Date:   Wed, 2 Nov 2016 08:20:32 +0100
From:   Jiri Pirko <jiri@...nulli.us>
To:     Roopa Prabhu <roopa@...ulusnetworks.com>
Cc:     Ido Schimmel <idosch@...sch.org>,
        Eric Dumazet <eric.dumazet@...il.com>, netdev@...r.kernel.org,
        davem@...emloft.net, jiri@...lanox.com, mlxsw@...lanox.com,
        dsa@...ulusnetworks.com, nikolay@...ulusnetworks.com,
        andy@...yhouse.net, vivien.didelot@...oirfairelinux.com,
        andrew@...n.ch, f.fainelli@...il.com, alexander.h.duyck@...el.com,
        kuznet@....inr.ac.ru, jmorris@...ei.org, yoshfuji@...ux-ipv6.org,
        kaber@...sh.net, Ido Schimmel <idosch@...lanox.com>
Subject: Re: [PATCH net-next v2] ipv4: fib: Replay events when registering
 FIB notifier

Wed, Nov 02, 2016 at 03:13:42AM CET, roopa@...ulusnetworks.com wrote:
>On 11/1/16, 10:03 AM, Ido Schimmel wrote:
>> Hi Roopa,
>>
>> On Tue, Nov 01, 2016 at 08:14:14AM -0700, Roopa Prabhu wrote:
>>>
>[snip]
>>> I have the same concern as Eric here.
>>>
>>> I understand why you need it, but can the driver request for an initial dump and that
>>> dump be made more efficient somehow ie not hold rtnl for the whole dump ?.
>>> instead of making the fib notifier registration code doing it.
>> We can do what we suggested in the last bi-weekly meeting, which is
>> still holding rtnl, but moving the hardware operation to delayed work.
>> This is possible because upper layers always assume operation was
>> successful and driver is responsible for invoking its abort mechanism in
>> case of failure.
>>
>>> these routing table sizes can be huge and an analogy for this in user-space:
>>> We do request a netlink dump of  routing tables at initialization (on driver starts or resets)...
>>> but, existing netlink routing table dumps for that scale don't hold rtnl for the whole dump.
>>> The dump is split into multiple responses to the user and hence it does not starve other rtnl users.
>> In my reply to Eric I mentioned that when we register and unregister
>> from this chain the tables aren't really huge, but instead quite small.
>> I understand your concerns, but I don't wish to make things more
>> complicated than they should be only to address concerns that aren't
>> really realistic.
>
>I understand..but, if you are adding some core infrastructure for switchdev ..it cannot be
>based on the number of simple use-cases or data you have today.
>
>I won't be surprised if tomorrow other switch drivers have a case where they need to
>reset the hw routing table state and reprogram all routes again. Re-registering the notifier to just
>get the routing state of the kernel will not scale. For the long term, since the driver does not maintain a cache,

Driver (mlxsw, rocker) maintain a cache. So I'm not sure why you say
otherwise.


>a pull api with efficient use of rtnl will be useful for other such cases as well.

How do you imagine this "pull API" should look like?


>
>
>If you don't want to get to the complexity of a new api right away because of the
>simple case of management interface routes you have, Can your driver register the notifier early  ?
>(I am sure you have probably already thought about this)

Register early? What it would resolve? I must be missing something. We
register as early as possible. But the thing is, we cannot register
in a past. And that is what this patch resolves.


>
>>
>> I believe current patch is quite simple and also consistent with other
>> notification chains in the kernel, such as the netdevice, where rtnl is
>> held during replay of events.
>> http://lxr.free-electrons.com/source/net/core/dev.c#L1535
>as you know, netdev and routing scale are not the same thing.
>Looking at the current code for netdevices (replay and rollback on failure),
>a pull api (equivalent to the netlink dump api) may end up being less complex...with an
>ability to batch in the future.
>
>
>
>
>
>
>

Powered by blists - more mailing lists