linux-kernel - Re: [PATCH net-next 2/9] net: bridge: offload initial and final port flags through switchdev

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <a22fee65-9cb5-bdd5-fd40-cdd57dc072f3@nvidia.com>
Date:   Mon, 8 Feb 2021 14:17:04 +0200
From:   Nikolay Aleksandrov <nikolay@...dia.com>
To:     Vladimir Oltean <olteanv@...il.com>
CC:     Jakub Kicinski <kuba@...nel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Andrew Lunn <andrew@...n.ch>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        <netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <bridge@...ts.linux-foundation.org>,
        "Roopa Prabhu" <roopa@...dia.com>, Jiri Pirko <jiri@...nulli.us>,
        Ido Schimmel <idosch@...sch.org>,
        Claudiu Manoil <claudiu.manoil@....com>,
        "Alexandre Belloni" <alexandre.belloni@...tlin.com>,
        <UNGLinuxDriver@...rochip.com>, Vadym Kochan <vkochan@...vell.com>,
        Taras Chornyi <tchornyi@...vell.com>,
        Grygorii Strashko <grygorii.strashko@...com>,
        Ioana Ciornei <ioana.ciornei@....com>,
        Ivan Vecera <ivecera@...hat.com>, <linux-omap@...r.kernel.org>
Subject: Re: [PATCH net-next 2/9] net: bridge: offload initial and final port
 flags through switchdev

On 08/02/2021 13:45, Vladimir Oltean wrote:
> On Mon, Feb 08, 2021 at 01:37:03PM +0200, Nikolay Aleksandrov wrote:
>> Hi Vladimir,
>> I think this patch potentially breaks some use cases. There are a few problems, I'll
>> start with the more serious one: before the ports would have a set of flags that were
>> always set when joining, now due to how nbp_flags_change() handles flag setting some might
>> not be set which would immediately change behaviour w.r.t software fwding. I'll use your
>> example of BR_BCAST_FLOOD: a lot of drivers will return an error for it and any broadcast
>> towards these ports will be dropped, we have mixed environments with software ports that
>> sometimes have traffic (e.g. decapped ARP requests) software forwarded which will stop working.
> 
> Yes, you're right. The only solution I can think of is to add a "bool ignore_errors"
> to nbp_flags_change, set to true from new_nbp and del_nbp, and to false from the
> netlink code.
> 

Indeed, I can't think of any better solution right now, but that would make it more or less
equal to the current situation where the flags are just set. You can read/restore them on add/del
of bridge port, but I guess that's what you'd like to avoid. :)
I don't mind adding the add/del_nbp() notifications, but both of them seem redundant with
the port add/del notifications which you can handle in the driver.

>> The other lesser issue is with the style below, I mean these three calls for each flag are
>> just ugly and look weird as you've also noted, since these APIs are internal can we do better?
> 
> Doing better would mean allowing nbp_flags_change() to have a bit mask with
> potentially more brport flags set, and to call br_switchdev_set_port_flag in
> a for_each_set_bit() loop?
> 

Sure, that sounds better for now. I think you've described the ideal case in your
commit message.