From: Jason Gunthorpe <jgg@mellanox.com>
To: Jerome Glisse <jglisse@redhat.com>
Cc: Logan Gunthorpe <logang@deltatee.com>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
"Rafael J . Wysocki" <rafael@kernel.org>,
Bjorn Helgaas <bhelgaas@google.com>,
Christian Koenig <christian.koenig@amd.com>,
Felix Kuehling <Felix.Kuehling@amd.com>,
"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>,
"dri-devel@lists.freedesktop.org"
<dri-devel@lists.freedesktop.org>, Christoph Hellwig <hch@lst.de>,
Marek Szyprowski <m.szyprowski@samsung.com>,
Robin Murphy <robin.murphy@arm.com>,
Joerg Roedel <jroedel@suse.de>,
"iommu@lists.linux-foundation.org"
<iommu@lists.linux-foundation.org>
Subject: Re: [RFC PATCH 3/5] mm/vma: add support for peer to peer to device vma
Date: Wed, 30 Jan 2019 04:30:27 +0000 [thread overview]
Message-ID: <20190130043020.GC30598@mellanox.com> (raw)
In-Reply-To: <20190130000805.GS3176@redhat.com>
On Tue, Jan 29, 2019 at 07:08:06PM -0500, Jerome Glisse wrote:
> On Tue, Jan 29, 2019 at 11:02:25PM +0000, Jason Gunthorpe wrote:
> > On Tue, Jan 29, 2019 at 03:44:00PM -0500, Jerome Glisse wrote:
> >
> > > > But this API doesn't seem to offer any control - I thought that
> > > > control was all coming from the mm/hmm notifiers triggering p2p_unmaps?
> > >
> > > The control is within the driver implementation of those callbacks.
> >
> > Seems like what you mean by control is 'the exporter gets to choose
> > the physical address at the instant of map' - which seems reasonable
> > for GPU.
> >
> >
> > > will only allow p2p map to succeed for objects that have been tagged by the
> > > userspace in some way ie the userspace application is in control of what
> > > can be map to peer device.
> >
> > I would have thought this means the VMA for the object is created
> > without the map/unmap ops? Or are GPU objects and VMAs unrelated?
>
> GPU object and VMA are unrelated in all open source GPU driver i am
> somewhat familiar with (AMD, Intel, NVidia). You can create a GPU
> object and never map it (and thus never have it associated with a
> vma) and in fact this is very common. For graphic you usualy only
> have hand full of the hundreds of GPU object your application have
> mapped.
I mean the other way does every VMA with a p2p_map/unmap point to
exactly one GPU object?
ie I'm surprised you say that p2p_map needs to have policy, I would
have though the policy is applied when the VMA is created (ie objects
that are not for p2p do not have p2p_map set), and even for GPU
p2p_map should really only have to do with window allocation and pure
'can I even do p2p' type functionality.
> Idea is that we can only ask exporter to be predictable and still allow
> them to fail if things are really going bad.
I think hot unplug / PCI error recovery is one of the 'really going
bad' cases..
> I think i put it in the comment above the ops but in any cases i should
> write something in documentation with example and thorough guideline.
> Note that there won't be any mmu notifier to mmap of a device file
> unless the device driver calls for it or there is a syscall like munmap
> or mremap or mprotect well any syscall that work on vma.
This is something we might need to explore, does calling
zap_vma_ptes() invoke enough notifiers that a MMU notifiers or HMM
mirror consumer will release any p2p maps on that VMA?
> If we ever want to support full pin then we might have to add a
> flag so that GPU driver can refuse an importer that wants things
> pin forever.
This would become interesting for VFIO and RDMA at least - I don't
think VFIO has anything like SVA so it would want to import a p2p_map
and indicate that it will not respond to MMU notifiers.
GPU can refuse, but maybe RDMA would allow it...
