content/browser/zygote_host/zygote_host_impl_linux.cc

// Copyright (c) 2012 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#include "content/browser/zygote_host/zygote_host_impl_linux.h"

#include <string.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <unistd.h>

#include "base/base_switches.h"
#include "base/command_line.h"
#include "base/environment.h"
#include "base/files/file_enumerator.h"
#include "base/files/file_util.h"
#include "base/files/scoped_file.h"
#include "base/linux_util.h"
#include "base/logging.h"
#include "base/memory/linked_ptr.h"
#include "base/memory/scoped_ptr.h"
#include "base/memory/scoped_vector.h"
#include "base/metrics/histogram.h"
#include "base/path_service.h"
#include "base/posix/eintr_wrapper.h"
#include "base/posix/unix_domain_socket_linux.h"
#include "base/process/launch.h"
#include "base/process/memory.h"
#include "base/process/process_handle.h"
#include "base/strings/string_number_conversions.h"
#include "base/strings/string_util.h"
#include "base/strings/utf_string_conversions.h"
#include "base/time/time.h"
#include "content/browser/renderer_host/render_sandbox_host_linux.h"
#include "content/common/child_process_sandbox_support_impl_linux.h"
#include "content/common/zygote_commands_linux.h"
#include "content/public/browser/content_browser_client.h"
#include "content/public/common/content_switches.h"
#include "content/public/common/result_codes.h"
#include "sandbox/linux/services/credentials.h"
#include "sandbox/linux/services/namespace_sandbox.h"
#include "sandbox/linux/services/namespace_utils.h"
#include "sandbox/linux/suid/client/setuid_sandbox_host.h"
#include "sandbox/linux/suid/common/sandbox.h"
#include "ui/base/ui_base_switches.h"
#include "ui/gfx/switches.h"

#if defined(USE_TCMALLOC)
#include "third_party/tcmalloc/chromium/src/gperftools/heap-profiler.h"
#endif

namespace content {

namespace {

// Receive a fixed message on fd and return the sender's PID.
// Returns true if the message received matches the expected message.
bool ReceiveFixedMessage(int fd,
                         const char* expect_msg,
                         size_t expect_len,
                         base::ProcessId* sender_pid) {
  char buf[expect_len + 1];
  ScopedVector<base::ScopedFD> fds_vec;

  const ssize_t len = UnixDomainSocket::RecvMsgWithPid(
      fd, buf, sizeof(buf), &fds_vec, sender_pid);
  if (static_cast<size_t>(len) != expect_len)
    return false;
  if (memcmp(buf, expect_msg, expect_len) != 0)
    return false;
  if (!fds_vec.empty())
    return false;
  return true;
}

}  // namespace

// static
ZygoteHost* ZygoteHost::GetInstance() {
  return ZygoteHostImpl::GetInstance();
}

ZygoteHostImpl::ZygoteHostImpl()
    : control_fd_(-1),
      control_lock_(),
      pid_(-1),
      init_(false),
      use_suid_sandbox_for_adj_oom_score_(false),
      sandbox_binary_(),
      have_read_sandbox_status_word_(false),
      sandbox_status_(0),
      child_tracking_lock_(),
      list_of_running_zygote_children_(),
      should_teardown_after_last_child_exits_(false) {}

ZygoteHostImpl::~ZygoteHostImpl() { TearDown(); }

// static
ZygoteHostImpl* ZygoteHostImpl::GetInstance() {
  return Singleton<ZygoteHostImpl>::get();
}

void ZygoteHostImpl::Init(const std::string& sandbox_cmd) {
  DCHECK(!init_);
  init_ = true;

  base::FilePath chrome_path;
  CHECK(PathService::Get(base::FILE_EXE, &chrome_path));
  base::CommandLine cmd_line(chrome_path);

  cmd_line.AppendSwitchASCII(switches::kProcessType, switches::kZygoteProcess);

