recipe_modules/nsjail/api.py

# Copyright 2022 The Fuchsia Authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.

from recipe_engine import recipe_api


class NsjailCmdBuilder:
    """NsjailCmdBuilder builds a command to run an nsjail sandbox."""

    def __init__(self, api, bin_path, dumb_init_path):
        self._api = api

        self._path = ""
        self._bin_path = bin_path
        self._dumb_init_path = str(dumb_init_path)
        self._ro_mounts = [self._dumb_init_path]
        self._rw_mounts = []
        self._symlinks = {}
        self._env = {}

    def _prepend_to_path(self, val):
        if self._path:
            self._path = "%s:%s" % (val, self._path)
        else:
            self._path = val

    def build(self, subcmd):
        """
        Build an nsjail command that wraps the given subcmd.

        Args:
            subcmd (list[str]): The subcommand to run inside the sandbox.

        Returns: A command, which takes the form list[str].
        """
        cmd = [
            self._bin_path,
            # Disable some of the lesser used namespaces.
            "--disable_clone_newipc",
            "--disable_clone_newuts",
            "--disable_clone_newcgroup",
            # Disable rlimits because our builds are single-tenant and can
            # safely use all of the system resources.
            "--disable_rlimits",
            # Increase the time limit to 12 hours.
            "--time_limit",
            "43200",
            "--log",
            # Direct nsjail output to a separate log to avoid mixing it in with
            # output of the underlying command.
            self._api.raw_io.output_text(name="nsjail_log", add_output_log=True),
        ]

        # Initialize the PATH variable.
        self._env["PATH"] = self._path

        # Create a tmpdir for use by this sandbox.
        # Ideally, we would just use a tmpfs, but nsjail limits the size of
        # a tmpfs mount to ~16MB, and we write far more than that to the
        # directory.
        tmp = self._api.path.mkdtemp("tmp")
        cmd.extend(["--bindmount", "%s:/tmp" % tmp])
        self._env["TMPDIR"] = "/tmp"

        # We need to sort mounts to make sure that we mount parent directories
        # before mounting their children. This must be done regardless of
        # whether the mount point is rw or ro.
        is_writable = {}
        for m in self._ro_mounts:
            is_writable[m] = False
        for m in self._rw_mounts:
            is_writable[m] = True

        for m in sorted(is_writable.keys()):
            if is_writable[m]:
                cmd.extend(["--bindmount", m])
            else:
                cmd.extend(["--bindmount_ro", m])

        # We sort the dictionaries to maintain stable ordering.
        for k in sorted(self._symlinks.keys()):
            cmd.extend(["--symlink", "%s:%s" % (k, self._symlinks[k])])

        for k in sorted(self._env.keys()):
            v = self._env[k]
            v = v.replace("$tmpdir", "/tmp")
            cmd.extend(["--env", "%s=%s" % (k, v)])

        cmd.append("--")
        cmd.append(self._dumb_init_path)
        cmd.extend(subcmd)
        return cmd

    def add_ro_mounts(self, mounts):
        """
        Adds a set of read-only mounts to the sandbox.

        Args:
            mounts (list[str]): A list of paths to mount read-only.
        """
        self._ro_mounts.extend(mounts)

    def add_rw_mounts(self, mounts):
        """
        Adds a set of read-write mounts to the sandbox.

        Args:
            mounts (list[str]): A list of paths to mount read-write.
        """
        self._rw_mounts.extend(mounts)

    def add_env_vars(self, env_vars):
        """
        Adds a set of environment variables to add to the sandbox.

        Args:
            env_vars (dict{str: str}): A map of key:value pairs.
        """
        self._env.update(env_vars)

    def add_symlinks(self, symlinks):
        """
        Adds a set of symlinks to add to the sandbox.

        Args:
            symlinks (dict{str: str}): A map of target:link_name pairs to
                symlink in the sandbox.
        """
        self._symlinks.update(symlinks)

    def add_luci_git(self):
        """
        Adds the paths containing the git wrapper and git binary provisioned
        by LUCI environments to the sandbox.
        """
        paths = self._api.nsjail._which(
            "find all git binaries",
            "git",
            include_all=True,
        )
        # Only mount if this is not system git.
        paths = [p for p in paths if not p.startswith(("/usr/bin", "/bin"))]
        self.add_ro_mounts(paths)
        for p in paths:
            self._prepend_to_path(self._api.path.dirname(p))

