Re: [PATCH] net: tcp: accept old ack during closing

[Eric Dumazet]

On Thu, Jan 11, 2024 at 11:06 AM Menglong Dong <menglong8.dong@xxxxxxxxx> wrote:
>
> On Wed, Jan 10, 2024 at 6:41 PM Eric Dumazet <edumazet@xxxxxxxxxx> wrote:
> >
> > On Wed, Jan 10, 2024 at 11:25 AM Eric Dumazet <edumazet@xxxxxxxxxx> wrote:
> > >
> > > On Wed, Jan 10, 2024 at 4:08 AM Menglong Dong <menglong8.dong@xxxxxxxxx> wrote:
> > > >
> > >
> > > >
> > > > Oops, It should be "SKB_DR_SET(reason, NOT_SPECIFIED);" here.
> > > > Sorry that I shouldn't be too confident to compile it.
> > > >
> > >
> > > net-next is closed, come back in ~two weeks, thanks.
>
> Okay, I'll send the V2 after two weeks.
>
> >
> > Also look at commit d0e1a1b5a833b625c ("tcp: better validation of
> > received ack sequences"), for some context.
>
> Yeah, I already analyzed this commit before. I think that the return
> value of tcp_ack() mean different thing to SYN_SEND and FIN_WAIT1
> if it is zero, and should be handled separately.
>
> Anyway, we can discuss this part when the net-next is opened.

Discussion can start now, but make sure to add RFC tag so that netdev
maintainers
can prioritize accordingly ;)