| 1 | //===- ConcatOutputSection.cpp --------------------------------------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | |
| 9 | #include "ConcatOutputSection.h" |
| 10 | #include "Config.h" |
| 11 | #include "OutputSegment.h" |
| 12 | #include "SymbolTable.h" |
| 13 | #include "Symbols.h" |
| 14 | #include "SyntheticSections.h" |
| 15 | #include "Target.h" |
| 16 | #include "lld/Common/CommonLinkerContext.h" |
| 17 | #include "llvm/BinaryFormat/MachO.h" |
| 18 | |
| 19 | using namespace llvm; |
| 20 | using namespace llvm::MachO; |
| 21 | using namespace lld; |
| 22 | using namespace lld::macho; |
| 23 | |
| 24 | MapVector<NamePair, ConcatOutputSection *> macho::concatOutputSections; |
| 25 | |
| 26 | void ConcatOutputSection::addInput(ConcatInputSection *input) { |
| 27 | assert(input->parent == this); |
| 28 | if (inputs.empty()) { |
| 29 | align = input->align; |
| 30 | flags = input->getFlags(); |
| 31 | } else { |
| 32 | align = std::max(a: align, b: input->align); |
| 33 | finalizeFlags(input); |
| 34 | } |
| 35 | inputs.push_back(x: input); |
| 36 | } |
| 37 | |
| 38 | // Branch-range extension can be implemented in two ways, either through ... |
| 39 | // |
| 40 | // (1) Branch islands: Single branch instructions (also of limited range), |
| 41 | // that might be chained in multiple hops to reach the desired |
| 42 | // destination. On ARM64, as 16 branch islands are needed to hop between |
| 43 | // opposite ends of a 2 GiB program. LD64 uses branch islands exclusively, |
| 44 | // even when it needs excessive hops. |
| 45 | // |
| 46 | // (2) Thunks: Instruction(s) to load the destination address into a scratch |
| 47 | // register, followed by a register-indirect branch. Thunks are |
| 48 | // constructed to reach any arbitrary address, so need not be |
| 49 | // chained. Although thunks need not be chained, a program might need |
| 50 | // multiple thunks to the same destination distributed throughout a large |
| 51 | // program so that all call sites can have one within range. |
| 52 | // |
| 53 | // The optimal approach is to mix islands for destinations within two hops, |
| 54 | // and use thunks for destinations at greater distance. For now, we only |
| 55 | // implement thunks. TODO: Adding support for branch islands! |
| 56 | // |
| 57 | // Internally -- as expressed in LLD's data structures -- a |
| 58 | // branch-range-extension thunk consists of: |
| 59 | // |
| 60 | // (1) new Defined symbol for the thunk named |
| 61 | // <FUNCTION>.thunk.<SEQUENCE>, which references ... |
| 62 | // (2) new InputSection, which contains ... |
| 63 | // (3.1) new data for the instructions to load & branch to the far address + |
| 64 | // (3.2) new Relocs on instructions to load the far address, which reference ... |
| 65 | // (4.1) existing Defined symbol for the real function in __text, or |
| 66 | // (4.2) existing DylibSymbol for the real function in a dylib |
| 67 | // |
| 68 | // Nearly-optimal thunk-placement algorithm features: |
| 69 | // |
| 70 | // * Single pass: O(n) on the number of call sites. |
| 71 | // |
| 72 | // * Accounts for the exact space overhead of thunks - no heuristics |
| 73 | // |
| 74 | // * Exploits the full range of call instructions - forward & backward |
| 75 | // |
| 76 | // Data: |
| 77 | // |
| 78 | // * DenseMap<Symbol *, ThunkInfo> thunkMap: Maps the function symbol |
