Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: Leon Romanovsky <leon@kernel.org>
To: Michael Margolin <mrgolin@amazon.com>
Cc: Ilia Kolominsky <ikolomin@amazon.com>,
	jgg@nvidia.com, linux-rdma@vger.kernel.org, sleybo@amazon.com,
	matua@amazon.com, gal.pressman@linux.dev
Subject: Re: [PATCH for-next 0/3] RDMA/efa: Add Completion Actions support
Date: Tue, 17 Nov 2026 14:53:25 +0200	[thread overview]
Message-ID: <20261117125325.GE124972@unreal> (raw)
In-Reply-To: <20261117081850.GA18565@dev-dsk-mrgolin-1c-b2091117.eu-west-1.amazon.com>

On Tue, Nov 17, 2026 at 08:18:50AM +0000, Michael Margolin wrote:
> On Mon, Nov 16, 2026 at 01:29:13PM +0200, Leon Romanovsky wrote:
> > On Mon, Nov 09, 2026 at 08:13:46AM +0000, Ilia Kolominsky wrote:
> > > EFA Completion actions let an application observe work request
> > > completions without polling a completion queue: when a work request
> > > completes, locally or at the remote side, the device performs an action
> > > that the work request names, such as writing a value to a user buffer.
> > 
> > Why are these verbs EFA-specific? This looks like extension of already
> > generic UVERBS_METHOD_COMP_CNTR_* code and probably should be general too.
> 
> It is indeed an extension of Completion Counters, and later we plan to
> connect the two by allowing a Completion Counter to be used as a
> Completion Action, but unlike counters, actions mostly matter for
> unordered transports where a completion count alone doesn't tell the
> application which requests have completed, so it needs an efficient way
> to track completion of specific requests.
> 
> Do you think these semantics can be useful for other providers at this
> stage, or can we start with an EFA-specific interface and add a common
> ABI later if required?

I think the idea and implementation is general enough to be picked by
anyone who will decide to implement it later. Let's do it in common ABI,
please.

Thanks

> 
> Michael
> 

      reply	other threads:[~2026-11-17 12:53 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-09  8:13 [PATCH for-next 0/3] RDMA/efa: Add Completion Actions support Ilia Kolominsky
2026-11-09  8:13 ` [PATCH for-next 1/3] RDMA/efa: Update device interface Ilia Kolominsky
2026-11-09  8:17   ` sashiko-bot
2026-11-09  8:13 ` [PATCH for-next 2/3] RDMA/efa: Report the admin RESOURCE_BUSY status as EBUSY Ilia Kolominsky
2026-11-09  8:19   ` sashiko-bot
2026-11-09  8:13 ` [PATCH for-next 3/3] RDMA/efa: Add Completion Actions support Ilia Kolominsky
2026-11-09  8:26   ` sashiko-bot
2026-11-12 11:06 ` [PATCH for-next 0/3] " Ilia Kolominsky
2026-11-16 11:29 ` Leon Romanovsky
2026-11-17  8:18   ` Michael Margolin
2026-11-17 12:53     ` Leon Romanovsky [this message]

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=20261117125325.GE124972@unreal \
    --to=leon@kernel.org \
    --cc=gal.pressman@linux.dev \
    --cc=ikolomin@amazon.com \
    --cc=jgg@nvidia.com \
    --cc=linux-rdma@vger.kernel.org \
    --cc=matua@amazon.com \
    --cc=mrgolin@amazon.com \
    --cc=sleybo@amazon.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