From: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
To: Olof Johansson <olof@lixom.net>, linux-kernel@vger.kernel.org
Cc: linux-accelerators@lists.ozlabs.org,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Frederic Barrat <fbarrat@linux.ibm.com>,
ogabbay@habana.ai, airlied@redhat.com, jglisse@redhat.com,
linuxppc-dev <linuxppc-dev@lists.ozlabs.org>
Subject: Re: [PATCH v2 1/5] drivers/accel: Introduce subsystem
Date: Sun, 27 Jan 2019 15:31:28 +1100 [thread overview]
Message-ID: <e3fc7e58-4d7e-f8aa-b96c-e515deb0494e@au1.ibm.com> (raw)
In-Reply-To: <20190125211335.65783-1-olof@lixom.net>
[+ linuxppc-dev, because cxl/ocxl are handled through powerpc - please
cc on future versions of this series]
On 26/1/19 8:13 am, Olof Johansson wrote:
> We're starting to see more of these kind of devices, the current
> upcoming wave will likely be around machine learning and inference
> engines. A few drivers have been added to drivers/misc for this, but
> it's timely to make it into a separate group of drivers/subsystem, to
> make it easier to find them, and to encourage collaboration between
> contributors.
>
> Over time, we expect to build shared frameworks that the drivers will
> make use of, but how that framework needs to look like to fill the needs
> is still unclear, and the best way to gain that knowledge is to give the
> disparate implementations a shared location.
>
> There has been some controversy around expectations for userspace
> stacks being open. The clear preference is to see that happen, and any
> driver and platform stack that is delivered like that will be given
> preferential treatment, and at some point in the future it might
> become the requirement. Until then, the bare minimum we need is an
> open low-level userspace such that the driver and HW interfaces can be
> exercised if someone is modifying the driver, even if the full details
> of the workload are not always available.
>
> Bootstrapping this with myself and Greg as maintainers (since the current
> drivers will be moving out of drivers/misc). Looking forward to expanding
> that group over time.
>
[snip]
> +
> +Hardware offload accelerator subsystem
> +======================================
> +
> +This is a brief overview of the subsystem (grouping) of hardware
> +accelerators kept under drivers/accel
> +
> +Types of hardware supported
> +---------------------------
> +
> + The general types of hardware supported are hardware devices that has
> + general interactions of sending commands and buffers to the hardware,
> + returning completions and possible filled buffers back, together
> + with the usual driver pieces around hardware control, setup, error
> + handling, etc.
> +
> + Drivers that fit into other subsystems are expected to be merged
> + there, and use the appropriate userspace interfaces of said functional
> + areas. We don't expect to see drivers for network, storage, graphics
> + and similar hardware implemented by drivers here.
> +
> +Expectations for contributions
> +------------------------------
> +
> + - Platforms and hardware that has fully open stacks, from Firmware to
> + Userspace, are always going to be given preferential treatment. These
> + platforms give the best insight for behavior and interaction of all
> + layers, including ability to improve implementation across the stack
> + over time.
> +
> + - If a platform is partially proprietary, it is still expected that the
> + portions that interact the driver can be shared in a form that allows
> + for exercising the hardware/driver and evolution of the interface over
> + time. This could be separated into a shared library and test/sample
> + programs, for example.
> +
> + - Over time, there is an expectation to converge drivers over to shared
> + frameworks and interfaces. Until then, the general rule is that no
> + more than one driver per vendor will be acceptable. For vendors that
> + aren't participating in the work towards shared frameworks over time,
> + we reserve the right to phase out support for the hardware.
How exactly do generic drivers for interconnect protocols, such as
cxl/ocxl, fit in here?
cxl and ocxl are not drivers for a specific device, they are generic
drivers which can be used with any device implementing the CAPI or
OpenCAPI protocol respectively - many of which will be FPGA boards
flashed with customer-designed accelerator cores for specific workloads,
some will be accelerators using ASICs or using FPGA images supplied by
vendors, some will be driven from userspace, others using the cxl/ocxl
kernel API, etc.