| 79 | // to its thunk bookkeeper. |
| 80 | // |
| 81 | // * struct ThunkInfo (bookkeeper): Call instructions have limited range, and |
| 82 | // distant call sites might be unable to reach the same thunk, so multiple |
| 83 | // thunks are necessary to serve all call sites in a very large program. A |
| 84 | // thunkInfo stores state for all thunks associated with a particular |
| 85 | // function: |
| 86 | // (a) thunk symbol |
| 87 | // (b) input section containing stub code, and |
| 88 | // (c) sequence number for the active thunk incarnation. |
| 89 | // When an old thunk goes out of range, we increment the sequence number and |
| 90 | // create a new thunk named <FUNCTION>.thunk.<SEQUENCE>. |
| 91 | // |
| 92 | // * A thunk consists of |
| 93 | // (a) a Defined symbol pointing to |
| 94 | // (b) an InputSection holding machine code (similar to a MachO stub), and |
| 95 | // (c) relocs referencing the real function for fixing up the stub code. |
| 96 | // |
| 97 | // * std::vector<InputSection *> MergedInputSection::thunks: A vector parallel |
| 98 | // to the inputs vector. We store new thunks via cheap vector append, rather |
| 99 | // than costly insertion into the inputs vector. |
| 100 | // |
| 101 | // Control Flow: |
| 102 | // |
| 103 | // * During address assignment, MergedInputSection::finalize() examines call |
| 104 | // sites by ascending address and creates thunks. When a function is beyond |
| 105 | // the range of a call site, we need a thunk. Place it at the largest |
| 106 | // available forward address from the call site. Call sites increase |
| 107 | // monotonically and thunks are always placed as far forward as possible; |
| 108 | // thus, we place thunks at monotonically increasing addresses. Once a thunk |
| 109 | // is placed, it and all previous input-section addresses are final. |
| 110 | // |
| 111 | // * ConcatInputSection::finalize() and ConcatInputSection::writeTo() merge |
| 112 | // the inputs and thunks vectors (both ordered by ascending address), which |
| 113 | // is simple and cheap. |
| 114 | |
| 115 | DenseMap<Symbol *, ThunkInfo> lld::macho::thunkMap; |
| 116 | |
| 117 | // Determine whether we need thunks, which depends on the target arch -- RISC |
| 118 | // (i.e., ARM) generally does because it has limited-range branch/call |
| 119 | // instructions, whereas CISC (i.e., x86) generally doesn't. RISC only needs |
| 120 | // thunks for programs so large that branch source & destination addresses |
| 121 | // might differ more than the range of branch instruction(s). |
| 122 | bool TextOutputSection::needsThunks() const { |
| 123 | if (!target->usesThunks()) |
| 124 | return false; |
| 125 | uint64_t isecAddr = addr; |
| 126 | for (ConcatInputSection *isec : inputs) |
| 127 | isecAddr = alignToPowerOf2(Value: isecAddr, Align: isec->align) + isec->getSize(); |
| 128 | // Other sections besides __text might be small enough to pass this |
| 129 | // test but nevertheless need thunks for calling into other sections. |
| 130 | // An imperfect heuristic to use in this case is that if a section |
| 131 | // we've already processed in this segment needs thunks, so do the |
| 132 | // rest. |
| 133 | bool needsThunks = parent && parent->needsThunks; |
| 134 | |
| 135 | // Calculate the total size of all branch target sections |
| 136 | uint64_t branchTargetsSize = in.stubs->getSize(); |
