Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 7 Jul 2022 20:15:07 +0300
From:   Vladimir Oltean <olteanv@gmail.com>
To:     Nikolay Aleksandrov <razor@blackwall.org>
Cc:     Hans Schultz <schultz.hans@gmail.com>, davem@davemloft.net,
        kuba@kernel.org, netdev@vger.kernel.org,
        Andrew Lunn <andrew@lunn.ch>,
        Vivien Didelot <vivien.didelot@gmail.com>,
        Florian Fainelli <f.fainelli@gmail.com>,
        Eric Dumazet <edumazet@google.com>,
        Paolo Abeni <pabeni@redhat.com>, Jiri Pirko <jiri@resnulli.us>,
        Ivan Vecera <ivecera@redhat.com>,
        Roopa Prabhu <roopa@nvidia.com>, Shuah Khan <shuah@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Ido Schimmel <idosch@nvidia.com>, linux-kernel@vger.kernel.org,
        bridge@lists.linux-foundation.org, linux-kselftest@vger.kernel.org
Subject: Re: [PATCH V3 net-next 1/4] net: bridge: add fdb flag to extent
 locked port feature
Message-ID: <20220707171507.pojkwqhwqk5u6mmn@skbuf>
References: <b78fb006-04c4-5a25-7ba5-94428cc9591a@blackwall.org>
 <86fskyggdo.fsf@gmail.com>
 <040a1551-2a9f-18d0-9987-f196bb429c1b@blackwall.org>
 <86v8tu7za3.fsf@gmail.com>
 <4bf1c80d-0f18-f444-3005-59a45797bcfd@blackwall.org>
 <20220706181316.r5l5rzjysxow2j7l@skbuf>
 <7cf30a3e-a562-d582-4391-072a2c98ab05@blackwall.org>
 <20220706202130.ehzxnnqnduaq3rmt@skbuf>
 <fe456fb0-4f68-f93e-d4a9-66e3bc56d547@blackwall.org>
 <37d59561-6ce8-6c5f-5d31-5c37a0a3d231@blackwall.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <37d59561-6ce8-6c5f-5d31-5c37a0a3d231@blackwall.org>
Precedence: bulk

Hi Nikolay,

On Thu, Jul 07, 2022 at 05:08:15PM +0300, Nikolay Aleksandrov wrote:
> On 07/07/2022 00:01, Nikolay Aleksandrov wrote:
> > On 06/07/2022 23:21, Vladimir Oltean wrote:
> >> On Wed, Jul 06, 2022 at 10:38:04PM +0300, Nikolay Aleksandrov wrote:
> [snip]
> > I already said it's ok to add hard configurable limits if they're done properly performance-wise.
> > Any distribution can choose to set some default limits after the option exists.
> > 
> 
> Just fyi, and to avoid duplicate efforts, I already have patches for global and per-port software
> fdb limits that I'll polish and submit soon (depending on time availability, of course). If I find
> more time I might add per-vlan limits as well to the set. They use embedded netlink attributes
> to config and dump, so we can easily extend them later (e.g. different action on limit hit, limit
> statistics etc).

So again, to repeat myself, it's nice to have limits on FDB size, but
those won't fix the software bridges that are now out in the open and
can't have their configuration scripts changed.

I haven't had the time to expand on this in a proper change yet, but I
was thinking more along the lines of adding an OOM handler with
register_oom_notifier() in br_fdb_init(), and on OOM, do something, like
flush the FDB from all bridges. There are going to be complications, it
will schedule switchdev, switchdev is going to allocate memory which
we're low on, the workqueues aren't created with WQ_MEM_RECLAIM, so this
isn't necessarily going to be a silver bullet either. But this is what
concerns me the most, the unconfigured bridge killing the kernel so
easily. As you can see, with an OOM handler I'm not so much trying to
impose a fixed limit on FDB size, but do something sensible such that
the bridge doesn't contribute to the kernel dying.