  int fds[2];
  CHECK(socketpair(AF_UNIX, SOCK_SEQPACKET, 0, fds) == 0);
  CHECK(UnixDomainSocket::EnableReceiveProcessId(fds[0]));
  base::FileHandleMappingVector fds_to_map;
  fds_to_map.push_back(std::make_pair(fds[1], kZygoteSocketPairFd));

  base::LaunchOptions options;
  const base::CommandLine& browser_command_line =
      *base::CommandLine::ForCurrentProcess();
  if (browser_command_line.HasSwitch(switches::kZygoteCmdPrefix)) {
    cmd_line.PrependWrapper(
        browser_command_line.GetSwitchValueNative(switches::kZygoteCmdPrefix));
  }
  // Append any switches from the browser process that need to be forwarded on
  // to the zygote/renderers.
  // Should this list be obtained from browser_render_process_host.cc?
  static const char* kForwardSwitches[] = {
    switches::kAllowSandboxDebugging,
    switches::kDisableSeccompFilterSandbox,
    switches::kEnableLogging,  // Support, e.g., --enable-logging=stderr.
    // Zygote process needs to know what resources to have loaded when it
    // becomes a renderer process.
    switches::kForceDeviceScaleFactor,
    switches::kLoggingLevel,
    switches::kNoSandbox,
    switches::kPpapiInProcess,
    switches::kRegisterPepperPlugins,
    switches::kV,
    switches::kVModule,
  };
  cmd_line.CopySwitchesFrom(browser_command_line, kForwardSwitches,
                            arraysize(kForwardSwitches));

  GetContentClient()->browser()->AppendExtraCommandLineSwitches(&cmd_line, -1);

  sandbox_binary_ = sandbox_cmd.c_str();

  const bool using_namespace_sandbox = ShouldUseNamespaceSandbox();
  // A non empty sandbox_cmd means we want a SUID sandbox.
  const bool using_suid_sandbox =
      !sandbox_cmd.empty() && !using_namespace_sandbox;

  // Use the SUID sandbox for adjusting OOM scores when we are using the setuid
  // or namespace sandbox. This is needed beacuse the processes are
  // non-dumpable, so /proc/pid/oom_score_adj can only be written by root.
  use_suid_sandbox_for_adj_oom_score_ =
      using_namespace_sandbox || using_suid_sandbox;

  // Start up the sandbox host process and get the file descriptor for the
  // renderers to talk to it.
  const int sfd = RenderSandboxHostLinux::GetInstance()->GetRendererSocket();
  fds_to_map.push_back(std::make_pair(sfd, GetSandboxFD()));

  base::ScopedFD dummy_fd;
  if (using_suid_sandbox) {
    scoped_ptr<sandbox::SetuidSandboxHost> sandbox_host(
        sandbox::SetuidSandboxHost::Create());
    sandbox_host->PrependWrapper(&cmd_line);
    sandbox_host->SetupLaunchOptions(&options, &fds_to_map, &dummy_fd);
    sandbox_host->SetupLaunchEnvironment();
  }

  options.fds_to_remap = &fds_to_map;
  base::Process process =
      using_namespace_sandbox
          ? sandbox::NamespaceSandbox::LaunchProcess(cmd_line, options)
          : base::LaunchProcess(cmd_line, options);
  CHECK(process.IsValid()) << "Failed to launch zygote process";

  dummy_fd.reset();

  if (using_suid_sandbox) {
    // The SUID sandbox will execute the zygote in a new PID namespace, and
    // the main zygote process will then fork from there.  Watch now our
    // elaborate dance to find and validate the zygote's PID.