| 137 | |
| 138 | // Add the size of __objc_stubs section if it exists |
| 139 | if (in.objcStubs && in.objcStubs->isNeeded()) |
| 140 | branchTargetsSize += in.objcStubs->getSize(); |
| 141 | |
| 142 | if (!needsThunks && |
| 143 | isecAddr - addr + branchTargetsSize <= |
| 144 | std::min(a: target->backwardBranchRange, b: target->forwardBranchRange)) |
| 145 | return false; |
| 146 | // Yes, this program is large enough to need thunks. |
| 147 | if (parent) { |
| 148 | parent->needsThunks = true; |
| 149 | } |
| 150 | for (ConcatInputSection *isec : inputs) { |
| 151 | for (Reloc &r : isec->relocs) { |
| 152 | if (!target->hasAttr(type: r.type, bit: RelocAttrBits::BRANCH)) |
| 153 | continue; |
| 154 | auto *sym = cast<Symbol *>(Val&: r.referent); |
| 155 | // Pre-populate the thunkMap and memoize call site counts for every |
| 156 | // InputSection and ThunkInfo. We do this for the benefit of |
| 157 | // estimateBranchTargetThresholdVA(). |
| 158 | ThunkInfo &thunkInfo = thunkMap[sym]; |
| 159 | // Knowing ThunkInfo call site count will help us know whether or not we |
| 160 | // might need to create more for this referent at the time we are |
| 161 | // estimating distance to __stubs in estimateBranchTargetThresholdVA(). |
| 162 | ++thunkInfo.callSiteCount; |
| 163 | // We can avoid work on InputSections that have no BRANCH relocs. |
| 164 | isec->hasCallSites = true; |
| 165 | } |
| 166 | } |
| 167 | return true; |
| 168 | } |
| 169 | |
| 170 | // Estimate the address beyond which branch targets (like __stubs and |
| 171 | // __objc_stubs) are within range of a simple forward branch. This is called |
| 172 | // exactly once, when the last input section has been finalized. |
| 173 | uint64_t |
| 174 | TextOutputSection::estimateBranchTargetThresholdVA(size_t callIdx) const { |
| 175 | // Tally the functions which still have call sites remaining to process, |
| 176 | // which yields the maximum number of thunks we might yet place. |
| 177 | size_t maxPotentialThunks = 0; |
| 178 | for (auto &tp : thunkMap) { |
| 179 | ThunkInfo &ti = tp.second; |
| 180 | // This overcounts: Only sections that are in forward jump range from the |
| 181 | // currently-active section get finalized, and all input sections are |
| 182 | // finalized when estimateBranchTargetThresholdVA() is called. So only |
| 183 | // backward jumps will need thunks, but we count all jumps. |
| 184 | if (ti.callSitesUsed < ti.callSiteCount) |
| 185 | maxPotentialThunks += 1; |
| 186 | } |
| 187 | // Tally the total size of input sections remaining to process. |
| 188 | uint64_t isecVA = inputs[callIdx]->getVA(); |
| 189 | uint64_t isecEnd = isecVA; |
| 190 | for (size_t i = callIdx; i < inputs.size(); i++) { |
| 191 | InputSection *isec = inputs[i]; |
| 192 | isecEnd = alignToPowerOf2(Value: isecEnd, Align: isec->align) + isec->getSize(); |
| 193 | } |
| 194 | |
| 195 | // Tally up any thunks that have already been placed that have VA higher than |
| 196 | // inputs[callIdx]. First, find the index of the first thunk that is beyond |
| 197 | // the current inputs[callIdx]. |
| 198 | auto itPostcallIdxThunks = |
| 199 | llvm::partition_point(Range: thunks, P: [isecVA](const ConcatInputSection *t) { |
| 200 | return t->getVA() <= isecVA; |
| 201 | }); |
| 202 | uint64_t existingForwardThunks = thunks.end() - itPostcallIdxThunks; |
| 203 | |
