Date: Mon, 7 Apr 2025 15:35:16 -0400 From: jaeyong yoo <y.jaeyong@gmail.com> To: Cheng Cui <cc@freebsd.org> Cc: freebsd-net@freebsd.org, freebsd-transport@freebsd.org Subject: Re: HyStart availability in FreeBSD stack Message-ID: <CANud0TFNgzwUh9gq7BTGh8m9Jn8=5vkwhigttm1pQefVQOnCeQ@mail.gmail.com> In-Reply-To: <CANud0TE8ta7psdoMMeMqK2PJa_SgBzOU-FrpVa6bdfFyhOEa8A@mail.gmail.com> References: <CANud0TGz-gG-9TSeWyQcUNXp5=KsNea8qA4vySy_UpsK6Qzv2g@mail.gmail.com> <D321CF65-424D-4397-8473-755D512AF373@freebsd.org> <CANud0TE8ta7psdoMMeMqK2PJa_SgBzOU-FrpVa6bdfFyhOEa8A@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
I was taking a look into the code for rttsample and can we add calling that rttsample callback at tcp_xmit_timer: - https://github.com/freebsd/freebsd-src/blob/main/sys/netinet/tcp_input.c#L3682 Looks like that place is to calculate t_srtt for TCP stack for feeding a new RTT from the recent ACK, which seems right for rttsample for HyStart. Thanks, Jaeyong 2025년 3월 24일 (월) 오후 4:17, jaeyong yoo <y.jaeyong@gmail.com>님이 작성: > > Thanks Cheng, > > Yes it looks like it implements newround callback for HyStart. > But shouldn't it also need for "rttsample" callback? > > Thanks, > Jaeyong > > 2025년 3월 24일 (월) 오후 3:34, Cheng Cui <cc@freebsd.org>님이 작성: > > > > > > > > On Mar 20, 2025, at 16:14, jaeyong yoo <y.jaeyong@gmail.com> wrote: > > > > Hi net/transport, > > > > We are looking for using HyStart but looks like it is only implemented > > in RACK stack but not in FreeBSD TCP Stack. > > Is there a plan to support this in the near future? > > > > Thanks, > > Jaeyong > > > > > > Hi Jaeyong, > > > > Sorry for the delayed response. There is a patch to enable Hystart++ for the default TCP stack. If you can help test it, I think it would be very helpful to speed up its acceptance. > > > > https://reviews.freebsd.org/D46425 > > > > > > Best Regards, > > Cheng Cui > > > > > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANud0TFNgzwUh9gq7BTGh8m9Jn8=5vkwhigttm1pQefVQOnCeQ>