--
Andrew Donnellan OzLabs, ADL Canberra
andrew.donnellan@au1.ibm.com IBM Australia Limited
next prev parent reply other threads:[~2019-01-27 4:38 UTC|newest]
Thread overview: 103+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-01-23 0:00 [PATCH 00/15] Habana Labs kernel driver Oded Gabbay
2019-01-23 0:00 ` [PATCH 01/15] habanalabs: add skeleton driver Oded Gabbay
2019-01-23 0:49 ` Joe Perches
2019-01-25 19:18 ` Oded Gabbay
2019-01-23 12:28 ` Mike Rapoport
2019-01-23 12:40 ` Greg KH
2019-01-23 12:55 ` Mike Rapoport
2019-01-25 20:09 ` Oded Gabbay
2019-01-25 20:05 ` Oded Gabbay
2019-01-26 16:05 ` Arnd Bergmann
2019-01-26 16:24 ` Oded Gabbay
2019-01-26 21:14 ` Arnd Bergmann
2019-01-26 21:48 ` Oded Gabbay
2019-01-27 8:32 ` gregkh
2019-01-29 22:49 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 03/15] habanalabs: add basic Goya support Oded Gabbay
2019-01-23 12:28 ` Mike Rapoport
2019-01-25 20:32 ` Oded Gabbay
2019-01-27 6:39 ` Mike Rapoport
2019-01-28 7:44 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 04/15] habanalabs: add context and ASID modules Oded Gabbay
2019-01-23 12:28 ` Mike Rapoport
2019-01-25 21:07 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 05/15] habanalabs: add command buffer module Oded Gabbay
2019-01-23 12:28 ` Mike Rapoport
2019-01-25 21:47 ` Oded Gabbay
2019-01-27 6:49 ` Mike Rapoport
2019-01-28 7:55 ` Oded Gabbay
2019-01-28 8:41 ` Mike Rapoport
2019-01-23 0:00 ` [PATCH 06/15] habanalabs: add basic Goya h/w initialization Oded Gabbay
2019-01-25 7:46 ` Mike Rapoport
2019-01-28 10:35 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 07/15] habanalabs: add h/w queues module Oded Gabbay
2019-01-25 7:50 ` Mike Rapoport
2019-01-28 10:50 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 08/15] habanalabs: add event queue and interrupts Oded Gabbay
2019-01-25 7:51 ` Mike Rapoport
2019-01-28 11:14 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 09/15] habanalabs: add sysfs and hwmon support Oded Gabbay
2019-01-25 7:54 ` Mike Rapoport
2019-01-28 11:26 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 10/15] habanalabs: add device reset support Oded Gabbay
2019-01-27 7:51 ` Mike Rapoport
2019-01-28 12:53 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 11/15] habanalabs: add command submission module Oded Gabbay
2019-01-27 15:11 ` Mike Rapoport
2019-01-28 13:51 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 12/15] habanalabs: add virtual memory and MMU modules Oded Gabbay
2019-01-27 16:13 ` Mike Rapoport
2019-01-30 10:34 ` Oded Gabbay
2019-01-23 0:00 ` [PATCH 13/15] habanalabs: implement INFO IOCTL Oded Gabbay
2019-01-23 0:00 ` [PATCH 14/15] habanalabs: add debugfs support Oded Gabbay
2019-01-23 0:00 ` [PATCH 15/15] Update MAINTAINERS and CREDITS with habanalabs info Oded Gabbay
2019-01-23 12:27 ` [PATCH 00/15] Habana Labs kernel driver Mike Rapoport
2019-01-23 22:43 ` Oded Gabbay
2019-01-23 21:52 ` Olof Johansson
2019-01-23 22:40 ` Oded Gabbay
2019-01-23 23:16 ` Olof Johansson
2019-01-24 1:03 ` Andrew Donnellan
2019-01-24 11:59 ` Jonathan Cameron
2019-01-25 17:13 ` Olof Johansson
2019-02-24 22:23 ` Pavel Machek
2019-01-23 21:57 ` Dave Airlie
2019-01-23 22:02 ` Dave Airlie
2019-01-23 22:31 ` Oded Gabbay
2019-01-23 22:45 ` Dave Airlie
2019-01-23 23:04 ` Olof Johansson
2019-01-23 23:20 ` Jerome Glisse
2019-01-23 23:35 ` Oded Gabbay
2019-01-23 23:41 ` Olof Johansson
2019-01-23 23:40 ` Olof Johansson
2019-01-23 23:48 ` Jerome Glisse
2019-01-24 7:35 ` Daniel Vetter
2019-01-24 9:50 ` Oded Gabbay
2019-01-24 10:22 ` Dave Airlie
2019-01-25 0:13 ` Olof Johansson
2019-01-25 7:43 ` Daniel Vetter
2019-01-25 15:02 ` Olof Johansson
2019-01-25 16:00 ` Daniel Vetter
2019-01-24 23:51 ` Olof Johansson
2019-01-23 23:23 ` Oded Gabbay
2019-01-25 7:37 ` Greg Kroah-Hartman
2019-01-25 15:33 ` Olof Johansson
2019-01-25 16:06 ` Greg Kroah-Hartman
2019-01-25 17:12 ` Olof Johansson
2019-01-25 18:16 ` [PATCH/RFC 0/5] HW accel subsystem Olof Johansson
2019-01-25 18:16 ` [PATCH 1/5] drivers/accel: Introduce subsystem Olof Johansson
2019-01-25 21:13 ` [PATCH v2 " Olof Johansson
2019-01-26 17:09 ` Randy Dunlap
2019-01-27 4:31 ` Andrew Donnellan [this message]
2019-01-28 19:36 ` Frederic Barrat
2019-01-25 22:23 ` [PATCH " Daniel Vetter
2019-01-27 16:31 ` Daniel Vetter
2019-01-25 18:16 ` [PATCH 2/5] cxl: Move to drivers/accel Olof Johansson
2019-01-25 18:16 ` [PATCH 3/5] drivers/accel: cxl: Move non-uapi include files Olof Johansson
2019-01-25 18:16 ` [PATCH 4/5] ocxl: Move to drivers/accel Olof Johansson
2019-01-25 18:16 ` [PATCH 5/5] drivers/accel: ocxl: Move non-uapi include files Olof Johansson
2019-01-26 13:51 ` Greg Kroah-Hartman
2019-01-26 21:11 ` [PATCH/RFC 0/5] HW accel subsystem Arnd Bergmann
2019-02-01 9:10 ` Kenneth Lee
2019-02-01 10:07 ` Greg Kroah-Hartman
2019-02-01 12:09 ` Kenneth Lee
2019-01-26 13:52 ` [PATCH 00/15] Habana Labs kernel driver Greg Kroah-Hartman
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=e3fc7e58-4d7e-f8aa-b96c-e515deb0494e@au1.ibm.com \
--to=andrew.donnellan@au1.ibm.com \
--cc=airlied@redhat.com \
--cc=fbarrat@linux.ibm.com \
--cc=gregkh@linuxfoundation.org \
--cc=jglisse@redhat.com \
--cc=linux-accelerators@lists.ozlabs.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=ogabbay@habana.ai \
--cc=olof@lixom.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;
as well as URLs for NNTP newsgroup(s).