tools/driver/driver.cpp

//===-- driver.cpp - Clang GCC-Compatible Driver --------------------------===//
//
// 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 is the entry point to the clang driver; it is a thin wrapper
// for functionality in the Driver clang library.
//
//===----------------------------------------------------------------------===//

#include "clang/Driver/Driver.h"
#include "clang/Basic/DiagnosticOptions.h"
#include "clang/Basic/Stack.h"
#include "clang/Config/config.h"
#include "clang/Driver/Compilation.h"
#include "clang/Driver/DriverDiagnostic.h"
#include "clang/Driver/Options.h"
#include "clang/Driver/ToolChain.h"
#include "clang/Frontend/ChainedDiagnosticConsumer.h"
#include "clang/Frontend/CompilerInvocation.h"
#include "clang/Frontend/SerializedDiagnosticPrinter.h"
#include "clang/Frontend/TextDiagnosticPrinter.h"
#include "clang/Frontend/Utils.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/Option/ArgList.h"
#include "llvm/Option/OptTable.h"
#include "llvm/Option/Option.h"
#include "llvm/Support/BuryPointer.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/CrashRecoveryContext.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/Host.h"
#include "llvm/Support/InitLLVM.h"
#include "llvm/Support/Path.h"
#include "llvm/Support/PrettyStackTrace.h"
#include "llvm/Support/Process.h"
#include "llvm/Support/Program.h"
#include "llvm/Support/Regex.h"
#include "llvm/Support/Signals.h"
#include "llvm/Support/StringSaver.h"
#include "llvm/Support/TargetSelect.h"
#include "llvm/Support/Timer.h"
#include "llvm/Support/raw_ostream.h"
#include <memory>
#include <set>
#include <system_error>
using namespace clang;
using namespace clang::driver;
using namespace llvm::opt;

std::string GetExecutablePath(const char *Argv0, bool CanonicalPrefixes) {
  if (!CanonicalPrefixes) {
    SmallString<128> ExecutablePath(Argv0);
    // Do a PATH lookup if Argv0 isn't a valid path.
    if (!llvm::sys::fs::exists(ExecutablePath))
      if (llvm::ErrorOr<std::string> P =
              llvm::sys::findProgramByName(ExecutablePath))
        ExecutablePath = *P;
    return std::string(ExecutablePath.str());
  }

  // This just needs to be some symbol in the binary; C++ doesn't
  // allow taking the address of ::main however.
  void *P = (void*) (intptr_t) GetExecutablePath;
  return llvm::sys::fs::getMainExecutable(Argv0, P);
}

static const char *GetStableCStr(std::set<std::string> &SavedStrings,
                                 StringRef S) {
  return SavedStrings.insert(std::string(S)).first->c_str();
}

/// ApplyQAOverride - Apply a list of edits to the input argument lists.
///
/// The input string is a space separate list of edits to perform,
/// they are applied in order to the input argument lists. Edits
/// should be one of the following forms:
///
///  '#': Silence information about the changes to the command line arguments.
///
///  '^': Add FOO as a new argument at the beginning of the command line.
///
///  '+': Add FOO as a new argument at the end of the command line.
///
///  's/XXX/YYY/': Substitute the regular expression XXX with YYY in the command
///  line.
///
///  'xOPTION': Removes all instances of the literal argument OPTION.
///
///  'XOPTION': Removes all instances of the literal argument OPTION,
///  and the following argument.
///
///  'Ox': Removes all flags matching 'O' or 'O[sz0-9]' and adds 'Ox'
///  at the end of the command line.
///
/// \param OS - The stream to write edit information to.
/// \param Args - The vector of command line arguments.
/// \param Edit - The override command to perform.
/// \param SavedStrings - Set to use for storing string representations.
static void ApplyOneQAOverride(raw_ostream &OS,
                               SmallVectorImpl<const char*> &Args,
                               StringRef Edit,
                               std::set<std::string> &SavedStrings) {
  // This does not need to be efficient.

