bwrap.xml

<?xml version="1.0"?>
<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.3//EN"
               "http://www.oasis-open.org/docbook/xml/4.3/docbookx.dtd" [
]>
<refentry id="bwrap">

<refentryinfo>
  <title>bwrap</title>
  <productname>Project Atomic</productname>
  <authorgroup>
    <author>
      <contrib>Developer</contrib>
      <firstname>Alexander</firstname>
      <surname>Larsson</surname>
    </author>
    <author>
      <contrib>Developer</contrib>
      <firstname>Colin</firstname>
      <surname>Walters</surname>
    </author>
  </authorgroup>
</refentryinfo>

<refmeta>
  <refentrytitle>bwrap</refentrytitle>
  <manvolnum>1</manvolnum>
  <refmiscinfo class="manual">User Commands</refmiscinfo>
</refmeta>

<refnamediv>
  <refname>bwrap</refname>
  <refpurpose>container setup utility</refpurpose>
</refnamediv>

<refsynopsisdiv>
<cmdsynopsis>
<command>bwrap</command>
<arg choice="opt" rep="repeat"><replaceable>OPTION</replaceable></arg>
<arg choice="opt"><replaceable>COMMAND</replaceable></arg>
</cmdsynopsis>
</refsynopsisdiv>

<refsect1><title>Description</title>
<para>
  <command>bwrap</command> is a privileged helper for container setup. You
  are unlikely to use it directly from the commandline, although that is possible.
</para>
<para>
  It works by creating a new, completely empty, filesystem namespace where the root
  is on a tmpfs that is invisible from the host, and which will be automatically
  cleaned up when the last process exits. You can then use commandline options to
  construct the root filesystem and process environment for the command to run in
  the namespace.
</para>
<para>
  By default, <command>bwrap</command> creates a new mount namespace for the sandbox.
  Optionally it also sets up new user, ipc, pid, network and uts namespaces (but note the
  user namespace is required if bwrap is not installed setuid root).
  The application in the sandbox can be made to run with a different UID and GID.
</para>
<para>
  If needed (e.g. when using a PID namespace) <command>bwrap</command>
  is running a minimal pid 1 process in the sandbox that is
  responsible for reaping zombies. It also detects when the initial
  application process (pid 2) dies and reports its exit status back to
  the original spawner.  The pid 1 process exits to clean up the
  sandbox when there are no other processes in the sandbox left.
</para>
</refsect1>