    // First we receive a message from the zygote boot process.
    base::ProcessId boot_pid;
    CHECK(ReceiveFixedMessage(
        fds[0], kZygoteBootMessage, sizeof(kZygoteBootMessage), &boot_pid));

    // Within the PID namespace, the zygote boot process thinks it's PID 1,
    // but its real PID can never be 1. This gives us a reliable test that
    // the kernel is translating the sender's PID to our namespace.
    CHECK_GT(boot_pid, 1)
        << "Received invalid process ID for zygote; kernel might be too old? "
           "See crbug.com/357670 or try using --"
        << switches::kDisableSetuidSandbox << " to workaround.";

    // Now receive the message that the zygote's ready to go, along with the
    // main zygote process's ID.
    CHECK(ReceiveFixedMessage(
        fds[0], kZygoteHelloMessage, sizeof(kZygoteHelloMessage), &pid_));
    CHECK_GT(pid_, 1);

    if (process.Pid() != pid_) {
      // Reap the sandbox.
      base::EnsureProcessGetsReaped(process.Pid());
    }
  } else {
    // Not using the SUID sandbox.
    // Note that ~base::Process() will reset the internal value, but there's no
    // real "handle" on POSIX so that is safe.
    pid_ = process.Pid();
  }

  close(fds[1]);
  control_fd_ = fds[0];

  Pickle pickle;
  pickle.WriteInt(kZygoteCommandGetSandboxStatus);
  if (!SendMessage(pickle, NULL))
    LOG(FATAL) << "Cannot communicate with zygote";
  // We don't wait for the reply. We'll read it in ReadReply.
}

void ZygoteHostImpl::TearDownAfterLastChild() {
  bool do_teardown = false;
  {
    base::AutoLock lock(child_tracking_lock_);
    should_teardown_after_last_child_exits_ = true;
    do_teardown = list_of_running_zygote_children_.empty();
  }
  if (do_teardown) {
    TearDown();
  }
}

// Note: this is also called from the destructor.
void ZygoteHostImpl::TearDown() {
  base::AutoLock lock(control_lock_);
  if (control_fd_ > -1) {
    // Closing the IPC channel will act as a notification to exit
    // to the Zygote.
    if (IGNORE_EINTR(close(control_fd_))) {
      PLOG(ERROR) << "Could not close Zygote control channel.";
      NOTREACHED();
    }
    control_fd_ = -1;
  }
}

void ZygoteHostImpl::ZygoteChildBorn(pid_t process) {
  base::AutoLock lock(child_tracking_lock_);
  bool new_element_inserted =
      list_of_running_zygote_children_.insert(process).second;
  DCHECK(new_element_inserted);
}

void ZygoteHostImpl::ZygoteChildDied(pid_t process) {
  bool do_teardown = false;
  {
    base::AutoLock lock(child_tracking_lock_);
    size_t num_erased = list_of_running_zygote_children_.erase(process);
    DCHECK_EQ(1U, num_erased);
    do_teardown = should_teardown_after_last_child_exits_ &&
                  list_of_running_zygote_children_.empty();
  }
  if (do_teardown) {
    TearDown();
  }
}

bool ZygoteHostImpl::SendMessage(const Pickle& data,
                                 const std::vector<int>* fds) {
  DCHECK_NE(-1, control_fd_);
  CHECK(data.size() <= kZygoteMaxMessageLength)
      << "Trying to send too-large message to zygote (sending " << data.size()
      << " bytes, max is " << kZygoteMaxMessageLength << ")";
  CHECK(!fds || fds->size() <= UnixDomainSocket::kMaxFileDescriptors)
      << "Trying to send message with too many file descriptors to zygote "
      << "(sending " << fds->size() << ", max is "
      << UnixDomainSocket::kMaxFileDescriptors << ")";

  return UnixDomainSocket::SendMsg(control_fd_,
                                   data.data(), data.size(),
                                   fds ? *fds : std::vector<int>());
}

