Date: Tue, 30 Aug 2016 14:19:12 +0000 From: "Cui, Cheng" <Cheng.Cui@netapp.com> To: "freebsd-transport@freebsd.org" <freebsd-transport@freebsd.org> Subject: Re: question about if a recent Linux patch on window scaling is required in FreeBSD Message-ID: <27FDDABD-4726-41B1-8A49-FF3274F9AAD3@netapp.com> In-Reply-To: <D3E4C4C9.147A0%Cheng.Cui@netapp.com> References: <D3E4C4C9.147A0%Cheng.Cui@netapp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Refresh this question. Can anyone make a comment? Thanks, --Cheng Cui NetApp Scale Out Networking On 8/25/16, 3:52 PM, "Cui, Cheng" <Cheng.Cui@netapp.com> wrote: Hello everyone, =20 I hope this email could reach you well, because I found related discussions about this topic on window scaling and the case of window shrinking (or retraction or loss of precision). And I try to make this question simple. =20 There is a recent Linux patch at receiver side to round-up advertised window due to precision loss of window scaling. It reaches my attention because the same problem could also happen between a pair of Linux and FreeBSD nodes, and I am not aware of any similar patch in FreeBSD yet. =20 The Linux patch is this: http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?= id=3D6 07bfbf2d55dd1cfe5368b41c2a81a8c9ccf4723 =20 And I quote some description of the Linux patch below: > If the sender uses up the entire window before it is shrunk, this can > have chaotic effects on the connection. When sending ACKs, > tcp_acceptable_seq() will notice that the window has been shrunk sinc= e > tcp_wnd_end() is before tp->snd_nxt, which makes it choose tcp_wnd_en= d() >as=20 > sequence number. This will fail the receivers checks in tcp_sequence(= ) >however=20 > since it is before it's tp->rcv_wup, making it respond with a dupack. =20 I think the Linux's behavior is right ("ACK-only packets should be sent with the largest in-window sequence number that has ever been sent." re= f: https://www.ietf.org/mail-archive/web/tcpm/current/msg10512.html), it actually chooses "tp->snd_una+tp->snd_wnd" (tcp_wnd_end()) instead of tp->snd_nxt, as it thought tp->snd_nxt is out of window, in case of precision loss which made the receiver's advertise-window smaller. But = at the=20 other side, if the other side is FreeBSD, I think FreeBSD will also fai= l the=20 check since "tp->snd_una+tp->snd_wnd" is before it's tp->rcv_nxt, and ignore=20 the sequence number in the packet. =20 I also sent an email to tcpm@ietf.org asking if this Linux patch is RFC 7323=20 (window scaling part) compliant, but I have not get any reply yet. =20 So my question here is: Is there any recent change in FreeBSD to accommodate the=20 Linux behavior ("tp->snd_una+tp->snd_wnd" as sequence number)? If not, = do we=20 consider to apply the same way as in the Linux patch? =20 Thanks and apologize in advance if I did not do enough research, --Cheng Cui =20 =20 =20 =20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?27FDDABD-4726-41B1-8A49-FF3274F9AAD3>