  if (Edit[0] == '^') {
    const char *Str =
      GetStableCStr(SavedStrings, Edit.substr(1));
    OS << "### Adding argument " << Str << " at beginning\n";
    Args.insert(Args.begin() + 1, Str);
  } else if (Edit[0] == '+') {
    const char *Str =
      GetStableCStr(SavedStrings, Edit.substr(1));
    OS << "### Adding argument " << Str << " at end\n";
    Args.push_back(Str);
  } else if (Edit[0] == 's' && Edit[1] == '/' && Edit.endswith("/") &&
             Edit.slice(2, Edit.size()-1).find('/') != StringRef::npos) {
    StringRef MatchPattern = Edit.substr(2).split('/').first;
    StringRef ReplPattern = Edit.substr(2).split('/').second;
    ReplPattern = ReplPattern.slice(0, ReplPattern.size()-1);

    for (unsigned i = 1, e = Args.size(); i != e; ++i) {
      // Ignore end-of-line response file markers
      if (Args[i] == nullptr)
        continue;
      std::string Repl = llvm::Regex(MatchPattern).sub(ReplPattern, Args[i]);

      if (Repl != Args[i]) {
        OS << "### Replacing '" << Args[i] << "' with '" << Repl << "'\n";
        Args[i] = GetStableCStr(SavedStrings, Repl);
      }
    }
  } else if (Edit[0] == 'x' || Edit[0] == 'X') {
    auto Option = Edit.substr(1);
    for (unsigned i = 1; i < Args.size();) {
      if (Option == Args[i]) {
        OS << "### Deleting argument " << Args[i] << '\n';
        Args.erase(Args.begin() + i);
        if (Edit[0] == 'X') {
          if (i < Args.size()) {
            OS << "### Deleting argument " << Args[i] << '\n';
            Args.erase(Args.begin() + i);
          } else
            OS << "### Invalid X edit, end of command line!\n";
        }
      } else
        ++i;
    }
  } else if (Edit[0] == 'O') {
    for (unsigned i = 1; i < Args.size();) {
      const char *A = Args[i];
      // Ignore end-of-line response file markers
      if (A == nullptr)
        continue;
      if (A[0] == '-' && A[1] == 'O' &&
          (A[2] == '\0' ||
           (A[3] == '\0' && (A[2] == 's' || A[2] == 'z' ||
                             ('0' <= A[2] && A[2] <= '9'))))) {
        OS << "### Deleting argument " << Args[i] << '\n';
        Args.erase(Args.begin() + i);
      } else
        ++i;
    }
    OS << "### Adding argument " << Edit << " at end\n";
    Args.push_back(GetStableCStr(SavedStrings, '-' + Edit.str()));
  } else {
    OS << "### Unrecognized edit: " << Edit << "\n";
  }
}

/// ApplyQAOverride - Apply a comma separate list of edits to the
/// input argument lists. See ApplyOneQAOverride.
static void ApplyQAOverride(SmallVectorImpl<const char*> &Args,
                            const char *OverrideStr,
                            std::set<std::string> &SavedStrings) {
  raw_ostream *OS = &llvm::errs();

  if (OverrideStr[0] == '#') {
    ++OverrideStr;
    OS = &llvm::nulls();
  }

  *OS << "### CCC_OVERRIDE_OPTIONS: " << OverrideStr << "\n";

  // This does not need to be efficient.

  const char *S = OverrideStr;
  while (*S) {
    const char *End = ::strchr(S, ' ');
    if (!End)
      End = S + strlen(S);
    if (End != S)
      ApplyOneQAOverride(*OS, Args, std::string(S, End), SavedStrings);
    S = End;
    if (*S != '\0')
      ++S;
  }
}

extern int cc1_main(ArrayRef<const char *> Argv, const char *Argv0,
                    void *MainAddr);
extern int cc1as_main(ArrayRef<const char *> Argv, const char *Argv0,
                      void *MainAddr);
extern int cc1gen_reproducer_main(ArrayRef<const char *> Argv,
                                  const char *Argv0, void *MainAddr);

static void insertTargetAndModeArgs(const ParsedClangName &NameParts,
                                    SmallVectorImpl<const char *> &ArgVector,
                                    std::set<std::string> &SavedStrings) {
  // Put target and mode arguments at the start of argument list so that
  // arguments specified in command line could override them. Avoid putting
  // them at index 0, as an option like '-cc1' must remain the first.
  int InsertionPoint = 0;
  if (ArgVector.size() > 0)
    ++InsertionPoint;

