[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CANn89iJ2aFYjQp639O6mDj7vhiNV5w_EVsMxM2jQGBHfHbOtOQ@mail.gmail.com>
Date: Sat, 2 Sep 2023 09:37:13 +0200
From: Eric Dumazet <edumazet@...gle.com>
To: r30009329 <renmingshuai@...wei.com>
Cc: netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
jhs@...atatu.com, xiyou.wangcong@...il.com, davem@...emloft.net,
jiri@...nulli.us, yanan@...wei.com, liaichun@...wei.com,
chenzhen126@...wei.com
Subject: Re: net/sched: Discuss about adding a new kernel parameter to set the
default value of flow_limit
On Sat, Sep 2, 2023 at 9:16 AM r30009329 <renmingshuai@...wei.com> wrote:
>
> How about adding a new kernel parameter to set the default value of flow_limit
> when the default qidsc is set to fq? Although We can use the tc to modify the
> default value of flow_limit, it is more convenient to use a kernel parameter to
> set the default value, especially in scenarios where the tc command is
> inconvenient or cannot be used.
Hmm, can you define 'inconvenient' ?
It seems to be an issue with the tool or the toolchain ?
The 'cannot be used' argument seems strange to me.
If I understand correctly, you want a generic mechanism for all qdisc
default parameters,
in case the 'default qdisc' is XXX instead of pfifo_fast.
kernel parameters should be reserved to specific cases when
programmers have no other way.
rtnetlink is the way to go really, sorry !
Powered by blists - more mailing lists