Apply -fmacro-prefix-map to __builtin_FILE()

[Svenny]

Summary

This will match the behavior of GCC.
Patch does not change remapping logic itself, so adding one simple smoke test should be enough.

[MaskRay]

Thanks!

clang/test/CodeGen/macro-prefix-map.c

1 ↗

(On Diff #363838)

The test may be placed in CodeGenCXX/builtin-source-location.cpp

[Svenny]

Moved new test to CodeGenCXX/builtin-source-location.cpp.

[Svenny]

Could you please land it? I don't have commit access.

Also, any chance this patch can get into 13.x release branch? Not sure if it counts as a feature or a bugfix :)

[MaskRay]

I can push it on your behalf, I want to wait a bit for more comments.
(You need to provide name/email so that you can get the proper attribution git commit --amend --author='...')

Looks like a good candidate for release/13.x for build reproducibility.
absl has a macro saying whether __builtin_FILE can be used.
I can find a few other places where people do something like -D__builtin_FILE()="file.cc" probably to work around Clang.

[MaskRay]

clang/include/clang/Driver/Options.td HelpText for -fmacro-prefix-map= needs an update.

https://gcc.gnu.org/onlinedocs/gcc/Preprocessor-Options.html has more documentation

[Svenny]

Great! My commit author string is Pavel Asyutchenko <[e-mail]>.

I have an even uglier hack to work around this: to offset the value of __builtin_FILE() by compile-time calculated location of some known substring like src or include.

[Svenny]

Updated ffile-prefix-map and fmacro-prefix-map in clang/include/clang/Driver/Options.td.

[MaskRay]

Created release/13.x merge request: https://bugs.llvm.org/show_bug.cgi?id=51350
The release manager makes the call.

clang/include/clang/Basic/LangOptions.h

public:
   std::vector<std::string> ModuleFeatures;
 
   /// Options for parsing comments.
   CommentOptions CommentOpts;
 
   /// A list of all -fno-builtin-* function names (e.g., memset).
   std::vector<std::string> NoBuiltinFuncs;
 
+  /// A prefix map for __FILE__, __BASE_FILE__ and __builtin_FILE().
+  std::map<std::string, std::string, std::greater<std::string>> MacroPrefixMap;
+
   /// Triples of the OpenMP targets that the host code codegen should
   /// take into account in order to generate accurate offloading descriptors.
   std::vector<llvm::Triple> OMPTargetTriples;
 
   /// Name of the IR file that contains the result of the OpenMP target
   /// host code generation.
   std::string OMPHostIRFile;
 
bool hasDWARFExceptions() const {
     return getExceptionHandling() == ExceptionHandlingKind::DwarfCFI;
   }
 
   bool hasWasmExceptions() const {
     return getExceptionHandling() == ExceptionHandlingKind::Wasm;
   }
 
   bool isSYCL() const { return SYCLIsDevice || SYCLIsHost; }
+
+  /// Remap path prefix according to -fmacro-prefix-path option.
+  void remapPathPrefix(SmallString<256> &Path) const;
 };
 
 /// Floating point control options
 class FPOptionsOverride;
 class FPOptions {
 public:
   // We start by defining the layout.
   using storage_type = uint16_t;

clang/include/clang/Lex/PreprocessorOptions.h

public:
   /// The set of modules that failed to build.
   ///
   /// This pointer will be shared among all of the compiler instances created
   /// to (re)build modules, so that once a module fails to build anywhere,
   /// other instances will see that the module has failed and won't try to
   /// build it again.
   std::shared_ptr<FailedModulesSet> FailedModules;
 
-  /// A prefix map for __FILE__ and __BASE_FILE__.
-  std::map<std::string, std::string, std::greater<std::string>> MacroPrefixMap;
-
   /// Contains the currently active skipped range mappings for skipping excluded
   /// conditional directives.
   ///
   /// The pointer is passed to the Preprocessor when it's constructed. The
   /// pointer is unowned, the client is responsible for its lifetime.
   ExcludedPreprocessorDirectiveSkipMapping
       *ExcludedConditionalDirectiveSkipMappings = nullptr;
 

clang/lib/AST/Expr.cpp

auto MakeStringLiteral = [&](StringRef Tmp) {
     using LValuePathEntry = APValue::LValuePathEntry;
     StringLiteral *Res = Ctx.getPredefinedStringLiteralFromCache(Tmp);
     // Decay the string to a pointer to the first character.
     LValuePathEntry Path[1] = {LValuePathEntry::ArrayIndex(0)};
     return APValue(Res, CharUnits::Zero(), Path, /*OnePastTheEnd=*/false);
   };
 
   switch (getIdentKind()) {
-  case SourceLocExpr::File:
-    return MakeStringLiteral(PLoc.getFilename());
+  case SourceLocExpr::File: {
+    SmallString<256> Path(PLoc.getFilename());
+    Ctx.getLangOpts().remapPathPrefix(Path);
+    return MakeStringLiteral(Path);
+  }
   case SourceLocExpr::Function: {
     const Decl *CurDecl = dyn_cast_or_null<Decl>(Context);
     return MakeStringLiteral(
         CurDecl ? PredefinedExpr::ComputeName(PredefinedExpr::Function, CurDecl)
                 : std::string(""));
   }
   case SourceLocExpr::Line:
   case SourceLocExpr::Column: {

clang/lib/Basic/LangOptions.cpp

 //===- LangOptions.cpp - C Language Family Language Options ---------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 //  This file defines the LangOptions class.
 //
 //===----------------------------------------------------------------------===//
 
