[Reproducers] Add file provider

This patch adds the file provider which is responsible for capturing
files used by LLDB.

When capturing a reproducer, we use a file collector that is very
similar to the one used in clang. For every file that we touch, we add
an entry with a mapping from its virtual to its real path. When we
decide to generate a reproducer we copy over the files and their
permission into to reproducer folder.

When replaying a reproducer, we load the VFS mapping and instantiate a
RedirectingFileSystem. The latter will transparently use the files
available in the reproducer.

I've tested this on two macOS machines with an artificial example.
Still, it is very likely that I missed some places where we (still) use
native file system calls. I'm hoping to flesh those out while testing
with more advanced examples. However, I will fix those things in
separate patches.

Differential revision: https://reviews.llvm.org/D54617

llvm-svn: 352538

lldb/include/lldb/Host/FileSystem.h

@@ -11,6 +11,7 @@
 
 #include "lldb/Host/File.h"
 #include "lldb/Utility/DataBufferLLVM.h"
+#include "lldb/Utility/FileCollector.h"
 #include "lldb/Utility/FileSpec.h"
 #include "lldb/Utility/Status.h"
 
@@ -30,15 +31,24 @@ class FileSystem {
   static const char *DEV_NULL;
   static const char *PATH_CONVERSION_ERROR;
 
-  FileSystem() : m_fs(llvm::vfs::getRealFileSystem()) {}
-  FileSystem(llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs) : m_fs(fs) {}
+  FileSystem()
+      : m_fs(llvm::vfs::getRealFileSystem()), m_collector(nullptr),
+        m_mapped(false) {}
+  FileSystem(FileCollector &collector)
+      : m_fs(llvm::vfs::getRealFileSystem()), m_collector(&collector),
+        m_mapped(false) {}
+  FileSystem(llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs,
+             bool mapped = false)
+      : m_fs(fs), m_mapped(mapped) {}
 
   FileSystem(const FileSystem &fs) = delete;
   FileSystem &operator=(const FileSystem &fs) = delete;
 
   static FileSystem &Instance();
 
   static void Initialize();
+  static void Initialize(FileCollector &collector);
+  static llvm::Error Initialize(const FileSpec &mapping);
   static void Initialize(llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs);
   static void Terminate();
 
@@ -167,9 +177,14 @@ class FileSystem {
   std::error_code GetRealPath(const llvm::Twine &path,
                               llvm::SmallVectorImpl<char> &output) const;
 
+  llvm::ErrorOr<std::string> GetExternalPath(const llvm::Twine &path);
+  llvm::ErrorOr<std::string> GetExternalPath(const FileSpec &file_spec);
+
 private:
   static llvm::Optional<FileSystem> &InstanceImpl();
   llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> m_fs;
+  FileCollector *m_collector;
+  bool m_mapped;
 };
 } // namespace lldb_private
 

lldb/include/lldb/Utility/FileCollector.h

@@ -0,0 +1,74 @@
+//===-- FileCollector.h -----------------------------------------*- C++ -*-===//
+//
+// 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
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLDB_UTILITY_FILE_COLLECTOR_H
+#define LLDB_UTILITY_FILE_COLLECTOR_H
+
+#include "lldb/Utility/FileSpec.h"
+
+#include "llvm/ADT/SmallVector.h"
+#include "llvm/ADT/StringMap.h"
+#include "llvm/ADT/StringSet.h"
+#include "llvm/ADT/Twine.h"
+#include "llvm/Support/VirtualFileSystem.h"
+
+#include <mutex>
+
+namespace lldb_private {
+
+/// Collects files into a directory and generates a mapping that can be used by
+/// the VFS.
+class FileCollector {
+public:
+  FileCollector(const FileSpec &root);
+
+  void AddFile(const llvm::Twine &file);
+  void AddFile(const FileSpec &file) { return AddFile(file.GetPath()); }
+
+  /// Write the yaml mapping (for the VFS) to the given file.
+  std::error_code WriteMapping(const FileSpec &mapping_file);
+
+  /// Copy the files into the root directory.
+  ///
+  /// When stop_on_error is true (the default) we abort as soon as one file
+  /// cannot be copied. This is relatively common, for example when a file was
+  /// removed after it was added to the mapping.
+  std::error_code CopyFiles(bool stop_on_error = true);
+
+protected:
+  void AddFileImpl(llvm::StringRef src_path);
+
+  bool MarkAsSeen(llvm::StringRef path) { return m_seen.insert(path).second; }
+
+  bool GetRealPath(llvm::StringRef src_path,
+                   llvm::SmallVectorImpl<char> &result);
+
+  void AddFileToMapping(llvm::StringRef virtual_path,
+                        llvm::StringRef real_path) {
+    m_vfs_writer.addFileMapping(virtual_path, real_path);
+  }
+
+  /// Synchronizes adding files.
+  std::mutex m_mutex;
+
+  /// The root directory where files are copied.
+  FileSpec m_root;
+
+  /// Tracks already seen files so they can be skipped.
+  llvm::StringSet<> m_seen;
+
+  /// The yaml mapping writer.
+  llvm::vfs::YAMLVFSWriter m_vfs_writer;
+
+  /// Caches real_path calls when resolving symlinks.
+  llvm::StringMap<std::string> m_symlink_map;
+};
+
+} // namespace lldb_private
+
+#endif // LLDB_UTILITY_FILE_COLLECTOR_H

