene Levianted30ce72016-11-28 09:58:04 +0000939const OutputSectionBase *LinkerScript<ELFT>::getOutputSection(const Twine &Loc,
940 StringRef Name) {
Eugene Leviantafaa9342016-11-16 09:49:39 +0000941 static OutputSectionBase FakeSec("", 0, 0);
George Rimar96659df2016-08-30 09:54:01 +0000942
Rafael Espindolae08e78d2016-11-09 23:23:45 +0000943 for (OutputSectionBase *Sec : *OutputSections)
Eugene Leviantb71d6f72016-10-06 09:39:28 +0000944 if (Sec->getName() == Name)
Eugene Leviantafaa9342016-11-16 09:49:39 +0000945 return Sec;
Eugene Levianted30ce72016-11-28 09:58:04 +0000946
947 error(Loc + ": undefined section " + Name);
Eugene Leviantafaa9342016-11-16 09:49:39 +0000948 return &FakeSec;
Eugene Leviant36fac7f2016-09-08 09:08:30 +0000949}
950
Rui Ueyamaedf75e72016-11-17 20:27:10 +0000951// This function is essentially the same as getOutputSection(Name)->Size,
952// but it won't print out an error message if a given section is not found.
953//
954// Linker script does not create an output section if its content is empty.
955// We want to allow SIZEOF(.foo) where .foo is a section which happened to
956// be empty. That is why this function is different from getOutputSection().
957template <class ELFT>
958uint64_t LinkerScript<ELFT>::getOutputSectionSize(StringRef Name) {
959 for (OutputSectionBase *Sec : *OutputSections)
960 if (Sec->getName() == Name)
961 return Sec->Size;
962 return 0;
963}
964
George Rimar884e7862016-09-08 08:19:13 +0000965template <class ELFT> uint64_t LinkerScript<ELFT>::getHeaderSize() {
Rafael Espindola0d4b6d52016-09-22 16:47:21 +0000966 return elf::getHeaderSize<ELFT>();
George Rimare32a3592016-08-10 07:59:34 +0000967}
968
Eugene Leviantf6aeed32016-12-22 13:13:12 +0000969template <class ELFT>
970uint64_t LinkerScript<ELFT>::getSymbolValue(const Twine &Loc, StringRef S) {
George Rimar884e7862016-09-08 08:19:13 +0000971 if (SymbolBody *B = Symtab<ELFT>::X->find(S))
972 return B->getVA<ELFT>();
Eugene Leviantf6aeed32016-12-22 13:13:12 +0000973 error(Loc + ": symbol not found: " + S);
George Rimar884e7862016-09-08 08:19:13 +0000974 return 0;
975}
976
George Rimarf34f45f2016-09-23 13:17:23 +0000977template <class ELFT> bool LinkerScript<ELFT>::isDefined(StringRef S) {
978 return Symtab<ELFT>::X->find(S) != nullptr;
979}
980
Rafael Espindola2f831dc2016-10-31 19:56:37 +0000981template <class ELFT> bool LinkerScript<ELFT>::isAbsolute(StringRef S) {
982 SymbolBody *Sym = Symtab<ELFT>::X->find(S);
983 auto *DR = dyn_cast_or_null<DefinedRegular<ELFT>>(Sym);
984 return DR && !DR->Section;
985}
986
Eugene Leviantafaa9342016-11-16 09:49:39 +0000987// Gets section symbol belongs to. Symbol "." doesn't belong to any
988// specific section but isn't absolute at the same time, so we try
989// to find suitable section for it as well.
990template <class ELFT>
991const OutputSectionBase *LinkerScript<ELFT>::getSymbolSection(StringRef S) {
Rafael Espindola06f47432017-02-06 22:21:46 +0000992 if (SymbolBody *Sym = Symtab<ELFT>::X->find(S))
993 return SymbolTableSection<ELFT>::getOutputSection(Sym);
994 return CurOutSec;
Eugene Leviantafaa9342016-11-16 09:49:39 +0000995}
996
Eugene Leviantbbe38602016-07-19 09:25:43 +0000997// Returns indices of ELF headers containing specific section, identified
998// by Name. Each index is a zero based number of ELF header listed within
999// PHDRS {} script block.
1000template <class ELFT>
Rui Ueyamaedebbdf2016-07-24 23:47:31 +00001001std::vector<size_t> LinkerScript<ELFT>::getPhdrIndices(StringRef SectionName) {
George Rimar076fe152016-07-21 06:43:01 +00001002 for (const std::unique_ptr<BaseCommand> &Base : Opt.Commands) {
1003 auto *Cmd = dyn_cast<OutputSectionCommand>(Base.get());
Rui Ueyamaedebbdf2016-07-24 23:47:31 +00001004 if (!Cmd || Cmd->Name != SectionName)
George Rimar31d842f2016-07-20 16:43:03 +00001005 continue;
1006
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001007 std::vector<size_t> Ret;
1008 for (StringRef PhdrName : Cmd->Phdrs)
Eugene Leviant2a942c42016-12-05 16:38:32 +00001009 Ret.push_back(getPhdrIndex(Cmd->Location, PhdrName));
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001010 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +00001011 }
George Rimar31d842f2016-07-20 16:43:03 +00001012 return {};
Eugene Leviantbbe38602016-07-19 09:25:43 +00001013}
1014
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001015template <class ELFT>
Eugene Leviant2a942c42016-12-05 16:38:32 +00001016size_t LinkerScript<ELFT>::getPhdrIndex(const Twine &Loc, StringRef PhdrName) {
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001017 size_t I = 0;
1018 for (PhdrsCommand &Cmd : Opt.PhdrsCommands) {
1019 if (Cmd.Name == PhdrName)
1020 return I;
1021 ++I;
1022 }
Eugene Leviant2a942c42016-12-05 16:38:32 +00001023 error(Loc + ": section header '" + PhdrName + "' is not listed in PHDRS");
Rui Ueyama29c5a2a2016-07-26 00:27:36 +00001024 return 0;
1025}
1026
Rui Ueyama794366a2017-02-14 04:47:05 +00001027class elf::ScriptParser final : public ScriptLexer {
George Rimarc3794e52016-02-24 09:21:47 +00001028 typedef void (ScriptParser::*Handler)();
1029
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001030public:
Rui Ueyama22375f22016-11-25 18:51:54 +00001031 ScriptParser(MemoryBufferRef MB)
Rui Ueyama794366a2017-02-14 04:47:05 +00001032 : ScriptLexer(MB),
Rui Ueyama22375f22016-11-25 18:51:54 +00001033 IsUnderSysroot(isUnderSysroot(MB.getBufferIdentifier())) {}
George Rimarf23b2322016-02-19 10:45:45 +00001034
George Rimar20b65982016-08-31 09:08:26 +00001035 void readLinkerScript();
1036 void readVersionScript();
Rafael Espindolad0ebd842016-12-08 17:54:26 +00001037 void readDynamicList();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001038
1039private:
Rui Ueyama52a15092015-10-11 03:28:42 +00001040 void addFile(StringRef Path);
1041
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001042 void readAsNeeded();
Denis Protivensky90c50992015-10-08 06:48:38 +00001043 void readEntry();
George Rimar83f406c2015-10-19 17:35:12 +00001044 void readExtern();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001045 void readGroup();
Rui Ueyama31aa1f82015-10-11 01:31:55 +00001046 void readInclude();
Meador Ingeb8897442017-01-24 02:34:00 +00001047 void readMemory();
Rui Ueyamaee592822015-10-07 00:25:09 +00001048 void readOutput();
Davide Italiano9159ce92015-10-12 21:50:08 +00001049 void readOutputArch();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001050 void readOutputFormat();
Eugene Leviantbbe38602016-07-19 09:25:43 +00001051 void readPhdrs();
Davide Italiano68a39a62015-10-08 17:51:41 +00001052 void readSearchDir();
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001053 void readSections();
Rui Ueyama95769b42016-08-31 20:03:54 +00001054 void readVersion();
1055 void readVersionScriptCommand();
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001056
Rui Ueyama113cdec2016-07-24 23:05:57 +00001057 SymbolAssignment *readAssignment(StringRef Name);
George Rimare38cbab2016-09-26 19:22:50 +00001058 BytesDataCommand *readBytesDataCommand(StringRef Tok);
Rui Ueyama16068ae2016-11-19 18:05:56 +00001059 uint32_t readFill();
Rui Ueyama10416562016-08-04 02:03:27 +00001060 OutputSectionCommand *readOutputSectionDescription(StringRef OutSec);
Rui Ueyama16068ae2016-11-19 18:05:56 +00001061 uint32_t readOutputSectionFiller(StringRef Tok);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001062 std::vector<StringRef> readOutputSectionPhdrs();
George Rimara2496cb2016-08-30 09:46:59 +00001063 InputSectionDescription *readInputSectionDescription(StringRef Tok);
Eugene Leviantdb688452016-11-03 10:54:58 +00001064 StringMatcher readFilePatterns();
George Rimar07171f22016-09-21 15:56:44 +00001065 std::vector<SectionPattern> readInputSectionsList();
George Rimara2496cb2016-08-30 09:46:59 +00001066 InputSectionDescription *readInputSectionRules(StringRef FilePattern);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001067 unsigned readPhdrType();
George Rimarbe394db2016-09-16 20:21:55 +00001068 SortSectionPolicy readSortKind();
Petr Hoseka35e39c2016-08-16 01:11:16 +00001069 SymbolAssignment *readProvideHidden(bool Provide, bool Hidden);
Rafael Espindolac96da112016-11-01 11:30:45 +00001070 SymbolAssignment *readProvideOrAssignment(StringRef Tok);
George Rimar03fc0102016-07-28 07:18:23 +00001071 void readSort();
George Rimareefa7582016-08-04 09:29:31 +00001072 Expr readAssert();
Rui Ueyama708019c2016-07-24 18:19:40 +00001073
Rui Ueyama24e626c2017-01-26 02:58:19 +00001074 uint64_t readMemoryAssignment(StringRef, StringRef, StringRef);
1075 std::pair<uint32_t, uint32_t> readMemoryAttributes();
1076
Rui Ueyama708019c2016-07-24 18:19:40 +00001077 Expr readExpr();
1078 Expr readExpr1(Expr Lhs, int MinPrec);
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001079 StringRef readParenLiteral();
Rui Ueyama708019c2016-07-24 18:19:40 +00001080 Expr readPrimary();
1081 Expr readTernary(Expr Cond);
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001082 Expr readParenExpr();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001083
George Rimar20b65982016-08-31 09:08:26 +00001084 // For parsing version script.
