From: Jiri Pirko <jiri@resnulli.us>
To: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Ariel Levkovich <lariel@mellanox.com>,
netdev@vger.kernel.org, kuba@kernel.org,
xiyou.wangcong@gmail.com, ast@kernel.org, daniel@iogearbox.net
Subject: Re: [PATCH net-next v2 0/3] ] TC datapath hash api
Date: Wed, 8 Jul 2020 16:45:08 +0200 [thread overview]
Message-ID: <20200708144508.GB3667@nanopsycho.orion> (raw)
In-Reply-To: <20877e09-45f2-fa89-d11c-4ae73c9a7310@mojatatu.com>
Wed, Jul 08, 2020 at 03:54:14PM CEST, jhs@mojatatu.com wrote:
>On 2020-07-07 6:05 a.m., Jiri Pirko wrote:
>> Fri, Jul 03, 2020 at 01:22:47PM CEST, jhs@mojatatu.com wrote:
>> > Hi,
>> >
>> > Several comments:
>> > 1) I agree with previous comments that you should
>> > look at incorporating this into skbedit.
>> > Unless incorporating into skbedit introduces huge
>> > complexity, IMO it belongs there.
>> >
>> > 2) I think it would make sense to create a skb hash classifier
>> > instead of tying this entirely to flower i.e i should not
>> > have to change u32 just so i can support hash classification.
>>
>> Well, we don't have multiple classifiers for each flower match, we have
>> them all in one classifier.
>
>Packet data matches, yes - makes sense. You could argue the same for
>the other classifiers.
>
>> It turned out to be very convenient and
>> intuitive for people to use one classifier to do the job for them.
>
>IMO:
>For this specific case where _offload_ is the main use case i think
>it is not a good idea because flower on ingress is slow.
Eh? What do you mean by that?
>The goal of offloading classifiers to hardware is so one can reduce
>consumed cpu cycles on the host. If the hardware
>has done the classification for me, a simple hash lookup of the
>32 bit skbhash(similar to fw) in the host would be a lot less
>compute intensive than running flower's algorithm.
It is totally up to the driver/fw how they decide to offload flower.
There are multiple ways. So I don't really follow what do you mean by
"flower's algorithm"
>
>I think there is a line for adding everything in one place,
>my main concern is that this feature this is needed
>by all classifiers and not just flower.
"All" is effectively only flower. Let's be clear about that.
>
>
>> Modularity is nice, but useability is I think more important in this
>> case. Flower turned out to do good job there.
>>
>
>For humans, agreed everything in one place is convinient.
>Note: your arguement could be used for "ls" to include "grep"
>functionality because in my scripts I do both most of the time.
>
>cheers,
>jamal
>
>
>
>> + Nothing stops you from creating separate classifier to match on hash
>> as you wanted to :)
>>
>>
next prev parent reply other threads:[~2020-07-08 14:45 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-01 18:47 [PATCH net-next v2 0/3] ] TC datapath hash api Ariel Levkovich
2020-07-01 18:47 ` [PATCH net-next v2 1/3] net/sched: Introduce action hash Ariel Levkovich
2020-07-01 22:20 ` kernel test robot
2020-07-02 20:52 ` Cong Wang
2020-07-01 18:47 ` [PATCH net-next v2 2/3] net/flow_dissector: add packet hash dissection Ariel Levkovich
2020-07-01 18:47 ` [PATCH net-next v2 3/3] net/sched: cls_flower: Add hash info to flow classification Ariel Levkovich
2020-07-03 11:22 ` [PATCH net-next v2 0/3] ] TC datapath hash api Jamal Hadi Salim
2020-07-05 17:26 ` Ariel Levkovich
2020-07-05 21:50 ` Jamal Hadi Salim
2020-07-06 0:28 ` Cong Wang
2020-07-09 13:52 ` Ariel Levkovich
2020-07-06 0:23 ` Cong Wang
2020-07-07 10:05 ` Jiri Pirko
2020-07-08 13:54 ` Jamal Hadi Salim
2020-07-08 14:45 ` Jiri Pirko [this message]
2020-07-09 11:00 ` Jamal Hadi Salim
2020-07-09 12:19 ` Jiri Pirko
2020-07-10 12:04 ` Jamal Hadi Salim
2020-08-07 10:41 ` Jamal Hadi Salim
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200708144508.GB3667@nanopsycho.orion \
--to=jiri@resnulli.us \
--cc=ast@kernel.org \
--cc=daniel@iogearbox.net \
--cc=jhs@mojatatu.com \
--cc=kuba@kernel.org \
--cc=lariel@mellanox.com \
--cc=netdev@vger.kernel.org \
--cc=xiyou.wangcong@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox