[lld][WebAssembly] Implement --why-extract flag from the ELF backend

See https://reviews.llvm.org/D109572 for the original ELF version.

Differential Revision: https://reviews.llvm.org/D145431

(cherry picked from commit 8aef04f)

Author: sbc100

lld/test/wasm/why-extract.s

@@ -0,0 +1,87 @@
+# RUN: rm -rf %t && split-file %s %t
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown %t/main.s -o %t/main.o
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown %t/a.s -o %t/a.o
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown %t/a_b.s -o %t/a_b.o
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown %t/b.s -o %t/b.o
+# RUN: llvm-ar rc %t/a.a %t/a.o
+# RUN: llvm-ar rc %t/a_b.a %t/a_b.o
+# RUN: llvm-ar rc %t/b.a %t/b.o
+# RUN: cd %t
+
+## Nothing is extracted from an archive. The file is created with just a header.
+# RUN: wasm-ld main.o a.o b.a -o /dev/null --why-extract=why1.txt
+# RUN: FileCheck %s --input-file=why1.txt --check-prefix=CHECK1 --match-full-lines --strict-whitespace
+
+#      CHECK1:reference	extracted	symbol
+#  CHECK1-NOT:{{.}}
+
+## Some archive members are extracted.
+# RUN: wasm-ld main.o a_b.a b.a -o /dev/null --why-extract=why2.txt
+# RUN: FileCheck %s --input-file=why2.txt --check-prefix=CHECK2 --match-full-lines --strict-whitespace
+
+#      CHECK2:reference	extracted	symbol
+# CHECK2-NEXT:main.o	a_b.a(a_b.o)	a
+# CHECK2-NEXT:a_b.a(a_b.o)	b.a(b.o)	b()
+
+## An undefined symbol error does not suppress the output.
+# RUN: not wasm-ld main.o a_b.a -o /dev/null --why-extract=why3.txt
+# RUN: FileCheck %s --input-file=why3.txt --check-prefix=CHECK3 --match-full-lines --strict-whitespace
+
+## Check that backward references are supported.
+## - means stdout.
+# RUN: wasm-ld b.a a_b.a main.o -o /dev/null --why-extract=- | FileCheck %s --check-prefix=CHECK4
+
+#      CHECK3:reference	extracted	symbol
+# CHECK3-NEXT:main.o	a_b.a(a_b.o)	a
+
+#      CHECK4:reference	extracted	symbol
+# CHECK4-NEXT:a_b.a(a_b.o)	b.a(b.o)	b()
+# CHECK4-NEXT:main.o	a_b.a(a_b.o)	a
+
+# RUN: wasm-ld main.o a_b.a b.a -o /dev/null --no-demangle --why-extract=- | FileCheck %s --check-prefix=MANGLED
+
+# MANGLED: a_b.a(a_b.o)	b.a(b.o)	_Z1bv
+
+# RUN: wasm-ld main.o a.a b.a -o /dev/null -u _Z1bv --why-extract=- | FileCheck %s --check-prefix=UNDEFINED
+
+## We insert -u symbol before processing other files, so its name is <internal>.
+## This is not ideal.
+# UNDEFINED: <internal>	b.a(b.o)	b()
+
+# RUN: wasm-ld main.o a.a b.a -o /dev/null -e _Z1bv --why-extract=- | FileCheck %s --check-prefix=ENTRY
+
+# ENTRY: --entry	b.a(b.o)	b()
+
+# SCRIPT: <internal>	b.a(b.o)	b()
+
+# RUN: not wasm-ld -shared main.o -o /dev/null --why-extract=/ 2>&1 | FileCheck %s --check-prefix=ERR
+
+# ERR: error: cannot open --why-extract= file /: {{.*}}
+
+#--- main.s
+.globl _start
+.functype a () -> ()
+_start:
+  .functype _start () -> ()
+  call a
+  end_function
+
+#--- a.s
+.globl a
+a:
+  .functype a () -> ()
+  end_function
+
+#--- a_b.s
+.functype _Z1bv () -> ()
+.globl a
+a:
+  .functype a () -> ()
+  call _Z1bv
+  end_function
+
+#--- b.s
+.globl _Z1bv
+_Z1bv:
+  .functype _Z1bv () -> ()
+  end_function