Rui Ueyama12450b22016-11-18 06:30:09 +00001085 std::vector<SymbolVersion> readVersionExtern();
1086 void readAnonymousDeclaration();
Rui Ueyama95769b42016-08-31 20:03:54 +00001087 void readVersionDeclaration(StringRef VerStr);
Rui Ueyama12450b22016-11-18 06:30:09 +00001088 std::vector<SymbolVersion> readSymbols();
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001089 void readLocals();
George Rimar20b65982016-08-31 09:08:26 +00001090
Rui Ueyama07320e42016-04-20 20:13:41 +00001091 ScriptConfiguration &Opt = *ScriptConfig;
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001092 bool IsUnderSysroot;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001093};
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001094
Rafael Espindolad0ebd842016-12-08 17:54:26 +00001095void ScriptParser::readDynamicList() {
1096 expect("{");
1097 readAnonymousDeclaration();
1098 if (!atEOF())
1099 setError("EOF expected, but got " + next());
1100}
1101
George Rimar20b65982016-08-31 09:08:26 +00001102void ScriptParser::readVersionScript() {
Rui Ueyama95769b42016-08-31 20:03:54 +00001103 readVersionScriptCommand();
1104 if (!atEOF())
1105 setError("EOF expected, but got " + next());
1106}
1107
1108void ScriptParser::readVersionScriptCommand() {
Rui Ueyama83043f22016-10-17 16:01:53 +00001109 if (consume("{")) {
Rui Ueyama12450b22016-11-18 06:30:09 +00001110 readAnonymousDeclaration();
George Rimar20b65982016-08-31 09:08:26 +00001111 return;
1112 }
1113
Rui Ueyama95769b42016-08-31 20:03:54 +00001114 while (!atEOF() && !Error && peek() != "}") {
George Rimar20b65982016-08-31 09:08:26 +00001115 StringRef VerStr = next();
1116 if (VerStr == "{") {
Rui Ueyama95769b42016-08-31 20:03:54 +00001117 setError("anonymous version definition is used in "
1118 "combination with other version definitions");
George Rimar20b65982016-08-31 09:08:26 +00001119 return;
1120 }
1121 expect("{");
Rui Ueyama95769b42016-08-31 20:03:54 +00001122 readVersionDeclaration(VerStr);
George Rimar20b65982016-08-31 09:08:26 +00001123 }
1124}
1125
Rui Ueyama95769b42016-08-31 20:03:54 +00001126void ScriptParser::readVersion() {
1127 expect("{");
1128 readVersionScriptCommand();
1129 expect("}");
1130}
1131
George Rimar20b65982016-08-31 09:08:26 +00001132void ScriptParser::readLinkerScript() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001133 while (!atEOF()) {
1134 StringRef Tok = next();
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001135 if (Tok == ";")
1136 continue;
1137
Eugene Leviant20d03192016-09-16 15:30:47 +00001138 if (Tok == "ASSERT") {
1139 Opt.Commands.emplace_back(new AssertCommand(readAssert()));
1140 } else if (Tok == "ENTRY") {
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001141 readEntry();
1142 } else if (Tok == "EXTERN") {
1143 readExtern();
1144 } else if (Tok == "GROUP" || Tok == "INPUT") {
1145 readGroup();
1146 } else if (Tok == "INCLUDE") {
1147 readInclude();
Meador Ingeb8897442017-01-24 02:34:00 +00001148 } else if (Tok == "MEMORY") {
1149 readMemory();
Rui Ueyamaa27eecc2016-09-02 18:52:41 +00001150 } else if (Tok == "OUTPUT") {
1151 readOutput();
1152 } else if (Tok == "OUTPUT_ARCH") {
1153 readOutputArch();
1154 } else if (Tok == "OUTPUT_FORMAT") {
1155 readOutputFormat();
1156 } else if (Tok == "PHDRS") {
1157 readPhdrs();
1158 } else if (Tok == "SEARCH_DIR") {
1159 readSearchDir();
1160 } else if (Tok == "SECTIONS") {
1161 readSections();
1162 } else if (Tok == "VERSION") {
1163 readVersion();
Rafael Espindolac96da112016-11-01 11:30:45 +00001164 } else if (SymbolAssignment *Cmd = readProvideOrAssignment(Tok)) {
Eugene Leviant20d03192016-09-16 15:30:47 +00001165 Opt.Commands.emplace_back(Cmd);
Petr Hoseke5d3ca52016-08-31 15:31:17 +00001166 } else {
George Rimar57610422016-03-11 14:43:02 +00001167 setError("unknown directive: " + Tok);
Petr Hoseke5d3ca52016-08-31 15:31:17 +00001168 }
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001169 }
1170}
1171
Rui Ueyama717677a2016-02-11 21:17:59 +00001172void ScriptParser::addFile(StringRef S) {
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001173 if (IsUnderSysroot && S.startswith("/")) {
Justin Bogner5af16872016-10-17 06:08:48 +00001174 SmallString<128> PathData;
1175 StringRef Path = (Config->Sysroot + S).toStringRef(PathData);
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001176 if (sys::fs::exists(Path)) {
Justin Bogner5af16872016-10-17 06:08:48 +00001177 Driver->addFile(Saver.save(Path));
Simon Atanasyan16b0cc92015-11-26 05:53:00 +00001178 return;
1179 }
1180 }
1181
Rui Ueyamaf03f3cc2015-10-13 00:09:21 +00001182 if (sys::path::is_absolute(S)) {
Rui Ueyama52a15092015-10-11 03:28:42 +00001183 Driver->addFile(S);
1184 } else if (S.startswith("=")) {
1185 if (Config->Sysroot.empty())
1186 Driver->addFile(S.substr(1));
1187 else
1188 Driver->addFile(Saver.save(Config->Sysroot + "/" + S.substr(1)));
1189 } else if (S.startswith("-l")) {
Rui Ueyama21eecb42016-02-02 21:13:09 +00001190 Driver->addLibrary(S.substr(2));
Simon Atanasyana1b8fc32015-11-26 20:23:46 +00001191 } else if (sys::fs::exists(S)) {
1192 Driver->addFile(S);
Rui Ueyama52a15092015-10-11 03:28:42 +00001193 } else {
Rui Ueyama061f9282016-11-19 19:23:58 +00001194 if (Optional<std::string> Path = findFromSearchPaths(S))
1195 Driver->addFile(Saver.save(*Path));
Rui Ueyama025d59b2016-02-02 20:27:59 +00001196 else
Rui Ueyama061f9282016-11-19 19:23:58 +00001197 setError("unable to find " + S);
Rui Ueyama52a15092015-10-11 03:28:42 +00001198 }
1199}
1200
Rui Ueyama717677a2016-02-11 21:17:59 +00001201void ScriptParser::readAsNeeded() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001202 expect("(");
Rui Ueyama35da9b62015-10-11 20:59:12 +00001203 bool Orig = Config->AsNeeded;
1204 Config->AsNeeded = true;
Rui Ueyama83043f22016-10-17 16:01:53 +00001205 while (!Error && !consume(")"))
George Rimarcd574a52016-09-09 14:35:36 +00001206 addFile(unquote(next()));
Rui Ueyama35da9b62015-10-11 20:59:12 +00001207 Config->AsNeeded = Orig;
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001208}
1209
Rui Ueyama717677a2016-02-11 21:17:59 +00001210void ScriptParser::readEntry() {
Denis Protivensky90c50992015-10-08 06:48:38 +00001211 // -e <symbol> takes predecence over ENTRY(<symbol>).
1212 expect("(");
1213 StringRef Tok = next();
1214 if (Config->Entry.empty())
1215 Config->Entry = Tok;
1216 expect(")");
1217}
1218
Rui Ueyama717677a2016-02-11 21:17:59 +00001219void ScriptParser::readExtern() {
George Rimar83f406c2015-10-19 17:35:12 +00001220 expect("(");
Rui Ueyama83043f22016-10-17 16:01:53 +00001221 while (!Error && !consume(")"))
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001222 Config->Undefined.push_back(next());
George Rimar83f406c2015-10-19 17:35:12 +00001223}
1224
Rui Ueyama717677a2016-02-11 21:17:59 +00001225void ScriptParser::readGroup() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001226 expect("(");
Rui Ueyama83043f22016-10-17 16:01:53 +00001227 while (!Error && !consume(")")) {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001228 StringRef Tok = next();
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001229 if (Tok == "AS_NEEDED")
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001230 readAsNeeded();
Rui Ueyamaa2acc932016-08-05 01:25:45 +00001231 else
George Rimarcd574a52016-09-09 14:35:36 +00001232 addFile(unquote(Tok));
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001233 }
1234}
1235
Rui Ueyama717677a2016-02-11 21:17:59 +00001236void ScriptParser::readInclude() {
George Rimard4500652016-12-21 09:42:25 +00001237 StringRef Tok = unquote(next());
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001238
George Rimard4500652016-12-21 09:42:25 +00001239 // https://sourceware.org/binutils/docs/ld/File-Commands.html:
1240 // The file will be searched for in the current directory, and in any
1241 // directory specified with the -L option.
