mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Edward Srouji <edwards@nvidia.com>
To: Leon Romanovsky <leon@kernel.org>, Jason Gunthorpe <jgg@ziepe.ca>,
	Sabrina Dubroca <sd@queasysnail.net>
Cc: linux-rdma@vger.kernel.org, linux-kernel@vger.kernel.org,
	netdev@vger.kernel.org, Akiva Goldberger <agoldberger@nvidia.com>,
	Michael Gur <michaelgur@nvidia.com>
Subject: Re: [PATCH rdma-next 00/11] RDMA/mlx5: Preparation for SD support
Date: Sun, 22 Nov 2026 12:32:38 +0200	[thread overview]
Message-ID: <0caf9acf-2640-40b5-b638-7d54db90f577@nvidia.com> (raw)
In-Reply-To: <20261117-mlx5-sd-prep-v1-0-8efd124c7895@nvidia.com>



On 11/17/2026 4:01 PM, Edward Srouji wrote:
> This series prepares mlx5_ib for Socket Direct (SD), where one IB device
> spans several core devices, one per PCIe function. Today the driver
> assumes a single core device per IB device and reaches it through
> dev->mdev all over the code.
> 
> Instead, each call site now finds its core device through the resource
> it works on: QPs, SRQs, AHs and MWs through their PD, and DEVX objects,
> UARs and mmaps through their ucontext. A CQ and an MR store the core
> device they were created on, since they cannot always reach their owner
> when they need it. The SD series can then change where a resource lives
> in the resolvers, without touching the call sites.
> 
> The resolvers follow one naming scheme: <resource>_to_mdev() returns a
> core device and <resource>_to_dev() an IB device. Where several resource
> types are accepted, a _Generic() dispatches to per-type helpers, which
> are prefixed with '__' and are not meant to be called directly.
> 
> State kept once per IB device, such as the ODP page-fault EQ, delay drop
> and local loopback, stays on dev->mdev. The SD series decides how each
> of them works across core devices.
> 
> This series only changes how each call site finds its core device. With
> one core device per IB device every resolver returns dev->mdev, so the
> behaviour is unchanged. Pre-existing issues in the touched code are
> intentionally left alone so that each patch stays a mechanical
> conversion.
> 
> Patches 1-2:  name device pointers consistently, dev for the IB device
>                and mdev for the core device.
> Patches 3-7:  resolve the core device through the PD.
> Patches 8-11: resolve the core device through the ucontext, and keep it
>                on the CQ.
> 
> Signed-off-by: Edward Srouji <edwards@nvidia.com>
Since this preparation series touches many files as part of a renaming 
process and some minor field changes, it is expected from Sashiko to 
flood us with many pre-existing issues.
I'll keep fixes for such issues out of this series and send them as 
separate patches with proper Fixes tags.
I'll reply briefly on each patch to note which findings we confirmed 
(and will fix separately) and which we believe can't happen.

      parent reply	other threads:[~2026-11-22 10:32 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 14:01 Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 01/11] RDMA/mlx5: Enforce dev/mdev naming for device pointers Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 02/11] RDMA/mlx5: Rename to_mdev() helpers that return an IB device Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 03/11] RDMA/mlx5: Select the PD and SRQ core device through the PD Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 04/11] RDMA/mlx5: Select the core device through PD for QP commands Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 05/11] RDMA/mlx5: Cache the MR core device and resolve the MW's via its PD Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 06/11] RDMA/mlx5: Select the AH core device through their PD Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 07/11] RDMA/mlx5: Select the QP core device through its PD Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 08/11] RDMA/mlx5: Resolve core device from ucontext for DEVX resources Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 09/11] RDMA/mlx5: Cache the CQ core device and resolve it via cq_to_mdev() Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 10/11] RDMA/mlx5: Resolve the bfreg UAR-index core device from the ucontext Edward Srouji
2026-11-17 14:01 ` [PATCH rdma-next 11/11] RDMA/mlx5: Resolve UAR and mmap core devices via " Edward Srouji
2026-11-22 10:32 ` Edward Srouji [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=0caf9acf-2640-40b5-b638-7d54db90f577@nvidia.com \
    --to=edwards@nvidia.com \
    --cc=agoldberger@nvidia.com \
    --cc=jgg@ziepe.ca \
    --cc=leon@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=michaelgur@nvidia.com \
    --cc=netdev@vger.kernel.org \
    --cc=sd@queasysnail.net \
    /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

all inboxes | Powered by JetHome®