Remove metadata generation from wasm-emscripten-finalize (#4863)

This is no longer needed by emscripten as of:
https://github.com/emscripten-core/emscripten/pull/16529

3 files changed, 3 insertions, 269 deletions

diff --git a/src/tools/wasm-emscripten-finalize.cpp b/src/tools/wasm-emscripten-finalize.cpp
index b4e27ca5c..99d03cec0 100644
--- a/src/tools/wasm-emscripten-finalize.cpp
+++ b/src/tools/wasm-emscripten-finalize.cpp
@@ -47,7 +47,6 @@ int main(int argc, const char* argv[]) {
   std::string outputSourceMapUrl;
   std::string dataSegmentFile;
   bool emitBinary = true;
-  bool emitMetadata = true;
   bool debugInfo = false;
   bool DWARF = false;
   bool sideModule = false;
@@ -95,13 +94,6 @@ int main(int argc, const char* argv[]) {
          WasmEmscriptenFinalizeOption,
          Options::Arguments::Zero,
          [&emitBinary](Options*, const std::string&) { emitBinary = false; })
-    .add(
-      "--no-emit-metadata",
-      "-n",
-      "Skip the writing to emscripten metadata JSON to stdout.",
-      WasmEmscriptenFinalizeOption,
-      Options::Arguments::Zero,
-      [&emitMetadata](Options*, const std::string&) { emitMetadata = false; })
     .add("--global-base",
          "",
          "The address at which static globals were placed",
@@ -228,9 +220,8 @@ int main(int argc, const char* argv[]) {
   options.applyFeatures(wasm);
   ModuleReader reader;
   // If we are not writing output then we definitely don't need to read debug
-  // info, as it does not affect the metadata we will emit. (However, if we
-  // emit output then definitely load the names section so that we roundtrip
-  // names properly.)
+  // info. However, if we emit output then definitely load the names section so
+  // that we roundtrip names properly.
   reader.setDebugInfo(writeOutput);
   reader.setDWARF(DWARF && writeOutput);
   if (!writeOutput) {
@@ -296,7 +287,6 @@ int main(int argc, const char* argv[]) {
                             : ABI::LegalizationLevel::Minimal));
   }
 
-  // Strip target features section (its information is in the metadata)
   passRunner.add("strip-target-features");
 
   // If DWARF is unused, strip it out. This avoids us keeping it alive
@@ -307,15 +297,7 @@ int main(int argc, const char* argv[]) {
 
   passRunner.run();
 
-  BYN_TRACE("generated metadata\n");
-  // Substantial changes to the wasm are done, enough to create the metadata.
-  std::string metadata;
-  if (emitMetadata) {
-    metadata = generator.generateEmscriptenMetadata();
-  }
-
-  // Finally, separate out data segments if relevant (they may have been needed
-  // for metadata).
+  // Finally, separate out data segments if relevant
   if (!dataSegmentFile.empty()) {
     Output memInitFile(dataSegmentFile, Flags::Binary);
     if (globalBase == INVALID_BASE) {
@@ -338,16 +320,6 @@ int main(int argc, const char* argv[]) {
       writer.setSourceMapUrl(outputSourceMapUrl);
     }
     writer.write(wasm, output);
-    if (emitMetadata && !emitBinary) {
-      output << "(;\n";
-      output << "--BEGIN METADATA --\n" << metadata << "-- END METADATA --\n";
-      output << ";)\n";
-    }
-  }
-  // If we emit text then we emitted the metadata together with that text
-  // earlier. Otherwise emit it to stdout.
-  if (emitMetadata && emitBinary) {
-    std::cout << metadata;
   }
   return 0;
 }
diff --git a/src/wasm-emscripten.h b/src/wasm-emscripten.h
index 694eb1368..4d20b8706 100644
--- a/src/wasm-emscripten.h
+++ b/src/wasm-emscripten.h
@@ -33,8 +33,6 @@ public:
     : wasm(wasm), builder(wasm), stackPointerOffset(stackPointerOffset),
       useStackPointerGlobal(stackPointerOffset == 0) {}
 
-  std::string generateEmscriptenMetadata();
-
   void fixInvokeFunctionNames();
 
   // Emits the data segments to a file. The file contains data from address base
diff --git a/src/wasm/wasm-emscripten.cpp b/src/wasm/wasm-emscripten.cpp
index f143310ce..0bc74580d 100644
--- a/src/wasm/wasm-emscripten.cpp
+++ b/src/wasm/wasm-emscripten.cpp
@@ -188,242 +188,6 @@ struct AsmConst {
   std::string code;
 };
 
