[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240729152938.GE3371438@nvidia.com>
Date: Mon, 29 Jul 2024 12:29:38 -0300
From: Jason Gunthorpe <jgg@...dia.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: James Bottomley <James.Bottomley@...senpartnership.com>,
Greg KH <gregkh@...uxfoundation.org>,
Laurent Pinchart <laurent.pinchart@...asonboard.com>,
Dan Williams <dan.j.williams@...el.com>, ksummit@...ts.linux.dev,
linux-cxl@...r.kernel.org, linux-rdma@...r.kernel.org,
netdev@...r.kernel.org
Subject: Re: [MAINTAINERS SUMMIT] Device Passthrough Considered Harmful?
On Mon, Jul 29, 2024 at 07:56:07AM -0700, Jakub Kicinski wrote:
> On Sun, 28 Jul 2024 11:49:44 -0400 James Bottomley wrote:
> > cross subsystem NAKs
>
> Could y'all please stop saying "cross subsystem NAKs"..
> It makes it sound like networking is nacking an addition to RDMA
> or storage.
Well, that is exactly what is happening. You are NAKing stuff we need
to operate RDMA, DPDK, and other parts of the device.
> The problem is that nVidia insists on making their proprietary
> gateway a "misc driver" usable in all subsystems.
Sigh, open code is not "proprietary", we've been over this.
fwctl manages the whole device, it is not tied to a subsystem and it
is not "usable" by any specific subsystem. This reflects the design of
HW/FW that is obviously not split up into per-Linux-subsystem
blocks. It is technically appropriate that is global and not somehow
weirdly split by subsystem.
I'm not going to backdoor generic full device fw access through RDMA
into the kernel.
> If they want to add something at the top level, all affected
> subsystems should have a say.
A say is fine, but it is not like netdev is suffering any real
technical harm to justify a full blown NAK fight like this.
Jason
Powered by blists - more mailing lists