  if (NameParts.DriverMode) {
    // Add the mode flag to the arguments.
    ArgVector.insert(ArgVector.begin() + InsertionPoint,
                     GetStableCStr(SavedStrings, NameParts.DriverMode));
  }

  if (NameParts.TargetIsValid) {
    const char *arr[] = {"-target", GetStableCStr(SavedStrings,
                                                  NameParts.TargetPrefix)};
    ArgVector.insert(ArgVector.begin() + InsertionPoint,
                     std::begin(arr), std::end(arr));
  }
}

static void getCLEnvVarOptions(std::string &EnvValue, llvm::StringSaver &Saver,
                               SmallVectorImpl<const char *> &Opts) {
  llvm::cl::TokenizeWindowsCommandLine(EnvValue, Saver, Opts);
  // The first instance of '#' should be replaced with '=' in each option.
  for (const char *Opt : Opts)
    if (char *NumberSignPtr = const_cast<char *>(::strchr(Opt, '#')))
      *NumberSignPtr = '=';
}

static void SetBackdoorDriverOutputsFromEnvVars(Driver &TheDriver) {
  auto CheckEnvVar = [](const char *EnvOptSet, const char *EnvOptFile,
                        std::string &OptFile) {
    bool OptSet = !!::getenv(EnvOptSet);
    if (OptSet) {
      if (const char *Var = ::getenv(EnvOptFile))
        OptFile = Var;
    }
    return OptSet;
  };

  TheDriver.CCPrintOptions =
      CheckEnvVar("CC_PRINT_OPTIONS", "CC_PRINT_OPTIONS_FILE",
                  TheDriver.CCPrintOptionsFilename);
  TheDriver.CCPrintHeaders =
      CheckEnvVar("CC_PRINT_HEADERS", "CC_PRINT_HEADERS_FILE",
                  TheDriver.CCPrintHeadersFilename);
  TheDriver.CCLogDiagnostics =
      CheckEnvVar("CC_LOG_DIAGNOSTICS", "CC_LOG_DIAGNOSTICS_FILE",
                  TheDriver.CCLogDiagnosticsFilename);
  TheDriver.CCPrintProcessStats =
      CheckEnvVar("CC_PRINT_PROC_STAT", "CC_PRINT_PROC_STAT_FILE",
                  TheDriver.CCPrintStatReportFilename);
}

static void FixupDiagPrefixExeName(TextDiagnosticPrinter *DiagClient,
                                   const std::string &Path) {
  // If the clang binary happens to be named cl.exe for compatibility reasons,
  // use clang-cl.exe as the prefix to avoid confusion between clang and MSVC.
  StringRef ExeBasename(llvm::sys::path::stem(Path));
  if (ExeBasename.equals_lower("cl"))
    ExeBasename = "clang-cl";
  DiagClient->setPrefix(std::string(ExeBasename));
}

// This lets us create the DiagnosticsEngine with a properly-filled-out
// DiagnosticOptions instance.
static DiagnosticOptions *
CreateAndPopulateDiagOpts(ArrayRef<const char *> argv, bool &UseNewCC1Process) {
  auto *DiagOpts = new DiagnosticOptions;
  unsigned MissingArgIndex, MissingArgCount;
  InputArgList Args = getDriverOptTable().ParseArgs(
      argv.slice(1), MissingArgIndex, MissingArgCount);
  // We ignore MissingArgCount and the return value of ParseDiagnosticArgs.
  // Any errors that would be diagnosed here will also be diagnosed later,
  // when the DiagnosticsEngine actually exists.
  (void)ParseDiagnosticArgs(*DiagOpts, Args);

  UseNewCC1Process =
      Args.hasFlag(clang::driver::options::OPT_fno_integrated_cc1,
                   clang::driver::options::OPT_fintegrated_cc1,
                   /*Default=*/CLANG_SPAWN_CC1);

  return DiagOpts;
}

static void SetInstallDir(SmallVectorImpl<const char *> &argv,
                          Driver &TheDriver, bool CanonicalPrefixes) {
  // Attempt to find the original path used to invoke the driver, to determine
  // the installed path. We do this manually, because we want to support that
  // path being a symlink.
  SmallString<128> InstalledPath(argv[0]);