lldb/include/lldb/Utility/Reproducer.h

@@ -9,6 +9,7 @@
 #ifndef LLDB_UTILITY_REPRODUCER_H
 #define LLDB_UTILITY_REPRODUCER_H
 
+#include "lldb/Utility/FileCollector.h"
 #include "lldb/Utility/FileSpec.h"
 
 #include "llvm/ADT/DenseMap.h"
@@ -81,6 +82,35 @@ template <typename ThisProviderT> class Provider : public ProviderBase {
   using ProviderBase::ProviderBase; // Inherit constructor.
 };
 
+struct FileInfo {
+  static const char *name;
+  static const char *file;
+};
+
+class FileProvider : public Provider<FileProvider> {
+public:
+  typedef FileInfo info;
+
+  FileProvider(const FileSpec &directory)
+      : Provider(directory),
+        m_collector(directory.CopyByAppendingPathComponent("root")) {}
+
+  FileCollector &GetFileCollector() { return m_collector; }
+
+  void Keep() override {
+    auto mapping = GetRoot().CopyByAppendingPathComponent(info::file);
+    // Temporary files that are removed during execution can cause copy errors.
+    if (auto ec = m_collector.CopyFiles(/*stop_on_error=*/false))
+      return;
+    m_collector.WriteMapping(mapping);
+  }
+
+  static char ID;
+
+private:
+  FileCollector m_collector;
+};
+
 /// The generator is responsible for the logic needed to generate a
 /// reproducer. For doing so it relies on providers, who serialize data that
 /// is necessary for reproducing  a failure.

lldb/lit/Reproducer/Inputs/FileCapture.in

@@ -0,0 +1,3 @@
+run
+reproducer status
+reproducer generate

lldb/lit/Reproducer/Inputs/FileReplay.in

@@ -0,0 +1,2 @@
+reproducer status
+run

lldb/lit/Reproducer/TestFileRepro.test

@@ -0,0 +1,20 @@
+# REQUIRES: system-darwin
+
+# This tests the replaying of GDB remote packets.
+#
+# We issue the same commands and ensure the output is identical to the original
+# process. To ensure we're not actually running the original binary we check
+# that the string "testing" is not printed.
+
+# RUN: %clang %S/Inputs/simple.c -g -o %t.out
+# RUN: %lldb -x -b -s %S/Inputs/FileCapture.in --capture %t.repro -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix CAPTURE
+# RUN: rm %t.out
+# RUN: %lldb -x -b -s %S/Inputs/FileReplay.in --replay %t.repro -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix REPLAY
+
+# CAPTURE: testing
+# REPLAY-NOT: testing
+
+# CHECK: Process {{.*}} exited
+
+# CAPTURE: Reproducer is in capture mode.
+# CAPTURE: Reproducer written

lldb/lit/Reproducer/TestGDBRemoteRepro.test

@@ -7,8 +7,8 @@
 # that the string "testing" is not printed.
 