Jason
next prev parent reply other threads:[~2019-01-30 4:30 UTC|newest]
Thread overview: 95+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-01-29 17:47 [RFC PATCH 0/5] Device peer to peer (p2p) through vma jglisse
2019-01-29 17:47 ` [RFC PATCH 1/5] pci/p2p: add a function to test peer to peer capability jglisse
2019-01-29 18:24 ` Logan Gunthorpe
2019-01-29 19:44 ` Greg Kroah-Hartman
2019-01-29 19:53 ` Jerome Glisse
2019-01-29 20:44 ` Logan Gunthorpe
2019-01-29 21:00 ` Jerome Glisse
2019-01-29 19:56 ` Alex Deucher
2019-01-29 20:00 ` Jerome Glisse
2019-01-29 20:24 ` Logan Gunthorpe
2019-01-29 21:28 ` Alex Deucher
2019-01-30 10:25 ` Christian König
2019-01-29 17:47 ` [RFC PATCH 2/5] drivers/base: " jglisse
2019-01-29 18:26 ` Logan Gunthorpe
2019-01-29 19:54 ` Jerome Glisse
2019-01-29 19:46 ` Greg Kroah-Hartman
2019-01-29 19:56 ` Jerome Glisse
2019-01-29 17:47 ` [RFC PATCH 3/5] mm/vma: add support for peer to peer to device vma jglisse
2019-01-29 18:36 ` Logan Gunthorpe
2019-01-29 19:11 ` Jerome Glisse
2019-01-29 19:24 ` Logan Gunthorpe
2019-01-29 19:44 ` Jerome Glisse
2019-01-29 20:43 ` Logan Gunthorpe
2019-01-30 7:52 ` Christoph Hellwig
2019-01-29 19:32 ` Jason Gunthorpe
2019-01-29 19:50 ` Jerome Glisse
2019-01-29 20:24 ` Jason Gunthorpe
2019-01-29 20:44 ` Jerome Glisse
2019-01-29 23:02 ` Jason Gunthorpe
2019-01-30 0:08 ` Jerome Glisse
2019-01-30 4:30 ` Jason Gunthorpe [this message]
2019-01-30 15:43 ` Jerome Glisse
2019-01-29 20:39 ` Logan Gunthorpe
2019-01-29 20:57 ` Jerome Glisse
2019-01-29 21:30 ` Logan Gunthorpe
2019-01-29 21:50 ` Jerome Glisse
2019-01-29 22:58 ` Logan Gunthorpe
2019-01-29 23:47 ` Jerome Glisse
2019-01-30 1:17 ` Logan Gunthorpe
2019-01-30 2:48 ` Jerome Glisse
2019-01-30 4:18 ` Jason Gunthorpe
2019-01-30 8:00 ` Christoph Hellwig
2019-01-30 15:49 ` Jerome Glisse
2019-01-30 19:06 ` Jason Gunthorpe
2019-01-30 19:45 ` Logan Gunthorpe
2019-01-30 19:59 ` Jason Gunthorpe
2019-01-30 21:01 ` Logan Gunthorpe
2019-01-30 21:50 ` Jason Gunthorpe
2019-01-30 22:52 ` Logan Gunthorpe
2019-01-30 23:30 ` Jason Gunthorpe
2019-01-31 8:13 ` Christoph Hellwig
2019-01-31 15:37 ` Jerome Glisse
2019-01-31 19:02 ` Jason Gunthorpe
2019-01-31 19:19 ` Logan Gunthorpe
2019-01-31 19:54 ` Jason Gunthorpe
2019-01-31 19:35 ` Jerome Glisse
2019-01-31 19:44 ` Logan Gunthorpe
2019-01-31 19:58 ` Jason Gunthorpe
2019-01-30 17:17 ` Logan Gunthorpe
2019-01-30 18:56 ` Jason Gunthorpe
2019-01-30 19:22 ` Jerome Glisse
2019-01-30 19:38 ` Jason Gunthorpe
2019-01-30 20:00 ` Logan Gunthorpe
2019-01-30 20:11 ` Jason Gunthorpe
2019-01-30 20:43 ` Jerome Glisse
2019-01-30 20:50 ` Jason Gunthorpe
2019-01-30 21:45 ` Jerome Glisse
2019-01-30 21:56 ` Jason Gunthorpe
2019-01-30 22:30 ` Jerome Glisse
2019-01-30 22:33 ` Jason Gunthorpe
2019-01-30 22:47 ` Jerome Glisse
2019-01-30 22:51 ` Jason Gunthorpe
2019-01-30 22:58 ` Jerome Glisse
2019-01-30 19:52 ` Logan Gunthorpe
2019-01-30 20:35 ` Jerome Glisse
2019-01-29 20:58 ` Jason Gunthorpe
2019-01-30 8:02 ` Christoph Hellwig
2019-01-30 10:33 ` Koenig, Christian
2019-01-30 15:55 ` Jerome Glisse
2019-01-30 17:26 ` Christoph Hellwig
2019-01-30 17:32 ` Logan Gunthorpe
2019-01-30 17:39 ` Jason Gunthorpe
2019-01-30 18:05 ` Jerome Glisse
2019-01-30 17:44 ` Jason Gunthorpe
2019-01-30 18:13 ` Logan Gunthorpe
2019-01-30 18:50 ` Jerome Glisse
2019-01-31 8:02 ` Christoph Hellwig
2019-01-31 15:03 ` Jerome Glisse
2019-01-30 19:19 ` Jason Gunthorpe
2019-01-30 19:48 ` Logan Gunthorpe
2019-01-30 20:44 ` Jason Gunthorpe
2019-01-31 8:05 ` Christoph Hellwig
2019-01-31 15:11 ` Jerome Glisse
2019-01-29 17:47 ` [RFC PATCH 4/5] mm/hmm: add support for peer to peer to HMM device memory jglisse
2019-01-29 17:47 ` [RFC PATCH 5/5] mm/hmm: add support for peer to peer to special device vma jglisse
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=20190130043020.GC30598@mellanox.com \
--to=jgg@mellanox.com \
--cc=Felix.Kuehling@amd.com \
--cc=bhelgaas@google.com \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=gregkh@linuxfoundation.org \
--cc=hch@lst.de \
--cc=iommu@lists.linux-foundation.org \
--cc=jglisse@redhat.com \
--cc=jroedel@suse.de \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-pci@vger.kernel.org \
--cc=logang@deltatee.com \
--cc=m.szyprowski@samsung.com \
--cc=rafael@kernel.org \
--cc=robin.murphy@arm.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).