[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20231017173448.3f1c35aa@kernel.org>
Date: Tue, 17 Oct 2023 17:34:48 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Alexander Duyck <alexander.duyck@...il.com>
Cc: Ahmed Zaki <ahmed.zaki@...el.com>, netdev@...r.kernel.org,
intel-wired-lan@...ts.osuosl.org, corbet@....net,
jesse.brandeburg@...el.com, anthony.l.nguyen@...el.com,
davem@...emloft.net, edumazet@...gle.com, pabeni@...hat.com,
vladimir.oltean@....com, andrew@...n.ch, horms@...nel.org,
mkubecek@...e.cz, willemdebruijn.kernel@...il.com,
linux-doc@...r.kernel.org, Wojciech Drewek <wojciech.drewek@...el.com>
Subject: Re: [PATCH net-next v4 1/6] net: ethtool: allow symmetric-xor RSS
hash for any flow type
On Tue, 17 Oct 2023 13:41:18 -0700 Alexander Duyck wrote:
> I am thinking of this from a software engineering perspective. This
> symmetric-xor aka simplified-toeplitz is actually much cheaper to
> implement in software than the original. As such I would want it to be
> considered a separate algorithm as I could make use of something like
> that when having to implement RSS in QEMU for instance.
That's exactly why XOR and CRC32 _algorithms_ already exist.
CPUs have instructions to do them word at a time.
ETH_RSS_HASH_TOP_BIT, /* Configurable RSS hash function -
Toeplitz */
ETH_RSS_HASH_XOR_BIT, /* Configurable RSS hash function - Xor */
ETH_RSS_HASH_CRC32_BIT, /* Configurable RSS hash function - Crc32 */
If efficient SW implementation is important why do some weird
bastardized para-toeplitz and not crc32? Hashes fairly well
from what I recall with the older NFPs. x86 has an instruction
for it, IIRC it was part of SSE but on normal registers.
> Based on earlier comments it doesn't change the inputs, it just
> changes how I have to handle the data and the key. It starts reducing
> things down to something like the Intel implementation of Flow
> Director in terms of how the key gets generated and hashed.
About Flow Director I know only that it is bad :)
Powered by blists - more mailing lists