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001242 if (sys::fs::exists(Tok)) {
1243 if (Optional<MemoryBufferRef> MB = readFile(Tok))
1244 tokenize(*MB);
Rui Ueyama025d59b2016-02-02 20:27:59 +00001245 return;
1246 }
Rui Ueyamaec1c75e2017-01-09 01:42:02 +00001247 if (Optional<std::string> Path = findFromSearchPaths(Tok)) {
1248 if (Optional<MemoryBufferRef> MB = readFile(*Path))
1249 tokenize(*MB);
1250 return;
1251 }
1252 setError("cannot open " + Tok);
Rui Ueyama31aa1f82015-10-11 01:31:55 +00001253}
1254
Rui Ueyama717677a2016-02-11 21:17:59 +00001255void ScriptParser::readOutput() {
Rui Ueyamaee592822015-10-07 00:25:09 +00001256 // -o <file> takes predecence over OUTPUT(<file>).
1257 expect("(");
1258 StringRef Tok = next();
1259 if (Config->OutputFile.empty())
George Rimarcd574a52016-09-09 14:35:36 +00001260 Config->OutputFile = unquote(Tok);
Rui Ueyamaee592822015-10-07 00:25:09 +00001261 expect(")");
1262}
1263
Rui Ueyama717677a2016-02-11 21:17:59 +00001264void ScriptParser::readOutputArch() {
George Rimar4e01c3e2017-02-08 09:59:06 +00001265 // OUTPUT_ARCH is ignored for now.
Davide Italiano9159ce92015-10-12 21:50:08 +00001266 expect("(");
George Rimar4e01c3e2017-02-08 09:59:06 +00001267 while (!Error && !consume(")"))
1268 skip();
Davide Italiano9159ce92015-10-12 21:50:08 +00001269}
1270
Rui Ueyama717677a2016-02-11 21:17:59 +00001271void ScriptParser::readOutputFormat() {
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001272 // Error checking only for now.
1273 expect("(");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001274 skip();
Davide Italiano6836c612015-10-12 21:08:41 +00001275 StringRef Tok = next();
1276 if (Tok == ")")
George Rimar6c55f0e2016-09-08 08:20:30 +00001277 return;
Rui Ueyama025d59b2016-02-02 20:27:59 +00001278 if (Tok != ",") {
George Rimar57610422016-03-11 14:43:02 +00001279 setError("unexpected token: " + Tok);
Rui Ueyama025d59b2016-02-02 20:27:59 +00001280 return;
1281 }
Justin Bogner5424e7c2016-10-17 06:21:13 +00001282 skip();
Davide Italiano6836c612015-10-12 21:08:41 +00001283 expect(",");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001284 skip();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00001285 expect(")");
1286}
1287
Eugene Leviantbbe38602016-07-19 09:25:43 +00001288void ScriptParser::readPhdrs() {
1289 expect("{");
Rui Ueyama83043f22016-10-17 16:01:53 +00001290 while (!Error && !consume("}")) {
Eugene Leviantbbe38602016-07-19 09:25:43 +00001291 StringRef Tok = next();
Eugene Leviant56b21c82016-09-09 09:46:16 +00001292 Opt.PhdrsCommands.push_back(
1293 {Tok, PT_NULL, false, false, UINT_MAX, nullptr});
Eugene Leviantbbe38602016-07-19 09:25:43 +00001294 PhdrsCommand &PhdrCmd = Opt.PhdrsCommands.back();
1295
1296 PhdrCmd.Type = readPhdrType();
1297 do {
1298 Tok = next();
1299 if (Tok == ";")
1300 break;
1301 if (Tok == "FILEHDR")
1302 PhdrCmd.HasFilehdr = true;
1303 else if (Tok == "PHDRS")
1304 PhdrCmd.HasPhdrs = true;
Eugene Leviant56b21c82016-09-09 09:46:16 +00001305 else if (Tok == "AT")
1306 PhdrCmd.LMAExpr = readParenExpr();
Eugene Leviant865bf862016-07-21 10:43:25 +00001307 else if (Tok == "FLAGS") {
1308 expect("(");
Rafael Espindolaeb685cd2016-08-02 22:14:57 +00001309 // Passing 0 for the value of dot is a bit of a hack. It means that
1310 // we accept expressions like ".|1".
1311 PhdrCmd.Flags = readExpr()(0);
Eugene Leviant865bf862016-07-21 10:43:25 +00001312 expect(")");
1313 } else
Eugene Leviantbbe38602016-07-19 09:25:43 +00001314 setError("unexpected header attribute: " + Tok);
1315 } while (!Error);
1316 }
1317}
1318
Rui Ueyama717677a2016-02-11 21:17:59 +00001319void ScriptParser::readSearchDir() {
Davide Italiano68a39a62015-10-08 17:51:41 +00001320 expect("(");
Rui Ueyama86c5fb82016-09-08 23:26:54 +00001321 StringRef Tok = next();
Rui Ueyama6c7ad132016-09-02 19:20:33 +00001322 if (!Config->Nostdlib)
George Rimarcd574a52016-09-09 14:35:36 +00001323 Config->SearchPaths.push_back(unquote(Tok));
Davide Italiano68a39a62015-10-08 17:51:41 +00001324 expect(")");
1325}
1326
Rui Ueyama717677a2016-02-11 21:17:59 +00001327void ScriptParser::readSections() {
Eugene Leviante05336ff2016-09-14 08:32:36 +00001328 Opt.HasSections = true;
George Rimar18a30962016-11-28 10:11:10 +00001329 // -no-rosegment is used to avoid placing read only non-executable sections in
1330 // their own segment. We do the same if SECTIONS command is present in linker
1331 // script. See comment for computeFlags().
1332 Config->SingleRoRx = true;
1333
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001334 expect("{");
Rui Ueyama83043f22016-10-17 16:01:53 +00001335 while (!Error && !consume("}")) {
Rui Ueyama113cdec2016-07-24 23:05:57 +00001336 StringRef Tok = next();
Rafael Espindolac96da112016-11-01 11:30:45 +00001337 BaseCommand *Cmd = readProvideOrAssignment(Tok);
Eugene Leviantceabe802016-08-11 07:56:43 +00001338 if (!Cmd) {
1339 if (Tok == "ASSERT")
1340 Cmd = new AssertCommand(readAssert());
1341 else
1342 Cmd = readOutputSectionDescription(Tok);
Rui Ueyama708019c2016-07-24 18:19:40 +00001343 }
Rui Ueyama10416562016-08-04 02:03:27 +00001344 Opt.Commands.emplace_back(Cmd);
George Rimar652852c2016-04-16 10:10:32 +00001345 }
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001346}
1347
Rui Ueyama708019c2016-07-24 18:19:40 +00001348static int precedence(StringRef Op) {
1349 return StringSwitch<int>(Op)
Rui Ueyama0120e3f2016-09-23 18:06:51 +00001350 .Cases("*", "/", 5)
1351 .Cases("+", "-", 4)
1352 .Cases("<<", ">>", 3)
Rui Ueyama9c4ac5f2016-09-23 22:22:34 +00001353 .Cases("<", "<=", ">", ">=", "==", "!=", 2)
Rui Ueyama0120e3f2016-09-23 18:06:51 +00001354 .Cases("&", "|", 1)
Rui Ueyama708019c2016-07-24 18:19:40 +00001355 .Default(-1);
1356}
1357
Eugene Leviantdb688452016-11-03 10:54:58 +00001358StringMatcher ScriptParser::readFilePatterns() {
Rui Ueyama10416562016-08-04 02:03:27 +00001359 std::vector<StringRef> V;
Rui Ueyama83043f22016-10-17 16:01:53 +00001360 while (!Error && !consume(")"))
Rui Ueyama10416562016-08-04 02:03:27 +00001361 V.push_back(next());
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001362 return StringMatcher(V);
George Rimar0702c4e2016-07-29 15:32:46 +00001363}
1364
George Rimarbe394db2016-09-16 20:21:55 +00001365SortSectionPolicy ScriptParser::readSortKind() {
Rui Ueyama83043f22016-10-17 16:01:53 +00001366 if (consume("SORT") || consume("SORT_BY_NAME"))
George Rimarbe394db2016-09-16 20:21:55 +00001367 return SortSectionPolicy::Name;
Rui Ueyama83043f22016-10-17 16:01:53 +00001368 if (consume("SORT_BY_ALIGNMENT"))
George Rimarbe394db2016-09-16 20:21:55 +00001369 return SortSectionPolicy::Alignment;
Rui Ueyama83043f22016-10-17 16:01:53 +00001370 if (consume("SORT_BY_INIT_PRIORITY"))
George Rimarbe394db2016-09-16 20:21:55 +00001371 return SortSectionPolicy::Priority;
Rui Ueyama83043f22016-10-17 16:01:53 +00001372 if (consume("SORT_NONE"))
Rui Ueyamab2a0abd2016-09-16 21:14:55 +00001373 return SortSectionPolicy::None;
1374 return SortSectionPolicy::Default;
George Rimarbe394db2016-09-16 20:21:55 +00001375}
1376
George Rimar395281c2016-09-16 17:42:10 +00001377// Method reads a list of sequence of excluded files and section globs given in
1378// a following form: ((EXCLUDE_FILE(file_pattern+))? section_pattern+)+
1379// Example: *(.foo.1 EXCLUDE_FILE (*a.o) .foo.2 EXCLUDE_FILE (*b.o) .foo.3)
George Rimaraf03be12016-09-17 19:17:25 +00001380// The semantics of that is next:
1381// * Include .foo.1 from every file.