lld/wasm/Config.h

@@ -19,6 +19,9 @@
 namespace lld {
 namespace wasm {
 
+class InputFile;
+class Symbol;
+
 // For --unresolved-symbols.
 enum class UnresolvedPolicy { ReportError, Warn, Ignore, ImportDynamic };
 
@@ -72,6 +75,7 @@ struct Configuration {
   llvm::StringRef mapFile;
   llvm::StringRef outputFile;
   llvm::StringRef thinLTOCacheDir;
+  llvm::StringRef whyExtract;
 
   llvm::StringSet<> allowUndefinedSymbols;
   llvm::StringSet<> exportedSymbols;
@@ -82,7 +86,8 @@ struct Configuration {
   std::optional<std::vector<std::string>> extraFeatures;
 
   // The following config options do not directly correspond to any
-  // particular command line options.
+  // particular command line options, and should probably be moved to seperate
+  // Ctx struct as in ELF/Config.h
 
   // True if we are creating position-independent code.
   bool isPic;
@@ -100,6 +105,11 @@ struct Configuration {
   // Will be set to true if bss data segments should be emitted. In most cases
   // this is not necessary.
   bool emitBssSegments = false;
+
+  // A tuple of (reference, extractedFile, sym). Used by --why-extract=.
+  llvm::SmallVector<std::tuple<std::string, const InputFile *, const Symbol &>,
+                    0>
+      whyExtractRecords;
 };
 
 // The only instance of Configuration struct.

lld/wasm/Driver.cpp

@@ -450,6 +450,7 @@ static void readConfigs(opt::InputArgList &args) {
       parseCachePruningPolicy(args.getLastArgValue(OPT_thinlto_cache_policy)),
       "--thinlto-cache-policy: invalid cache policy");
   config->unresolvedSymbols = getUnresolvedSymbolPolicy(args);
+  config->whyExtract = args.getLastArgValue(OPT_why_extract);
   errorHandler().verbose = args.hasArg(OPT_verbose);
   LLVM_DEBUG(errorHandler().verbose = true);
 
@@ -631,7 +632,7 @@ static const char *getReproduceOption(opt::InputArgList &args) {
 }
 
 // Force Sym to be entered in the output. Used for -u or equivalent.
-static Symbol *handleUndefined(StringRef name) {
+static Symbol *handleUndefined(StringRef name, const char *option) {
   Symbol *sym = symtab->find(name);
   if (!sym)
     return nullptr;
@@ -640,8 +641,11 @@ static Symbol *handleUndefined(StringRef name) {
   // eliminate it. Mark the symbol as "used" to prevent it.
   sym->isUsedInRegularObj = true;
 
-  if (auto *lazySym = dyn_cast<LazySymbol>(sym))
+  if (auto *lazySym = dyn_cast<LazySymbol>(sym)) {
     lazySym->fetch();
+    if (!config->whyExtract.empty())
+      config->whyExtractRecords.emplace_back(option, sym->getFile(), *sym);
+  }
 
   return sym;
 }
@@ -653,8 +657,31 @@ static void handleLibcall(StringRef name) {
 
   if (auto *lazySym = dyn_cast<LazySymbol>(sym)) {
     MemoryBufferRef mb = lazySym->getMemberBuffer();
-    if (isBitcode(mb))
+    if (isBitcode(mb)) {
+      if (!config->whyExtract.empty())
+        config->whyExtractRecords.emplace_back("<libcall>", sym->getFile(),
+                                               *sym);
       lazySym->fetch();
+    }
+  }
+}
+
+static void writeWhyExtract() {
+  if (config->whyExtract.empty())
+    return;
+
+  std::error_code ec;
+  raw_fd_ostream os(config->whyExtract, ec, sys::fs::OF_None);
+  if (ec) {
+    error("cannot open --why-extract= file " + config->whyExtract + ": " +
+          ec.message());
+    return;
+  }
+
+  os << "reference\textracted\tsymbol\n";
+  for (auto &entry : config->whyExtractRecords) {
+    os << std::get<0>(entry) << '\t' << toString(std::get<1>(entry)) << '\t'
+       << toString(std::get<2>(entry)) << '\n';
   }
 }
 