<refsect1><title>Options</title>
  <para>
    When options are used multiple times, the last option wins, unless otherwise
    specified.
  </para>
  <para>General options:</para>
  <variablelist>
    <varlistentry>
      <term><option>--help</option></term>
      <listitem><para>Print help and exit</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--version</option></term>
      <listitem><para>Print version</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--args <arg choice="plain">FD</arg></option></term>
      <listitem><para>
        Parse nul-separated arguments from the given file descriptor.
        This option can be used multiple times to parse options from
        multiple sources.
      </para></listitem>
    </varlistentry>
  </variablelist>
  <para>Options related to kernel namespaces:</para>
  <variablelist>
    <varlistentry>
      <term><option>--unshare-user</option></term>
      <listitem><para>Create a new user namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-user-try</option></term>
      <listitem><para>Create a new user namespace if possible else skip it</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-ipc</option></term>
      <listitem><para>Create a new ipc namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-pid</option></term>
      <listitem><para>Create a new pid namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-net</option></term>
      <listitem><para>Create a new network namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-uts</option></term>
      <listitem><para>Create a new uts namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-cgroup</option></term>
      <listitem><para>Create a new cgroup namespace</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-cgroup-try</option></term>
      <listitem><para>Create a new cgroup namespace if possible else skip it</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unshare-all</option></term>
      <listitem><para>Unshare all possible namespaces. Currently equivalent with: <option>--unshare-user-try</option> <option>--unshare-ipc</option> <option>--unshare-pid</option> <option>--unshare-net</option> <option>--unshare-uts</option> <option>--unshare-cgroup-try</option></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--share-net</option></term>
      <listitem><para>Retain the network namespace, overriding an earlier <option>--unshare-all</option> or <option>--unshare-net</option></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--userns <arg choice="plain">FD</arg></option></term>
      <listitem><para>Use an existing user namespace instead of creating a new one. The namespace must fulfil the permission requirements for setns(), which generally means that it must be a descendant of the currently active user namespace, owned by the same user. </para>
      <para>This is incompatible with --unshare-user, and doesn't work in the setuid version of bubblewrap.</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--userns2 <arg choice="plain">FD</arg></option></term>
      <listitem><para>After setting up the new namespace, switch into the specified namespace. For this to work the specified namespace must be a descendant of the user namespace used for the setup, so this is only useful in combination with --userns.</para>
      <para>This is useful because sometimes bubblewrap itself creates nested user namespaces (to work around some kernel issues) and --userns2 can be used to enter these.</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--pidns <arg choice="plain">FD</arg></option></term>
      <listitem><para>Use an existing pid namespace instead of creating one. This is often used with --userns, because the pid namespace must be owned by the same user namespace that bwrap uses. </para>
      <para>Note that this can be combined with --unshare-pid, and in that case it means that the sandbox will be in its own pid namespace, which is a child of the passed in one.</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--uid <arg choice="plain">UID</arg></option></term>
      <listitem><para>Use a custom user id in the sandbox (requires <option>--unshare-user</option>)</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--gid <arg choice="plain">GID</arg></option></term>
      <listitem><para>Use a custom group id in the sandbox (requires <option>--unshare-user</option>)</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--hostname <arg choice="plain">HOSTNAME</arg></option></term>
      <listitem><para>Use a custom hostname in the sandbox (requires <option>--unshare-uts</option>)</para></listitem>
    </varlistentry>
  </variablelist>
  <para>Options about environment setup:</para>
  <variablelist>
    <varlistentry>
      <term><option>--chdir <arg choice="plain">DIR</arg></option></term>
      <listitem><para>Change directory to <arg choice="plain">DIR</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--setenv <arg choice="plain">VAR</arg> <arg choice="plain">VALUE</arg></option></term>
      <listitem><para>Set an environment variable</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--unsetenv <arg choice="plain">VAR</arg></option></term>
      <listitem><para>Unset an environment variable</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--clearenv</option></term>
      <listitem><para>Unset all environment variables, except for
        <envar>PWD</envar> and any that are subsequently set by
        <option>--setenv</option></para></listitem>
    </varlistentry>
  </variablelist>
  <para>Options for monitoring the sandbox from the outside:</para>
  <variablelist>
    <varlistentry>
      <term><option>--lock-file <arg choice="plain">DEST</arg></option></term>
      <listitem><para>
        Take a lock on <arg choice="plain">DEST</arg> while the sandbox is running.
        This option can be used multiple times to take locks on multiple files.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--sync-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>Keep this file descriptor open while the sandbox is running</para></listitem>
    </varlistentry>
  </variablelist>
  <para>
    Filesystem related options. These are all operations that modify the filesystem directly, or
    mounts stuff in the filesystem. These are applied in the order they are given as arguments.
  </para>
  <para>
    Any missing parent directories that are required to create a specified destination are
    automatically created as needed. Their permissions are normally set to 0755
    (rwxr-xr-x). However, if a <option>--perms</option> option is in effect, and
    it sets the permissions for group or other to zero, then newly-created
    parent directories will also have their corresponding permission set to zero.
    <option>--size</option> modifies the size of the created mount when preceding a
    <option>--tmpfs</option> action; <option>--perms</option> and <option>--size</option>
    can be combined.
