public inbox for linux-rdma@vger.kernel.org
 help / color / mirror / Atom feed
From: Leon Romanovsky <leon@kernel.org>
To: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Cc: jgg@ziepe.ca, Dean Luick <dean.luick@cornelisnetworks.com>,
	Breandan Cunningham <brendan.cunningham@cornelisnetworks.com>,
	linux-rdma@vger.kernel.org
Subject: Re: [PATCH for-next] RDMA/hfi2: Consolidate ABI files and setup uverbs access
Date: Tue, 17 Mar 2026 11:51:48 +0200	[thread overview]
Message-ID: <20260317095148.GS61385@unreal> (raw)
In-Reply-To: <2ce72f8e-3a7d-46c6-9b1e-68f99c91a6d2@cornelisnetworks.com>

On Mon, Mar 16, 2026 at 05:31:40PM -0400, Dennis Dalessandro wrote:
> On 3/16/26 10:27 AM, Leon Romanovsky wrote:
> > On Wed, Mar 11, 2026 at 01:33:57PM -0400, Dennis Dalessandro wrote:
> > > hfi1 driver is being replaced eventually with an hfi2 driver. Until that
> > > happens rather than have all the duplicated code in header files, make hfi1
> > > use hfi2 variants where it can. When compatibility breaks we'll keep a
> > > separate hfi1 version.
> > > 
> > > This is the case for the <dev>_status struture. The hfi1 varaint is single
> > > port and uses a freezemsg char array while the new hfi2 chip provides
> > > multiple ports and thus needs and array of ports.
> > > 
> > > Likewise the tid info struct is expanded for hfi2 so we include both an
> > > hfi1 and hfi2 vaiant.
> > > 
> > > There is a naming conflict with the trace_hfi1_ctxt_info() call. It has been
> > > renamed to remove the 1 from the function name to keep the code readable
> > > but allow it to compile due to the #define in hfi1_ioctl.h.
> > > 
> > > The big departure from hfi1 is that we are no longer supporting access from
> > > users through a private character device. Instead we define two custom
> > > verbs ojects. dv0/1, which proivdes methods for what in hfi1 are individual
> > > IOCTLs. We have added an additional method to get stats related to page
> > > pinning done by the driver.
> > > 
> > > The reason we are not removing the hfi1_ioctl.h and hfi1_user.h header
> > > files is user application compatibility. User apps depend on having these
> > > files available. Once user apps have converted and hfi1 is removed these
> > > files will be deleted as well.
> > 
> > What are the applications that use include/uapi/rdma/hfi/hfi1_* directly?
> > I have hard time to find any application on github which includes them.
> > 
> 
> rdma-core (submitted PR but have some checkpatch type stuff to fix that was
> missed), psm2, and libfabric (opx) all use these.

At least in rdma-core, the include/uapi/rdma files are copied rather than
used directly. You will probably need to apply the same approach in the
other libraries.

My point is that you most likely do not need to keep the old
include/uapi/rdma/hfi/hfi1_* files.

Thanks

> 
> -Denny
> 

  reply	other threads:[~2026-03-17  9:51 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-11 17:33 [PATCH for-next] RDMA/hfi2: Consolidate ABI files and setup uverbs access Dennis Dalessandro
2026-03-16 14:27 ` Leon Romanovsky
2026-03-16 21:31   ` Dennis Dalessandro
2026-03-17  9:51     ` Leon Romanovsky [this message]
2026-03-17 19:55       ` Dennis Dalessandro

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=20260317095148.GS61385@unreal \
    --to=leon@kernel.org \
    --cc=brendan.cunningham@cornelisnetworks.com \
    --cc=dean.luick@cornelisnetworks.com \
    --cc=dennis.dalessandro@cornelisnetworks.com \
    --cc=jgg@ziepe.ca \
    --cc=linux-rdma@vger.kernel.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