[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <630433b7.630a0220.6ca31.e01f@mx.google.com>
Date: Tue, 23 Aug 2022 01:56:06 +0000
From: CGEL <cgel.zte@...il.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: davem@...emloft.net, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org, xu.xin16@....com.cn
Subject: Re: [PATCH 0/2] Namespaceify two sysctls related with route
On Wed, Aug 17, 2022 at 08:52:37PM -0700, Jakub Kicinski wrote:
> On Tue, 16 Aug 2022 02:25:22 +0000 cgel.zte@...il.com wrote:
> > Different netns has different requirements on the setting of error_cost
> > and error_burst, which are related with limiting the frequency of sending
> > ICMP_DEST_UNREACH packets or outputing error message to dmesg.
>
> Could you add a bit more detail about why you need this knob per netns?
Yes, I have sent new-version patches which update the commit lot at
https://lore.kernel.org/all/20220822045310.203649-1-xu.xin16@zte.com.cn/
The sysctls of error_cost and error_burst are important knobs to control
the sending frequency of ICMP_DEST_UNREACH packet for ipv4. When different
containers has requirements on the tuning of error_cost and error_burst,
for host's security, the sysctls should exist per network namespace so
not to bother the host's sysctl setting.
> The code looks fine, no objections there, what I'm confused by is that
> we don't have this knob for IPv6. So is it somehow important enough for
> v4 to be per-ns and yet not important enough to exist at all on v6?
>
Sorry, but I'm not familiar with IPv6 implementation.
> Could you add Documentation in Documentation/admin-guide/sysctl/net.rst
> while at it, and use READ_ONCE / WRITE_ONCE when accessing the sysctl?
>
Yes, done.
> Please make sure to CC the relevant maintainers. IP maintainers were
> not CCed here. The get_maintainers script will tell you who to CC,
> please use it.
Fine, done.
Powered by blists - more mailing lists