Re: [Virtio-fs] [PATCH for-5.1 2/3] virtiofsd: add container-friendly -o chroot sandboxing option

[Vivek Goyal <vgoyal@redhat.com>]

On Wed, Jul 22, 2020 at 02:02:05PM +0100, Stefan Hajnoczi wrote:
> virtiofsd cannot run in an unprivileged container because CAP_SYS_ADMIN
> is required to create namespaces.
> 
> Introduce a weaker sandbox that is sufficient in container environments
> because the container runtime already sets up namespaces. Use chroot to
> restrict path traversal to the shared directory.
> 
> virtiofsd loses the following:
> 
> 1. Mount namespace. The process chroots to the shared directory but
>    leaves the mounts in place. Seccomp rejects mount(2)/umount(2)
>    syscalls.
> 
> 2. Pid namespace. This should be fine because virtiofsd is the only
>    process running in the container.
> 
> 3. Network namespace. This should be fine because seccomp already
>    rejects the connect(2) syscall, but an additional layer of security
>    is lost. Container runtime-specific network security policies can be
>    used drop network traffic (except for the vhost-user UNIX domain
>    socket).
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  tools/virtiofsd/helper.c         |  3 +++
>  tools/virtiofsd/passthrough_ll.c | 44 ++++++++++++++++++++++++++++++--
>  2 files changed, 45 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/virtiofsd/helper.c b/tools/virtiofsd/helper.c
> index 3105b6c23a..7421c9ca1a 100644
> --- a/tools/virtiofsd/helper.c
> +++ b/tools/virtiofsd/helper.c
> @@ -151,6 +151,9 @@ void fuse_cmdline_help(void)
>             "    -o cache=<mode>            cache mode. could be one of \"auto, "
>             "always, none\"\n"
>             "                               default: auto\n"
> +           "    -o chroot|no_chroot        use container-friendly chroot instead\n"
> +           "                               of stronger mount namespace sandbox\n"
> +           "                               default: false\n"

This option name disabling namespace setup is little confusing to me.

Will it make sense to provide another option to disable/enable
namespaces. "-o no-namespaces" and that disables setting up
namespaces.

Thanks
Vivek

>             "    -o flock|no_flock          enable/disable flock\n"
>             "                               default: no_flock\n"
>             "    -o log_level=<level>       log level, default to \"info\"\n"
> diff --git a/tools/virtiofsd/passthrough_ll.c b/tools/virtiofsd/passthrough_ll.c
> index 50a164a599..990c0a8a70 100644
> --- a/tools/virtiofsd/passthrough_ll.c
> +++ b/tools/virtiofsd/passthrough_ll.c
> @@ -139,6 +139,7 @@ enum {
>  
>  struct lo_data {
>      pthread_mutex_t mutex;
> +    int chroot; /* 1 - use chroot, 0 - use mount namespace */
>      int debug;
>      int writeback;
>      int flock;
> @@ -162,6 +163,8 @@ struct lo_data {
>  };
>  
>  static const struct fuse_opt lo_opts[] = {
> +    { "chroot", offsetof(struct lo_data, chroot), 1 },
> +    { "no_chroot", offsetof(struct lo_data, chroot), 0 },
>      { "writeback", offsetof(struct lo_data, writeback), 1 },
>      { "no_writeback", offsetof(struct lo_data, writeback), 0 },
>      { "source=%s", offsetof(struct lo_data, source), 0 },
> @@ -2665,6 +2668,37 @@ static void setup_capabilities(char *modcaps_in)
>      pthread_mutex_unlock(&cap.mutex);
>  }
>  
> +/*
> + * Use chroot as a weaker sandbox for environment where the process is launched
> + * without CAP_SYS_ADMIN.
> + */
> +static void setup_chroot(struct lo_data *lo)
> +{
> +    lo->proc_self_fd = open("/proc/self/fd", O_PATH);
> +    if (lo->proc_self_fd == -1) {
> +        fuse_log(FUSE_LOG_ERR, "open(\"/proc/self/fd\", O_PATH): %m\n");
> +        exit(1);
> +    }
> +
> +    /*
> +     * Make the shared directory the file system root so that FUSE_OPEN
> +     * (lo_open()) cannot escape the shared directory by opening a symlink.
> +     *
> +     * It's still possible to escape the chroot via lo->proc_self_fd but that
> +     * requires gaining control of the process first.
> +     */
> +    if (chroot(lo->source) != 0) {
> +        fuse_log(FUSE_LOG_ERR, "chroot(\"%s\"): %m\n", lo->source);
> +        exit(1);
> +    }
> +
> +    /* Move into the chroot */
> +    if (chdir("/") != 0) {
> +        fuse_log(FUSE_LOG_ERR, "chdir(\"/\"): %m\n");
> +        exit(1);
> +    }
> +}
> +
>  /*
>   * Lock down this process to prevent access to other processes or files outside
>   * source directory.  This reduces the impact of arbitrary code execution bugs.
> @@ -2672,8 +2706,13 @@ static void setup_capabilities(char *modcaps_in)
>  static void setup_sandbox(struct lo_data *lo, struct fuse_session *se,
>                            bool enable_syslog)
>  {
> -    setup_namespaces(lo, se);
> -    setup_mounts(lo->source);
> +    if (lo->chroot) {
> +        setup_chroot(lo);
> +    } else {
> +        setup_namespaces(lo, se);
> +        setup_mounts(lo->source);
> +    }
> +
>      setup_seccomp(enable_syslog);
>      setup_capabilities(g_strdup(lo->modcaps));
>  }
> @@ -2820,6 +2859,7 @@ int main(int argc, char *argv[])
>      struct fuse_session *se;
>      struct fuse_cmdline_opts opts;
>      struct lo_data lo = {
> +        .chroot = 0,
>          .debug = 0,
>          .writeback = 0,
>          .posix_lock = 1,
> -- 
> 2.26.2
> 
> _______________________________________________
> Virtio-fs mailing list
> Virtio-fs@redhat.com
> https://www.redhat.com/mailman/listinfo/virtio-fs