-static Address getExportedAddress(Module& wasm, Export* export_) {
-  Global* g = wasm.getGlobal(export_->value);
-  auto* addrConst = g->init->dynCast<Const>();
-  return addrConst->value.getUnsigned();
-}
-
-static std::vector<AsmConst> findEmAsmConsts(Module& wasm,
-                                             bool minimizeWasmChanges) {
-  // Newer version of emscripten/llvm export these symbols so we can use them to
-  // find all the EM_ASM constants.   Sadly __start_em_asm and __stop_em_asm
-  // don't alwasy mark the start and end of segment because in dynamic linking
-  // we merge all data segments into one.
-  Export* start = wasm.getExportOrNull("__start_em_asm");
-  Export* end = wasm.getExportOrNull("__stop_em_asm");
-  if (!start && !end) {
-    BYN_TRACE("findEmAsmConsts: no start/stop symbols\n");
-    return {};
-  }
-
-  if (!start || !end) {
-    Fatal() << "Found only one of __start_em_asm and __stop_em_asm";
-  }
-
-  StringConstantTracker stringTracker(wasm);
-  Address startAddress = getExportedAddress(wasm, start);
-  Address endAddress = getExportedAddress(wasm, end);
-  for (Index i = 0; i < wasm.dataSegments.size(); i++) {
-    Address segmentStart = stringTracker.segmentOffsets[i];
-    size_t segmentSize = wasm.dataSegments[i]->data.size();
-    if (segmentStart <= startAddress &&
-        segmentStart + segmentSize >= endAddress) {
-      std::vector<AsmConst> asmConsts;
-      Address address = startAddress;
-      while (address < endAddress) {
-        auto code = stringTracker.stringAtAddr(address);
-        asmConsts.push_back({address, code});
-        address.addr += strlen(code) + 1;
-      }
-      assert(asmConsts.size());
-      return asmConsts;
-    }
-  }
-  Fatal() << "Segment data not found between symbols __start_em_asm and "
-             "__stop_em_asm";
-}
-
-struct EmJsWalker : public PostWalker<EmJsWalker> {
-  Module& wasm;
-  StringConstantTracker stringTracker;
-
-  std::map<std::string, std::string> codeByName;
-
-  EmJsWalker(Module& _wasm) : wasm(_wasm), stringTracker(_wasm) {}
-
-  void visitExport(Export* curr) {
-    if (!curr->name.startsWith(EM_JS_PREFIX.str)) {
-      return;
-    }
-
-    Address address;
-    if (curr->kind == ExternalKind::Global) {
-      auto* global = wasm.getGlobal(curr->value);
-      Const* const_ = global->init->cast<Const>();
-      address = const_->value.getUnsigned();
-    } else if (curr->kind == ExternalKind::Function) {
-      auto* func = wasm.getFunction(curr->value);
-      // An EM_JS has a single const in the body. Typically it is just returned,
-      // but in unoptimized code it might be stored to a local and loaded from
-      // there, and in relocatable code it might get added to __memory_base etc.
-      FindAll<Const> consts(func->body);
-      if (consts.list.size() != 1) {
-        Fatal() << "Unexpected generated __em_js__ function body: "
-                << curr->name;
-      }
-      auto* addrConst = consts.list[0];
-      address = addrConst->value.getUnsigned();
-    } else {
-      return;
-    }
-
-    auto code = stringTracker.stringAtAddr(address);
-    auto funcName = std::string(curr->name.stripPrefix(EM_JS_PREFIX.str));
-    codeByName[funcName] = code;
-  }
-};
-
-EmJsWalker findEmJsFuncsAndReturnWalker(Module& wasm) {
-  EmJsWalker walker(wasm);
-  walker.walkModule(&wasm);
-  return walker;
-}
-
-std::string EmscriptenGlueGenerator::generateEmscriptenMetadata() {
-  bool commaFirst;
-  auto nextElement = [&commaFirst]() {
-    if (commaFirst) {
-      commaFirst = false;
-      return "\n    ";
-    } else {
-      return ",\n    ";
-    }
-  };
-
-  std::stringstream meta;
-  meta << "{\n";
-
-  std::vector<AsmConst> asmConsts = findEmAsmConsts(wasm, minimizeWasmChanges);
-
-  // print
-  commaFirst = true;
-  if (!asmConsts.empty()) {
-    meta << "  \"asmConsts\": {";
-    for (auto& asmConst : asmConsts) {
-      meta << nextElement();
-      meta << '"' << asmConst.id << "\": \"" << escape(asmConst.code) << "\"";
-    }
-    meta << "\n  },\n";
-  }
-
-  EmJsWalker emJsWalker = findEmJsFuncsAndReturnWalker(wasm);
