From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 168D0A0093; Wed, 27 May 2020 13:43:31 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id E27B01D71F; Wed, 27 May 2020 13:43:29 +0200 (CEST) Received: from mail-io1-f47.google.com (mail-io1-f47.google.com [209.85.166.47]) by dpdk.org (Postfix) with ESMTP id 29F9D1D5D6 for ; Wed, 27 May 2020 13:43:28 +0200 (CEST) Received: by mail-io1-f47.google.com with SMTP id y5so2691685iob.12 for ; Wed, 27 May 2020 04:43:28 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=jrh5PgHgH0HEBcAqfcSNjHk5YA/p+X6XeLcM6mgtGYI=; b=Jvc+roTXUvkyQoeMsi658hxqFC7+0YYQ5E5lMxVdxGK/jVcTCV7iLUoY6UjIOyBoiE p6xV4mJIH2l7oAh3uCeOKPNAMwTJjxOoDOIXaHWphgfGUNfozaurLf+fcAbMkG7L7yZl rYrVTsTYNRFHMNfjhbCclUm6l7UctSeJtRwd2Irmdp5Z2mf4aTXdXLeL7vlaeIQZnx/C EBdHegQApuA+vak8VogPWS/tMz+faSF5bni92fD6svkxjwdXrp5SmcBTprvFdx9DNsYD k+ZltEiWY0ae+lcbvEn73UQJHAk4qTmnaiqqNxcHGufVuGOD93qX3OQM0rXZRw8rsg7V Sdxg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=jrh5PgHgH0HEBcAqfcSNjHk5YA/p+X6XeLcM6mgtGYI=; b=Qk7JQfqFu8DNT0cOE4h2wSkCaVP3i+OhOIePMyDrU8UjI3+oUpANczb2sERVEK5BCc igf6jr3vvrltjG9oNTMv/eR0P16ckrjtf4ITHGbtuiL2D/iMI2xB4y1YFOLPpfBP1tgM N8+HgRYBivq8GH9Uu25ga3bQxvZc/P+8hTW0OghjmdIWbyPMw2ByzyrMEPxf+aCGZygM KIlTnI5GVX+SxskCDXBWH/aS8lFbsvIxS31b9VN+cvvdlzDTMMj+O9hLZJP51woyuguV IGV7TmgiwUE/2LTq/qKeJSXsWjS2rb1EyB33zXujdMx4Gp2g2t8TSD2SldZpO/YdHOWR J1FA== X-Gm-Message-State: AOAM533H9sgfvpKfHCh83FuooZuA9eBc4kc46uRV9TIyIhuTIAk0vtqC lq6lrYpn8WkNwQHMEMuq4+cSLhOvNaspTXFviUI= X-Google-Smtp-Source: ABdhPJzLA9cKTZobCl5PHFLNZkm3/k8/yOa/PkqnRcEim+1XHKGWvgnWC5PiiIVl6xGkcrdxHVHc8kpwFVde6yRXrJU= X-Received: by 2002:a6b:f40b:: with SMTP id i11mr20645398iog.59.1590579807217; Wed, 27 May 2020 04:43:27 -0700 (PDT) MIME-Version: 1.0 References: <20200525212415.3173817-1-thomas@monjalon.net> <20200527070958.GE2554@platinum> <2617275.OaBZfIjOFF@thomas> In-Reply-To: <2617275.OaBZfIjOFF@thomas> From: Jerin Jacob Date: Wed, 27 May 2020 17:13:11 +0530 Message-ID: To: Thomas Monjalon Cc: Jerin Jacob , Olivier Matz , dpdk-dev , David Marchand , Nithin Dabilpuram , Krzysztof Kanas , Andrew Rybchenko , Ferruh Yigit , "Richardson, Bruce" , John McNamara , Marko Kovacevic Content-Type: text/plain; charset="UTF-8" Subject: Re: [dpdk-dev] [PATCH] mbuf: document rule for new fields and flags X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Wed, May 27, 2020 at 3:21 PM Thomas Monjalon wrote: > > 27/05/2020 09:31, Jerin Jacob: > > On Wed, May 27, 2020 at 12:39 PM Olivier Matz wrote: > > > On Tue, May 26, 2020 at 09:59:45PM +0530, Jerin Jacob wrote: > > > > On Tue, May 26, 2020 at 9:36 PM Olivier Matz wrote: > > > > > On Tue, May 26, 2020 at 01:09:37PM +0530, Jerin Jacob wrote: > > > > > > On Tue, May 26, 2020 at 2:54 AM Thomas Monjalon wrote: > > > > > > > > > > > > > > Since dynamic fields and flags were added in 19.11, > > > > > > > the idea was to use them for new features, not only PMD-specific. > > > > > > > > > > > > > > The rule is made more explicit in doxygen, in the mbuf guide, > > > > > > > and in the contribution design guidelines. > > > > > > > > > > > > > > For more information about the original design, see the presentation > > > > > > > https://www.dpdk.org/wp-content/uploads/sites/35/2019/10/DynamicMbuf.pdf > > > > > > > > > > > > > > Signed-off-by: Thomas Monjalon > > > > > > > --- > > > > > > > doc/guides/contributing/design.rst | 13 +++++++++++++ > > > > > > > doc/guides/prog_guide/mbuf_lib.rst | 23 +++++++++++++++++++++++ > > > > > > > lib/librte_mbuf/rte_mbuf_core.h | 2 ++ > > > > > > > 3 files changed, 38 insertions(+) > > > > > > > > > > > > > > diff --git a/doc/guides/contributing/design.rst b/doc/guides/contributing/design.rst > > > > > > > index d3dd694b65..508115d5bd 100644 > > > > > > > --- a/doc/guides/contributing/design.rst > > > > > > > +++ b/doc/guides/contributing/design.rst > > > > > > > +Mbuf features > > > > > > > +------------- > > > > > > > + > > > > > > > +The ``rte_mbuf`` structure