irgenerator/name-mangling.spice
| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | // Copyright (c) 2021-2026 ChilliBits. All rights reserved. | ||
| 2 | |||
| 3 | // Std imports | ||
| 4 | import "std/data/vector"; | ||
| 5 | import "std/text/stringstream"; | ||
| 6 | import "std/type/type-conversion"; | ||
| 7 | |||
| 8 | // Own imports | ||
| 9 | import "@/global/type-name-disambiguator"; | ||
| 10 | import "@/model/function"; | ||
| 11 | import "@/model/generic-type"; | ||
| 12 | import "@/model/interface"; | ||
| 13 | import "@/model/struct"; | ||
| 14 | import "@/model/struct-base"; | ||
| 15 | import "@/model/union"; | ||
| 16 | import "@/symboltablebuilder/qual-type"; | ||
| 17 | import "@/symboltablebuilder/type-qualifiers"; | ||
| 18 | import "@/symboltablebuilder/super-type"; | ||
| 19 | import "@/symboltablebuilder/type"; | ||
| 20 | import "@/symboltablebuilder/type-chain"; | ||
| 21 | |||
| 22 | /** | ||
| 23 | * Helper for name mangling. | ||
| 24 | * This should be mostly compatible with the C++ Itanium ABI name mangling scheme. | ||
| 25 | * | ||
| 26 | * Spice has no static methods, so the helpers are free functions, prefixed with 'mangle'. | ||
| 27 | * | ||
| 28 | * Names: | ||
| 29 | * - P: pointer | ||
| 30 | * - R: reference | ||
| 31 | * - A: array | ||
| 32 | * - d: double | ||
| 33 | * - i: int | ||
| 34 | * - j: unsigned int | ||
| 35 | * - s: short | ||
| 36 | * - t: unsigned short | ||
| 37 | * - l: long | ||
| 38 | * - m: unsigned long | ||
| 39 | * - a: byte | ||
| 40 | * - h: unsigned byte / unsigned char | ||
| 41 | * - c: char | ||
| 42 | * - PKc: string (const char*) | ||
| 43 | * - b: bool | ||
| 44 | * - v: void | ||
| 45 | * - K: const | ||
| 46 | * - C: capturing | ||
| 47 | * | ||
| 48 | * Reserved: | ||
| 49 | * - I | ||
| 50 | * - E | ||
| 51 | */ | ||
| 52 | |||
| 53 | /** | ||
| 54 | * Mangle a function or procedure. | ||
| 55 | * This should be mostly compatible with the C++ Itanium ABI name mangling scheme. | ||
| 56 | * | ||
| 57 | * @param spiceFunc Input function | ||
| 58 | * @return Mangled name | ||
| 59 | */ | ||
| 60 | 174404 | public f<String> mangleFunction(const Function& spiceFunc) { | |
| 61 | // Check if we have a predefined mangled name, specified by e.g. a function attribute | ||
| 62 | 174404 | if !spiceFunc.predefinedMangledName.isEmpty() { | |
| 63 | ✗ | return String(spiceFunc.predefinedMangledName); | |
| 64 | } | ||
| 65 | |||
| 66 | // Check if mangling is required | ||
| 67 | 174404 | if !spiceFunc.mangleFunctionName { | |
| 68 | ✗ | return String(spiceFunc.name); | |
| 69 | } | ||
| 70 | |||
| 71 | 174404 | StringStream mangledName; | |
| 72 | 174404 | mangledName << "_Z"; | |
| 73 | |||
| 74 | // This type | ||
| 75 | 174404 | if spiceFunc.isMethod() { | |
| 76 | 113002 | mangledName << "N"; | |
| 77 | 113002 | mangleType(mangledName, spiceFunc.thisType); | |
| 78 | 113002 | } | |
| 79 | |||
| 80 | // Function name | ||
| 81 | 174404 | const String name = spiceFunc.name + spiceFunc.mangleSuffix; | |
| 82 | 174404 | mangledName << toString(name.getLength()) << name; | |
| 83 | |||
| 84 | // Template types | ||
| 85 | 174404 | bool isSelfGeneric = !spiceFunc.templateTypes.isEmpty(); | |