-  if (!emJsWalker.codeByName.empty()) {
-    meta << "  \"emJsFuncs\": {";
-    commaFirst = true;
-    for (auto& [name, code] : emJsWalker.codeByName) {
-      meta << nextElement();
-      meta << '"' << name << "\": \"" << escape(code) << '"';
-    }
-    meta << "\n  },\n";
-  }
-
-  // Avoid adding duplicate imports to `declares' or `invokeFuncs`.  Even
-  // though we might import the same function multiple times (i.e. with
-  // different sigs) we only need to list is in the metadata once.
-  std::set<std::string> declares;
-  std::set<std::string> invokeFuncs;
-
-  // We use the `base` rather than the `name` of the imports here and below
-  // becasue this is the externally visible name that the embedder (JS) will
-  // see.
-  meta << "  \"declares\": [";
-  commaFirst = true;
-  ModuleUtils::iterImportedFunctions(wasm, [&](Function* import) {
-    if (emJsWalker.codeByName.count(import->base.str) == 0 &&
-        !import->base.startsWith("invoke_")) {
-      if (declares.insert(import->base.str).second) {
-        meta << nextElement() << '"' << import->base.str << '"';
-      }
-    }
-  });
-  meta << "\n  ],\n";
-
-  meta << "  \"globalImports\": [";
-  commaFirst = true;
-  ModuleUtils::iterImportedGlobals(wasm, [&](Global* import) {
-    meta << nextElement() << '"' << import->base.str << '"';
-  });
-  meta << "\n  ],\n";
-
-  if (!wasm.exports.empty()) {
-    meta << "  \"exports\": [";
-    commaFirst = true;
-    for (const auto& ex : wasm.exports) {
-      if (ex->kind == ExternalKind::Function || ex->kind == ExternalKind::Tag) {
-        meta << nextElement() << '"' << ex->name.str << '"';
-      }
-    }
-    meta << "\n  ],\n";
-
-    meta << "  \"namedGlobals\": {";
-    commaFirst = true;
-    for (const auto& ex : wasm.exports) {
-      if (ex->kind == ExternalKind::Global) {
-        const Global* g = wasm.getGlobal(ex->value);
-        assert(g->type == Type::i32 || g->type == Type::i64);
-        Const* init = g->init->cast<Const>();
-        uint64_t addr = init->value.getInteger();
-        meta << nextElement() << '"' << ex->name.str << "\" : \"" << addr
-             << '"';
-      }
-    }
-    meta << "\n  },\n";
-  }
-
-  meta << "  \"invokeFuncs\": [";
-  commaFirst = true;
-  ModuleUtils::iterImportedFunctions(wasm, [&](Function* import) {
-    if (import->module == ENV && import->base.startsWith("invoke_")) {
-      if (invokeFuncs.insert(import->base.str).second) {
-        meta << nextElement() << '"' << import->base.str << '"';
-      }
-    }
-  });
-  meta << "\n  ],\n";
-
-  // In normal mode we attempt to determine if main takes argumnts or not
-  // In standalone mode we export _start instead and rely on the presence
-  // of the __wasi_args_get and __wasi_args_sizes_get syscalls allow us to
-  // DCE to the argument handling JS code instead.
-  if (!standalone) {
-    auto mainReadsParams = false;
-    auto* exp = wasm.getExportOrNull("main");
-    if (!exp) {
-      exp = wasm.getExportOrNull("__main_argc_argv");
-    }
-    if (exp) {
-      if (exp->kind == ExternalKind::Function) {
-        auto* main = wasm.getFunction(exp->value);
-        mainReadsParams = main->getNumParams() > 0;
-        if (mainReadsParams) {
-          // Main could also be stub that just calls __original_main with
-          // no parameters.
-          // TODO(sbc): Remove this once https://reviews.llvm.org/D75277
-          // lands.
-          if (auto* call = main->body->dynCast<Call>()) {
-            if (call->operands.empty()) {
-              mainReadsParams = false;
-            }
-          }
-        }
-      }
-    }
-    meta << "  \"mainReadsParams\": " << int(mainReadsParams) << ",\n";
-  }
-
-  meta << "  \"features\": [";
-  commaFirst = true;
-  wasm.features.iterFeatures([&](FeatureSet::Feature f) {
-    meta << nextElement() << "\"--enable-" << FeatureSet::toString(f) << '"';
-  });
-  meta << "\n  ]\n";
-
-  meta << "}\n";
-
-  return meta.str();
-}
-
 void EmscriptenGlueGenerator::separateDataSegments(Output* outfile,
                                                    Address base) {
   size_t lastEnd = 0;