linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: John Hubbard <jhubbard@nvidia.com>
To: Michal Hocko <mhocko@kernel.org>
Cc: "Jason Gunthorpe" <jgg@mellanox.com>,
	"Ira Weiny" <ira.weiny@intel.com>,
	"Ralph Campbell" <rcampbell@nvidia.com>,
	"linux-nvdimm@lists.01.org" <linux-nvdimm@lists.01.org>,
	"nouveau@lists.freedesktop.org" <nouveau@lists.freedesktop.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"dri-devel@lists.freedesktop.org"
	<dri-devel@lists.freedesktop.org>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>,
	"Jérôme Glisse" <jglisse@redhat.com>,
	"Ben Skeggs" <bskeggs@redhat.com>,
	"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>,
	"Christoph Hellwig" <hch@lst.de>
Subject: Re: [PATCH 18/22] mm: mark DEVICE_PUBLIC as broken
Date: Tue, 25 Jun 2019 23:07:13 -0700	[thread overview]
Message-ID: <f71f16a4-d411-a540-fc71-34d15f4f02d6@nvidia.com> (raw)
In-Reply-To: <20190626054554.GA17798@dhcp22.suse.cz>

On 6/25/19 10:45 PM, Michal Hocko wrote:
> On Tue 25-06-19 20:15:28, John Hubbard wrote:
>> On 6/19/19 12:27 PM, Jason Gunthorpe wrote:
>>> On Thu, Jun 13, 2019 at 06:23:04PM -0700, John Hubbard wrote:
>>>> On 6/13/19 5:43 PM, Ira Weiny wrote:
>>>>> On Thu, Jun 13, 2019 at 07:58:29PM +0000, Jason Gunthorpe wrote:
>>>>>> On Thu, Jun 13, 2019 at 12:53:02PM -0700, Ralph Campbell wrote:
>>>>>>>
>>>> ...
>>>>> So I think it is ok.  Frankly I was wondering if we should remove the public
>>>>> type altogether but conceptually it seems ok.  But I don't see any users of it
>>>>> so...  should we get rid of it in the code rather than turning the config off?
>>>>>
>>>>> Ira
>>>>
>>>> That seems reasonable. I recall that the hope was for those IBM Power 9
>>>> systems to use _PUBLIC, as they have hardware-based coherent device (GPU)
>>>> memory, and so the memory really is visible to the CPU. And the IBM team
>>>> was thinking of taking advantage of it. But I haven't seen anything on
>>>> that front for a while.
>>>
>>> Does anyone know who those people are and can we encourage them to
>>> send some patches? :)
>>>
>>
>> I asked about this, and it seems that the idea was: DEVICE_PUBLIC was there
>> in order to provide an alternative way to do things (such as migrate memory
>> to and from a device), in case the combination of existing and near-future
>> NUMA APIs was insufficient. This probably came as a follow-up to the early
>> 2017-ish conversations about NUMA, in which the linux-mm recommendation was
>> "try using HMM mechanisms, and if those are inadequate, then maybe we can
>> look at enhancing NUMA so that it has better handling of advanced (GPU-like)
>> devices".
> 
> Yes that was the original idea. It sounds so much better to use a common
> framework rather than awkward special cased cpuless NUMA nodes with
> a weird semantic. User of the neither of the two has shown up so I guess
> that the envisioned HW just didn't materialized. Or has there been a
> completely different approach chosen?

The HW showed up, alright: it's the IBM Power 9, which provides HW-based
memory coherency between its CPUs and GPUs. So on this system, the CPU is
allowed to access GPU memory, which *could* be modeled as DEVICE_PUBLIC.

However, what happened was that the system worked well enough with a combination
of the device driver, plus NUMA APIs, plus heaven knows what sort of /proc tuning
might have also gone on. :) No one saw the need to reach for the DEVICE_PUBLIC
functionality.