| 86 | 174404 | if spiceFunc.isMethod() { | |
| 87 | 113002 | const QualTypeList thisTemplateTypes = spiceFunc.thisType.getTemplateTypes(); | |
| 88 | 113002 | isSelfGeneric = spiceFunc.templateTypes.getSize() > thisTemplateTypes.getSize(); | |
| 89 | 113002 | } | |
| 90 | 174404 | if isSelfGeneric { | |
| 91 | 12764 | mangledName << "I"; | |
| 92 | // Template types themselves | ||
| 93 | 28714 | foreach const GenericType& genericTemplateType : spiceFunc.templateTypes { | |
| 94 | 15950 | const String subType = genericTemplateType.qualType.getSubType(); | |
| 95 | 15950 | assert spiceFunc.typeMapping.contains(subType); | |
| 96 | 15950 | const QualType& actualType = spiceFunc.typeMapping.get(subType); | |
| 97 | 15950 | mangleType(mangledName, actualType); | |
| 98 | 15950 | } | |
| 99 | 12764 | mangledName << "E"; | |
| 100 | |||
| 101 | // Insert second end marker to end the nested type | ||
| 102 | 12764 | if spiceFunc.isMethod() { | |
| 103 | 2505 | mangledName << "E"; | |
| 104 | 2505 | } | |
| 105 | |||
| 106 | // Return type | ||
| 107 | 12764 | if spiceFunc.isFunction() { | |
| 108 | 6586 | mangleType(mangledName, spiceFunc.returnType); | |
| 109 | 6586 | } else { | |
| 110 | 6178 | mangledName << "v"; | |
| 111 | 6178 | } | |
| 112 | 174404 | } else if spiceFunc.isMethod() { | |
| 113 | 110497 | mangledName << "E"; | |
| 114 | 161640 | } | |
| 115 | |||
| 116 | // Parameter types | ||
| 117 | 337436 | foreach const Param& param : spiceFunc.paramList { | |
| 118 | 163032 | assert !param.isOptional; | |
| 119 | 163032 | mangleType(mangledName, param.qualType); | |
| 120 | 163032 | } | |
| 121 | 174404 | if spiceFunc.paramList.isEmpty() { | |
| 122 | 67060 | mangledName << "v"; | |
| 123 | 67060 | } | |
| 124 | |||
| 125 | 348808 | return String(mangledName.str()); | |
| 126 | 174404 | } | |
| 127 | |||
| 128 | /** | ||
| 129 | * Mangle a struct | ||
| 130 | * This should be mostly compatible with the C++ Itanium ABI name mangling scheme. | ||
| 131 | * | ||
| 132 | * @param spiceStruct Input struct | ||
| 133 | * @return Mangled name | ||
| 134 | */ | ||
| 135 | 9114 | public f<String> mangleStruct(const Struct& spiceStruct) { | |
| 136 | 18228 | return String("struct.") + spiceStruct.name; | |
| 137 | 9114 | } | |
| 138 | |||
| 139 | /** | ||
| 140 | * Mangle an interface | ||
| 141 | * | ||
| 142 | * @param spiceInterface Input interface | ||
| 143 | * @return Mangled name | ||
| 144 | */ | ||
| 145 | 823 | public f<String> mangleInterface(const Interface& spiceInterface) { | |
| 146 | 1646 | return String("interface.") + spiceInterface.name; | |
| 147 | 823 | } | |
| 148 | |||
| 149 | /** | ||
| 150 | * Mangle a union | ||
| 151 | * | ||
| 152 | * @param spiceUnion Input union | ||
| 153 | * @return Mangled name | ||
| 154 | */ | ||
| 155 | 27 | public f<String> mangleUnion(const Union& spiceUnion) { | |
| 156 | 54 | return String("union.") + spiceUnion.name; | |
| 157 | 27 | } | |
| 158 | |||
| 159 | /** | ||
| 160 | * Mangle a fully qualified name like e.g. test::s1::calledMethod to 4test2s112calledMethod | ||
| 161 | * This should be mostly compatible with the C++ Itanium ABI name mangling scheme. | ||
| 162 | * | ||
| 163 | * @param out Output string stream | ||