| 204 | uint64_t forwardBranchRange = target->forwardBranchRange; |
| 205 | assert(isecEnd > forwardBranchRange && |
| 206 | "should not run thunk insertion if all code fits in jump range" ); |
| 207 | assert(isecEnd - isecVA <= forwardBranchRange && |
| 208 | "should only finalize sections in jump range" ); |
| 209 | |
| 210 | // Estimate the maximum size of the code, right before the branch target |
| 211 | // sections. |
| 212 | uint64_t maxTextSize = 0; |
| 213 | // Add the size of all the inputs, including the unprocessed ones. |
| 214 | maxTextSize += isecEnd; |
| 215 | |
| 216 | // Add the size of the thunks that have already been created that are ahead of |
| 217 | // inputs[callIdx]. These are already created thunks that will be interleaved |
| 218 | // with inputs[callIdx...end]. |
| 219 | maxTextSize += existingForwardThunks * target->thunkSize; |
| 220 | |
| 221 | // Add the size of the thunks that may be created in the future. Since |
| 222 | // 'maxPotentialThunks' overcounts, this is an estimate of the upper limit. |
| 223 | maxTextSize += maxPotentialThunks * target->thunkSize; |
| 224 | |
| 225 | // Calculate the total size of all late branch target sections |
| 226 | uint64_t branchTargetsSize = 0; |
| 227 | |
| 228 | // Add the size of __stubs section |
| 229 | branchTargetsSize += in.stubs->getSize(); |
| 230 | |
| 231 | // Add the size of __objc_stubs section if it exists |
| 232 | if (in.objcStubs && in.objcStubs->isNeeded()) |
| 233 | branchTargetsSize += in.objcStubs->getSize(); |
| 234 | |
| 235 | // Estimated maximum VA of the last branch target. |
| 236 | uint64_t maxVAOfLastBranchTarget = maxTextSize + branchTargetsSize; |
| 237 | |
| 238 | // Estimate the address after which call sites can safely call branch targets |
| 239 | // directly rather than through intermediary thunks. |
| 240 | uint64_t branchTargetThresholdVA = |
| 241 | maxVAOfLastBranchTarget - forwardBranchRange; |
| 242 | |
| 243 | log(msg: "thunks = " + std::to_string(val: thunkMap.size()) + |
| 244 | ", potential = " + std::to_string(val: maxPotentialThunks) + |
| 245 | ", stubs = " + std::to_string(val: in.stubs->getSize()) + |
| 246 | (in.objcStubs && in.objcStubs->isNeeded() |
| 247 | ? ", objc_stubs = " + std::to_string(val: in.objcStubs->getSize()) |
| 248 | : "" ) + |
| 249 | ", isecVA = " + utohexstr(X: isecVA) + ", threshold = " + |
| 250 | utohexstr(X: branchTargetThresholdVA) + ", isecEnd = " + utohexstr(X: isecEnd) + |
| 251 | ", tail = " + utohexstr(X: isecEnd - isecVA) + |
| 252 | ", slop = " + utohexstr(X: forwardBranchRange - (isecEnd - isecVA))); |
| 253 | return branchTargetThresholdVA; |
| 254 | } |
| 255 | |
| 256 | void ConcatOutputSection::finalizeOne(ConcatInputSection *isec) { |
| 257 | size = alignToPowerOf2(Value: size, Align: isec->align); |
| 258 | fileSize = alignToPowerOf2(Value: fileSize, Align: isec->align); |
| 259 | isec->outSecOff = size; |
| 260 | isec->isFinal = true; |
| 261 | size += isec->getSize(); |
| 262 | fileSize += isec->getFileSize(); |
| 263 | } |
| 264 | |
| 265 | void ConcatOutputSection::finalizeContents() { |
| 266 | for (ConcatInputSection *isec : inputs) |
| 267 | finalizeOne(isec); |
| 268 | } |
| 269 | |
| 270 | void TextOutputSection::finalize() { |
| 271 | if (!needsThunks()) { |
| 272 | for (ConcatInputSection *isec : inputs) |
| 273 | finalizeOne(isec); |
| 274 | return; |
| 275 | } |