 #include "clang/Basic/LangOptions.h"
+#include "llvm/ADT/SmallString.h"
+#include "llvm/Support/Path.h"
 
 using namespace clang;
 
 LangOptions::LangOptions() : LangStd(LangStandard::lang_unspecified) {
 #define LANGOPT(Name, Bits, Default, Description) Name = Default;
 #define ENUM_LANGOPT(Name, Type, Bits, Default, Description) set##Name(Default);
 #include "clang/Basic/LangOptions.def"
 }
bool LangOptions::isNoBuiltinFunc(StringRef FuncName) const {
   return false;
 }
 
 VersionTuple LangOptions::getOpenCLVersionTuple() const {
   const int Ver = OpenCLCPlusPlus ? OpenCLCPlusPlusVersion : OpenCLVersion;
   return VersionTuple(Ver / 100, (Ver % 100) / 10);
 }
 
+void LangOptions::remapPathPrefix(SmallString<256> &Path) const {
+  for (const auto &Entry : MacroPrefixMap)
+    if (llvm::sys::path::replace_path_prefix(Path, Entry.first, Entry.second))
+      break;
+}
+
 FPOptions FPOptions::defaultWithoutTrailingStorage(const LangOptions &LO) {
   FPOptions result(LO);
   return result;
 }
 
 LLVM_DUMP_METHOD void FPOptions::dump() {
 #define OPTION(NAME, TYPE, WIDTH, PREVIOUS)                                    \
   llvm::errs() << "\n " #NAME " " << get##NAME();

clang/lib/CodeGen/CodeGenModule.cpp

CodeGenModule::CodeGenModule(ASTContext &C, const HeaderSearchOptions &HSO,
   if (CodeGenOpts.CoverageMapping)
     CoverageMapping.reset(new CoverageMappingModuleGen(*this, *CoverageInfo));
 
   // Generate the module name hash here if needed.
   if (CodeGenOpts.UniqueInternalLinkageNames &&
       !getModule().getSourceFileName().empty()) {
     std::string Path = getModule().getSourceFileName();
     // Check if a path substitution is needed from the MacroPrefixMap.
-    for (const auto &Entry : PPO.MacroPrefixMap)
+    for (const auto &Entry : LangOpts.MacroPrefixMap)
       if (Path.rfind(Entry.first, 0) != std::string::npos) {
         Path = Entry.second + Path.substr(Entry.first.size());
         break;
       }
     llvm::MD5 Md5;
     Md5.update(Path);
     llvm::MD5::MD5Result R;
     Md5.final(R);

clang/lib/Frontend/CompilerInvocation.cpp

#undef LANG_OPTION_WITH_MARSHALLING
   if (Opts.CXXABI)
     GenerateArg(Args, OPT_fcxx_abi_EQ, TargetCXXABI::getSpelling(*Opts.CXXABI),
                 SA);
 
   if (Opts.RelativeCXXABIVTables)
     GenerateArg(Args, OPT_fexperimental_relative_cxx_abi_vtables, SA);
   else
     GenerateArg(Args, OPT_fno_experimental_relative_cxx_abi_vtables, SA);
+
+  for (const auto &MP : Opts.MacroPrefixMap)
+    GenerateArg(Args, OPT_fmacro_prefix_map_EQ, MP.first + "=" + MP.second, SA);
 }
 
 bool CompilerInvocation::ParseLangArgs(LangOptions &Opts, ArgList &Args,
                                        InputKind IK, const llvm::Triple &T,
                                        std::vector<std::string> &Includes,
                                        DiagnosticsEngine &Diags) {
   unsigned NumErrorsBefore = Diags.getNumErrors();
 
if (!CXXABI.empty()) {
     }
   }
 
   Opts.RelativeCXXABIVTables =
       Args.hasFlag(options::OPT_fexperimental_relative_cxx_abi_vtables,
                    options::OPT_fno_experimental_relative_cxx_abi_vtables,
                    TargetCXXABI::usesRelativeVTables(T));
 
+  for (const auto &A : Args.getAllArgValues(OPT_fmacro_prefix_map_EQ)) {
+    auto Split = StringRef(A).split('=');
+    Opts.MacroPrefixMap.insert(
+        {std::string(Split.first), std::string(Split.second)});
+  }
+
   return Diags.getNumErrors() == NumErrorsBefore;
 }
 
 static bool isStrictlyPreprocessorAction(frontend::ActionKind Action) {
   switch (Action) {
   case frontend::ASTDeclList:
   case frontend::ASTDump:
   case frontend::ASTPrint:

