[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <20171108.122855.673986719935863302.davem@davemloft.net>
Date: Wed, 08 Nov 2017 12:28:55 +0900 (KST)
From: David Miller <davem@...emloft.net>
To: jiri@...nulli.us
Cc: netdev@...r.kernel.org, nogahf@...lanox.com, jhs@...atatu.com,
xiyou.wangcong@...il.com, mlxsw@...lanox.com, andrew@...n.ch,
vivien.didelot@...oirfairelinux.com, f.fainelli@...il.com,
michael.chan@...adcom.com, ganeshgr@...lsio.com,
saeedm@...lanox.com, matanb@...lanox.com, leonro@...lanox.com,
idosch@...lanox.com, jakub.kicinski@...ronome.com,
simon.horman@...ronome.com, pieter.jansenvanvuuren@...ronome.com,
john.hurley@...ronome.com, alexander.h.duyck@...el.com,
ogerlitz@...lanox.com, john.fastabend@...il.com
Subject: Re: [patch net-next 0/9] qdisc RED offload
From: Jiri Pirko <jiri@...nulli.us>
Date: Mon, 6 Nov 2017 07:23:40 +0100
> From: Jiri Pirko <jiri@...lanox.com>
>
> Nogah says:
>
> Add an offload support for RED qdisc for mlxsw driver.
> The first patch adds the ability to offload RED qdisc by using
> ndo_setup_tc. It gives RED three commands, to offload, change or delete
> the qdisc, to get the qdisc generic stats and to get it's RED xstats.
> There is no enforcement on a driver to offload or not offload the qdisc and
> it is up to the driver to decide.
> RED qdisc is first being created and only later graft to a parent (unless
> it is a root qdisc). For that reason the return value of the offload
> replace command that is called in the init process doesn't reflect actual
> offload state. The offload state is determined in the dump function so it
> can be reflected to the user. This function is also responsible for stats
> update.
>
> The patchses 2-3 change the name of TC_SETUP_MQPRIO & TC_SETUP_CBS to match
> with the new convention of QDISC prefix.
> The rest of the patchset is driver support for the qdisc. Currently only
> as root qdisc that is being set on the default traffic class. It supports
> only the following parameters of RED: min, max, probability and ECN mode.
> Limit and burst size related params are being ignored at this moment.
Series applied, thanks!
Powered by blists - more mailing lists