netdev - Re: icmp6_send() is too expensive

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] [day] [month] [year] [list]

Message-Id: <20181004.123343.1000066789406204771.davem@davemloft.net>
Date:   Thu, 04 Oct 2018 12:33:43 -0700 (PDT)
From:   David Miller <davem@...emloft.net>
To:     0xeffeff@...il.com
Cc:     netdev@...r.kernel.org
Subject: Re: icmp6_send() is too expensive

From: Jeff Barnhill <0xeffeff@...il.com>
Date: Thu, 4 Oct 2018 15:02:39 -0400

> As mentioned here:
> https://www.spinics.net/lists/netdev/msg505054.html
> 
> icmp6_send() can be expensive when there are a lot of devices and
> anycast addresses. One solution I've prototyped is adding a global
> hash table to store and allow more efficient searches for anycast
> addresses.  This works and prevents the long delays I've seen when
> many neighbors are invalidated in large numbers.
> 
> A colleague mentioned that another alternative may be to do something
> like an rt6_lookup on the address and check for the RTF_ANYCAST flag.
> This is much simpler, and I'm in the process of testing it now.
> 
> Is there any preference or recommendation as to which way to proceed
> before I submit a patch?

Generally speaking, using existing tables and infrastructure is
preferrable to adding new stuff.