 #undef PREPROCESSOR_OPTION_WITH_MARSHALLING
 
   if (Opts.PCHWithHdrStop && !Opts.PCHWithHdrStopCreate)
     GenerateArg(Args, OPT_pch_through_hdrstop_use, SA);
 
   for (const auto &D : Opts.DeserializedPCHDeclsToErrorOn)
     GenerateArg(Args, OPT_error_on_deserialized_pch_decl, D, SA);
 
-  for (const auto &MP : Opts.MacroPrefixMap)
-    GenerateArg(Args, OPT_fmacro_prefix_map_EQ, MP.first + "=" + MP.second, SA);
-
   if (Opts.PrecompiledPreambleBytes != std::make_pair(0u, false))
     GenerateArg(Args, OPT_preamble_bytes_EQ,
                 Twine(Opts.PrecompiledPreambleBytes.first) + "," +
                     (Opts.PrecompiledPreambleBytes.second ? "1" : "0"),
                 SA);
 
   for (const auto &M : Opts.Macros) {
     // Don't generate __CET__ macro definitions. They are implied by the

 #undef PREPROCESSOR_OPTION_WITH_MARSHALLING
 
   Opts.PCHWithHdrStop = Args.hasArg(OPT_pch_through_hdrstop_create) ||
                         Args.hasArg(OPT_pch_through_hdrstop_use);
 
   for (const auto *A : Args.filtered(OPT_error_on_deserialized_pch_decl))
     Opts.DeserializedPCHDeclsToErrorOn.insert(A->getValue());
 
-  for (const auto &A : Args.getAllArgValues(OPT_fmacro_prefix_map_EQ)) {
-    auto Split = StringRef(A).split('=');
-    Opts.MacroPrefixMap.insert(
-        {std::string(Split.first), std::string(Split.second)});
-  }
-
   if (const Arg *A = Args.getLastArg(OPT_preamble_bytes_EQ)) {
     StringRef Value(A->getValue());
     size_t Comma = Value.find(',');
     unsigned Bytes = 0;
     unsigned EndOfLine = 0;
 
     if (Comma == StringRef::npos ||
         Value.substr(0, Comma).getAsInteger(10, Bytes) ||

clang/lib/Lex/PPMacroExpansion.cpp

 /// Implements the __is_target_environment builtin macro.
 static bool isTargetEnvironment(const TargetInfo &TI,
                                 const IdentifierInfo *II) {
   std::string EnvName = (llvm::Twine("---") + II->getName().lower()).str();
   llvm::Triple Env(EnvName);
   return TI.getTriple().getEnvironment() == Env.getEnvironment();
 }
 
-static void remapMacroPath(
-    SmallString<256> &Path,
-    const std::map<std::string, std::string, std::greater<std::string>>
-        &MacroPrefixMap) {
-  for (const auto &Entry : MacroPrefixMap)
-    if (llvm::sys::path::replace_path_prefix(Path, Entry.first, Entry.second))
-      break;
-}
-
 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded
 /// as a builtin macro, handle it and return the next token as 'Tok'.
 void Preprocessor::ExpandBuiltinMacro(Token &Tok) {
   // Figure out which token this is.
   IdentifierInfo *II = Tok.getIdentifierInfo();
   assert(II && "Can't be a macro without id info!");
 
   // If this is an _Pragma or Microsoft __pragma directive, expand it,
if (PLoc.isValid()) {
         StringRef PLFileName = llvm::sys::path::filename(PLoc.getFilename());
         if (PLFileName != "")
           FN += PLFileName;
         else
           FN += PLoc.getFilename();
       } else {
         FN += PLoc.getFilename();
       }
-      remapMacroPath(FN, PPOpts->MacroPrefixMap);
+      getLangOpts().remapPathPrefix(FN);
       Lexer::Stringify(FN);
       OS << '"' << FN << '"';
     }
     Tok.setKind(tok::string_literal);
   } else if (II == Ident__DATE__) {
     Diag(Tok.getLocation(), diag::warn_pp_date_time);
     if (!DATELoc.isValid())
       ComputeDATE_TIME(DATELoc, TIMELoc, *this);

clang/test/CodeGenCXX/builtin-source-location.cpp

 // RUN: %clang_cc1 -std=c++2a -fblocks %s -triple x86_64-unknown-unknown -emit-llvm -o %t.ll
 
+// This needs to be performed before #line directives which alter filename
+// RUN: %clang_cc1 -fmacro-prefix-map=%p=/UNLIKELY/PATH -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-PREFIX-MAP
+//
+// CHECK-PREFIX-MAP: /UNLIKELY/PATH{{/|\\\\}}builtin-source-location.cpp
+void testRemap() {
+  const char *file = __builtin_FILE();
+}
+
 #line 8 "builtin-source-location.cpp"
 
 struct source_location {
 private:
   unsigned int __m_line = 0;
   unsigned int __m_col = 0;
   const char *__m_file = nullptr;
   const char *__m_func = nullptr;