| 164 | * @param name Input name | ||
| 165 | * @param nestedType True if the name is a nested type | ||
| 166 | */ | ||
| 167 | 195837 | p mangleName(StringStream& out, const String& name, bool& nestedType) { | |
| 168 | 195837 | Vector<String> fragments; | |
| 169 | 195837 | const Vector<String> tokens = splitLikeGetline(name, ':'); | |
| 170 | 391674 | foreach const String& token : tokens { | |
| 171 | 195837 | const Vector<String> subTokens = splitLikeGetline(token, '/'); | |
| 172 | 391674 | foreach const String& subToken : subTokens { | |
| 173 | 195837 | fragments.pushBack(subToken); | |
| 174 | 195837 | } | |
| 175 | 195837 | } | |
| 176 | |||
| 177 | // Start a nested type if needed. The caller needs to emit the end marker. | ||
| 178 | 195837 | if fragments.getSize() > 1l { | |
| 179 | ✗ | out << "N"; | |
| 180 | ✗ | nestedType = true; | |
| 181 | ✗ | } | |
| 182 | |||
| 183 | // Process each fragment and append it to the result | ||
| 184 | 391674 | foreach const String& currentFragment : fragments { | |
| 185 | 195837 | out << toString(currentFragment.getLength()) << currentFragment; | |
| 186 | 195837 | } | |
| 187 | 195837 | } | |
| 188 | |||
| 189 | /** | ||
| 190 | * Split a string at the given delimiter with the semantics of repeated std::getline calls: empty tokens between two | ||
| 191 | * delimiters are kept, but there is no trailing empty token (and none at all for an empty input). | ||
| 192 | * | ||
| 193 | * @param input Input string | ||
| 194 | * @param delimiter Delimiter char | ||
| 195 | * @return Tokens | ||
| 196 | */ | ||
| 197 | 391674 | f<Vector<String>> splitLikeGetline(const String& input, char delimiter) { | |
| 198 | 391674 | Vector<String> tokens; | |
| 199 | 391674 | String token; | |
| 200 | 391674 | const unsigned long length = input.getLength(); | |
| 201 | 3146728 | for unsigned long i = 0l; i < length; i++ { | |
| 202 | 2755054 | const char c = input[i]; | |
| 203 | 2755054 | if c == delimiter { | |
| 204 | ✗ | tokens.pushBack(token); | |
| 205 | ✗ | token.clear(); | |
| 206 | ✗ | } else { | |
| 207 | 2755054 | token += c; | |
| 208 | 2755054 | } | |
| 209 | 2755054 | } | |
| 210 | 391674 | if !token.isEmpty() { | |
| 211 | 391674 | tokens.pushBack(token); | |
| 212 | 391674 | } | |
| 213 | 783348 | return tokens; | |
| 214 | 391674 | } | |
| 215 | |||
| 216 | /** | ||
| 217 | * Mangle a symbol qualType | ||
| 218 | * This should be mostly compatible with the C++ Itanium ABI name mangling scheme. | ||
| 219 | * | ||
| 220 | * @param out Output string stream | ||
| 221 | * @param qualType Input symbol qualType | ||
| 222 | */ | ||
| 223 | 381954 | p mangleType(StringStream& out, const QualType& qualType) { | |
| 224 | 381954 | const Type* rawType = qualType.getType(); | |
| 225 | 381954 | assert !qualType.hasAnyGenericParts(); | |
| 226 | |||
| 227 | // Unwrap qualType chain | ||
| 228 | 381954 | assert !rawType.typeChain.isEmpty(); | |
| 229 | 455507 | for unsigned long i = rawType.typeChain.getSize() - 1l; i >= 1l; i-- { | |
| 230 | 73553 | mangleTypeChainElement(out, rawType.typeChain.get(i), false); | |
| 231 | 73553 | } | |
| 232 | |||
| 233 | // Qualifiers | ||
