linux-rdma.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Leon Romanovsky <leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
To: Adit Ranadive <aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
Cc: dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org,
	linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	pv-drivers-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org
Subject: Re: [PATCH for-next v1 2/2] RDMA/vmw_pvrdma: Update device query parameters and port caps
Date: Thu, 24 Aug 2017 07:32:04 +0300	[thread overview]
Message-ID: <20170824043204.GA1724@mtr-leonro.local> (raw)
In-Reply-To: <6cffa7b7-63ee-ccd8-558d-af4538012159-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>

[-- Attachment #1: Type: text/plain, Size: 4133 bytes --]

On Wed, Aug 23, 2017 at 12:29:21PM -0700, Adit Ranadive wrote:
> On Wed, Aug 23, 2017 at 11:53:26AM -0700, Leon Romanovsky wrote:
> > On Wed, Aug 23, 2017 at 11:08:01AM -0700, Adit Ranadive wrote:
> > > On Wed, Aug 23, 2017 at 12:09:20AM +0300, Leon Romanovsky wrote:
> > > > On Tue, Aug 22, 2017 at 11:19:01PM -0700, Adit Ranadive wrote:
> > > > > Added support for two device caps - max_sge_rd, max_fast_reg_page_list_len
> > > > > and the IP_BASED_GIDS port cap flag.
> > > > >
> > > > > Reviewed-by: Jorgen Hansen <jhansen-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
> > > > > Reviewed-by: Bryan Tan <bryantan-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
> > > > > Reviewed-by: Aditya Sarwade <asarwade-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
> > > > > Signed-off-by: Adit Ranadive <aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
> > > > > ---
> > > > >  drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h | 9 ++++++++-
> > > > >  drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c   | 9 +++++++++
> > > > >  2 files changed, 17 insertions(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h
> > > > > index 3a308ff..df0a6b5 100644
> > > > > --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h
> > > > > +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h
> > > > > @@ -149,6 +149,13 @@
> > > > >  	((_dev->dsr->caps.mode == PVRDMA_DEVICE_MODE_ROCE) &&		\
> > > > >  	 (PVRDMA_IS_VERSION17(_dev) || PVRDMA_IS_VERSION18(_dev)))
> > > > >
> > > > > +/*
> > > > > + * Get capability values based on device version.
> > > > > + */
> > > > > +
> > > > > +#define PVRDMA_GET_CAP(_dev, _old_val, _val) \
> > > > > +	((PVRDMA_IS_VERSION18(_dev)) ? _val : _old_val)
> > > > > +
> > > >
> > > > The current macro implementation will require you to go and update all
> > > > the places once you will move to new version. In simple case, everything
> > > > is supported and you will change your check of PVRDMA_IS_VERSION18 to
> > > > something PVRDMA_IS_VERSIONXXX and it will work.
> > >
> > > Thanks for taking a look. That's the intention I wrote the macro with so it
> > > should pick up the value based on the current device version. If you are
> > > concerned that we will change device caps without changing the PVRDMA
> > > version, that's something we avoid for compatibility reasons.
> >
> > I have hard time to understand how will you extend this macro once new
> > feature is introduced.
> >
> > Let's for an example conduct the following experiment:
> > 1. IB/core gets new exciting ROCE_V3
> > 2. You are adding support to it and increasing the version to be VERSION19.
> > 3. You now have code for ROCE_V2 and ROCE_V3.
> >
> > So the question is: how will your VRDMA_GET_CAP() look in such case?
> > RECE_V2 is supported in VERSION18 and above, while ROCE_V3 is supported
> > in VERSION19 only.
>
> The values for the capabilities come from our DSR (device shared region) itself.
> This would have nothing to do with the exact RoCE version but is based
> on the device version itself. So, if we add a VERSION19, we can change the
> macro to something like PVRDMA_IS_VERSION_NEXT_GEN, it will just pick up the
> new values defined in the DSR rather than based on RoCE versions.
> If we have removed support for that capability for VERSION19, that should
> simply be 0.

Thanks, I got the idea and looking forward to see how it handles over time.

>
> To make this more clear, I can put the RoCE v1/v2 type checks, each in own
> macro. Drop them from the PVRDMA_IS_VERSION checks. Rename the
> PVRDMA_IS_VERSION18 to PVRDMA_IS_NEXTVERSION so we will always pick up the
> cap values correctly for future versions and we are not dependent on the
> RoCE version itself.

I don't think that it is needed at this stage.

>
> > And if we add more than one new feature, how will it look?
> >
> > Thanks
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

  parent reply	other threads:[~2017-08-24  4:32 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-08-23  6:18 [PATCH for-next v1 0/2] RDMA/vmw_pvrdma: Add RoCEv2 support Adit Ranadive
     [not found] ` <cover.1503468979.git.aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
2017-08-23  6:19   ` [PATCH for-next v1 1/2] " Adit Ranadive
     [not found]     ` <91bf53451b8618acc731f6b8b4235103abae7687.1503468979.git.aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
2017-08-24  4:33       ` Leon Romanovsky
2017-08-24 13:51       ` Yuval Shaia
2017-08-23  6:19   ` [PATCH for-next v1 2/2] RDMA/vmw_pvrdma: Update device query parameters and port caps Adit Ranadive
     [not found]     ` <a31ee424df9ad51371ff38e4f8239a390df13947.1503468979.git.aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
2017-08-23  9:09       ` Leon Romanovsky
     [not found]         ` <20170823090920.GN1724-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-08-23 18:08           ` Adit Ranadive
     [not found]             ` <34d28cff-79a5-b604-fc1e-8d083f53225d-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
2017-08-23 18:53               ` Leon Romanovsky
     [not found]                 ` <20170823185326.GX1724-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-08-23 19:29                   ` Adit Ranadive
     [not found]                     ` <6cffa7b7-63ee-ccd8-558d-af4538012159-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org>
2017-08-24  4:32                       ` Leon Romanovsky [this message]
2017-08-24  4:32       ` Leon Romanovsky
2017-08-24 14:33       ` Yuval Shaia
2017-08-24 21:47         ` Adit Ranadive
2017-08-24 21:36   ` [PATCH for-next v1 0/2] RDMA/vmw_pvrdma: Add RoCEv2 support Doug Ledford

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=20170824043204.GA1724@mtr-leonro.local \
    --to=leon-dgejt+ai2ygdnm+yrofe0a@public.gmane.org \
    --cc=aditr-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org \
    --cc=dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=pv-drivers-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org \
    /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;
as well as URLs for NNTP newsgroup(s).