sandboxed-api/sandboxed_api/sandbox2/policy.cc

// Copyright 2019 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

// Implementation of the sandbox2::Policy class.

#include "sandboxed_api/sandbox2/policy.h"

#include <fcntl.h>
#include <linux/audit.h>
#include <linux/bpf_common.h>
#include <linux/filter.h>
#include <linux/seccomp.h>
#include <sched.h>
#include <syscall.h>

#include <cstdint>
#include <limits>
#include <optional>
#include <string>
#include <vector>

#include "absl/flags/flag.h"
#include "absl/log/log.h"
#include "absl/strings/string_view.h"
#include "sandboxed_api/config.h"
#include "sandboxed_api/sandbox2/bpfdisassembler.h"
#include "sandboxed_api/sandbox2/comms.h"
#include "sandboxed_api/sandbox2/syscall.h"
#include "sandboxed_api/sandbox2/util/bpf_helper.h"
#include "sandboxed_api/util/raw_logging.h"

#ifndef SECCOMP_FILTER_FLAG_NEW_LISTENER
#define SECCOMP_FILTER_FLAG_NEW_LISTENER (1UL << 3)
#endif

#ifndef SECCOMP_RET_USER_NOTIF
#define SECCOMP_RET_USER_NOTIF 0x7fc00000U /* notifies userspace */
#endif

#define DO_USER_NOTIF BPF_STMT(BPF_RET + BPF_K, SECCOMP_RET_USER_NOTIF)

ABSL_FLAG(bool, sandbox2_danger_danger_permit_all, false,
          "Allow all syscalls, useful for testing");
ABSL_FLAG(std::string, sandbox2_danger_danger_permit_all_and_log, "",
          "Allow all syscalls and log them into specified file");

namespace sandbox2 {

// The final policy is the concatenation of:
//   1. default policy (GetDefaultPolicy, private),
//   2. user policy (user_policy_, public),
//   3. default KILL action (avoid failing open if user policy did not do it).
std::vector<sock_filter> Policy::GetPolicy(bool user_notif) const {
  if (absl::GetFlag(FLAGS_sandbox2_danger_danger_permit_all) ||
      !absl::GetFlag(FLAGS_sandbox2_danger_danger_permit_all_and_log).empty()) {
    return GetTrackingPolicy();
  }

  // Now we can start building the policy.
  // 1. Start with the default policy (e.g. syscall architecture checks).
  auto policy = GetDefaultPolicy(user_notif);
  VLOG(3) << "Default policy:\n" << bpf::Disasm(policy);

  // 2. Append user policy.
  VLOG(3) << "User policy:\n" << bpf::Disasm(user_policy_);
  // Add default syscall_nr loading in case the user forgets.
  policy.push_back(LOAD_SYSCALL_NR);
  policy.insert(policy.end(), user_policy_.begin(), user_policy_.end());

  // 3. Finish with default KILL action.
  policy.push_back(KILL);

  // In seccomp_unotify mode replace all KILLS with unotify
  if (user_notif) {
    for (sock_filter& filter : policy) {
      if (filter.code == BPF_RET + BPF_K && filter.k == SECCOMP_RET_KILL) {
        filter = DO_USER_NOTIF;
      }
    }
  }

  VLOG(2) << "Final policy:\n" << bpf::Disasm(policy);
  return policy;
}

// If you modify this function, you should also modify.
// Monitor::LogAccessViolation to keep them in sync.
//
// Produces a policy which returns SECCOMP_RET_TRACE instead of SECCOMP_RET_KILL
// for the __NR_execve syscall, so the tracer can make a decision to allow or
// disallow it depending on which occurrence of __NR_execve it was.
// LINT.IfChange
std::vector<sock_filter> Policy::GetDefaultPolicy(bool user_notif) const {
  bpf_labels l = {0};

  std::vector<sock_filter> policy;
  if (user_notif) {
    policy = {
        // If compiled arch is different from the runtime one, inform the
        // Monitor.
        LOAD_ARCH,
        JNE32(Syscall::GetHostAuditArch(), DENY),
        LOAD_SYSCALL_NR,
        // TODO(b/271400371) Use NOTIF_FLAG_CONTINUE once generally available
        JNE32(__NR_seccomp, JUMP(&l, past_seccomp_l)),
        ARG_32(3),
        JNE32(internal::kExecveMagic, JUMP(&l, past_seccomp_l)),
        ALLOW,
        LABEL(&l, past_seccomp_l),
        LOAD_SYSCALL_NR,
        JNE32(__NR_execveat, JUMP(&l, past_execveat_l)),
        ARG_32(4),
        JNE32(AT_EMPTY_PATH, JUMP(&l, past_execveat_l)),
        ARG_32(5),
        JNE32(internal::kExecveMagic, JUMP(&l, past_execveat_l)),
        ALLOW,
        LABEL(&l, past_execveat_l),

        LOAD_SYSCALL_NR,
    };
  } else {
    policy = {
      // If compiled arch is different from the runtime one, inform the Monitor.
      LOAD_ARCH,
      JEQ32(Syscall::GetHostAuditArch(), JUMP(&l, past_arch_check_l)),
#if defined(SAPI_X86_64)
      JEQ32(AUDIT_ARCH_I386, TRACE(sapi::cpu::kX86)),  // 32-bit sandboxee
#endif
      TRACE(sapi::cpu::kUnknown),
      LABEL(&l, past_arch_check_l),