must be kept small (128 bytes). > > > > > > > + > > > > > > > +In order to add new features without wasting buffer space for unused features, > > > > > > > +some fields and flags can be registered dynamically in a shared area. > > > > > > > > > > > > I think, instead of "can", it should be "must" > > > > > > > > > > > > > +The "dynamic" mbuf area is the default choice for the new features. > > > > > > > > > > In my opinion, Thomas' proposal is correct, with the next sentence > > > > > saying it is the default choice for new features. > > > > > > > > > > Giving guidelines is a good thing (thanks Thomas for documenting it), > > > > > but I don't think we should be too strict: the door remains open for > > > > > technical debate and exceptions. > > > > > > > > If you are open for the exception then it must be mention in what case > > > > the exception is allowed and what are the criteria of the exception? > > > > > > > > For example, Why did n't we choose the following patch as expectation > > > > http://patches.dpdk.org/patch/68733/ even if only one bit used. > > > > > > > > If we are not not defining the criteria, IMO, This patch serve no purpose than > > > > the existing situation. > > > > > > > > Do you think, any case where the dynamic scheme can not be used as a replacement > > > > for static other than performance hit. > > > > > > I don't think it is possible to anticipate all criteria in the > > > documentation. With Thomas' proposal, it gives a direction is and a > > > global view, but it must not completly replace reflection and > > > discussion. > > > > I don't think, we need to anticipate all the criteria in the documentation. > > At least ONE should be given as an example of an exception. > > I think it is too early to be more specific in the guidelines. > Do we agree this patch is a first good step in the documentation? IMO, there is a gap. The subject says the rule, but no rule here. We are just giving some guideline and following info in the patch given by Olivier is not expressed if we read the patch. " I don't think we should be too strict: the door remains open for technical debate and exceptions. " > We can extend the guidelines a bit later after having some > discussions on specific cases, and probably in the techboard too. > > > > I would say, > > a) If a feature takes only one bit and its part of normative API spec > > and it used in fastpath we should consider the static scheme. > > b) Adding an exception to the existing list needs approval at least > > from three maintainers > > > > For me, it is a very legitimate case to have support for > > http://patches.dpdk.org/patch/68733/ to the static scheme > > as it takes 1 bit for a feature and it is part of the normative spec. > > I don't get in explanation in the ml, why > > we can not make it as the static scheme for this case. > > We can continue discussion about this specific case in the right thread. Yes. The email thread[1] provided all the details. We have optimized to one bit for this feature. We are expecting Olivier to comment on the new proposal. [1] http://patches.dpdk.org/patch/68733/ > Note: I don't have a definitive opinion on it, I need to read it carefully. Please read it carefully and please provide any technical opinions if you have any. > > > > My worry is, if we are keeping as open-ended means, we are giving room > > for the disparity among the vendors/feature > > as I dont think, There is use case where dynamic scheme can not be > > used as a replacement > > for static other than performance hit.(Could think of any use case?) > > So open ended boils down to preference to specific feature/vendor. I > > think,that path should be avoided. > > Of course all rules and decisions have to be fair. > It's not even a question. Yes. But I dont think, this patch is not enforcing anything such, instead it makes it as an open-ended for more confusion. IMO, if it not black and white then better to not express the rule. > >