> 
>> In the end, however, _PUBLIC was never used, nor does anyone in the local
>> (NVIDIA + IBM) kernel vicinity seem to have plans to use it.  So it really
>> does seem safe to remove, although of course it's good to start with 
>> BROKEN and see if anyone pops up and complains.
> 
> Well, I do not really see much of a difference. Preserving an unused
> code which doesn't have any user in sight just adds a maintenance burden
> whether the code depends on BROKEN or not. We can always revert patches
> which remove the code once a real user shows up.

Sure, I don't see much difference either. Either way seems fine.

thanks,
-- 
John Hubbard
NVIDIA

  reply	other threads:[~2019-06-26  6:07 UTC|newest]

Thread overview: 103+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-13  9:43 dev_pagemap related cleanups Christoph Hellwig
2019-06-13  9:43 ` [PATCH 01/22] mm: remove the unused ARCH_HAS_HMM_DEVICE Kconfig option Christoph Hellwig
2019-06-13 18:30   ` Jason Gunthorpe
2019-06-13  9:43 ` [PATCH 02/22] mm: remove the struct hmm_device infrastructure Christoph Hellwig
2019-06-13 18:46   ` Jason Gunthorpe
2019-06-13 23:06   ` [Nouveau] " John Hubbard
2019-06-13  9:43 ` [PATCH 03/22] mm: remove hmm_devmem_add_resource Christoph Hellwig
2019-06-13 18:52   ` Jason Gunthorpe
2019-06-14  6:19     ` Christoph Hellwig
2019-06-14  0:54   ` [Nouveau] " John Hubbard
2019-06-20 19:32   ` Michal Hocko
2019-06-13  9:43 ` [PATCH 04/22] mm: don't clear ->mapping in hmm_devmem_free Christoph Hellwig
2019-06-13 19:05   ` Jason Gunthorpe
2019-06-14  6:21     ` Christoph Hellwig
2019-06-14  1:46   ` John Hubbard
2019-06-20 19:36   ` Michal Hocko
2019-06-13  9:43 ` [PATCH 05/22] mm: export alloc_pages_vma Christoph Hellwig
2019-06-14  1:47   ` John Hubbard
2019-06-14  6:23     ` Christoph Hellwig
2019-06-20 19:17   ` Michal Hocko
2019-06-24 18:24     ` Dan Williams
2019-06-25  7:23       ` Christoph Hellwig
2019-06-25 15:00         ` Michal Hocko
2019-06-25 18:03           ` Dan Williams
2019-06-25 19:00             ` Michal Hocko
2019-06-25 19:52               ` Dan Williams
2019-06-26  5:46                 ` Michal Hocko
2019-06-26 16:14                   ` Dan Williams
2019-06-27  6:41                     ` Michal Hocko
2019-06-13  9:43 ` [PATCH 06/22] mm: factor out a devm_request_free_mem_region helper Christoph Hellwig
2019-06-13 19:16   ` Jason Gunthorpe
2019-06-14  6:24     ` Christoph Hellwig
2019-06-15  2:21   ` John Hubbard
2019-06-15 14:30     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 07/22] memremap: move dev_pagemap callbacks into a separate structure Christoph Hellwig
2019-06-13 19:18   ` Jason Gunthorpe
2019-06-13 20:14   ` Logan Gunthorpe
2019-06-13  9:43 ` [PATCH 08/22] memremap: pass a struct dev_pagemap to ->kill Christoph Hellwig
2019-06-13 19:26   ` Jason Gunthorpe
2019-06-13 20:12   ` Logan Gunthorpe
2019-06-13 20:15     ` Dan Williams
2019-06-13  9:43 ` [PATCH 09/22] memremap: lift the devmap_enable manipulation into devm_memremap_pages Christoph Hellwig
2019-06-13 19:34   ` Jason Gunthorpe
2019-06-13 20:13     ` Dan Williams
2019-06-14  6:28     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 10/22] memremap: add a migrate callback to struct dev_pagemap_ops Christoph Hellwig
2019-06-13 23:42   ` Ralph Campbell
2019-06-14  6:33     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 11/22] memremap: remove the data field in struct dev_pagemap Christoph Hellwig
