netdev - Re: [PATCH net] netdevsim: Restore per-network namespace accounting for fib entries

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: <20190813153401.GR2428@nanopsycho>
Date:   Tue, 13 Aug 2019 17:34:01 +0200
From:   Jiri Pirko <jiri@...nulli.us>
To:     David Ahern <dsahern@...il.com>
Cc:     David Miller <davem@...emloft.net>, dsahern@...nel.org,
        netdev@...r.kernel.org
Subject: Re: [PATCH net] netdevsim: Restore per-network namespace accounting
 for fib entries

Tue, Aug 13, 2019 at 04:41:18PM CEST, dsahern@...il.com wrote:
>On 8/13/19 1:14 AM, Jiri Pirko wrote:
>> Mon, Aug 12, 2019 at 05:28:02PM CEST, davem@...emloft.net wrote:
>>> From: Jiri Pirko <jiri@...nulli.us>
>>> Date: Mon, 12 Aug 2019 10:36:35 +0200
>>>
>>>> I understand it with real devices, but dummy testing device, who's
>>>> purpose is just to test API. Why?
>>>
>>> Because you'll break all of the wonderful testing infrastructure
>>> people like David have created.
>>  
>> Are you referring to selftests? There is no such test there :(
>
>I  have one now and will be submitting it after net merges with net-next.
>
>> But if it would be, could implement the limitations
>> properly (like using cgroups), change the tests and remove this
>> code from netdevsim?
>
>The intent of this code and test is to have a s/w model similar to how
>mlxsw works - responding to notifiers and deciding to reject a change.
>You are currently adding (or trying to) more devlink based s/w tests, so
>you must see the value of netdevsim as a source of testing.

Sure I do. Not sure makes sence to repeat myself again, but why not:
The way you use netdevsim with netnamespace limitation is nothing like
it is done in hardware. Devlink resources should limit the resources of
the device, not network namespace. You abused netdevsim and devlink for
that. Not cool :(

To be in sync with mlxsw, netdevsim should track fibs added to the ports
and apply the resource limitations to that. That is the correct
behaviour. Exacly like mlxsw does.

Frankly I don't really understand why you keep pushing your broken
design. Why the limitation applied only for fibs related to netdevsim
ports is not enough for testing??? Would that work for you? Please?

This is keeping me awake at night. Sigh :(