 # RUN: %clang %S/Inputs/simple.c -g -o %t.out
-# RUN: %lldb -x -b -s %S/Inputs/GDBRemoteCapture.in --capture %T/reproducer -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix CAPTURE
-# RUN: %lldb -x -b -s %S/Inputs/GDBRemoteReplay.in --replay %T/reproducer -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix REPLAY
+# RUN: %lldb -x -b -s %S/Inputs/GDBRemoteCapture.in --capture %t.repro -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix CAPTURE
+# RUN: %lldb -x -b -s %S/Inputs/GDBRemoteReplay.in --replay %t.repro -- %t.out | FileCheck %s --check-prefix CHECK --check-prefix REPLAY
 
 # CHECK: Breakpoint 1
 # CHECK: Process {{.*}} stopped

lldb/source/Host/common/FileSystem.cpp

@@ -11,7 +11,9 @@
 #include "lldb/Utility/LLDBAssert.h"
 #include "lldb/Utility/TildeExpressionResolver.h"
 
+#include "llvm/Support/Errc.h"
 #include "llvm/Support/Errno.h"
+#include "llvm/Support/Error.h"
 #include "llvm/Support/FileSystem.h"
 #include "llvm/Support/Path.h"
 #include "llvm/Support/Program.h"
@@ -47,6 +49,26 @@ void FileSystem::Initialize() {
   InstanceImpl().emplace();
 }
 
+void FileSystem::Initialize(FileCollector &collector) {
+  lldbassert(!InstanceImpl() && "Already initialized.");
+  InstanceImpl().emplace(collector);
+}
+
+llvm::Error FileSystem::Initialize(const FileSpec &mapping) {
+  lldbassert(!InstanceImpl() && "Already initialized.");
+
+  llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> buffer =
+      llvm::vfs::getRealFileSystem()->getBufferForFile(mapping.GetPath());
+
+  if (!buffer)
+    return llvm::errorCodeToError(buffer.getError());
+
+  InstanceImpl().emplace(
+      llvm::vfs::getVFSFromYAML(std::move(buffer.get()), nullptr, ""), true);
+
+  return llvm::Error::success();
+}
+
 void FileSystem::Initialize(IntrusiveRefCntPtr<vfs::FileSystem> fs) {
   lldbassert(!InstanceImpl() && "Already initialized.");
   InstanceImpl().emplace(fs);
@@ -249,18 +271,25 @@ void FileSystem::Resolve(FileSpec &file_spec) {
 std::shared_ptr<DataBufferLLVM>
 FileSystem::CreateDataBuffer(const llvm::Twine &path, uint64_t size,
                              uint64_t offset) {
+  if (m_collector)
+    m_collector->AddFile(path);
+
   const bool is_volatile = !IsLocal(path);
+  const ErrorOr<std::string> external_path = GetExternalPath(path);
+
+  if (!external_path)
+    return nullptr;
 
   std::unique_ptr<llvm::WritableMemoryBuffer> buffer;
   if (size == 0) {
     auto buffer_or_error =
-        llvm::WritableMemoryBuffer::getFile(path, -1, is_volatile);
+        llvm::WritableMemoryBuffer::getFile(*external_path, -1, is_volatile);
     if (!buffer_or_error)
       return nullptr;
     buffer = std::move(*buffer_or_error);
   } else {
     auto buffer_or_error = llvm::WritableMemoryBuffer::getFileSlice(
-        path, size, offset, is_volatile);
+        *external_path, size, offset, is_volatile);
     if (!buffer_or_error)
       return nullptr;
     buffer = std::move(*buffer_or_error);
@@ -380,16 +409,22 @@ static mode_t GetOpenMode(uint32_t permissions) {
 
 Status FileSystem::Open(File &File, const FileSpec &file_spec, uint32_t options,
                         uint32_t permissions) {
+  if (m_collector)
+    m_collector->AddFile(file_spec);
+
   if (File.IsValid())
     File.Close();
 
   const int open_flags = GetOpenFlags(options);
   const mode_t open_mode =
       (open_flags & O_CREAT) ? GetOpenMode(permissions) : 0;
-  const std::string path = file_spec.GetPath();
+
+  auto path = GetExternalPath(file_spec);
+  if (!path)
+    return Status(path.getError());
 
   int descriptor = llvm::sys::RetryAfterSignal(
-      -1, OpenWithFS, *this, path.c_str(), open_flags, open_mode);
+      -1, OpenWithFS, *this, path->c_str(), open_flags, open_mode);
 