1382// * Include .foo.2 from every file but a.o
1383// * Include .foo.3 from every file but b.o
George Rimar07171f22016-09-21 15:56:44 +00001384std::vector<SectionPattern> ScriptParser::readInputSectionsList() {
1385 std::vector<SectionPattern> Ret;
George Rimar601e9892016-09-21 08:53:21 +00001386 while (!Error && peek() != ")") {
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001387 StringMatcher ExcludeFilePat;
Rui Ueyama83043f22016-10-17 16:01:53 +00001388 if (consume("EXCLUDE_FILE")) {
George Rimar395281c2016-09-16 17:42:10 +00001389 expect("(");
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001390 ExcludeFilePat = readFilePatterns();
George Rimar395281c2016-09-16 17:42:10 +00001391 }
1392
George Rimar601e9892016-09-21 08:53:21 +00001393 std::vector<StringRef> V;
1394 while (!Error && peek() != ")" && peek() != "EXCLUDE_FILE")
1395 V.push_back(next());
1396
1397 if (!V.empty())
Rui Ueyamaf91282e2016-11-03 17:57:38 +00001398 Ret.push_back({std::move(ExcludeFilePat), StringMatcher(V)});
George Rimar601e9892016-09-21 08:53:21 +00001399 else
1400 setError("section pattern is expected");
George Rimar395281c2016-09-16 17:42:10 +00001401 }
George Rimar07171f22016-09-21 15:56:44 +00001402 return Ret;
George Rimar395281c2016-09-16 17:42:10 +00001403}
1404
Rui Ueyamaf8f6f1e2016-11-18 07:03:56 +00001405// Reads contents of "SECTIONS" directive. That directive contains a
1406// list of glob patterns for input sections. The grammar is as follows.
1407//
1408// <patterns> ::= <section-list>
1409// | <sort> "(" <section-list> ")"
1410// | <sort> "(" <sort> "(" <section-list> ")" ")"
1411//
1412// <sort> ::= "SORT" | "SORT_BY_NAME" | "SORT_BY_ALIGNMENT"
1413// | "SORT_BY_INIT_PRIORITY" | "SORT_NONE"
1414//
1415// <section-list> is parsed by readInputSectionsList().
George Rimara2496cb2016-08-30 09:46:59 +00001416InputSectionDescription *
1417ScriptParser::readInputSectionRules(StringRef FilePattern) {
George Rimarc91930a2016-09-02 21:17:20 +00001418 auto *Cmd = new InputSectionDescription(FilePattern);
Davide Italiano0ed42b02016-07-25 21:47:13 +00001419 expect("(");
Rui Ueyamaf373dd72016-11-24 01:43:21 +00001420 while (!Error && !consume(")")) {
George Rimar07171f22016-09-21 15:56:44 +00001421 SortSectionPolicy Outer = readSortKind();
1422 SortSectionPolicy Inner = SortSectionPolicy::Default;
1423 std::vector<SectionPattern> V;
1424 if (Outer != SortSectionPolicy::Default) {
George Rimar350ece42016-08-03 08:35:59 +00001425 expect("(");
George Rimar07171f22016-09-21 15:56:44 +00001426 Inner = readSortKind();
1427 if (Inner != SortSectionPolicy::Default) {
1428 expect("(");
1429 V = readInputSectionsList();
1430 expect(")");
1431 } else {
1432 V = readInputSectionsList();
1433 }
George Rimar350ece42016-08-03 08:35:59 +00001434 expect(")");
1435 } else {
George Rimar07171f22016-09-21 15:56:44 +00001436 V = readInputSectionsList();
George Rimar350ece42016-08-03 08:35:59 +00001437 }
George Rimar0702c4e2016-07-29 15:32:46 +00001438
George Rimar07171f22016-09-21 15:56:44 +00001439 for (SectionPattern &Pat : V) {
1440 Pat.SortInner = Inner;
1441 Pat.SortOuter = Outer;
1442 }
1443
1444 std::move(V.begin(), V.end(), std::back_inserter(Cmd->SectionPatterns));
1445 }
Rui Ueyama10416562016-08-04 02:03:27 +00001446 return Cmd;
Davide Italianoe7282792016-07-27 01:44:01 +00001447}
1448
George Rimara2496cb2016-08-30 09:46:59 +00001449InputSectionDescription *
1450ScriptParser::readInputSectionDescription(StringRef Tok) {
George Rimar06598002016-07-28 21:51:30 +00001451 // Input section wildcard can be surrounded by KEEP.
1452 // https://sourceware.org/binutils/docs/ld/Input-Section-Keep.html#Input-Section-Keep
George Rimara2496cb2016-08-30 09:46:59 +00001453 if (Tok == "KEEP") {
George Rimar06598002016-07-28 21:51:30 +00001454 expect("(");
George Rimara2496cb2016-08-30 09:46:59 +00001455 StringRef FilePattern = next();
1456 InputSectionDescription *Cmd = readInputSectionRules(FilePattern);
George Rimar06598002016-07-28 21:51:30 +00001457 expect(")");
Eugene Leviantcf43f172016-10-05 09:36:59 +00001458 Opt.KeptSections.push_back(Cmd);
Rui Ueyama10416562016-08-04 02:03:27 +00001459 return Cmd;
George Rimar06598002016-07-28 21:51:30 +00001460 }
George Rimara2496cb2016-08-30 09:46:59 +00001461 return readInputSectionRules(Tok);
Davide Italiano0ed42b02016-07-25 21:47:13 +00001462}
1463
George Rimar03fc0102016-07-28 07:18:23 +00001464void ScriptParser::readSort() {
1465 expect("(");
1466 expect("CONSTRUCTORS");
1467 expect(")");
1468}
1469
George Rimareefa7582016-08-04 09:29:31 +00001470Expr ScriptParser::readAssert() {
1471 expect("(");
1472 Expr E = readExpr();
1473 expect(",");
George Rimarcd574a52016-09-09 14:35:36 +00001474 StringRef Msg = unquote(next());
George Rimareefa7582016-08-04 09:29:31 +00001475 expect(")");
1476 return [=](uint64_t Dot) {
George Rimar60f1fe82017-02-21 07:33:38 +00001477 if (!E(Dot))
George Rimareefa7582016-08-04 09:29:31 +00001478 error(Msg);
George Rimar60f1fe82017-02-21 07:33:38 +00001479 return Dot;
George Rimareefa7582016-08-04 09:29:31 +00001480 };
1481}
1482
Rui Ueyama25150e82016-09-06 17:46:43 +00001483// Reads a FILL(expr) command. We handle the FILL command as an
1484// alias for =fillexp section attribute, which is different from
1485// what GNU linkers do.
1486// https://sourceware.org/binutils/docs/ld/Output-Section-Data.html
Rui Ueyama16068ae2016-11-19 18:05:56 +00001487uint32_t ScriptParser::readFill() {
George Rimarff1f29e2016-09-06 13:51:57 +00001488 expect("(");
Rui Ueyama16068ae2016-11-19 18:05:56 +00001489 uint32_t V = readOutputSectionFiller(next());
George Rimarff1f29e2016-09-06 13:51:57 +00001490 expect(")");
1491 expect(";");
1492 return V;
1493}
1494
Rui Ueyama10416562016-08-04 02:03:27 +00001495OutputSectionCommand *
1496ScriptParser::readOutputSectionDescription(StringRef OutSec) {
George Rimar076fe152016-07-21 06:43:01 +00001497 OutputSectionCommand *Cmd = new OutputSectionCommand(OutSec);
Eugene Leviant2a942c42016-12-05 16:38:32 +00001498 Cmd->Location = getCurrentLocation();
George Rimar58e5c4d2016-07-25 08:29:46 +00001499
1500 // Read an address expression.
1501 // https://sourceware.org/binutils/docs/ld/Output-Section-Address.html#Output-Section-Address
1502 if (peek() != ":")
1503 Cmd->AddrExpr = readExpr();
1504
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001505 expect(":");
Davide Italiano246f6812016-07-22 03:36:24 +00001506
Rui Ueyama83043f22016-10-17 16:01:53 +00001507 if (consume("AT"))
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001508 Cmd->LMAExpr = readParenExpr();
Rui Ueyama83043f22016-10-17 16:01:53 +00001509 if (consume("ALIGN"))
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001510 Cmd->AlignExpr = readParenExpr();
Rui Ueyama83043f22016-10-17 16:01:53 +00001511 if (consume("SUBALIGN"))
George Rimardb24d9c2016-08-19 15:18:23 +00001512 Cmd->SubalignExpr = readParenExpr();
George Rimar630c6172016-07-26 18:06:29 +00001513
Davide Italiano246f6812016-07-22 03:36:24 +00001514 // Parse constraints.
Rui Ueyama83043f22016-10-17 16:01:53 +00001515 if (consume("ONLY_IF_RO"))
Rui Ueyamaefc40662016-07-25 22:00:10 +00001516 Cmd->Constraint = ConstraintKind::ReadOnly;
Rui Ueyama83043f22016-10-17 16:01:53 +00001517 if (consume("ONLY_IF_RW"))
Rui Ueyamaefc40662016-07-25 22:00:10 +00001518 Cmd->Constraint = ConstraintKind::ReadWrite;
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001519 expect("{");
Rui Ueyama8ec77e62016-04-21 22:00:51 +00001520
Rui Ueyama83043f22016-10-17 16:01:53 +00001521 while (!Error && !consume("}")) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001522 StringRef Tok = next();
George Rimar2fe07922017-01-31 08:50:11 +00001523 if (Tok == ";") {
George Rimar69750752017-02-01 09:14:22 +00001524 // Empty commands are allowed. Do nothing here.