  </para>
  <variablelist>
    <varlistentry>
      <term><option>--perms <arg choice="plain">OCTAL</arg></option></term>
      <listitem><para>This option does nothing on its own, and must be followed
        by one of the options that it affects. It sets the permissions
        for the next operation to <arg choice="plain">OCTAL</arg>.
        Subsequent operations are not affected: for example,
        <literal>--perms 0700 --tmpfs /a --tmpfs /b</literal> will mount
        <filename>/a</filename> with permissions 0700, then return to
        the default permissions for <filename>/b</filename>.
        Note that <option>--perms</option> and <option>--size</option> can be
        combined: <literal>--perms 0700 --size 10485760 --tmpfs /s</literal> will apply
        permissions as well as a maximum size to the created tmpfs.</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--size <arg choice="plain">BYTES</arg></option></term>
      <listitem><para>This option does nothing on its own, and must be followed
        by <literal>--tmpfs</literal>. It sets the size in bytes for the next tmpfs.
        For example, <literal>--size 10485760 --tmpfs /tmp</literal> will create a tmpfs
        at <filename>/tmp</filename> of size 10MiB. Subsequent operations are not
        affected: for example,
        <literal>--size 10485760 --tmpfs /a --tmpfs /b</literal> will mount
        <filename>/a</filename> with size 10MiB, then return to the default size for
        <filename>/b</filename>.
        Note that <option>--perms</option> and <option>--size</option> can be
        combined: <literal>--size 10485760 --perms 0700 --tmpfs /s</literal> will apply
        permissions as well as a maximum size to the created tmpfs.</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--bind <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Bind mount the host path <arg choice="plain">SRC</arg> on <arg choice="plain">DEST</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--bind-try <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Equal to <option>--bind</option> but ignores non-existent <arg choice="plain">SRC</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--dev-bind <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Bind mount the host path <arg choice="plain">SRC</arg> on <arg choice="plain">DEST</arg>, allowing device access</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--dev-bind-try <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Equal to <option>--dev-bind</option> but ignores non-existent <arg choice="plain">SRC</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--ro-bind <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Bind mount the host path <arg choice="plain">SRC</arg> readonly on <arg choice="plain">DEST</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--ro-bind-try <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Equal to <option>--ro-bind</option> but ignores non-existent <arg choice="plain">SRC</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--remount-ro <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Remount the path <arg choice="plain">DEST</arg> as readonly.  It works only on the specified mount point, without changing any other mount point under the specified path</para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--proc <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Mount procfs on <arg choice="plain">DEST</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--dev <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Mount new devtmpfs on <arg choice="plain">DEST</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--tmpfs <arg choice="plain">DEST</arg></option></term>
      <listitem>
        <para>Mount new tmpfs on <arg choice="plain">DEST</arg>.
          If the previous option was <option>--perms</option>, it sets the
          mode of the tmpfs. Otherwise, the tmpfs has mode 0755.
          If the previous option was <option>--size</option>, it sets the
          size in bytes of the tmpfs. Otherwise, the tmpfs has the default size.</para>
      </listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--mqueue <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Mount new mqueue on <arg choice="plain">DEST</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--dir <arg choice="plain">DEST</arg></option></term>
      <listitem>
        <para>Create a directory at <arg choice="plain">DEST</arg>.
          If the directory already exists, its permissions are unmodified,
          ignoring <option>--perms</option> (use <option>--chmod</option>
          if the permissions of an existing directory need to be changed).
          If the directory is newly created and the previous option was
          <option>--perms</option>, it sets the mode of the directory.
          Otherwise, newly-created directories have mode 0755.</para>
      </listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--file <arg choice="plain">FD</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem>
        <para>Copy from the file descriptor <arg choice="plain">FD</arg> to
          <arg choice="plain">DEST</arg>.
          If the previous option was <option>--perms</option>, it sets the
          mode of the new file. Otherwise, the file has mode 0666
          (note that this is not the same as <option>--bind-data</option>).</para>
      </listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--bind-data <arg choice="plain">FD</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem>
        <para>Copy from the file descriptor <arg choice="plain">FD</arg> to
          a file which is bind-mounted on <arg choice="plain">DEST</arg>.
          If the previous option was <option>--perms</option>, it sets the
          mode of the new file. Otherwise, the file has mode 0600
          (note that this is not the same as <option>--file</option>).</para>
      </listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--ro-bind-data <arg choice="plain">FD</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem>
        <para>Copy from the file descriptor <arg choice="plain">FD</arg> to
          a file which is bind-mounted read-only on
          <arg choice="plain">DEST</arg>.
          If the previous option was <option>--perms</option>, it sets the
          mode of the new file. Otherwise, the file has mode 0600
          (note that this is not the same as <option>--file</option>).</para>
      </listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--symlink <arg choice="plain">SRC</arg> <arg choice="plain">DEST</arg></option></term>
      <listitem><para>Create a symlink at <arg choice="plain">DEST</arg> with target <arg choice="plain">SRC</arg></para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--chmod <arg choice="plain">OCTAL</arg> <arg choice="plain">PATH</arg></option></term>
      <listitem>
        <para>
