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  PHC 
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:   Mon, 1 Aug 2022 17:24:19 +0300
From:   Roi Dayan <>
To:     Baowen Zheng <>,
        Jakub Kicinski <>,
        Simon Horman <>
Cc:     Simon Horman <>,
        Saeed Mahameed <>,
        "David S. Miller" <>,
        Paolo Abeni <>,
        Eric Dumazet <>,
        Saeed Mahameed <>,
        "" <>,
        Tariq Toukan <>,
        Jianbo Liu <>, Oz Shlomo <>
Subject: Re: [net-next 06/15] net/mlx5e: TC, Support tc action api for police

On 2022-07-30 7:58 AM, Baowen Zheng wrote:
>> Subject: Re: [net-next 06/15] net/mlx5e: TC, Support tc action api for police
>> On Fri, 29 Jul 2022 17:51:05 +0100 Simon Horman wrote:
>>> my reading of things is that the handling of offload of police (meter)
>>> actions in flower rules by the mlx5 driver is such that it can handle
>>> offloading actions by index - actions that it would now be possible to
>>> add to hardware with this patch in place.
>>> My reasoning assumes that mlx5e_tc_add_flow_meter() is called to
>>> offload police (meter) actions in flower rules. And that it calls
>>> mlx5e_tc_meter_get(), which can find actions based on an index.
>>> I could, however, be mistaken as I have so much knowledge of the mlx5
>>> driver. And rather than dive deeper I wanted to respond as above - I
>>> am mindful of the point we are in the development cycle.
>>> I would be happy to dive deeper into this as a mater of priority if
>>> desired.
>> Thank you! No very deep dives necessary from my perspective, just wanted for
>> the authors of the action offload API to look over.
> Hi Jakub, thanks for noticing us about this change, for this patch, it seems good to me.
> I just have a tiny doubt, since Nvida just add a validation for a police offload, why it is not applied to this meter offload patch?
> Since I do not know much details about Nvdia meter implement, I guess not all the police action result is supported, so maybe validation is necessary.
> Thanks


as Simon already mentioned. offloading action meter is done in
mlx5e_tc_add_flow_meter() and adding a tc rule with meter will use
mlx5e_tc_meter_get() to find the existing meter by index if exists.

as for Baowen Zhang comment. you are right. it seems I missed a
validation part. since its merged I will send a fix commit.


Powered by blists - more mailing lists