George Rimar2fe07922017-01-31 08:50:11 +00001525 } else if (SymbolAssignment *Assignment = readProvideOrAssignment(Tok)) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001526 Cmd->Commands.emplace_back(Assignment);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001527 } else if (BytesDataCommand *Data = readBytesDataCommand(Tok)) {
George Rimare38cbab2016-09-26 19:22:50 +00001528 Cmd->Commands.emplace_back(Data);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001529 } else if (Tok == "ASSERT") {
1530 Cmd->Commands.emplace_back(new AssertCommand(readAssert()));
1531 expect(";");
George Rimar8e2eca22017-01-23 09:36:19 +00001532 } else if (Tok == "CONSTRUCTORS") {
1533 // CONSTRUCTORS is a keyword to make the linker recognize C++ ctors/dtors
1534 // by name. This is for very old file formats such as ECOFF/XCOFF.
1535 // For ELF, we should ignore.
Meador Ingeb2d99d62016-11-22 18:01:50 +00001536 } else if (Tok == "FILL") {
George Rimarff1f29e2016-09-06 13:51:57 +00001537 Cmd->Filler = readFill();
Meador Ingeb2d99d62016-11-22 18:01:50 +00001538 } else if (Tok == "SORT") {
George Rimar03fc0102016-07-28 07:18:23 +00001539 readSort();
Meador Ingeb2d99d62016-11-22 18:01:50 +00001540 } else if (peek() == "(") {
George Rimara2496cb2016-08-30 09:46:59 +00001541 Cmd->Commands.emplace_back(readInputSectionDescription(Tok));
Meador Ingeb2d99d62016-11-22 18:01:50 +00001542 } else {
Eugene Leviantceabe802016-08-11 07:56:43 +00001543 setError("unknown command " + Tok);
Meador Ingeb2d99d62016-11-22 18:01:50 +00001544 }
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001545 }
Meador Ingeb8897442017-01-24 02:34:00 +00001546
1547 if (consume(">"))
1548 Cmd->MemoryRegionName = next();
1549
George Rimar076fe152016-07-21 06:43:01 +00001550 Cmd->Phdrs = readOutputSectionPhdrs();
George Rimar4ebc5622016-09-23 13:29:20 +00001551
Rui Ueyama83043f22016-10-17 16:01:53 +00001552 if (consume("="))
George Rimar4ebc5622016-09-23 13:29:20 +00001553 Cmd->Filler = readOutputSectionFiller(next());
1554 else if (peek().startswith("="))
George Rimarff1f29e2016-09-06 13:51:57 +00001555 Cmd->Filler = readOutputSectionFiller(next().drop_front());
George Rimar4ebc5622016-09-23 13:29:20 +00001556
George Rimar7185a1a2017-01-17 15:32:12 +00001557 // Consume optional comma following output section command.
1558 consume(",");
1559
Rui Ueyama10416562016-08-04 02:03:27 +00001560 return Cmd;
Rui Ueyamaf71caa22016-07-29 06:14:07 +00001561}
Rui Ueyama8ec77e62016-04-21 22:00:51 +00001562
Rui Ueyama2c8f1f02016-08-29 22:01:21 +00001563// Read "=<number>" where <number> is an octal/decimal/hexadecimal number.
1564// https://sourceware.org/binutils/docs/ld/Output-Section-Fill.html
1565//
1566// ld.gold is not fully compatible with ld.bfd. ld.bfd handles
1567// hexstrings as blobs of arbitrary sizes, while ld.gold handles them
1568// as 32-bit big-endian values. We will do the same as ld.gold does
1569// because it's simpler than what ld.bfd does.
Rui Ueyama16068ae2016-11-19 18:05:56 +00001570uint32_t ScriptParser::readOutputSectionFiller(StringRef Tok) {
Rui Ueyama965827d2016-08-03 23:25:15 +00001571 uint32_t V;
Rui Ueyama16068ae2016-11-19 18:05:56 +00001572 if (!Tok.getAsInteger(0, V))
1573 return V;
1574 setError("invalid filler expression: " + Tok);
1575 return 0;
Denis Protivensky8e3b38a2015-11-12 09:52:08 +00001576}
1577
Petr Hoseka35e39c2016-08-16 01:11:16 +00001578SymbolAssignment *ScriptParser::readProvideHidden(bool Provide, bool Hidden) {
Eugene Levianta31c91b2016-07-22 07:38:40 +00001579 expect("(");
Rui Ueyama174e0a12016-07-29 00:29:25 +00001580 SymbolAssignment *Cmd = readAssignment(next());
Petr Hoseka35e39c2016-08-16 01:11:16 +00001581 Cmd->Provide = Provide;
Rui Ueyama174e0a12016-07-29 00:29:25 +00001582 Cmd->Hidden = Hidden;
Eugene Levianta31c91b2016-07-22 07:38:40 +00001583 expect(")");
1584 expect(";");
Rui Ueyama10416562016-08-04 02:03:27 +00001585 return Cmd;
Eugene Levianteda81a12016-07-12 06:39:48 +00001586}
1587
Rafael Espindolac96da112016-11-01 11:30:45 +00001588SymbolAssignment *ScriptParser::readProvideOrAssignment(StringRef Tok) {
Eugene Leviantceabe802016-08-11 07:56:43 +00001589 SymbolAssignment *Cmd = nullptr;
1590 if (peek() == "=" || peek() == "+=") {
1591 Cmd = readAssignment(Tok);
1592 expect(";");
1593 } else if (Tok == "PROVIDE") {
Petr Hoseka35e39c2016-08-16 01:11:16 +00001594 Cmd = readProvideHidden(true, false);
1595 } else if (Tok == "HIDDEN") {
1596 Cmd = readProvideHidden(false, true);
Eugene Leviantceabe802016-08-11 07:56:43 +00001597 } else if (Tok == "PROVIDE_HIDDEN") {
Petr Hoseka35e39c2016-08-16 01:11:16 +00001598 Cmd = readProvideHidden(true, true);
Eugene Leviantceabe802016-08-11 07:56:43 +00001599 }
1600 return Cmd;
1601}
1602
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001603static uint64_t getSymbolValue(const Twine &Loc, StringRef S, uint64_t Dot) {
George Rimar30835ea2016-07-28 21:08:56 +00001604 if (S == ".")
1605 return Dot;
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001606 return ScriptBase->getSymbolValue(Loc, S);
George Rimare32a3592016-08-10 07:59:34 +00001607}
1608
Rafael Espindola2f831dc2016-10-31 19:56:37 +00001609static bool isAbsolute(StringRef S) {
1610 if (S == ".")
1611 return false;
1612 return ScriptBase->isAbsolute(S);
1613}
1614
George Rimar30835ea2016-07-28 21:08:56 +00001615SymbolAssignment *ScriptParser::readAssignment(StringRef Name) {
1616 StringRef Op = next();
Eugene Leviantdb741e72016-09-07 07:08:43 +00001617 Expr E;
George Rimar30835ea2016-07-28 21:08:56 +00001618 assert(Op == "=" || Op == "+=");
Rui Ueyama83043f22016-10-17 16:01:53 +00001619 if (consume("ABSOLUTE")) {
Rui Ueyama731a66a2017-02-15 19:58:17 +00001620 E = readExpr();
Rui Ueyama009d1742016-11-18 06:49:09 +00001621 E.IsAbsolute = [] { return true; };
Eugene Leviantdb741e72016-09-07 07:08:43 +00001622 } else {
1623 E = readExpr();
1624 }
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001625 if (Op == "+=") {
1626 std::string Loc = getCurrentLocation();
1627 E = [=](uint64_t Dot) {
1628 return getSymbolValue(Loc, Name, Dot) + E(Dot);
1629 };
1630 }
George Rimar2ee2d2d2017-02-21 14:50:38 +00001631 return new SymbolAssignment(Name, E, getCurrentLocation());
George Rimar30835ea2016-07-28 21:08:56 +00001632}
1633
1634// This is an operator-precedence parser to parse a linker
1635// script expression.
Rui Ueyama731a66a2017-02-15 19:58:17 +00001636Expr ScriptParser::readExpr() {
1637 // Our lexer is context-aware. Set the in-expression bit so that
1638 // they apply different tokenization rules.
1639 bool Orig = InExpr;
1640 InExpr = true;
1641 Expr E = readExpr1(readPrimary(), 0);
1642 InExpr = Orig;
1643 return E;
1644}
George Rimar30835ea2016-07-28 21:08:56 +00001645
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001646static Expr combine(StringRef Op, Expr L, Expr R) {
George Rimarcc4d3e52017-02-01 09:01:16 +00001647 auto IsAbs = [=] { return L.IsAbsolute() && R.IsAbsolute(); };
1648 auto GetOutSec = [=] {
1649 const OutputSectionBase *S = L.Section();
1650 return S ? S : R.Section();
1651 };
1652
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001653 if (Op == "*")
1654 return [=](uint64_t Dot) { return L(Dot) * R(Dot); };
1655 if (Op == "/") {
1656 return [=](uint64_t Dot) -> uint64_t {
1657 uint64_t RHS = R(Dot);
1658 if (RHS == 0) {
1659 error("division by zero");
1660 return 0;
1661 }
1662 return L(Dot) / RHS;
1663 };
1664 }
1665 if (Op == "+")
George Rimarcc4d3e52017-02-01 09:01:16 +00001666 return {[=](uint64_t Dot) { return L(Dot) + R(Dot); }, IsAbs, GetOutSec};
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001667 if (Op == "-")
George Rimarcc4d3e52017-02-01 09:01:16 +00001668 return {[=](uint64_t Dot) { return L(Dot) - R(Dot); }, IsAbs, GetOutSec};
George Rimarc8ccd1f2016-09-23 13:13:55 +00001669 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); };
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001673 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); };
1679 if (Op == "<=")
1680 return [=](uint64_t Dot) { return L(Dot) <= R(Dot); };
1681 if (Op == "==")
1682 return [=](uint64_t Dot) { return L(Dot) == R(Dot); };
1683 if (Op == "!=")
1684 return [=](uint64_t Dot) { return L(Dot) != R(Dot); };
1685 if (Op == "&")
1686 return [=](uint64_t Dot) { return L(Dot) & R(Dot); };
Rafael Espindolacc3dd622016-08-22 21:33:35 +00001687 if (Op == "|")
1688 return [=](uint64_t Dot) { return L(Dot) | R(Dot); };
Rui Ueyama36c1cd22016-08-05 01:04:59 +00001689 llvm_unreachable("invalid operator");
1690}
1691
Rui Ueyama708019c2016-07-24 18:19:40 +00001692// This is a part of the operator-precedence parser. This function
1693// assumes that the remaining token stream starts with an operator.