| 276 | |
| 277 | uint64_t forwardBranchRange = target->forwardBranchRange; |
| 278 | uint64_t backwardBranchRange = target->backwardBranchRange; |
| 279 | uint64_t branchTargetThresholdVA = TargetInfo::outOfRangeVA; |
| 280 | size_t thunkSize = target->thunkSize; |
| 281 | size_t relocCount = 0; |
| 282 | size_t callSiteCount = 0; |
| 283 | size_t thunkCallCount = 0; |
| 284 | size_t thunkCount = 0; |
| 285 | |
| 286 | // Walk all sections in order. Finalize all sections that are less than |
| 287 | // forwardBranchRange in front of it. |
| 288 | // isecVA is the address of the current section. |
| 289 | // addr + size is the start address of the first non-finalized section. |
| 290 | |
| 291 | // inputs[finalIdx] is for finalization (address-assignment) |
| 292 | size_t finalIdx = 0; |
| 293 | // Kick-off by ensuring that the first input section has an address |
| 294 | for (size_t callIdx = 0, endIdx = inputs.size(); callIdx < endIdx; |
| 295 | ++callIdx) { |
| 296 | if (finalIdx == callIdx) |
| 297 | finalizeOne(isec: inputs[finalIdx++]); |
| 298 | ConcatInputSection *isec = inputs[callIdx]; |
| 299 | assert(isec->isFinal); |
| 300 | uint64_t isecVA = isec->getVA(); |
| 301 | |
| 302 | // Assign addresses up-to the forward branch-range limit. |
| 303 | // Every call instruction needs a small number of bytes (on Arm64: 4), |
| 304 | // and each inserted thunk needs a slightly larger number of bytes |
| 305 | // (on Arm64: 12). If a section starts with a branch instruction and |
| 306 | // contains several branch instructions in succession, then the distance |
| 307 | // from the current position to the position where the thunks are inserted |
| 308 | // grows. So leave room for a bunch of thunks. |
| 309 | unsigned slop = 256 * thunkSize; |
| 310 | while (finalIdx < endIdx) { |
| 311 | uint64_t expectedNewSize = |
| 312 | alignToPowerOf2(Value: addr + size, Align: inputs[finalIdx]->align) + |
| 313 | inputs[finalIdx]->getSize(); |
| 314 | if (expectedNewSize >= isecVA + forwardBranchRange - slop) |
| 315 | break; |
| 316 | finalizeOne(isec: inputs[finalIdx++]); |
| 317 | } |
| 318 | |
| 319 | if (!isec->hasCallSites) |
| 320 | continue; |
| 321 | |
| 322 | if (finalIdx == endIdx && |
| 323 | branchTargetThresholdVA == TargetInfo::outOfRangeVA) { |
| 324 | // When we have finalized all input sections, branch target sections (like |
| 325 | // __stubs and __objc_stubs) (destined to follow __text) come within range |
| 326 | // of forward branches and we can estimate the threshold address after |
| 327 | // which we can reach any branch target with a forward branch. Note that |
| 328 | // although it sits in the middle of a loop, this code executes only once. |
| 329 | // It is in the loop because we need to call it at the proper |
| 330 | // time: the earliest call site from which the end of __text |
| 331 | // (and start of branch target sections) comes within range of a forward |
| 332 | // branch. |
| 333 | branchTargetThresholdVA = estimateBranchTargetThresholdVA(callIdx); |
| 334 | } |
| 335 | // Process relocs by ascending address, i.e., ascending offset within isec |
| 336 | std::vector<Reloc> &relocs = isec->relocs; |
| 337 | // FIXME: This property does not hold for object files produced by ld64's |
| 338 | // `-r` mode. |
| 339 | assert(is_sorted(relocs, |
| 340 | [](Reloc &a, Reloc &b) { return a.offset > b.offset; })); |