  // Do a PATH lookup, if there are no directory components.
  if (llvm::sys::path::filename(InstalledPath) == InstalledPath)
    if (llvm::ErrorOr<std::string> Tmp = llvm::sys::findProgramByName(
            llvm::sys::path::filename(InstalledPath.str())))
      InstalledPath = *Tmp;

  // FIXME: We don't actually canonicalize this, we just make it absolute.
  if (CanonicalPrefixes)
    llvm::sys::fs::make_absolute(InstalledPath);

  StringRef InstalledPathParent(llvm::sys::path::parent_path(InstalledPath));
  if (llvm::sys::fs::exists(InstalledPathParent))
    TheDriver.setInstalledDir(InstalledPathParent);
}

static int ExecuteCC1Tool(SmallVectorImpl<const char *> &ArgV) {
  // If we call the cc1 tool from the clangDriver library (through
  // Driver::CC1Main), we need to clean up the options usage count. The options
  // are currently global, and they might have been used previously by the
  // driver.
  llvm::cl::ResetAllOptionOccurrences();

  llvm::BumpPtrAllocator A;
  llvm::StringSaver Saver(A);
  llvm::cl::ExpandResponseFiles(Saver, &llvm::cl::TokenizeGNUCommandLine, ArgV,
                                /*MarkEOLs=*/false);
  StringRef Tool = ArgV[1];
  void *GetExecutablePathVP = (void *)(intptr_t)GetExecutablePath;
  if (Tool == "-cc1")
    return cc1_main(makeArrayRef(ArgV).slice(1), ArgV[0], GetExecutablePathVP);
  if (Tool == "-cc1as")
    return cc1as_main(makeArrayRef(ArgV).slice(2), ArgV[0],
                      GetExecutablePathVP);
  if (Tool == "-cc1gen-reproducer")
    return cc1gen_reproducer_main(makeArrayRef(ArgV).slice(2), ArgV[0],
                                  GetExecutablePathVP);
  // Reject unknown tools.
  llvm::errs() << "error: unknown integrated tool '" << Tool << "'. "
               << "Valid tools include '-cc1' and '-cc1as'.\n";
  return 1;
}

int main(int Argc, const char **Argv) {
  noteBottomOfStack();
  llvm::InitLLVM X(Argc, Argv);
  llvm::setBugReportMsg("PLEASE submit a bug report to " BUG_REPORT_URL
                        " and include the crash backtrace, preprocessed "
                        "source, and associated run script.\n");
  SmallVector<const char *, 256> Args(Argv, Argv + Argc);

  if (llvm::sys::Process::FixupStandardFileDescriptors())
    return 1;

  llvm::InitializeAllTargets();
  auto TargetAndMode = ToolChain::getTargetAndModeFromProgramName(Args[0]);

  llvm::BumpPtrAllocator A;
  llvm::StringSaver Saver(A);

  // Parse response files using the GNU syntax, unless we're in CL mode. There
  // are two ways to put clang in CL compatibility mode: Args[0] is either
  // clang-cl or cl, or --driver-mode=cl is on the command line. The normal
  // command line parsing can't happen until after response file parsing, so we
  // have to manually search for a --driver-mode=cl argument the hard way.
  // Finally, our -cc1 tools don't care which tokenization mode we use because
  // response files written by clang will tokenize the same way in either mode.
  bool ClangCLMode = false;
  if (StringRef(TargetAndMode.DriverMode).equals("--driver-mode=cl") ||
      llvm::find_if(Args, [](const char *F) {
        return F && strcmp(F, "--driver-mode=cl") == 0;
      }) != Args.end()) {
    ClangCLMode = true;
  }
  enum { Default, POSIX, Windows } RSPQuoting = Default;
  for (const char *F : Args) {
    if (strcmp(F, "--rsp-quoting=posix") == 0)
      RSPQuoting = POSIX;
    else if (strcmp(F, "--rsp-quoting=windows") == 0)
      RSPQuoting = Windows;
  }