   Status error;
   if (!File::DescriptorIsValid(descriptor)) {
@@ -401,3 +436,28 @@ Status FileSystem::Open(File &File, const FileSpec &file_spec, uint32_t options,
   }
   return error;
 }
+
+ErrorOr<std::string> FileSystem::GetExternalPath(const llvm::Twine &path) {
+  if (!m_mapped)
+    return path.str();
+
+  // If VFS mapped we know the underlying FS is a RedirectingFileSystem.
+  ErrorOr<vfs::RedirectingFileSystem::Entry *> E =
+      static_cast<vfs::RedirectingFileSystem &>(*m_fs).lookupPath(path);
+  if (!E) {
+    if (E.getError() == llvm::errc::no_such_file_or_directory) {
+      return path.str();
+    }
+    return E.getError();
+  }
+
+  auto *F = dyn_cast<vfs::RedirectingFileSystem::RedirectingFileEntry>(*E);
+  if (!F)
+    return make_error_code(llvm::errc::not_supported);
+
+  return F->getExternalContentsPath().str();
+}
+
+ErrorOr<std::string> FileSystem::GetExternalPath(const FileSpec &file_spec) {
+  return GetExternalPath(file_spec.GetPath());
+}

lldb/source/Host/macosx/objcxx/Host.mm

@@ -1299,12 +1299,15 @@ static bool ShouldLaunchUsingXPC(ProcessLaunchInfo &launch_info) {
 
   lldb::pid_t pid = LLDB_INVALID_PROCESS_ID;
 
-  if (ShouldLaunchUsingXPC(launch_info)) {
-    error = LaunchProcessXPC(exe_spec.GetPath().c_str(), launch_info, pid);
-  } else {
-    error =
-        LaunchProcessPosixSpawn(exe_spec.GetPath().c_str(), launch_info, pid);
-  }
+  // From now on we'll deal with the external (devirtualized) path.
+  auto exe_path = fs.GetExternalPath(exe_spec);
+  if (!exe_path)
+    return Status(exe_path.getError());
+
+  if (ShouldLaunchUsingXPC(launch_info))
+    error = LaunchProcessXPC(exe_path->c_str(), launch_info, pid);
+  else
+    error = LaunchProcessPosixSpawn(exe_path->c_str(), launch_info, pid);
 
   if (pid != LLDB_INVALID_PROCESS_ID) {
     // If all went well, then set the process ID into the launch info

lldb/source/Initialization/SystemInitializerCommon.cpp

@@ -65,6 +65,7 @@ SystemInitializerCommon::Initialize(const InitializerOptions &options) {
   }
 #endif
 
+  // Initialize the reproducer.
   ReproducerMode mode = ReproducerMode::Off;
   if (options.reproducer_capture)
     mode = ReproducerMode::Capture;
@@ -74,7 +75,23 @@ SystemInitializerCommon::Initialize(const InitializerOptions &options) {
   if (auto e = Reproducer::Initialize(mode, FileSpec(options.reproducer_path)))
     return e;
 
-  FileSystem::Initialize();
+  // Initialize the file system.
+  auto &r = repro::Reproducer::Instance();
+  if (repro::Loader *loader = r.GetLoader()) {
+    FileSpec vfs_mapping = loader->GetFile<FileInfo>();
+    if (vfs_mapping) {
+      if (llvm::Error e = FileSystem::Initialize(vfs_mapping))
+        return e;
+    } else {
+      FileSystem::Initialize();
+    }
+  } else if (repro::Generator *g = r.GetGenerator()) {
+    repro::FileProvider &fp = g->GetOrCreate<repro::FileProvider>();
+    FileSystem::Initialize(fp.GetFileCollector());
+  } else {
+    FileSystem::Initialize();
+  }
+
   Log::Initialize();
   HostInfo::Initialize();
   static Timer::Category func_cat(LLVM_PRETTY_FUNCTION);

lldb/source/Utility/CMakeLists.txt

@@ -54,6 +54,7 @@ add_lldb_library(lldbUtility
   DataEncoder.cpp
   DataExtractor.cpp
   Environment.cpp
+  FileCollector.cpp
   Event.cpp
   FileSpec.cpp
   IOObject.cpp