| 341 | for (Reloc &r : reverse(C&: relocs)) { |
| 342 | ++relocCount; |
| 343 | if (!target->hasAttr(type: r.type, bit: RelocAttrBits::BRANCH)) |
| 344 | continue; |
| 345 | ++callSiteCount; |
| 346 | // Calculate branch reachability boundaries |
| 347 | uint64_t callVA = isecVA + r.offset; |
| 348 | uint64_t lowVA = |
| 349 | backwardBranchRange < callVA ? callVA - backwardBranchRange : 0; |
| 350 | uint64_t highVA = callVA + forwardBranchRange; |
| 351 | // Calculate our call referent address |
| 352 | auto *funcSym = cast<Symbol *>(Val&: r.referent); |
| 353 | ThunkInfo &thunkInfo = thunkMap[funcSym]; |
| 354 | // The referent is not reachable, so we need to use a thunk ... |
| 355 | if ((funcSym->isInStubs() || |
| 356 | (in.objcStubs && in.objcStubs->isNeeded() && |
| 357 | ObjCStubsSection::isObjCStubSymbol(sym: funcSym))) && |
| 358 | callVA >= branchTargetThresholdVA) { |
| 359 | assert(callVA != TargetInfo::outOfRangeVA); |
| 360 | // ... Oh, wait! We are close enough to the end that branch target |
| 361 | // sections (__stubs, __objc_stubs) are now within range of a simple |
| 362 | // forward branch. |
| 363 | continue; |
| 364 | } |
| 365 | uint64_t funcVA = funcSym->resolveBranchVA(); |
| 366 | ++thunkInfo.callSitesUsed; |
| 367 | if (lowVA <= funcVA && funcVA <= highVA) { |
| 368 | // The referent is reachable with a simple call instruction. |
| 369 | continue; |
| 370 | } |
| 371 | ++thunkInfo.thunkCallCount; |
| 372 | ++thunkCallCount; |
| 373 | // If an existing thunk is reachable, use it ... |
| 374 | if (thunkInfo.sym) { |
| 375 | uint64_t thunkVA = thunkInfo.isec->getVA(); |
| 376 | if (lowVA <= thunkVA && thunkVA <= highVA) { |
| 377 | r.referent = thunkInfo.sym; |
| 378 | continue; |
| 379 | } |
| 380 | } |
| 381 | // ... otherwise, create a new thunk. |
| 382 | if (addr + size > highVA) { |
| 383 | // There were too many consecutive branch instructions for `slop` |
| 384 | // above. If you hit this: For the current algorithm, just bumping up |
| 385 | // slop above and trying again is probably simplest. (See also PR51578 |
| 386 | // comment 5). |
| 387 | fatal(msg: Twine(__FUNCTION__) + ": FIXME: thunk range overrun" ); |
| 388 | } |
| 389 | thunkInfo.isec = |
| 390 | makeSyntheticInputSection(segName: isec->getSegName(), sectName: isec->getName()); |
| 391 | thunkInfo.isec->parent = this; |
| 392 | assert(thunkInfo.isec->live); |
| 393 | |
| 394 | StringRef thunkName = saver().save(S: funcSym->getName() + ".thunk." + |
| 395 | std::to_string(val: thunkInfo.sequence++)); |
| 396 | if (!isa<Defined>(Val: funcSym) || cast<Defined>(Val: funcSym)->isExternal()) { |
| 397 | r.referent = thunkInfo.sym = symtab->addDefined( |
| 398 | name: thunkName, /*file=*/nullptr, thunkInfo.isec, /*value=*/0, size: thunkSize, |
| 399 | /*isWeakDef=*/false, /*isPrivateExtern=*/true, |
| 400 | /*isReferencedDynamically=*/false, /*noDeadStrip=*/false, |
| 401 | /*isWeakDefCanBeHidden=*/false); |
| 402 | } else { |
| 403 | r.referent = thunkInfo.sym = make<Defined>( |
| 404 | args&: thunkName, /*file=*/args: nullptr, args&: thunkInfo.isec, /*value=*/args: 0, args&: thunkSize, |
| 405 | /*isWeakDef=*/args: false, /*isExternal=*/args: false, /*isPrivateExtern=*/args: true, |
| 406 | /*includeInSymtab=*/args: true, /*isReferencedDynamically=*/args: false, |
| 407 | /*noDeadStrip=*/args: false, /*isWeakDefCanBeHidden=*/args: false); |