| 234 | 381954 | const TypeQualifiers& qualifiers = qualType.qualifiers; | |
| 235 | 381954 | assert qualifiers.isSigned == !qualifiers.isUnsigned; | |
| 236 | 381954 | const bool signedness = qualifiers.isSigned; | |
| 237 | // The 'heap' qualifier has no equivalent in the C++ Itanium ABI, but it changes the semantics of the type (ownership), so | ||
| 238 | // it must be reflected in the mangled name. Otherwise e.g. 'byte*' and 'heap byte*' manifestations of the same generic | ||
| 239 | // function would collide on the same mangled name. Piggyback on the 'volatile' cv-qualifier slot ('V', ordered before | ||
| 240 | // 'K' per the Itanium <CV-qualifiers> grammar) so demangling still succeeds. | ||
| 241 | 381954 | if qualifiers.isHeap { | |
| 242 | 15089 | out << "V"; | |
| 243 | 15089 | } | |
| 244 | 381954 | if qualifiers.isConst && rawType.typeChain.getSize() > 1l { | |
| 245 | 35277 | out << "K"; | |
| 246 | 35277 | } | |
| 247 | |||
| 248 | // Base chain element | ||
| 249 | 381954 | mangleTypeChainElement(out, rawType.typeChain.front(), signedness); | |
| 250 | 381954 | } | |
| 251 | |||
| 252 | /** | ||
| 253 | * Mangle a type chain element | ||
| 254 | * | ||
| 255 | * @param out Output string stream | ||
| 256 | * @param chainElement Input type chain element | ||
| 257 | * @param signedness Signedness of the type | ||
| 258 | */ | ||
| 259 | 455507 | p mangleTypeChainElement(StringStream& out, const TypeChainElement& chainElement, bool signedness) { | |
| 260 | 455507 | switch chainElement.superType { | |
| 261 | case SuperType::TY_PTR: { | ||
| 262 | 23385 | out << "P"; | |
| 263 | 23385 | } | |
| 264 | case SuperType::TY_ARRAY: { | ||
| 265 | 560 | if cast<unsigned long>(chainElement.data.arraySize) == ARRAY_SIZE_UNKNOWN { | |
| 266 | 457 | out << "P"; | |
| 267 | 457 | } else { | |
| 268 | 103 | out << "A" << toString(chainElement.data.arraySize) << "_"; | |
| 269 | 103 | } | |
| 270 | 560 | } | |
| 271 | case SuperType::TY_REF: { | ||
| 272 | 49608 | out << "R"; | |
| 273 | 49608 | } | |
| 274 | case SuperType::TY_DOUBLE: { | ||
| 275 | 3413 | assert signedness; // Unsigned double types are forbidden | |
| 276 | 3413 | out << "d"; | |
| 277 | 3413 | } | |
| 278 | case SuperType::TY_INT: { | ||
| 279 | 46609 | out << (signedness ? "i" : "j"); | |
| 280 | 46609 | } | |
| 281 | case SuperType::TY_SHORT: { | ||
| 282 | 3605 | out << (signedness ? "s" : "t"); | |
| 283 | 3605 | } | |
| 284 | case SuperType::TY_LONG: { | ||
| 285 | 29831 | out << (signedness ? "l" : "m"); | |
| 286 | 29831 | } | |
| 287 | case SuperType::TY_BYTE: { | ||
| 288 | 18926 | out << (signedness ? "a" : "h"); | |
| 289 | 18926 | } | |
| 290 | case SuperType::TY_CHAR: { | ||
| 291 | 25795 | out << "c"; | |
| 292 | 25795 | } | |
| 293 | case SuperType::TY_STRING: { | ||
| 294 | 46103 | out << "PKc"; | |
| 295 | 46103 | } | |
| 296 | case SuperType::TY_BOOL: { | ||
| 297 | 8537 | assert !signedness; // Signed bool types are forbidden | |
| 298 | 8537 | out << "b"; | |
| 299 | 8537 | } | |
| 300 | case SuperType::TY_STRUCT, SuperType::TY_INTERFACE, SuperType::TY_UNION: { | ||
| 301 | 195837 | bool nestedType = false; | |
| 302 | // Append a disambiguation suffix for same-named but distinct structs/interfaces/unions (see issue #1253), so | ||
