Re: [PATCH net-next 2/3] bpf: add TCP_BPF_RTO_MAX for bpf_setsockopt

[Jason Xing <kerneljasonxing@xxxxxxxxx>]

On Thu, Feb 13, 2025 at 10:32 AM Kuniyuki Iwashima <kuniyu@xxxxxxxxxx> wrote:
>
> From: Kuniyuki Iwashima <kuniyu@xxxxxxxxxx>
> Date: Thu, 13 Feb 2025 11:25:01 +0900
> > > diff --git a/net/core/filter.c b/net/core/filter.c
> > > index 2ec162dd83c4..a21a147e0a86 100644
> > > --- a/net/core/filter.c
> > > +++ b/net/core/filter.c
> > > @@ -5303,6 +5303,12 @@ static int bpf_sol_tcp_setsockopt(struct sock *sk, int optname,
> > >                     return -EINVAL;
> > >             tp->bpf_sock_ops_cb_flags = val;
> > >             break;
> > > +   case TCP_BPF_RTO_MAX:
> > > +           if (val > TCP_RTO_MAX_SEC * MSEC_PER_SEC ||
> > > +               val < TCP_RTO_MAX_MIN_SEC * MSEC_PER_SEC)
> > > +                   return -EINVAL;
> > > +           inet_csk(sk)->icsk_rto_max = msecs_to_jiffies(val);
> > > +           break;
> > >     default:
> > >             return -EINVAL;
> > >     }
> >
> > You need not define TCP_BPF_RTO_MAX because TCP_RTO_MAX is not a
> > BPF specific option, and you can just reuse do_tcp_setsockopt(),
> > then bpf_getsockopt() also works.
> >
> > ---8<---
> > diff --git a/net/core/filter.c b/net/core/filter.c
> > index 2ec162dd83c4..77732f10097c 100644
> > --- a/net/core/filter.c
> > +++ b/net/core/filter.c
> > @@ -5382,6 +5382,7 @@ static int sol_tcp_sockopt(struct sock *sk, int optname,
> >       case TCP_USER_TIMEOUT:
> >       case TCP_NOTSENT_LOWAT:
> >       case TCP_SAVE_SYN:
> > +     case TCP_RTO_MAX:
>
> s/TCP_RTO_MAX/TCP_RTO_MAX_MS/ :)

Thanks for spotting this simpler approach. It works.

Thanks,
Jason

>
>
> >               if (*optlen != sizeof(int))
> >                       return -EINVAL;
> >               break;
> > ---8<---