Received-SPF: pass (google.com: domain of linux-kernel+bounces-105827-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20240311122401.6549-1-lizheng043@gmail.com> <20240311135117.GA1244788@maili.marvell.com>
In-Reply-To: <20240311135117.GA1244788@maili.marvell.com>
From: James Lee <lizheng043@gmail.com>
Date: Mon, 18 Mar 2024 16:39:21 +0800
Message-ID: <CAPCnf4zS=FN0MHM2tQV0b468zN0yqRHbaNMsk3cDQ7Vu8wiHKA@mail.gmail.com>
Subject: Re: [PATCH] neighbour: guarantee the localhost connections be
 established successfully even the ARP table is full
To: Ratheesh Kannoth <rkannoth@marvell.com>
Cc: linux-kernel@vger.kernel.org, netdev@vger.kernel.org, 
	nhorman@tuxdriver.com, davem@davemloft.net, jmorris@namei.org, 
	James.Z.Li@dell.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

loopback neigh is a special device in the neighbour system which is
used by all local communications and state is NUD_NOARP.
Any setting value of gc_thresh3 might encounter arp table be full,
manually increasing gc_thresh3 can resolve this issue for every time,
but we hope this issue automatically be resolved in Linux kernel for
all local communications whenever ARP table is full, rather than
manually operation as a workaround.


Ratheesh Kannoth <rkannoth@marvell.com> =E4=BA=8E2024=E5=B9=B43=E6=9C=8811=
=E6=97=A5=E5=91=A8=E4=B8=80 21:51=E5=86=99=E9=81=93=EF=BC=9A
>
> On 2024-03-11 at 17:54:01, Zheng Li (lizheng043@gmail.com) wrote:
> >
> > Inter-process communication on localhost should be established successf=
ully even the ARP table is full,
> > many processes on server machine use the localhost to communicate such =
as command-line interface (CLI),
> > servers hope all CLI commands can be executed successfully even the arp=
 table is full.
> > Right now CLI commands got timeout when the arp table is full.
> > Set the parameter of exempt_from_gc to be true for LOOPBACK net device =
to
> > keep localhost neigh in arp table, not removed by gc.
> >
> > the steps of reproduced:
> > server with "gc_thresh3 =3D 1024" setting, ping server from more than 1=
024 IPv4 addresses,
> > run "ssh localhost" on console interface, then the command will get tim=
eout.
> it does not look correct to me. why gc has to behave differently for loop=
back devices.
> why can't a higher gc_thresh3 value (fine tuned to your use case) wont so=
lve the issue ?
> can't you add localhost arp entry statically and get rid of gc issue ?
>
> >
> > Signed-off-by: Zheng Li <James.Z.Li@Dell.com>
> > ---
> >  net/core/neighbour.c | 5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> >
> > diff --git a/net/core/neighbour.c b/net/core/neighbour.c
> > index 552719c3bbc3..d96dee3d4af6 100644
> > --- a/net/core/neighbour.c
> > +++ b/net/core/neighbour.c
> > @@ -734,7 +734,10 @@ ___neigh_create(struct neigh_table *tbl, const voi=
d *pkey,
> >  struct neighbour *__neigh_create(struct neigh_table *tbl, const void *=
pkey,
> >                                struct net_device *dev, bool want_ref)
> >  {
> > -     return ___neigh_create(tbl, pkey, dev, 0, false, want_ref);
> > +     if (dev->flags & IFF_LOOPBACK)
> > +             return ___neigh_create(tbl, pkey, dev, 0, true, want_ref)=
;
> > +     else
> > +             return ___neigh_create(tbl, pkey, dev, 0, false, want_ref=
);
> >  }
> >  EXPORT_SYMBOL(__neigh_create);
> >
> > --
> > 2.17.1
> >