      // After the policy is uploaded, forkserver will execve the sandboxee. We
      // need to allow this execve but not others. Since BPF does not have
      // state, we need to inform the Monitor to decide, and for that we use a
      // magic value in syscall args 5. Note that this value is not supposed to
      // be secret, but just an optimization so that the monitor is not
      // triggered on every call to execveat.
      LOAD_SYSCALL_NR,
      JNE32(__NR_execveat, JUMP(&l, past_execveat_l)),
      ARG_32(4),
      JNE32(AT_EMPTY_PATH, JUMP(&l, past_execveat_l)),
      ARG_32(5),
      JNE32(internal::kExecveMagic, JUMP(&l, past_execveat_l)),
      SANDBOX2_TRACE,
      LABEL(&l, past_execveat_l),

      LOAD_SYSCALL_NR,
    };
  }

  // Forbid ptrace because it's unsafe or too risky. The user policy can only
  // block (i.e. return an error instead of killing the process) but not allow
  // ptrace. This uses LOAD_SYSCALL_NR from above.
  if (!user_policy_handles_ptrace_) {
    policy.insert(policy.end(), {JEQ32(__NR_ptrace, DENY)});
  }

  // If user policy doesn't mention it, then forbid bpf because it's unsafe or
  // too risky.  This uses LOAD_SYSCALL_NR from above.
  if (!user_policy_handles_bpf_) {
    policy.insert(policy.end(), {JEQ32(__NR_bpf, DENY)});
  }
#ifndef CLONE_NEWCGROUP
#define CLONE_NEWCGROUP 0x02000000
#endif
  constexpr uintptr_t kNewNamespacesFlags =
      CLONE_NEWNS | CLONE_NEWUSER | CLONE_NEWNET | CLONE_NEWUTS |
      CLONE_NEWCGROUP | CLONE_NEWIPC | CLONE_NEWPID;
  static_assert(kNewNamespacesFlags <= std::numeric_limits<uint32_t>::max());
  constexpr uintptr_t kUnsafeCloneFlags = kNewNamespacesFlags | CLONE_UNTRACED;
  static_assert(kUnsafeCloneFlags <= std::numeric_limits<uint32_t>::max());
  policy.insert(policy.end(),
                {
#ifdef __NR_clone3
                    // Disallow clone3. Errno instead of DENY so that libraries
                    // can fallback to regular clone/clone2.
                    JEQ32(__NR_clone3, ERRNO(ENOSYS)),
#endif
                    // Disallow clone3 and clone with unsafe flags.  This uses
                    // LOAD_SYSCALL_NR from above.
                    JNE32(__NR_clone, JUMP(&l, past_clone_unsafe_l)),
                    // Regardless of arch, we only care about the lower 32-bits
                    // of the flags.
                    ARG_32(0),
                    JA32(kUnsafeCloneFlags, DENY),
                    LABEL(&l, past_clone_unsafe_l),
                    // Disallow unshare with unsafe flags.
                    LOAD_SYSCALL_NR,
                    JNE32(__NR_unshare, JUMP(&l, past_unshare_unsafe_l)),
                    // Regardless of arch, we only care about the lower 32-bits
                    // of the flags.
                    ARG_32(0),
                    JA32(kNewNamespacesFlags, DENY),
                    LABEL(&l, past_unshare_unsafe_l),
                    // Disallow seccomp with SECCOMP_FILTER_FLAG_NEW_LISTENER
                    // flag.
                    LOAD_SYSCALL_NR,
                    JNE32(__NR_seccomp, JUMP(&l, past_seccomp_new_listener)),
                    // Regardless of arch, we only care about the lower 32-bits
                    // of the flags.
                    ARG_32(1),
                    JA32(SECCOMP_FILTER_FLAG_NEW_LISTENER, DENY),
                    LABEL(&l, past_seccomp_new_listener),
                });

  if (bpf_resolve_jumps(&l, policy.data(), policy.size()) != 0) {
    LOG(FATAL) << "Cannot resolve bpf jumps";
  }

  return policy;
}
// LINT.ThenChange(monitor_ptrace.cc)

std::vector<sock_filter> Policy::GetTrackingPolicy() const {
  return {
    LOAD_ARCH,
#if defined(SAPI_X86_64)
        JEQ32(AUDIT_ARCH_X86_64, TRACE(sapi::cpu::kX8664)),
        JEQ32(AUDIT_ARCH_I386, TRACE(sapi::cpu::kX86)),
#elif defined(SAPI_PPC64_LE)
        JEQ32(AUDIT_ARCH_PPC64LE, TRACE(sapi::cpu::kPPC64LE)),
#elif defined(SAPI_ARM64)
        JEQ32(AUDIT_ARCH_AARCH64, TRACE(sapi::cpu::kArm64)),
#elif defined(SAPI_ARM)
        JEQ32(AUDIT_ARCH_ARM, TRACE(sapi::cpu::kArm)),
#endif
        TRACE(sapi::cpu::kUnknown),
  };
}

bool Policy::SendPolicy(Comms* comms, bool user_notif) const {
  auto policy = GetPolicy(user_notif);
  if (!comms->SendBytes(
          reinterpret_cast<uint8_t*>(policy.data()),
          static_cast<uint64_t>(policy.size()) * sizeof(sock_filter))) {
    LOG(ERROR) << "Couldn't send policy";
    return false;
  }

  return true;
}

void Policy::GetPolicyDescription(PolicyDescription* policy) const {
  policy->set_user_bpf_policy(user_policy_.data(),
                              user_policy_.size() * sizeof(sock_filter));
  if (policy_builder_description_) {
    *policy->mutable_policy_builder_description() =
        *policy_builder_description_;
  }

  if (namespace_) {
    namespace_->GetNamespaceDescription(
        policy->mutable_namespace_description());
  }
}

}  // namespace sandbox2