  // Determines whether we want nullptr markers in Args to indicate response
  // files end-of-lines. We only use this for the /LINK driver argument with
  // clang-cl.exe on Windows.
  bool MarkEOLs = ClangCLMode;

  llvm::cl::TokenizerCallback Tokenizer;
  if (RSPQuoting == Windows || (RSPQuoting == Default && ClangCLMode))
    Tokenizer = &llvm::cl::TokenizeWindowsCommandLine;
  else
    Tokenizer = &llvm::cl::TokenizeGNUCommandLine;

  if (MarkEOLs && Args.size() > 1 && StringRef(Args[1]).startswith("-cc1"))
    MarkEOLs = false;
  llvm::cl::ExpandResponseFiles(Saver, Tokenizer, Args, MarkEOLs);

  // Handle -cc1 integrated tools, even if -cc1 was expanded from a response
  // file.
  auto FirstArg = std::find_if(Args.begin() + 1, Args.end(),
                               [](const char *A) { return A != nullptr; });
  if (FirstArg != Args.end() && StringRef(*FirstArg).startswith("-cc1")) {
    // If -cc1 came from a response file, remove the EOL sentinels.
    if (MarkEOLs) {
      auto newEnd = std::remove(Args.begin(), Args.end(), nullptr);
      Args.resize(newEnd - Args.begin());
    }
    return ExecuteCC1Tool(Args);
  }

  // Handle options that need handling before the real command line parsing in
  // Driver::BuildCompilation()
  bool CanonicalPrefixes = true;
  for (int i = 1, size = Args.size(); i < size; ++i) {
    // Skip end-of-line response file markers
    if (Args[i] == nullptr)
      continue;
    if (StringRef(Args[i]) == "-no-canonical-prefixes") {
      CanonicalPrefixes = false;
      break;
    }
  }

  // Handle CL and _CL_ which permits additional command line options to be
  // prepended or appended.
  if (ClangCLMode) {
    // Arguments in "CL" are prepended.
    llvm::Optional<std::string> OptCL = llvm::sys::Process::GetEnv("CL");
    if (OptCL.hasValue()) {
      SmallVector<const char *, 8> PrependedOpts;
      getCLEnvVarOptions(OptCL.getValue(), Saver, PrependedOpts);

      // Insert right after the program name to prepend to the argument list.
      Args.insert(Args.begin() + 1, PrependedOpts.begin(), PrependedOpts.end());
    }
    // Arguments in "_CL_" are appended.
    llvm::Optional<std::string> Opt_CL_ = llvm::sys::Process::GetEnv("_CL_");
    if (Opt_CL_.hasValue()) {
      SmallVector<const char *, 8> AppendedOpts;
      getCLEnvVarOptions(Opt_CL_.getValue(), Saver, AppendedOpts);

      // Insert at the end of the argument list to append.
      Args.append(AppendedOpts.begin(), AppendedOpts.end());
    }
  }

  std::set<std::string> SavedStrings;
  // Handle CCC_OVERRIDE_OPTIONS, used for editing a command line behind the
  // scenes.
  if (const char *OverrideStr = ::getenv("CCC_OVERRIDE_OPTIONS")) {
    // FIXME: Driver shouldn't take extra initial argument.
    ApplyQAOverride(Args, OverrideStr, SavedStrings);
  }

  std::string Path = GetExecutablePath(Args[0], CanonicalPrefixes);

  // Whether the cc1 tool should be called inside the current process, or if we
  // should spawn a new clang subprocess (old behavior).
  // Not having an additional process saves some execution time of Windows,
  // and makes debugging and profiling easier.
  bool UseNewCC1Process;

  IntrusiveRefCntPtr<DiagnosticOptions> DiagOpts =
      CreateAndPopulateDiagOpts(Args, UseNewCC1Process);

  TextDiagnosticPrinter *DiagClient
    = new TextDiagnosticPrinter(llvm::errs(), &*DiagOpts);
  FixupDiagPrefixExeName(DiagClient, Path);

  IntrusiveRefCntPtr<DiagnosticIDs> DiagID(new DiagnosticIDs());