1694Expr ScriptParser::readExpr1(Expr Lhs, int MinPrec) {
1695 while (!atEOF() && !Error) {
1696 // Read an operator and an expression.
Rui Ueyama46247b82016-11-18 06:49:07 +00001697 if (consume("?"))
Rui Ueyama708019c2016-07-24 18:19:40 +00001698 return readTernary(Lhs);
Rui Ueyama46247b82016-11-18 06:49:07 +00001699 StringRef Op1 = peek();
Rui Ueyama708019c2016-07-24 18:19:40 +00001700 if (precedence(Op1) < MinPrec)
Eugene Levianteda81a12016-07-12 06:39:48 +00001701 break;
Justin Bogner5424e7c2016-10-17 06:21:13 +00001702 skip();
Rui Ueyama708019c2016-07-24 18:19:40 +00001703 Expr Rhs = readPrimary();
1704
1705 // Evaluate the remaining part of the expression first if the
1706 // next operator has greater precedence than the previous one.
1707 // For example, if we have read "+" and "3", and if the next
1708 // operator is "*", then we'll evaluate 3 * ... part first.
1709 while (!atEOF()) {
1710 StringRef Op2 = peek();
1711 if (precedence(Op2) <= precedence(Op1))
1712 break;
1713 Rhs = readExpr1(Rhs, precedence(Op2));
1714 }
1715
1716 Lhs = combine(Op1, Lhs, Rhs);
Eugene Levianteda81a12016-07-12 06:39:48 +00001717 }
Rui Ueyama708019c2016-07-24 18:19:40 +00001718 return Lhs;
1719}
1720
1721uint64_t static getConstant(StringRef S) {
Michael J. Spencere2cc07b2016-08-17 02:10:51 +00001722 if (S == "COMMONPAGESIZE")
Rui Ueyama708019c2016-07-24 18:19:40 +00001723 return Target->PageSize;
Michael J. Spencere2cc07b2016-08-17 02:10:51 +00001724 if (S == "MAXPAGESIZE")
Petr Hosek997f8832016-09-28 15:20:47 +00001725 return Config->MaxPageSize;
Rui Ueyama708019c2016-07-24 18:19:40 +00001726 error("unknown constant: " + S);
1727 return 0;
1728}
1729
Rui Ueyama626e0b02016-09-02 18:19:00 +00001730// Parses Tok as an integer. Returns true if successful.
1731// It recognizes hexadecimal (prefixed with "0x" or suffixed with "H")
1732// and decimal numbers. Decimal numbers may have "K" (kilo) or
1733// "M" (mega) prefixes.
George Rimar9f2f7ad2016-09-02 16:01:42 +00001734static bool readInteger(StringRef Tok, uint64_t &Result) {
Rui Ueyama46247b82016-11-18 06:49:07 +00001735 // Negative number
Simon Atanasyaneaeafb22016-09-02 21:54:35 +00001736 if (Tok.startswith("-")) {
1737 if (!readInteger(Tok.substr(1), Result))
1738 return false;
1739 Result = -Result;
1740 return true;
1741 }
Rui Ueyama46247b82016-11-18 06:49:07 +00001742
1743 // Hexadecimal
George Rimar9f2f7ad2016-09-02 16:01:42 +00001744 if (Tok.startswith_lower("0x"))
1745 return !Tok.substr(2).getAsInteger(16, Result);
1746 if (Tok.endswith_lower("H"))
1747 return !Tok.drop_back().getAsInteger(16, Result);
1748
Rui Ueyama46247b82016-11-18 06:49:07 +00001749 // Decimal
George Rimar9f2f7ad2016-09-02 16:01:42 +00001750 int Suffix = 1;
1751 if (Tok.endswith_lower("K")) {
1752 Suffix = 1024;
1753 Tok = Tok.drop_back();
1754 } else if (Tok.endswith_lower("M")) {
1755 Suffix = 1024 * 1024;
1756 Tok = Tok.drop_back();
1757 }
1758 if (Tok.getAsInteger(10, Result))
1759 return false;
1760 Result *= Suffix;
1761 return true;
1762}
1763
George Rimare38cbab2016-09-26 19:22:50 +00001764BytesDataCommand *ScriptParser::readBytesDataCommand(StringRef Tok) {
1765 int Size = StringSwitch<unsigned>(Tok)
1766 .Case("BYTE", 1)
1767 .Case("SHORT", 2)
1768 .Case("LONG", 4)
1769 .Case("QUAD", 8)
1770 .Default(-1);
1771 if (Size == -1)
1772 return nullptr;
1773
Meador Inge95c7d8d2016-12-08 23:21:30 +00001774 return new BytesDataCommand(readParenExpr(), Size);
George Rimare38cbab2016-09-26 19:22:50 +00001775}
1776
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001777StringRef ScriptParser::readParenLiteral() {
1778 expect("(");
1779 StringRef Tok = next();
1780 expect(")");
1781 return Tok;
1782}
1783
Rui Ueyama708019c2016-07-24 18:19:40 +00001784Expr ScriptParser::readPrimary() {
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001785 if (peek() == "(")
1786 return readParenExpr();
Rui Ueyama708019c2016-07-24 18:19:40 +00001787
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001788 StringRef Tok = next();
Rui Ueyamab5f1c3e2016-12-01 04:36:49 +00001789 std::string Location = getCurrentLocation();
Rui Ueyama708019c2016-07-24 18:19:40 +00001790
Simon Atanasyaneaeafb22016-09-02 21:54:35 +00001791 if (Tok == "~") {
1792 Expr E = readPrimary();
1793 return [=](uint64_t Dot) { return ~E(Dot); };
1794 }
1795 if (Tok == "-") {
1796 Expr E = readPrimary();
1797 return [=](uint64_t Dot) { return -E(Dot); };
1798 }
1799
Rui Ueyama708019c2016-07-24 18:19:40 +00001800 // Built-in functions are parsed here.
1801 // https://sourceware.org/binutils/docs/ld/Builtin-Functions.html.
George Rimar96659df2016-08-30 09:54:01 +00001802 if (Tok == "ADDR") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001803 StringRef Name = readParenLiteral();
Eugene Levianted30ce72016-11-28 09:58:04 +00001804 return {[=](uint64_t Dot) {
1805 return ScriptBase->getOutputSection(Location, Name)->Addr;
1806 },
1807 [=] { return false; },
1808 [=] { return ScriptBase->getOutputSection(Location, Name); }};
George Rimar96659df2016-08-30 09:54:01 +00001809 }
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001810 if (Tok == "LOADADDR") {
1811 StringRef Name = readParenLiteral();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001812 return [=](uint64_t Dot) {
Eugene Levianted30ce72016-11-28 09:58:04 +00001813 return ScriptBase->getOutputSection(Location, Name)->getLMA();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001814 };
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001815 }
George Rimareefa7582016-08-04 09:29:31 +00001816 if (Tok == "ASSERT")
1817 return readAssert();
Rui Ueyama708019c2016-07-24 18:19:40 +00001818 if (Tok == "ALIGN") {
Rui Ueyama5d804dc2016-12-16 18:19:35 +00001819 expect("(");
1820 Expr E = readExpr();
1821 if (consume(",")) {
1822 Expr E2 = readExpr();
1823 expect(")");
1824 return [=](uint64_t Dot) { return alignTo(E(Dot), E2(Dot)); };
1825 }
1826 expect(")");
Rui Ueyama708019c2016-07-24 18:19:40 +00001827 return [=](uint64_t Dot) { return alignTo(Dot, E(Dot)); };
1828 }
1829 if (Tok == "CONSTANT") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001830 StringRef Name = readParenLiteral();
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001831 return [=](uint64_t Dot) { return getConstant(Name); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001832 }
George Rimarf34f45f2016-09-23 13:17:23 +00001833 if (Tok == "DEFINED") {
Rui Ueyama0ee25a62016-11-17 03:52:14 +00001834 StringRef Name = readParenLiteral();
1835 return [=](uint64_t Dot) { return ScriptBase->isDefined(Name) ? 1 : 0; };
George Rimarf34f45f2016-09-23 13:17:23 +00001836 }
Rafael Espindola54c145c2016-07-28 18:16:24 +00001837 if (Tok == "SEGMENT_START") {
1838 expect("(");
Justin Bogner5424e7c2016-10-17 06:21:13 +00001839 skip();
Rafael Espindola54c145c2016-07-28 18:16:24 +00001840 expect(",");
George Rimar8c658bf2016-09-17 18:14:56 +00001841 Expr E = readExpr();
Rafael Espindola54c145c2016-07-28 18:16:24 +00001842 expect(")");
George Rimar8c658bf2016-09-17 18:14:56 +00001843 return [=](uint64_t Dot) { return E(Dot); };
Rafael Espindola54c145c2016-07-28 18:16:24 +00001844 }
Rui Ueyama708019c2016-07-24 18:19:40 +00001845 if (Tok == "DATA_SEGMENT_ALIGN") {
1846 expect("(");
1847 Expr E = readExpr();
1848 expect(",");
1849 readExpr();
1850 expect(")");
Rui Ueyamaf7791bb2016-07-26 19:34:10 +00001851 return [=](uint64_t Dot) { return alignTo(Dot, E(Dot)); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001852 }
1853 if (Tok == "DATA_SEGMENT_END") {
1854 expect("(");
1855 expect(".");
1856 expect(")");
1857 return [](uint64_t Dot) { return Dot; };
1858 }
George Rimar276b4e62016-07-26 17:58:44 +00001859 // GNU linkers implements more complicated logic to handle
1860 // DATA_SEGMENT_RELRO_END. We instead ignore the arguments and just align to
1861 // the next page boundary for simplicity.
