netdev - Re: [net-next v1 1/2] net: sched: use queue

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: <CAMDZJNVnvAXfqFSah4wgXri1c3jnQhxCdBVo41uP37e0L3BUAg@mail.gmail.com>
Date:   Tue, 7 Dec 2021 10:10:22 +0800
From:   Tonghao Zhang <xiangxia.m.yue@...il.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     Linux Kernel Network Developers <netdev@...r.kernel.org>,
        Jamal Hadi Salim <jhs@...atatu.com>,
        Cong Wang <xiyou.wangcong@...il.com>,
        Jiri Pirko <jiri@...nulli.us>,
        "David S. Miller" <davem@...emloft.net>,
        Jonathan Lemon <jonathan.lemon@...il.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Alexander Lobakin <alobakin@...me>,
        Paolo Abeni <pabeni@...hat.com>,
        Talal Ahmad <talalahmad@...gle.com>,
        Kevin Hao <haokexin@...il.com>,
        Ilias Apalodimas <ilias.apalodimas@...aro.org>,
        Kees Cook <keescook@...omium.org>,
        Kumar Kartikeya Dwivedi <memxor@...il.com>,
        Antoine Tenart <atenart@...nel.org>,
        Wei Wang <weiwan@...gle.com>, Arnd Bergmann <arnd@...db.de>,
        Alexander Duyck <alexander.duyck@...il.com>
Subject: Re: [net-next v1 1/2] net: sched: use queue_mapping to pick tx queue

On Tue, Dec 7, 2021 at 4:40 AM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Mon,  6 Dec 2021 16:05:11 +0800 xiangxia.m.yue@...il.com wrote:
> >   +----+      +----+      +----+
> >   | P1 |      | P2 |      | Pn |
> >   +----+      +----+      +----+
> >     |           |           |
> >     +-----------+-----------+
> >                 |
> >                 | clsact/skbedit
> >                 |    MQ
> >                 v
> >     +-----------+-----------+
> >     | q0        | q1        | qn
> >     v           v           v
> >    HTB         HTB   ...   FIFO
Hi Jakub, thanks for your comments
> The usual suggestion these days is to try to use FQ + EDT to
> implement efficient policies. You don't need dedicated qdiscs,
> just modulate transmission time appropriately on egress of the
> container.
FQ+EDT is good solution. But this patch should be used on another scenario.
1. the containers  which outbound traffic is not limited, want to use
the fifo qdisc.
If this traffic share the FQ/HTB Qdisc, the qdisc lock will affect the
performance and latency.
2. we can support user to select tx queue, range from A to B. skb hash
or cgroup classid is good to do load balance.
patch 2/2: https://patchwork.kernel.org/project/netdevbpf/patch/20211206080512.36610-3-xiangxia.m.yue@gmail.com/

> In general recording the decision in the skb seems a little heavy
> handed. We just need to carry the information from the egress hook
> to the queue selection a few lines below. Or in fact maybe egress
Yes, we can refactor netdev_core_pick_tx to
1. select queue_index and invoke skb_set_queue_mapping, but don't
return the txq.
2. after egress hook, use skb_get_queue_mapping/netdev_get_tx_queue to get txq.
> hook shouldn't be used for this in the first place, and we need
> a more appropriate root qdisc than simple mq?
I have no idea about mq, I think clsact may make the things more flexible.
and act_bpf can also support to change sk queue_mapping. queue_mapping
was included in __sk_buff.

> Not sure. What I am sure of is that you need to fix these warnings:
Ok
> include/linux/skbuff.h:937: warning: Function parameter or member 'tc_skip_txqueue' not described in 'sk_buff'
>
> ERROR: spaces required around that '=' (ctx:VxW)
> #103: FILE: net/sched/act_skbedit.c:42:
> +       queue_mapping= (queue_mapping & 0xff) + hash % mapping_mod;
>                      ^
>
> ;)



-- 
Best regards, Tonghao