Netdev List
 help / color / mirror / Atom feed
From: Pablo Neira Ayuso <pablo@netfilter.org>
To: Jakub Kicinski <jakub.kicinski@netronome.com>
Cc: netfilter-devel@vger.kernel.org, davem@davemloft.net,
	netdev@vger.kernel.org, vishal@chelsio.com, saeedm@mellanox.com,
	jiri@resnulli.us
Subject: Re: [PATCH 0/4 net-next] flow_offload: update mangle action representation
Date: Fri, 30 Aug 2019 11:07:10 +0200	[thread overview]
Message-ID: <20190830090710.g7q2chf3qulfs5e4@salvia> (raw)
In-Reply-To: <20190829185448.0b502af8@cakuba.netronome.com>

On Thu, Aug 29, 2019 at 06:54:48PM -0700, Jakub Kicinski wrote:
> On Fri, 30 Aug 2019 02:53:32 +0200, Pablo Neira Ayuso wrote:
> > * Offsets do not need to be on the 32-bits boundaries anymore. This
> >   patchset adds front-end code to adjust the offset and length coming
> >   from the tc pedit representation, so drivers get an exact header field
> >   offset and length.
> 
> But drivers use offsetof(start of field) to match headers, and I don't
> see you changing that. So how does this work then?

Drivers can only use offsetof() for fields that are on the 32-bits
boundary.

Before this patchset, if you want to mangle the destination port, then
the driver needs to refer to the source port offset and the length is
4 bytes, so the mask is telling what needs to be mangled.

After this patchset, the offset is set to the destination port, the
length is set to 2-bytes, and the mask is telling what bytes of the
destination port field you specifically want to update.

It's just 100 LOC of preprocessing that is simplifying driver
codebase.

> Say - I want to change the second byte of an IPv4 address.

Then, the front-end sets the offset to IPv4 address header field, and
the mask tells what to update.

> > * The front-end coalesces consecutive pedit actions into one single
> >   word, so drivers can mangle IPv6 and ethernet address fields in one
> >   single go.
> 
> You still only coalesce up to 16 bytes, no?

You only have to rise FLOW_ACTION_MANGLE_MAXLEN coming in this patch
if you need more. I don't know of any packet field larger than 16
bytes. If there is a use-case for this, it should be easy to rise that
definition.

> As I said previously drivers will continue to implement mangle action
> merge code if that's the case. It'd be nice if core did the coalescing,
> and mark down first and last action, in case there is a setup cost for
> rewrite group.

In this patchset, the core front-end is doing the coalescing.

  reply	other threads:[~2019-08-30  9:07 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-30  0:53 [PATCH 0/4 net-next] flow_offload: update mangle action representation Pablo Neira Ayuso
2019-08-30  0:53 ` [PATCH net-next 1/4] net: flow_offload: flip mangle action mask Pablo Neira Ayuso
2019-08-30  0:53 ` [PATCH net-next 2/4] net: flow_offload: bitwise AND on mangle action value field Pablo Neira Ayuso
2019-08-30  0:53 ` [PATCH net-next 3/4] net: flow_offload: mangle action at byte level Pablo Neira Ayuso
2019-08-30 15:28   ` Vlad Buslov
2019-08-30  0:53 ` [PATCH net-next 4/4] netfilter: nft_payload: packet mangling offload support Pablo Neira Ayuso
2019-08-30  1:54 ` [PATCH 0/4 net-next] flow_offload: update mangle action representation Jakub Kicinski
2019-08-30  9:07   ` Pablo Neira Ayuso [this message]
2019-08-30 22:33     ` Jakub Kicinski
2019-08-31 14:22       ` Pablo Neira Ayuso
2019-09-01 20:47         ` Jakub Kicinski
2019-09-03  0:05           ` Pablo Neira Ayuso

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=20190830090710.g7q2chf3qulfs5e4@salvia \
    --to=pablo@netfilter.org \
    --cc=davem@davemloft.net \
    --cc=jakub.kicinski@netronome.com \
    --cc=jiri@resnulli.us \
    --cc=netdev@vger.kernel.org \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=saeedm@mellanox.com \
    --cc=vishal@chelsio.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