Received-SPF: pass (google.com: domain of linux-kernel+bounces-151400-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Date: Fri, 19 Apr 2024 09:13:57 -0300
From: Jason Gunthorpe <jgg@ziepe.ca>
To: Konstantin Taranov <kotaranov@microsoft.com>
Cc: Konstantin Taranov <kotaranov@linux.microsoft.com>,
	"sharmaajay@microsoft.com" <sharmaajay@microsoft.com>,
	Long Li <longli@microsoft.com>, "leon@kernel.org" <leon@kernel.org>,
	"linux-rdma@vger.kernel.org" <linux-rdma@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH rdma-next 2/2] RDMA/mana_ib: Implement get_dma_mr
Message-ID: <20240419121357.GB223006@ziepe.ca>
References: <1713363659-30156-1-git-send-email-kotaranov@linux.microsoft.com>
 <1713363659-30156-3-git-send-email-kotaranov@linux.microsoft.com>
 <20240417145106.GV223006@ziepe.ca>
 <PAXPR83MB055789D898814B9F73B15C3BB40D2@PAXPR83MB0557.EURPRD83.prod.outlook.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <PAXPR83MB055789D898814B9F73B15C3BB40D2@PAXPR83MB0557.EURPRD83.prod.outlook.com>

On Fri, Apr 19, 2024 at 09:14:14AM +0000, Konstantin Taranov wrote:
> > From: Jason Gunthorpe <jgg@ziepe.ca>
> > On Wed, Apr 17, 2024 at 07:20:59AM -0700, Konstantin Taranov wrote:
> > > From: Konstantin Taranov <kotaranov@microsoft.com>
> > >
> > > Implement allocation of DMA-mapped memory regions.
> > >
> > > Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
> > > ---
> > >  drivers/infiniband/hw/mana/device.c |  1 +
> > >  drivers/infiniband/hw/mana/mr.c     | 36
> > +++++++++++++++++++++++++++++
> > >  include/net/mana/gdma.h             |  5 ++++
> > >  3 files changed, 42 insertions(+)
> > 
> > What is the point of doing this without supporting enough verbs to allow a
> > kernel ULP?
> > 
> 
> True, the proposed code is useless at this state.
> Nevertheless, mana_ib team aims to send kernel ULP patches after we are done
> with uverbs pathes (i.e., udata is not null). As this change does not conflict with the
> current effort, I decided to send this patch now. I can extend the series to make
> it more useful.
> 
> Jason, could  you suggest a minimal list of ib_device_ops methods, that includes
> get_dma_mr, which can be approved?

Is there any chance you can send a single series to support a
ULP. NVMe or something like?

Jason