          Set the permissions of <arg choice="plain">PATH</arg>, which
          must already exist, to <arg choice="plain">OCTAL</arg>.
        </para>
      </listitem>
    </varlistentry>
  </variablelist>
  <para>Lockdown options:</para>
  <variablelist>
    <varlistentry>
      <term><option>--seccomp <arg choice="plain">FD</arg></option></term>
      <listitem><para>
        Load and use seccomp rules from <arg choice="plain">FD</arg>.
        The rules need to be in the form of a compiled cBPF program,
        as generated by seccomp_export_bpf.
        If this option is given more than once, only the last one is used.
        Use <option>--add-seccomp-fd</option> if multiple seccomp programs
        are needed.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--add-seccomp-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>
        Load and use seccomp rules from <arg choice="plain">FD</arg>.
        The rules need to be in the form of a compiled cBPF program,
        as generated by seccomp_export_bpf.
        This option can be repeated, in which case all the seccomp
        programs will be loaded in the order given (note that the kernel
        will evaluate them in reverse order, so the last program on the
        bwrap command-line is evaluated first). All of them, except
        possibly the last, must allow use of the PR_SET_SECCOMP prctl.
        This option cannot be combined with <option>--seccomp</option>.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--exec-label <arg choice="plain">LABEL</arg></option></term>
      <listitem><para>
	Exec Label from the sandbox. On an SELinux system you can specify the SELinux
	context for the sandbox process(s).
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--file-label <arg choice="plain">LABEL</arg></option></term>
      <listitem><para>
	File label for temporary sandbox content. On an SELinux system you can specify
	the SELinux context for the sandbox content.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--block-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>
	Block the sandbox on reading from FD until some data is available.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--userns-block-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>
	Do not initialize the user namespace but wait on FD until it is ready.  This allow
        external processes (like newuidmap/newgidmap) to setup the user namespace before it
        is used by the sandbox process.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--info-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>
	Write information in JSON format about the sandbox to FD.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--json-status-fd <arg choice="plain">FD</arg></option></term>
      <listitem><para>
        Multiple JSON documents are written to <arg choice="plain">FD</arg>,
        one per line (<ulink url="https://jsonlines.org/">"JSON lines" format</ulink>).
        Each line is a single JSON object.
        After <command>bwrap</command> has started the child process inside the sandbox,
        it writes an object with a <literal>child-pid</literal> member to the
        <option>--json-status-fd</option> (this duplicates the older <option>--info-fd</option>).
        The corresponding value is the process ID of the child process in the pid namespace from
        which <command>bwrap</command> was run.
        If available, the namespace IDs are also included in the object with the <literal>child-pid</literal>;
        again, this duplicates the older <option>--info-fd</option>.
        When the child process inside the sandbox exits, <command>bwrap</command> writes an object
        with an exit-code member, and then closes the <option>--json-status-fd</option>. The value
        corresponding to <literal>exit-code</literal> is the exit status of the child, in the usual
        shell encoding (n if it exited normally with status n, or 128+n if it was killed by signal n).
        Other members may be added to those objects in future versions of <command>bwrap</command>,
        and other JSON objects may be added before or after the current objects, so readers must
        ignore members and objects that they do not understand.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--new-session</option></term>
      <listitem><para>
	Create a new terminal session for the sandbox (calls setsid()). This
        disconnects the sandbox from the controlling terminal which means
        the sandbox can't for instance inject input into the terminal.
        </para><para>
        Note: In a general sandbox, if you don't use --new-session, it is
        recommended to use seccomp to disallow the TIOCSTI ioctl, otherwise
        the application can feed keyboard input to the terminal.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--die-with-parent</option></term>
      <listitem><para>
    Ensures child process (COMMAND) dies when bwrap's parent dies. Kills (SIGKILL)
        all bwrap sandbox processes in sequence from parent to child
        including COMMAND process when bwrap or bwrap's parent dies.
        See prctl, PR_SET_PDEATHSIG.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--as-pid-1</option></term>
      <listitem><para>
	Do not create a process with PID=1 in the sandbox to reap child processes.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--cap-add <arg choice="plain">CAP</arg></option></term>
      <listitem><para>
        Add the specified capability when running as privileged user.  It accepts
        the special value ALL to add all the permitted caps.
      </para></listitem>
    </varlistentry>
    <varlistentry>
      <term><option>--cap-drop <arg choice="plain">CAP</arg></option></term>
      <listitem><para>
        Drop the specified capability when running as privileged user.  It accepts
        the special value ALL to drop all the caps.

        By default no caps are left in the sandboxed process.  The
        <option>--cap-add</option> and <option>--cap-drop</option>
        options are processed in the order they are specified on the
        command line.  Please be careful to the order they are specified.
      </para></listitem>
    </varlistentry>
  </variablelist>
</refsect1>

<refsect1>
  <title>Environment</title>

  <variablelist>
    <varlistentry>
      <term><envar>HOME</envar></term>
      <listitem><para>
        Used as the cwd in the sandbox if <option>--chdir</option> has not been
        explicitly specified and the current cwd is not present inside the sandbox.
        The <option>--setenv</option> option can be used to override the value
        that is used here.
      </para></listitem>
    </varlistentry>
  </variablelist>
</refsect1>

<refsect1>
  <title>Exit status</title>

  <para>
    The <command>bwrap</command> command returns the exit status of the
    initial application process (pid 2 in the sandbox).
  </para>
</refsect1>

</refentry>