ssize_t ZygoteHostImpl::ReadReply(void* buf, size_t buf_len) {
  DCHECK_NE(-1, control_fd_);
  // At startup we send a kZygoteCommandGetSandboxStatus request to the zygote,
  // but don't wait for the reply. Thus, the first time that we read from the
  // zygote, we get the reply to that request.
  if (!have_read_sandbox_status_word_) {
    if (HANDLE_EINTR(read(control_fd_, &sandbox_status_,
                          sizeof(sandbox_status_))) !=
        sizeof(sandbox_status_)) {
      return -1;
    }
    have_read_sandbox_status_word_ = true;
  }

  return HANDLE_EINTR(read(control_fd_, buf, buf_len));
}

pid_t ZygoteHostImpl::ForkRequest(const std::vector<std::string>& argv,
                                  scoped_ptr<FileDescriptorInfo> mapping,
                                  const std::string& process_type) {
  DCHECK(init_);
  Pickle pickle;

  int raw_socks[2];
  PCHECK(0 == socketpair(AF_UNIX, SOCK_SEQPACKET, 0, raw_socks));
  base::ScopedFD my_sock(raw_socks[0]);
  base::ScopedFD peer_sock(raw_socks[1]);
  CHECK(UnixDomainSocket::EnableReceiveProcessId(my_sock.get()));

  pickle.WriteInt(kZygoteCommandFork);
  pickle.WriteString(process_type);
  pickle.WriteInt(argv.size());
  for (std::vector<std::string>::const_iterator
       i = argv.begin(); i != argv.end(); ++i)
    pickle.WriteString(*i);

  // Fork requests contain one file descriptor for the PID oracle, and one
  // more for each file descriptor mapping for the child process.
  const size_t num_fds_to_send = 1 + mapping->GetMappingSize();
  pickle.WriteInt(num_fds_to_send);

  std::vector<int> fds;

  // First FD to send is peer_sock.
  // TODO(morrita): Ideally, this should be part of the mapping so that
  // FileDescriptorInfo can manages its lifetime.
  fds.push_back(peer_sock.get());

  // The rest come from mapping.
  for (size_t i = 0; i < mapping->GetMappingSize(); ++i) {
    pickle.WriteUInt32(mapping->GetIDAt(i));
    fds.push_back(mapping->GetFDAt(i));
  }

  // Sanity check that we've populated |fds| correctly.
  DCHECK_EQ(num_fds_to_send, fds.size());

  pid_t pid;
  {
    base::AutoLock lock(control_lock_);
    if (!SendMessage(pickle, &fds))
      return base::kNullProcessHandle;
    mapping.reset();
    peer_sock.reset();

    {
      char buf[sizeof(kZygoteChildPingMessage) + 1];
      ScopedVector<base::ScopedFD> recv_fds;
      base::ProcessId real_pid;

      ssize_t n = UnixDomainSocket::RecvMsgWithPid(
          my_sock.get(), buf, sizeof(buf), &recv_fds, &real_pid);
      if (n != sizeof(kZygoteChildPingMessage) ||
          0 != memcmp(buf,
                      kZygoteChildPingMessage,
                      sizeof(kZygoteChildPingMessage))) {
        // Zygote children should still be trustworthy when they're supposed to
        // ping us, so something's broken if we don't receive a valid ping.
        LOG(ERROR) << "Did not receive ping from zygote child";
        NOTREACHED();
        real_pid = -1;
      }
      my_sock.reset();

      // Always send PID back to zygote.
      Pickle pid_pickle;
      pid_pickle.WriteInt(kZygoteCommandForkRealPID);
      pid_pickle.WriteInt(real_pid);
      if (!SendMessage(pid_pickle, NULL))
        return base::kNullProcessHandle;
    }

    // Read the reply, which pickles the PID and an optional UMA enumeration.
    static const unsigned kMaxReplyLength = 2048;
    char buf[kMaxReplyLength];
    const ssize_t len = ReadReply(buf, sizeof(buf));

    Pickle reply_pickle(buf, len);
    PickleIterator iter(reply_pickle);
    if (len <= 0 || !iter.ReadInt(&pid))
      return base::kNullProcessHandle;