1862 if (Tok == "DATA_SEGMENT_RELRO_END") {
1863 expect("(");
Rafael Espindola97bdc722016-09-14 19:14:01 +00001864 readExpr();
George Rimar276b4e62016-07-26 17:58:44 +00001865 expect(",");
1866 readExpr();
1867 expect(")");
1868 return [](uint64_t Dot) { return alignTo(Dot, Target->PageSize); };
1869 }
George Rimar9e694502016-07-29 16:18:47 +00001870 if (Tok == "SIZEOF") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001871 StringRef Name = readParenLiteral();
Rui Ueyamaedf75e72016-11-17 20:27:10 +00001872 return [=](uint64_t Dot) { return ScriptBase->getOutputSectionSize(Name); };
George Rimar9e694502016-07-29 16:18:47 +00001873 }
Eugene Leviant36fac7f2016-09-08 09:08:30 +00001874 if (Tok == "ALIGNOF") {
Eugene Leviantb71d6f72016-10-06 09:39:28 +00001875 StringRef Name = readParenLiteral();
Eugene Leviantafaa9342016-11-16 09:49:39 +00001876 return [=](uint64_t Dot) {
Eugene Levianted30ce72016-11-28 09:58:04 +00001877 return ScriptBase->getOutputSection(Location, Name)->Addralign;
Eugene Leviantafaa9342016-11-16 09:49:39 +00001878 };
Eugene Leviant36fac7f2016-09-08 09:08:30 +00001879 }
George Rimare32a3592016-08-10 07:59:34 +00001880 if (Tok == "SIZEOF_HEADERS")
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001881 return [=](uint64_t Dot) { return ScriptBase->getHeaderSize(); };
Rui Ueyama708019c2016-07-24 18:19:40 +00001882
George Rimar9f2f7ad2016-09-02 16:01:42 +00001883 // Tok is a literal number.
1884 uint64_t V;
1885 if (readInteger(Tok, V))
Rafael Espindolab0de56b2016-10-31 21:36:23 +00001886 return [=](uint64_t Dot) { return V; };
George Rimar9f2f7ad2016-09-02 16:01:42 +00001887
1888 // Tok is a symbol name.
1889 if (Tok != "." && !isValidCIdentifier(Tok))
1890 setError("malformed number: " + Tok);
Eugene Leviantf6aeed32016-12-22 13:13:12 +00001891 return {[=](uint64_t Dot) { return getSymbolValue(Location, Tok, Dot); },
Rui Ueyama009d1742016-11-18 06:49:09 +00001892 [=] { return isAbsolute(Tok); },
1893 [=] { return ScriptBase->getSymbolSection(Tok); }};
Rui Ueyama708019c2016-07-24 18:19:40 +00001894}
1895
1896Expr ScriptParser::readTernary(Expr Cond) {
Rui Ueyama708019c2016-07-24 18:19:40 +00001897 Expr L = readExpr();
1898 expect(":");
1899 Expr R = readExpr();
1900 return [=](uint64_t Dot) { return Cond(Dot) ? L(Dot) : R(Dot); };
1901}
1902
Rui Ueyama6ad7dfc2016-08-17 18:59:16 +00001903Expr ScriptParser::readParenExpr() {
1904 expect("(");
1905 Expr E = readExpr();
1906 expect(")");
1907 return E;
1908}
1909
Eugene Leviantbbe38602016-07-19 09:25:43 +00001910std::vector<StringRef> ScriptParser::readOutputSectionPhdrs() {
1911 std::vector<StringRef> Phdrs;
1912 while (!Error && peek().startswith(":")) {
1913 StringRef Tok = next();
George Rimarda841c12016-11-14 10:03:54 +00001914 Phdrs.push_back((Tok.size() == 1) ? next() : Tok.substr(1));
Eugene Leviantbbe38602016-07-19 09:25:43 +00001915 }
1916 return Phdrs;
1917}
1918
George Rimar95dd7182016-10-18 10:49:50 +00001919// Read a program header type name. The next token must be a
1920// name of a program header type or a constant (e.g. "0x3").
Eugene Leviantbbe38602016-07-19 09:25:43 +00001921unsigned ScriptParser::readPhdrType() {
Eugene Leviantbbe38602016-07-19 09:25:43 +00001922 StringRef Tok = next();
George Rimar95dd7182016-10-18 10:49:50 +00001923 uint64_t Val;
1924 if (readInteger(Tok, Val))
1925 return Val;
1926
Rui Ueyamab0f6c592016-07-20 19:36:38 +00001927 unsigned Ret = StringSwitch<unsigned>(Tok)
George Rimar6c55f0e2016-09-08 08:20:30 +00001928 .Case("PT_NULL", PT_NULL)
1929 .Case("PT_LOAD", PT_LOAD)
1930 .Case("PT_DYNAMIC", PT_DYNAMIC)
1931 .Case("PT_INTERP", PT_INTERP)
1932 .Case("PT_NOTE", PT_NOTE)
1933 .Case("PT_SHLIB", PT_SHLIB)
1934 .Case("PT_PHDR", PT_PHDR)
1935 .Case("PT_TLS", PT_TLS)
1936 .Case("PT_GNU_EH_FRAME", PT_GNU_EH_FRAME)
1937 .Case("PT_GNU_STACK", PT_GNU_STACK)
1938 .Case("PT_GNU_RELRO", PT_GNU_RELRO)
George Rimar270173f2016-10-14 13:02:22 +00001939 .Case("PT_OPENBSD_RANDOMIZE", PT_OPENBSD_RANDOMIZE)
George Rimarcc6e5672016-10-14 10:34:36 +00001940 .Case("PT_OPENBSD_WXNEEDED", PT_OPENBSD_WXNEEDED)
George Rimara2a32c22016-12-06 17:57:42 +00001941 .Case("PT_OPENBSD_BOOTDATA", PT_OPENBSD_BOOTDATA)
George Rimar6c55f0e2016-09-08 08:20:30 +00001942 .Default(-1);
Eugene Leviantbbe38602016-07-19 09:25:43 +00001943
Rui Ueyamab0f6c592016-07-20 19:36:38 +00001944 if (Ret == (unsigned)-1) {
1945 setError("invalid program header type: " + Tok);
1946 return PT_NULL;
1947 }
1948 return Ret;
Eugene Leviantbbe38602016-07-19 09:25:43 +00001949}
1950
Rui Ueyama12450b22016-11-18 06:30:09 +00001951// Reads a list of symbols, e.g. "{ global: foo; bar; local: *; };".
1952void ScriptParser::readAnonymousDeclaration() {
1953 // Read global symbols first. "global:" is default, so if there's
1954 // no label, we assume global symbols.
Rafael Espindola45242682017-02-03 13:24:01 +00001955 if (peek() != "local") {
1956 if (consume("global"))
1957 expect(":");
Peter Collingbourne904c5ed2017-02-13 18:31:12 +00001958 for (SymbolVersion V : readSymbols())
1959 Config->VersionScriptGlobals.push_back(V);
Rafael Espindola45242682017-02-03 13:24:01 +00001960 }
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001961 readLocals();
Rui Ueyama12450b22016-11-18 06:30:09 +00001962 expect("}");
1963 expect(";");
1964}
1965
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001966void ScriptParser::readLocals() {
Rafael Espindola45242682017-02-03 13:24:01 +00001967 if (!consume("local"))
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001968 return;
Rafael Espindola45242682017-02-03 13:24:01 +00001969 expect(":");
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001970 std::vector<SymbolVersion> Locals = readSymbols();
1971 for (SymbolVersion V : Locals) {
1972 if (V.Name == "*") {
1973 Config->DefaultSymbolVersion = VER_NDX_LOCAL;
1974 continue;
1975 }
1976 Config->VersionScriptLocals.push_back(V);
1977 }
1978}
1979
Rui Ueyama12450b22016-11-18 06:30:09 +00001980// Reads a list of symbols, e.g. "VerStr { global: foo; bar; local: *; };".
Rui Ueyama95769b42016-08-31 20:03:54 +00001981void ScriptParser::readVersionDeclaration(StringRef VerStr) {
George Rimar20b65982016-08-31 09:08:26 +00001982 // Identifiers start at 2 because 0 and 1 are reserved
1983 // for VER_NDX_LOCAL and VER_NDX_GLOBAL constants.
Rui Ueyamada805c42016-11-17 03:39:21 +00001984 uint16_t VersionId = Config->VersionDefinitions.size() + 2;
George Rimar20b65982016-08-31 09:08:26 +00001985 Config->VersionDefinitions.push_back({VerStr, VersionId});
1986
Rui Ueyama12450b22016-11-18 06:30:09 +00001987 // Read global symbols.