2019-06-13 19:37   ` Jason Gunthorpe
2019-06-13  9:43 ` [PATCH 12/22] memremap: provide an optional internal refcount " Christoph Hellwig
2019-06-13  9:43 ` [PATCH 13/22] device-dax: use the dev_pagemap internal refcount Christoph Hellwig
2019-06-14  0:22   ` Ira Weiny
2019-06-14  6:35     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 14/22] nouveau: use alloc_page_vma directly Christoph Hellwig
2019-06-13 19:39   ` Jason Gunthorpe
2019-06-13  9:43 ` [PATCH 15/22] nouveau: use devm_memremap_pages directly Christoph Hellwig
2019-06-13  9:43 ` [PATCH 16/22] mm: remove hmm_vma_alloc_locked_page Christoph Hellwig
2019-06-13  9:43 ` [PATCH 17/22] mm: remove hmm_devmem_add Christoph Hellwig
2019-06-13 19:42   ` Jason Gunthorpe
2019-06-14  6:39     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 18/22] mm: mark DEVICE_PUBLIC as broken Christoph Hellwig
2019-06-13 19:44   ` Jason Gunthorpe
2019-06-13 19:53     ` Ralph Campbell
2019-06-13 19:58       ` Jason Gunthorpe
2019-06-14  0:43         ` Ira Weiny
2019-06-14  1:23           ` John Hubbard
2019-06-19 19:27             ` Jason Gunthorpe
2019-06-19 19:46               ` Dan Williams
2019-06-26  3:15               ` John Hubbard
2019-06-26  5:45                 ` Michal Hocko
2019-06-26  6:07                   ` John Hubbard [this message]
2019-06-14  6:43           ` Christoph Hellwig
2019-06-20 19:26   ` Michal Hocko
2019-06-25  7:29     ` Christoph Hellwig
2019-06-25 11:44       ` Jason Gunthorpe
2019-06-25 11:59         ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 19/22] mm: simplify ZONE_DEVICE page private data Christoph Hellwig
2019-06-13  9:43 ` [PATCH 20/22] mm: sort out the DEVICE_PRIVATE Kconfig mess Christoph Hellwig
2019-06-13 19:55   ` Jason Gunthorpe
2019-06-13  9:43 ` [PATCH 21/22] mm: remove the HMM config option Christoph Hellwig
2019-06-13 20:01   ` Jason Gunthorpe
2019-06-14  6:47     ` Christoph Hellwig
2019-06-13  9:43 ` [PATCH 22/22] mm: don't select MIGRATE_VMA_HELPER from HMM_MIRROR Christoph Hellwig
2019-06-13 20:04   ` Jason Gunthorpe
2019-06-14  1:53   ` [Nouveau] " John Hubbard
2019-06-14  6:48     ` Christoph Hellwig
2019-06-13 14:16 ` dev_pagemap related cleanups Jason Gunthorpe
2019-06-14  6:12   ` Christoph Hellwig
2019-06-13 18:27 ` Dan Williams
2019-06-13 20:17   ` Logan Gunthorpe
2019-06-13 20:21     ` Dan Williams
2019-06-13 20:24       ` Logan Gunthorpe
2019-06-13 20:48         ` Andrew Morton
2019-06-13 20:40   ` Jason Gunthorpe
2019-06-13 21:21     ` Christoph Hellwig
2019-06-13 23:10       ` Jason Gunthorpe
2019-06-14  6:14         ` Christoph Hellwig
2019-06-14  0:31     ` Ira Weiny
2019-06-14  6:13   ` Christoph Hellwig
2019-06-15  1:14     ` Dan Williams
2019-06-15  8:33       ` Christoph Hellwig
2019-06-15 18:09         ` Dan Williams

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=f71f16a4-d411-a540-fc71-34d15f4f02d6@nvidia.com \
    --to=jhubbard@nvidia.com \
    --cc=bskeggs@redhat.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=hch@lst.de \
    --cc=ira.weiny@intel.com \
    --cc=jgg@mellanox.com \
    --cc=jglisse@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-nvdimm@lists.01.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=mhocko@kernel.org \
    --cc=nouveau@lists.freedesktop.org \
    --cc=rcampbell@nvidia.com \
    /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).