| 408 | } |
| 409 | thunkInfo.sym->used = true; |
| 410 | target->populateThunk(thunk: thunkInfo.isec, funcSym); |
| 411 | finalizeOne(isec: thunkInfo.isec); |
| 412 | thunks.push_back(x: thunkInfo.isec); |
| 413 | ++thunkCount; |
| 414 | } |
| 415 | } |
| 416 | |
| 417 | log(msg: "thunks for " + parent->name + "," + name + |
| 418 | ": funcs = " + std::to_string(val: thunkMap.size()) + |
| 419 | ", relocs = " + std::to_string(val: relocCount) + |
| 420 | ", all calls = " + std::to_string(val: callSiteCount) + |
| 421 | ", thunk calls = " + std::to_string(val: thunkCallCount) + |
| 422 | ", thunks = " + std::to_string(val: thunkCount)); |
| 423 | } |
| 424 | |
| 425 | void ConcatOutputSection::writeTo(uint8_t *buf) const { |
| 426 | for (ConcatInputSection *isec : inputs) |
| 427 | isec->writeTo(buf: buf + isec->outSecOff); |
| 428 | } |
| 429 | |
| 430 | void TextOutputSection::writeTo(uint8_t *buf) const { |
| 431 | // Merge input sections from thunk & ordinary vectors |
| 432 | size_t i = 0, ie = inputs.size(); |
| 433 | size_t t = 0, te = thunks.size(); |
| 434 | while (i < ie || t < te) { |
| 435 | while (i < ie && (t == te || inputs[i]->empty() || |
| 436 | inputs[i]->outSecOff < thunks[t]->outSecOff)) { |
| 437 | inputs[i]->writeTo(buf: buf + inputs[i]->outSecOff); |
| 438 | ++i; |
| 439 | } |
| 440 | while (t < te && (i == ie || thunks[t]->outSecOff < inputs[i]->outSecOff)) { |
| 441 | thunks[t]->writeTo(buf: buf + thunks[t]->outSecOff); |
| 442 | ++t; |
| 443 | } |
| 444 | } |
| 445 | } |
| 446 | |
| 447 | void ConcatOutputSection::finalizeFlags(InputSection *input) { |
| 448 | switch (sectionType(flags: input->getFlags())) { |
| 449 | default /*type-unspec'ed*/: |
| 450 | // FIXME: Add additional logic here when supporting emitting obj files. |
| 451 | break; |
| 452 | case S_4BYTE_LITERALS: |
| 453 | case S_8BYTE_LITERALS: |
| 454 | case S_16BYTE_LITERALS: |
| 455 | case S_CSTRING_LITERALS: |
| 456 | case S_ZEROFILL: |
| 457 | case S_LAZY_SYMBOL_POINTERS: |
| 458 | case S_MOD_TERM_FUNC_POINTERS: |
| 459 | case S_THREAD_LOCAL_REGULAR: |
| 460 | case S_THREAD_LOCAL_ZEROFILL: |
| 461 | case S_THREAD_LOCAL_VARIABLES: |
| 462 | case S_THREAD_LOCAL_INIT_FUNCTION_POINTERS: |
| 463 | case S_THREAD_LOCAL_VARIABLE_POINTERS: |
| 464 | case S_NON_LAZY_SYMBOL_POINTERS: |
| 465 | case S_SYMBOL_STUBS: |
| 466 | flags |= input->getFlags(); |
| 467 | break; |
| 468 | } |
| 469 | } |
| 470 | |
| 471 | ConcatOutputSection * |
| 472 | ConcatOutputSection::getOrCreateForInput(const InputSection *isec) { |
| 473 | NamePair names = maybeRenameSection(key: {isec->getSegName(), isec->getName()}); |
| 474 | ConcatOutputSection *&osec = concatOutputSections[names]; |
| 475 | if (!osec) { |
| 476 | if (isec->getSegName() == segment_names::text && |
| 477 | isec->getName() != section_names::gccExceptTab && |
| 478 | isec->getName() != section_names::ehFrame) |
| 479 | osec = make<TextOutputSection>(args&: names.second); |
| 480 | else |
| 481 | osec = make<ConcatOutputSection>(args&: names.second); |
| 482 | } |
| 483 | return osec; |
| 484 | } |
| 485 | |
| 486 | NamePair macho::maybeRenameSection(NamePair key) { |
| 487 | auto newNames = config->sectionRenameMap.find(Val: key); |
| 488 | if (newNames != config->sectionRenameMap.end()) |
| 489 | return newNames->second; |
| 490 | return key; |
| 491 | } |
| 492 | |