netdev - Re: [PATCH] net: Add ndo_gso

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]

Date:	Tue, 7 Oct 2014 11:15:32 -0700
From:	Alexei Starovoitov <alexei.starovoitov@...il.com>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	Tom Herbert <therbert@...gle.com>, Jesse Gross <jesse@...ira.com>,
	Or Gerlitz <gerlitz.or@...il.com>,
	Alexander Duyck <alexander.h.duyck@...el.com>,
	John Fastabend <john.r.fastabend@...el.com>,
	Jeff Kirsher <jeffrey.t.kirsher@...el.com>,
	David Miller <davem@...emloft.net>,
	Linux Netdev List <netdev@...r.kernel.org>,
	Thomas Graf <tgraf@...g.ch>,
	Pravin Shelar <pshelar@...ira.com>,
	Andy Zhou <azhou@...ira.com>
Subject: Re: [PATCH] net: Add ndo_gso_check

On Tue, Oct 7, 2014 at 10:23 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> On Tue, 2014-10-07 at 09:50 -0700, Alexei Starovoitov wrote:
>
>> it's definitely more difficult to properly implement
>> CHECKSUM_UNNECESSARY in HW, but it's worth it.
>> CHECKSUM_COMPLETE is a burden on software. Old NICs
>> used to do that, but overhead of recomputing csum for every
>> step of packet parsing and header modifications is too high.
>> sw routers, bridges and < L4 networking devices are
>> simpler and faster with CHECKSUM_UNNECESSARY.
>
> Really this is wrong. Once you validated/pulled a header,
> adjusting the complete checksum is in the order of 10 cycles or so,
> ie less than 1% of the other costs.

correct, but there is also postpull() cost for every pull.
and there are many of them for encapsulated traffic.
It's small, but it's not zero.

> UNNECESSARY usually requests complex NIC firmware, and usually the NIC
> has fewer cores than the host.
>
> It mostly worked for basic IP+TCP kind of traffic, but once you want
> complex cloud models, it is a major pain.
>
> If we need to optimize csum_partial() for short lengths, lets do it,
> instead of pushing hardware vendors adding more and more schemes.

csum_partial() is in asm already. probably not much more can be squeezed.
I'm not suggesting that NICs must always do UNNECESSARY.
COMPLETE is a low bar which is magnitude better than NONE,
but if HW has programmable parser it should be able to take advantage
of it and UNNECESSARY is an established model.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html