Rafael Espindola45242682017-02-03 13:24:01 +00001988 if (peek() != "local") {
1989 if (consume("global"))
1990 expect(":");
Rui Ueyama12450b22016-11-18 06:30:09 +00001991 Config->VersionDefinitions.back().Globals = readSymbols();
Rafael Espindola45242682017-02-03 13:24:01 +00001992 }
Rafael Espindolae999ddb2017-01-10 16:37:24 +00001993 readLocals();
George Rimar20b65982016-08-31 09:08:26 +00001994 expect("}");
1995
Rui Ueyama12450b22016-11-18 06:30:09 +00001996 // Each version may have a parent version. For example, "Ver2"
1997 // defined as "Ver2 { global: foo; local: *; } Ver1;" has "Ver1"
1998 // as a parent. This version hierarchy is, probably against your
1999 // instinct, purely for hint; the runtime doesn't care about it
2000 // at all. In LLD, we simply ignore it.
2001 if (peek() != ";")
Justin Bogner5424e7c2016-10-17 06:21:13 +00002002 skip();
George Rimar20b65982016-08-31 09:08:26 +00002003 expect(";");
2004}
2005
Rui Ueyama12450b22016-11-18 06:30:09 +00002006// Reads a list of symbols for a versions cript.
2007std::vector<SymbolVersion> ScriptParser::readSymbols() {
2008 std::vector<SymbolVersion> Ret;
George Rimare0fc2422016-11-16 17:59:10 +00002009 for (;;) {
Rafael Espindola1ef90d22016-12-09 16:44:05 +00002010 if (consume("extern")) {
Rui Ueyama12450b22016-11-18 06:30:09 +00002011 for (SymbolVersion V : readVersionExtern())
2012 Ret.push_back(V);
Rafael Espindola1ef90d22016-12-09 16:44:05 +00002013 continue;
2014 }
George Rimare0fc2422016-11-16 17:59:10 +00002015
Dmitry Mikulinf3965c02017-02-07 19:50:47 +00002016 if (peek() == "}" || (peek() == "local" && peek(1) == ":") || Error)
Rui Ueyama12450b22016-11-18 06:30:09 +00002017 break;
Rui Ueyama0ee25a62016-11-17 03:52:14 +00002018 StringRef Tok = next();
Rui Ueyama12450b22016-11-18 06:30:09 +00002019 Ret.push_back({unquote(Tok), false, hasWildcard(Tok)});
George Rimare0fc2422016-11-16 17:59:10 +00002020 expect(";");
2021 }
Rui Ueyama12450b22016-11-18 06:30:09 +00002022 return Ret;
George Rimare0fc2422016-11-16 17:59:10 +00002023}
2024
Rui Ueyama12450b22016-11-18 06:30:09 +00002025// Reads an "extern C++" directive, e.g.,
2026// "extern "C++" { ns::*; "f(int, double)"; };"
2027std::vector<SymbolVersion> ScriptParser::readVersionExtern() {
Rafael Espindola7e714152016-12-08 17:26:53 +00002028 StringRef Tok = next();
2029 bool IsCXX = Tok == "\"C++\"";
2030 if (!IsCXX && Tok != "\"C\"")
Rafael Espindolad0ebd842016-12-08 17:54:26 +00002031 setError("Unknown language");
George Rimar20b65982016-08-31 09:08:26 +00002032 expect("{");
2033
Rui Ueyama12450b22016-11-18 06:30:09 +00002034 std::vector<SymbolVersion> Ret;
Rui Ueyama0ee25a62016-11-17 03:52:14 +00002035 while (!Error && peek() != "}") {
2036 StringRef Tok = next();
2037 bool HasWildcard = !Tok.startswith("\"") && hasWildcard(Tok);
Rafael Espindola7e714152016-12-08 17:26:53 +00002038 Ret.push_back({unquote(Tok), IsCXX, HasWildcard});
George Rimar20b65982016-08-31 09:08:26 +00002039 expect(";");
2040 }
2041
2042 expect("}");
2043 expect(";");
Rui Ueyama12450b22016-11-18 06:30:09 +00002044 return Ret;
George Rimar20b65982016-08-31 09:08:26 +00002045}
2046
Rui Ueyama24e626c2017-01-26 02:58:19 +00002047uint64_t ScriptParser::readMemoryAssignment(
2048 StringRef S1, StringRef S2, StringRef S3) {
2049 if (!(consume(S1) || consume(S2) || consume(S3))) {
2050 setError("expected one of: " + S1 + ", " + S2 + ", or " + S3);
2051 return 0;
2052 }
2053 expect("=");
2054
2055 // TODO: Fully support constant expressions.
2056 uint64_t Val;
2057 if (!readInteger(next(), Val))
2058 setError("nonconstant expression for "+ S1);
2059 return Val;
2060}
2061
2062// Parse the MEMORY command as specified in:
2063// https://sourceware.org/binutils/docs/ld/MEMORY.html
2064//
2065// MEMORY { name [(attr)] : ORIGIN = origin, LENGTH = len ... }
Meador Ingeb8897442017-01-24 02:34:00 +00002066void ScriptParser::readMemory() {
2067 expect("{");
2068 while (!Error && !consume("}")) {
2069 StringRef Name = next();
Rui Ueyama24e626c2017-01-26 02:58:19 +00002070
Meador Ingeb8897442017-01-24 02:34:00 +00002071 uint32_t Flags = 0;
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002072 uint32_t NegFlags = 0;
Meador Ingeb8897442017-01-24 02:34:00 +00002073 if (consume("(")) {
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002074 std::tie(Flags, NegFlags) = readMemoryAttributes();
Meador Ingeb8897442017-01-24 02:34:00 +00002075 expect(")");
2076 }
2077 expect(":");
2078
Rui Ueyama24e626c2017-01-26 02:58:19 +00002079 uint64_t Origin = readMemoryAssignment("ORIGIN", "org", "o");
Meador Ingeb8897442017-01-24 02:34:00 +00002080 expect(",");
Rui Ueyama24e626c2017-01-26 02:58:19 +00002081 uint64_t Length = readMemoryAssignment("LENGTH", "len", "l");
Meador Ingeb8897442017-01-24 02:34:00 +00002082
Meador Ingeb8897442017-01-24 02:34:00 +00002083 // Add the memory region to the region map (if it doesn't already exist).
2084 auto It = Opt.MemoryRegions.find(Name);
2085 if (It != Opt.MemoryRegions.end())
2086 setError("region '" + Name + "' already defined");
2087 else
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002088 Opt.MemoryRegions[Name] = {Name, Origin, Length, Origin, Flags, NegFlags};
Meador Ingeb8897442017-01-24 02:34:00 +00002089 }
2090}
2091
2092// This function parses the attributes used to match against section
2093// flags when placing output sections in a memory region. These flags
2094// are only used when an explicit memory region name is not used.
2095std::pair<uint32_t, uint32_t> ScriptParser::readMemoryAttributes() {
2096 uint32_t Flags = 0;
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002097 uint32_t NegFlags = 0;
Meador Ingeb8897442017-01-24 02:34:00 +00002098 bool Invert = false;
Rui Ueyama481ac992017-01-26 02:58:39 +00002099
2100 for (char C : next().lower()) {
Meador Ingeb8897442017-01-24 02:34:00 +00002101 uint32_t Flag = 0;
2102 if (C == '!')
2103 Invert = !Invert;
Rui Ueyama481ac992017-01-26 02:58:39 +00002104 else if (C == 'w')
Meador Ingeb8897442017-01-24 02:34:00 +00002105 Flag = SHF_WRITE;
Rui Ueyama481ac992017-01-26 02:58:39 +00002106 else if (C == 'x')
Meador Ingeb8897442017-01-24 02:34:00 +00002107 Flag = SHF_EXECINSTR;
Rui Ueyama481ac992017-01-26 02:58:39 +00002108 else if (C == 'a')
Meador Ingeb8897442017-01-24 02:34:00 +00002109 Flag = SHF_ALLOC;
Rui Ueyama481ac992017-01-26 02:58:39 +00002110 else if (C != 'r')
Meador Ingeb8897442017-01-24 02:34:00 +00002111 setError("invalid memory region attribute");
Rui Ueyama481ac992017-01-26 02:58:39 +00002112
Meador Ingeb8897442017-01-24 02:34:00 +00002113 if (Invert)
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002114 NegFlags |= Flag;
Meador Ingeb8897442017-01-24 02:34:00 +00002115 else
2116 Flags |= Flag;
2117 }
Rui Ueyama8a8a9532017-01-26 02:58:59 +00002118 return {Flags, NegFlags};
Meador Ingeb8897442017-01-24 02:34:00 +00002119}
2120
Rui Ueyama07320e42016-04-20 20:13:41 +00002121void elf::readLinkerScript(MemoryBufferRef MB) {
Rui Ueyama22375f22016-11-25 18:51:54 +00002122 ScriptParser(MB).readLinkerScript();
George Rimar20b65982016-08-31 09:08:26 +00002123}
2124
2125void elf::readVersionScript(MemoryBufferRef MB) {
Rui Ueyama22375f22016-11-25 18:51:54 +00002126 ScriptParser(MB).readVersionScript();
Rui Ueyamaf7c5fbb2015-09-30 17:23:26 +00002127}
Rui Ueyama1ebc8ed2016-02-12 21:47:28 +00002128
Rafael Espindolad0ebd842016-12-08 17:54:26 +00002129void elf::readDynamicList(MemoryBufferRef MB) {
2130 ScriptParser(MB).readDynamicList();
2131}
2132
Rui Ueyama07320e42016-04-20 20:13:41 +00002133template class elf::LinkerScript<ELF32LE>;
2134template class elf::LinkerScript<ELF32BE>;
2135template class elf::LinkerScript<ELF64LE>;
2136template class elf::LinkerScript<ELF64BE>;