    // If there is a nonempty UMA name string, then there is a UMA
    // enumeration to record.
    std::string uma_name;
    int uma_sample;
    int uma_boundary_value;
    if (iter.ReadString(&uma_name) &&
        !uma_name.empty() &&
        iter.ReadInt(&uma_sample) &&
        iter.ReadInt(&uma_boundary_value)) {
      // We cannot use the UMA_HISTOGRAM_ENUMERATION macro here,
      // because that's only for when the name is the same every time.
      // Here we're using whatever name we got from the other side.
      // But since it's likely that the same one will be used repeatedly
      // (even though it's not guaranteed), we cache it here.
      static base::HistogramBase* uma_histogram;
      if (!uma_histogram || uma_histogram->histogram_name() != uma_name) {
        uma_histogram = base::LinearHistogram::FactoryGet(
            uma_name, 1,
            uma_boundary_value,
            uma_boundary_value + 1,
            base::HistogramBase::kUmaTargetedHistogramFlag);
      }
      uma_histogram->Add(uma_sample);
    }

    if (pid <= 0)
      return base::kNullProcessHandle;
  }

#if !defined(OS_OPENBSD)
  // This is just a starting score for a renderer or extension (the
  // only types of processes that will be started this way).  It will
  // get adjusted as time goes on.  (This is the same value as
  // chrome::kLowestRendererOomScore in chrome/chrome_constants.h, but
  // that's not something we can include here.)
  const int kLowestRendererOomScore = 300;
  AdjustRendererOOMScore(pid, kLowestRendererOomScore);
#endif

  ZygoteChildBorn(pid);
  return pid;
}

#if !defined(OS_OPENBSD)
void ZygoteHostImpl::AdjustRendererOOMScore(base::ProcessHandle pid,
                                            int score) {
  // 1) You can't change the oom_score_adj of a non-dumpable process
  //    (EPERM) unless you're root. Because of this, we can't set the
  //    oom_adj from the browser process.
  //
  // 2) We can't set the oom_score_adj before entering the sandbox
  //    because the zygote is in the sandbox and the zygote is as
  //    critical as the browser process. Its oom_adj value shouldn't
  //    be changed.
  //
  // 3) A non-dumpable process can't even change its own oom_score_adj
  //    because it's root owned 0644. The sandboxed processes don't
  //    even have /proc, but one could imagine passing in a descriptor
  //    from outside.
  //
  // So, in the normal case, we use the SUID binary to change it for us.
  // However, Fedora (and other SELinux systems) don't like us touching other
  // process's oom_score_adj (or oom_adj) values
  // (https://bugzilla.redhat.com/show_bug.cgi?id=581256).
  //
  // The offical way to get the SELinux mode is selinux_getenforcemode, but I
  // don't want to add another library to the build as it's sure to cause
  // problems with other, non-SELinux distros.
  //
  // So we just check for files in /selinux. This isn't foolproof, but it's not
  // bad and it's easy.

  static bool selinux;
  static bool selinux_valid = false;

  if (!selinux_valid) {
    const base::FilePath kSelinuxPath("/selinux");
    base::FileEnumerator en(kSelinuxPath, false, base::FileEnumerator::FILES);
    bool has_selinux_files = !en.Next().empty();

    selinux = access(kSelinuxPath.value().c_str(), X_OK) == 0 &&
              has_selinux_files;
    selinux_valid = true;
  }

  if (use_suid_sandbox_for_adj_oom_score_ && !selinux) {
#if defined(USE_TCMALLOC)
    // If heap profiling is running, these processes are not exiting, at least
    // on ChromeOS. The easiest thing to do is not launch them when profiling.
    // TODO(stevenjb): Investigate further and fix.
    if (IsHeapProfilerRunning())
      return;
#endif
    std::vector<std::string> adj_oom_score_cmdline;
    adj_oom_score_cmdline.push_back(sandbox_binary_);
    adj_oom_score_cmdline.push_back(sandbox::kAdjustOOMScoreSwitch);
    adj_oom_score_cmdline.push_back(base::Int64ToString(pid));
    adj_oom_score_cmdline.push_back(base::IntToString(score));

    base::Process sandbox_helper_process;
    base::LaunchOptions options;