    def add_linux_tools(self):
        """
        Adds a set of linux tools we expect most programs in infra to use.
        """
        tools = [
            "awk",
            "basename",
            "cat",
            "chmod",
            "cmp",
            "comm",
            "cp",
            "cut",
            "date",
            "diff",
            "dirname",
            "egrep",
            "env",
            "expr",
            "file",
            "find",
            "grep",
            "gzip",
            "head",
            "ldd",
            "ls",
            "mkdir",
            "mktemp",
            "mv",
            "readlink",
            "rm",
            "sed",
            "sort",
            "stat",
            "tail",
            "tar",
            "tee",
            "touch",
            "tr",
            "uname",
            "wc",
            "which",
        ]
        mounts = [
            # We special case bash here because the fuchsia build often hardcodes
            # /bin/bash in shebangs.
            "/bin/bash",
            # These are required for bash to work.
            "/lib",
            "/lib64",
        ]
        # The set of directories to add to PATH.
        path_additions = {"/bin": True}

        tool_paths = self._api.nsjail._which(
            "find tool paths",
            *tools,
            test_output_lines=["/usr/bin/%s" % tool for tool in tools],
        )
        for tool_path in tool_paths:
            mounts.append(tool_path)
            path_additions[self._api.path.dirname(tool_path)] = True

        self.add_ro_mounts(mounts)

        # Sort and add all path additions.
        paths = sorted(path_additions.keys())
        for p in paths:
            self._prepend_to_path(p)


class NsjailApi(recipe_api.RecipeApi):
    """NsjailApi allows callers to run steps in lightweight sandboxes."""

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        # Cache repeated calls to `_which()`.
        self._which_cache = {}

    @property
    def _nsjail_tool(self):
        """Fetch and provision nsjail from CIPD."""
        return self.m.ensure_tool("nsjail", self.resource("nsjail/tool_manifest.json"))

    @property
    def _dumb_init_tool(self):
        """Fetch and provision dumb_init from CIPD."""
        return self.m.ensure_tool(
            "dumb_init", self.resource("dumb_init/tool_manifest.json")
        )

    def sandboxed_step(
        self,
        step_name,
        cmd,
        env=None,
        symlinks=None,
        ro_mounts=None,
        rw_mounts=None,
        use_linux_tools=False,
        use_luci_git=False,
        **kwargs,
    ):
        """
        Constructs and returns a step that runs the given cmd in a sandbox.

        Args:
            step_name (str): The name of the step.
            cmd (list[str]): The command to run inside the sandbox.
            env (dict{str: str}): A map of key:value pairs to add to the env.
            symlinks (dict{str: str}): A map of target:link_name pairs to add
                to the sandbox.
            ro_mounts (list[str]): A list of read-only mounts.
            rw_mounts (list[str]): A list of read-write mounts.
            use_linux_tools (bool): Whether to mount a canonical set of linux
                system tools.
            use_luci_git (bool): Whether to mount the git wrapper provisioned
                by LUCI.
            **kwargs (dict): Passed through as kwargs to step.

        Returns: A step that runs the given cmd inside an Nsjail sandbox.
        """
        b = NsjailCmdBuilder(self.m, self._nsjail_tool, self._dumb_init_tool)
        if env:
            b.add_env_vars(env)
        if symlinks:
            b.add_symlinks(symlinks)
        if ro_mounts:
            b.add_ro_mounts(ro_mounts)
        if rw_mounts:
            b.add_rw_mounts(rw_mounts)
        if use_linux_tools:
            b.add_linux_tools()
        if use_luci_git:
            b.add_luci_git()
        sandboxed_cmd = b.build(cmd)
        return self.m.step(
            step_name,
            sandboxed_cmd,
            **kwargs,
        )

    def _which(self, name, *tools, include_all=False, test_output_lines=()):
        """Look up the absolute paths to executables on $PATH."""
        cmd = ["which"]
        if include_all:
            cmd.append("-a")
        cmd.extend(tools)

        cache_key = tuple(cmd)
        if cache_key not in self._which_cache:
            output = self.m.step(
                name,
                cmd,
                stdout=self.m.raw_io.output_text(),
                step_test_data=lambda: self.m.raw_io.test_api.stream_output_text(
                    "\n".join(test_output_lines) + "\n"
                ),
            ).stdout.strip()
            self._which_cache[cache_key] = [
                l for l in output.splitlines() if l  # Filter empty lines.
            ]
        return self._which_cache[cache_key]