Re: [PATCH bpf 1/2] bpf: Forbid bpf_ktime_get_coarse_ns and bpf_timer_* in tracing progs

[Alexei Starovoitov <alexei.starovoitov@gmail.com>]

On Mon, Nov 8, 2021 at 8:46 AM Dmitrii Banshchikov <me@ubique.spb.ru> wrote:
>
> bpf_ktime_get_coarse_ns() helper uses ktime_get_coarse_ns() time
> accessor that isn't safe for any context.
> This results in locking issues:

Please trim the cc. vger thinks it's a spam.
The patch didn't reach the mailing list or patchwork.

>  const struct bpf_func_proto bpf_ktime_get_coarse_ns_proto = {
>         .func           = bpf_ktime_get_coarse_ns,
>         .gpl_only       = false,
>         .ret_type       = RET_INTEGER,
> +       .allowed        = bpf_ktime_get_coarse_ns_allowed,

I think it's easier to move to networking prog types
instead of going through a callback.

>  static const struct bpf_func_proto bpf_timer_init_proto = {
>         .func           = bpf_timer_init,
>         .gpl_only       = true,
> @@ -1147,6 +1173,7 @@ static const struct bpf_func_proto bpf_timer_init_proto = {
>         .arg1_type      = ARG_PTR_TO_TIMER,
>         .arg2_type      = ARG_CONST_MAP_PTR,
>         .arg3_type      = ARG_ANYTHING,
> +       .allowed        = bpf_timer_allowed,

I think disabling timers in check_map_prog_compatibility
(similar to how bpf_spin_lock is restricted) would be cleaner.