Re: [PATCH v5 bpf-next 00/11] Socket migration for SO_REUSEPORT.

From: Martin KaFai Lau
Date: Fri May 14 2021 - 02:27:04 EST


On Fri, May 14, 2021 at 08:23:00AM +0900, Kuniyuki Iwashima wrote:
> From: Andrii Nakryiko <andrii.nakryiko@xxxxxxxxx>
> Date: Thu, 13 May 2021 14:27:13 -0700
> > On Sun, May 9, 2021 at 8:45 PM Kuniyuki Iwashima <kuniyu@xxxxxxxxxxxx> wrote:
> > >
> > > The SO_REUSEPORT option allows sockets to listen on the same port and to
> > > accept connections evenly. However, there is a defect in the current
> > > implementation [1]. When a SYN packet is received, the connection is tied
> > > to a listening socket. Accordingly, when the listener is closed, in-flight
> > > requests during the three-way handshake and child sockets in the accept
> > > queue are dropped even if other listeners on the same port could accept
> > > such connections.
> [...]
> >
> > One test is failing in CI ([0]), please take a look.
> >
> > [0] https://travis-ci.com/github/kernel-patches/bpf/builds/225784969
>
> Thank you for checking.
>
> The test needs to drop SYN+ACK and currently it is done by iptables or
> ip6tables. But it seems that I should not use them. Should this be done
> by XDP?
or drop it at a bpf_prog@tc-egress.

I also don't have iptables in my kconfig and I had to add them
to run this test. None of the test_progs depends on iptables also.

>
> ---8<---
> iptables v1.8.5 (legacy): can't initialize iptables table `filter': Table does not exist (do you need to insmod?)
> Perhaps iptables or your kernel needs to be upgraded.
> ip6tables v1.8.5 (legacy): can't initialize ip6tables table `filter': Table does not exist (do you need to insmod?)
> Perhaps ip6tables or your kernel needs to be upgraded.
> ---8<---
>