    // sandbox_helper_process is a setuid binary.
    options.allow_new_privs = true;

    sandbox_helper_process =
        base::LaunchProcess(adj_oom_score_cmdline, options);
    if (sandbox_helper_process.IsValid())
      base::EnsureProcessGetsReaped(sandbox_helper_process.Pid());
  } else if (!use_suid_sandbox_for_adj_oom_score_) {
    if (!base::AdjustOOMScore(pid, score))
      PLOG(ERROR) << "Failed to adjust OOM score of renderer with pid " << pid;
  }
}
#endif

void ZygoteHostImpl::EnsureProcessTerminated(pid_t process) {
  DCHECK(init_);
  Pickle pickle;

  pickle.WriteInt(kZygoteCommandReap);
  pickle.WriteInt(process);
  if (!SendMessage(pickle, NULL))
    LOG(ERROR) << "Failed to send Reap message to zygote";
  ZygoteChildDied(process);
}

base::TerminationStatus ZygoteHostImpl::GetTerminationStatus(
    base::ProcessHandle handle,
    bool known_dead,
    int* exit_code) {
  DCHECK(init_);
  Pickle pickle;
  pickle.WriteInt(kZygoteCommandGetTerminationStatus);
  pickle.WriteBool(known_dead);
  pickle.WriteInt(handle);

  static const unsigned kMaxMessageLength = 128;
  char buf[kMaxMessageLength];
  ssize_t len;
  {
    base::AutoLock lock(control_lock_);
    if (!SendMessage(pickle, NULL))
      LOG(ERROR) << "Failed to send GetTerminationStatus message to zygote";
    len = ReadReply(buf, sizeof(buf));
  }

  // Set this now to handle the error cases.
  if (exit_code)
    *exit_code = RESULT_CODE_NORMAL_EXIT;
  int status = base::TERMINATION_STATUS_NORMAL_TERMINATION;

  if (len == -1) {
    LOG(WARNING) << "Error reading message from zygote: " << errno;
  } else if (len == 0) {
    LOG(WARNING) << "Socket closed prematurely.";
  } else {
    Pickle read_pickle(buf, len);
    int tmp_status, tmp_exit_code;
    PickleIterator iter(read_pickle);
    if (!iter.ReadInt(&tmp_status) || !iter.ReadInt(&tmp_exit_code)) {
      LOG(WARNING)
          << "Error parsing GetTerminationStatus response from zygote.";
    } else {
      if (exit_code)
        *exit_code = tmp_exit_code;
      status = tmp_status;
    }
  }

  if (status != base::TERMINATION_STATUS_STILL_RUNNING) {
    ZygoteChildDied(handle);
  }
  return static_cast<base::TerminationStatus>(status);
}

pid_t ZygoteHostImpl::GetPid() const {
  return pid_;
}

int ZygoteHostImpl::GetSandboxStatus() const {
  if (have_read_sandbox_status_word_)
    return sandbox_status_;
  return 0;
}

bool ZygoteHostImpl::ShouldUseNamespaceSandbox() {
  const base::CommandLine& command_line =
      *base::CommandLine::ForCurrentProcess();
  if (command_line.HasSwitch(switches::kNoSandbox)) {
    return false;
  }

  if (command_line.HasSwitch(switches::kDisableNamespaceSandbox)) {
    return false;
  }

  if (!sandbox::Credentials::CanCreateProcessInNewUserNS()) {
    return false;
  }

  return true;
}

}  // namespace content