| 303 | // that two independent types sharing a name do not end up with the same mangled name (and thus clash at link | ||
| 304 | // time). | ||
| 305 | 195837 | const String name = chainElement.subType + getDisambiguationSuffix(chainElement.subType, chainElement.typeId); | |
| 306 | 195837 | mangleName(out, name, nestedType); | |
| 307 | 195837 | if !chainElement.templateTypes.isEmpty() { | |
| 308 | 64458 | out << "I"; | |
| 309 | 138846 | foreach const QualType& templateType : chainElement.templateTypes { | |
| 310 | 74388 | mangleType(out, templateType); | |
| 311 | 74388 | } | |
| 312 | 64458 | out << "E"; | |
| 313 | 64458 | } | |
| 314 | 195837 | if nestedType { | |
| 315 | ✗ | out << "E"; | |
| 316 | ✗ | } | |
| 317 | 195837 | } | |
| 318 | case SuperType::TY_ENUM: { | ||
| 319 | ✗ | bool nestedType = false; | |
| 320 | ✗ | mangleName(out, chainElement.subType, nestedType); | |
| 321 | ✗ | if nestedType { | |
| 322 | ✗ | out << "E"; | |
| 323 | ✗ | } | |
| 324 | ✗ | } | |
| 325 | case SuperType::TY_FUNCTION: { | ||
| 326 | 184 | out << (chainElement.data.hasCaptures ? "PFC" : "PF"); | |
| 327 | 595 | foreach const QualType& paramType : chainElement.paramTypes { | |
| 328 | 411 | mangleType(out, paramType); | |
| 329 | 411 | } | |
| 330 | 184 | out << "E"; | |
| 331 | 184 | } | |
| 332 | case SuperType::TY_PROCEDURE: { | ||
| 333 | 3114 | out << (chainElement.data.hasCaptures ? "PFCv" : "PFv"); | |
| 334 | 4763 | for unsigned long i = 1l; i < chainElement.paramTypes.getSize(); i++ { | |
| 335 | 1649 | mangleType(out, chainElement.paramTypes.get(i)); | |
| 336 | 1649 | } | |
| 337 | 3114 | out << "E"; | |
| 338 | 3114 | } | |
| 339 | default: { | ||
| 340 | ✗ | const String message = "Type " + chainElement.getName(false, true, true) + " cannot be mangled"; | |
| 341 | ✗ | panic(Error(message.getRaw())); // Internal error | |
| 342 | ✗ | } | |
| 343 | } | ||
| 344 | 455507 | } | |
| 345 | |||
| 346 | 1554 | public f<String> mangleTypeInfoName(const StructBase* structBase) { | |
| 347 | 1554 | StringStream out; | |
| 348 | 1554 | out << "_ZTS"; | |
| 349 | 1554 | mangleType(out, structBase.entry.getQualType()); | |
| 350 | 3108 | return String(out.str()); | |
| 351 | 1554 | } | |
| 352 | |||
| 353 | 1554 | public f<String> mangleTypeInfoValue(const String& value) { | |
| 354 | 3108 | return toString(value.getLength()) + value; | |
| 355 | 1554 | } | |
| 356 | |||
| 357 | 2260 | public f<String> mangleTypeInfo(const StructBase* structBase) { | |
| 358 | 2260 | StringStream out; | |
| 359 | 2260 | out << "_ZTI"; | |
| 360 | 2260 | mangleType(out, structBase.entry.getQualType()); | |
| 361 | 4520 | return String(out.str()); | |
| 362 | 2260 | } | |
| 363 | |||
| 364 | 3122 | public f<String> mangleVTable(const StructBase* structBase) { | |
| 365 | 3122 | StringStream out; | |
| 366 | 3122 | out << "_ZTV"; | |
| 367 | 3122 | mangleType(out, structBase.entry.getQualType()); | |
| 368 | 6244 | return String(out.str()); | |
| 369 | 3122 | } | |
| 370 | |||
| 371 | 1394 | public f<String> mangleVTable(const String& typeName) { | |
| 372 | 2788 | return "_ZTV" + toString(typeName.getLength()) + typeName; | |
| 373 | 1394 | } | |
| 374 |