@@ -1035,16 +1062,16 @@ void LinkerDriver::linkerMain(ArrayRef<const char *> argsArr) {
 
   // Handle the `--undefined <sym>` options.
   for (auto *arg : args.filtered(OPT_undefined))
-    handleUndefined(arg->getValue());
+    handleUndefined(arg->getValue(), "<internal>");
 
   // Handle the `--export <sym>` options
   // This works like --undefined but also exports the symbol if its found
   for (auto &iter : config->exportedSymbols)
-    handleUndefined(iter.first());
+    handleUndefined(iter.first(), "--export");
 
   Symbol *entrySym = nullptr;
   if (!config->relocatable && !config->entry.empty()) {
-    entrySym = handleUndefined(config->entry);
+    entrySym = handleUndefined(config->entry, "--entry");
     if (entrySym && entrySym->isDefined())
       entrySym->forceExport = true;
     else
@@ -1061,7 +1088,8 @@ void LinkerDriver::linkerMain(ArrayRef<const char *> argsArr) {
       !WasmSym::callCtors->isUsedInRegularObj &&
       WasmSym::callCtors->getName() != config->entry &&
       !config->exportedSymbols.count(WasmSym::callCtors->getName())) {
-    if (Symbol *callDtors = handleUndefined("__wasm_call_dtors")) {
+    if (Symbol *callDtors =
+            handleUndefined("__wasm_call_dtors", "<internal>")) {
       if (auto *callDtorsFunc = dyn_cast<DefinedFunction>(callDtors)) {
         if (callDtorsFunc->signature &&
             (!callDtorsFunc->signature->Params.empty() ||
@@ -1096,6 +1124,8 @@ void LinkerDriver::linkerMain(ArrayRef<const char *> argsArr) {
   if (errorCount())
     return;
 
+  writeWhyExtract();
+
   // Do link-time optimization if given files are LLVM bitcode files.
   // This compiles bitcode files into real object files.
   symtab->compileBitcodeFiles();

lld/wasm/Options.td

@@ -221,6 +221,8 @@ defm whole_archive: B<"whole-archive",
     "Force load of all members in a static library",
     "Do not force load of all members in a static library (default)">;
 
+def why_extract: JJ<"why-extract=">, HelpText<"Print to a file about why archive members are extracted">;
+
 defm check_features: BB<"check-features",
     "Check feature compatibility of linked objects (default)",
     "Ignore feature compatibility of linked objects">;

lld/wasm/SymbolTable.cpp

@@ -524,6 +524,9 @@ Symbol *SymbolTable::addUndefinedFunction(StringRef name,
       lazy->signature = sig;
     } else {
       lazy->fetch();
+      if (!config->whyExtract.empty())
+        config->whyExtractRecords.emplace_back(toString(file), s->getFile(),
+                                               *s);
     }
   } else {
     auto existingFunction = dyn_cast<FunctionSymbol>(s);
@@ -748,7 +751,10 @@ void SymbolTable::addLazy(ArchiveFile *file, const Archive::Symbol *sym) {
   }
 
   LLVM_DEBUG(dbgs() << "replacing existing undefined\n");
+  const InputFile *oldFile = s->getFile();
   file->addMember(sym);
+  if (!config->whyExtract.empty())
+    config->whyExtractRecords.emplace_back(toString(oldFile), s->getFile(), *s);
 }
 
 bool SymbolTable::addComdat(StringRef name) {