  DiagnosticsEngine Diags(DiagID, &*DiagOpts, DiagClient);

  if (!DiagOpts->DiagnosticSerializationFile.empty()) {
    auto SerializedConsumer =
        clang::serialized_diags::create(DiagOpts->DiagnosticSerializationFile,
                                        &*DiagOpts, /*MergeChildRecords=*/true);
    Diags.setClient(new ChainedDiagnosticConsumer(
        Diags.takeClient(), std::move(SerializedConsumer)));
  }

  ProcessWarningOptions(Diags, *DiagOpts, /*ReportDiags=*/false);

  Driver TheDriver(Path, llvm::sys::getDefaultTargetTriple(), Diags);
  SetInstallDir(Args, TheDriver, CanonicalPrefixes);
  TheDriver.setTargetAndMode(TargetAndMode);

  insertTargetAndModeArgs(TargetAndMode, Args, SavedStrings);

  SetBackdoorDriverOutputsFromEnvVars(TheDriver);

  if (!UseNewCC1Process) {
    TheDriver.CC1Main = &ExecuteCC1Tool;
    // Ensure the CC1Command actually catches cc1 crashes
    llvm::CrashRecoveryContext::Enable();
  }

  std::unique_ptr<Compilation> C(TheDriver.BuildCompilation(Args));
  int Res = 1;
  bool IsCrash = false;
  if (C && !C->containsError()) {
    SmallVector<std::pair<int, const Command *>, 4> FailingCommands;
    Res = TheDriver.ExecuteCompilation(*C, FailingCommands);

    // Force a crash to test the diagnostics.
    if (TheDriver.GenReproducer) {
      Diags.Report(diag::err_drv_force_crash)
        << !::getenv("FORCE_CLANG_DIAGNOSTICS_CRASH");

      // Pretend that every command failed.
      FailingCommands.clear();
      for (const auto &J : C->getJobs())
        if (const Command *C = dyn_cast<Command>(&J))
          FailingCommands.push_back(std::make_pair(-1, C));

      // Print the bug report message that would be printed if we did actually
      // crash, but only if we're crashing due to FORCE_CLANG_DIAGNOSTICS_CRASH.
      if (::getenv("FORCE_CLANG_DIAGNOSTICS_CRASH"))
        llvm::dbgs() << llvm::getBugReportMsg();
    }

    for (const auto &P : FailingCommands) {
      int CommandRes = P.first;
      const Command *FailingCommand = P.second;
      if (!Res)
        Res = CommandRes;

      // If result status is < 0, then the driver command signalled an error.
      // If result status is 70, then the driver command reported a fatal error.
      // On Windows, abort will return an exit code of 3.  In these cases,
      // generate additional diagnostic information if possible.
      IsCrash = CommandRes < 0 || CommandRes == 70;
#ifdef _WIN32
      IsCrash |= CommandRes == 3;
#endif
#if LLVM_ON_UNIX
      // When running in integrated-cc1 mode, the CrashRecoveryContext returns
      // the same codes as if the program crashed. See section "Exit Status for
      // Commands":
      // https://pubs.opengroup.org/onlinepubs/9699919799/xrat/V4_xcu_chap02.html
      IsCrash |= CommandRes > 128;
#endif
      if (IsCrash) {
        TheDriver.generateCompilationDiagnostics(*C, *FailingCommand);
        break;
      }
    }
  }

  Diags.getClient()->finish();

  if (!UseNewCC1Process && IsCrash) {
    // When crashing in -fintegrated-cc1 mode, bury the timer pointers, because
    // the internal linked list might point to already released stack frames.
    llvm::BuryPointer(llvm::TimerGroup::aquireDefaultGroup());
  } else {
    // If any timers were active but haven't been destroyed yet, print their
    // results now.  This happens in -disable-free mode.
    llvm::TimerGroup::printAll(llvm::errs());
    llvm::TimerGroup::clearAll();
  }

#ifdef _WIN32
  // Exit status should not be negative on Win32, unless abnormal termination.
  // Once abnormal termination was caught, negative status should not be
  // propagated.
  if (Res < 0)
    Res = 1;
#endif

  // If we have multiple failing commands, we return the result of the first
  // failing command.
  return Res;
}