KVM Archive on lore.kernel.org
 help / color / Atom feed
From: Daniel Vetter <daniel.vetter@ffwll.ch>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: "Tomasz Figa" <tfiga@chromium.org>,
	linux-s390 <linux-s390@vger.kernel.org>,
	linux-samsung-soc <linux-samsung-soc@vger.kernel.org>,
	"Jan Kara" <jack@suse.cz>, "Kees Cook" <keescook@chromium.org>,
	"KVM list" <kvm@vger.kernel.org>, "Linux MM" <linux-mm@kvack.org>,
	"Mauro Carvalho Chehab" <mchehab+huawei@kernel.org>,
	"John Hubbard" <jhubbard@nvidia.com>,
	LKML <linux-kernel@vger.kernel.org>,
	"DRI Development" <dri-devel@lists.freedesktop.org>,
	"Jason Gunthorpe" <jgg@ziepe.ca>,
	"Jérôme Glisse" <jglisse@redhat.com>,
	"Daniel Vetter" <daniel.vetter@intel.com>,
	"Dan Williams" <dan.j.williams@intel.com>,
	"Linus Torvalds" <torvalds@linux-foundation.org>,
	"Andrew Morton" <akpm@linux-foundation.org>,
	"Linux ARM" <linux-arm-kernel@lists.infradead.org>,
	"open list:DMA BUFFER SHARING FRAMEWORK"
	<linux-media@vger.kernel.org>
Subject: Re: [PATCH v2 09/17] mm: Add unsafe_follow_pfn
Date: Sat, 10 Oct 2020 23:50:27 +0200
Message-ID: <CAKMK7uGhq+BiaJ5jD+bkO4VOaCPuUZ_empA3Ojr1AsvwNef6QQ@mail.gmail.com> (raw)
In-Reply-To: <20201010213554.GD3939@pendragon.ideasonboard.com>

On Sat, Oct 10, 2020 at 11:36 PM Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
>
> Hi Tomasz,
>
> On Sat, Oct 10, 2020 at 07:22:48PM +0200, Tomasz Figa wrote:
> > On Fri, Oct 9, 2020 at 7:52 PM Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> > > On Fri, Oct 9, 2020 at 2:48 PM Jason Gunthorpe <jgg@ziepe.ca> wrote:
> > > > On Fri, Oct 09, 2020 at 02:37:23PM +0200, Mauro Carvalho Chehab wrote:
> > > >
> > > > > I'm not a mm/ expert, but, from what I understood from Daniel's patch
> > > > > description is that this is unsafe *only if*  __GFP_MOVABLE is used.
> > > >
> > > > No, it is unconditionally unsafe. The CMA movable mappings are
> > > > specific VMAs that will have bad issues here, but there are other
> > > > types too.
> > > >
> > > > The only way to do something at a VMA level is to have a list of OK
> > > > VMAs, eg because they were creatd via a special mmap helper from the
> > > > media subsystem.
> > > >
> > > > > Well, no drivers inside the media subsystem uses such flag, although
> > > > > they may rely on some infrastructure that could be using it behind
> > > > > the bars.
> > > >
> > > > It doesn't matter, nothing prevents the user from calling media APIs
> > > > on mmaps it gets from other subsystems.
> > >
> > > I think a good first step would be to disable userptr of non struct
> > > page backed storage going forward for any new hw support. Even on
> > > existing drivers. dma-buf sharing has been around for long enough now
> > > that this shouldn't be a problem. Unfortunately right now this doesn't
> > > seem to exist, so the entire problem keeps getting perpetuated.
> > >
> > > > > If this is the case, the proper fix seems to have a GFP_NOT_MOVABLE
> > > > > flag that it would be denying the core mm code to set __GFP_MOVABLE.
> > > >
> > > > We can't tell from the VMA these kinds of details..
> > > >
> > > > It has to go the other direction, evey mmap that might be used as a
> > > > userptr here has to be found and the VMA specially created to allow
> > > > its use. At least that is a kernel only change, but will need people
> > > > with the HW to do this work.
> > >
> > > I think the only reasonable way to keep this working is:
> > > - add a struct dma_buf *vma_tryget_dma_buf(struct vm_area_struct *vma);
> > > - add dma-buf export support to fbdev and v4l
> >
> > I assume you mean V4L2 and not the obsolete V4L that is emulated in
> > the userspace by libv4l. If so, every video device that uses videobuf2
> > gets DMA-buf export for free and there is nothing needed to enable it.

Yeah. And I missed that v4l2 added dma-buf export too.

> > We probably still have a few legacy drivers using videobuf (non-2),
> > but IMHO those should be safe to put behind some disabled-by-default
> > Kconfig symbol or even completely drop, as the legacy framework has
> > been deprecated for many years already.
>
> There's 8 drivers left, and they support a very large number of devices.
> I expect unhappy users distros stop shipping them. On the other hand,
> videobuf has been deprecated for a loooooooong time, so there has been
> plenty of time to convert the remaining drivers to videobuf2. If nobody
> can do it, then we'll have to drop support for these devices given the
> security issues.

Again, the issue here is _only_ with follow_pfn. For videobuf1 this
means videbuf-dma-contig.c userptr support is broken. Unlike videobuf2
it means it's broken for all usage (not just zero-copy userptr),
because videbuf-dma-contig.c lacks the pin_user_pages path. But that
would be easy to add if this poses a  problem I think - we just need
to carry over the pin_user_pages_fast logic from videbuf2, no driver
changes required. But of course I don't think we should do that before
someone reports the regression, since videobuf1 userptr is doubly
deprecated :-)

Everything else keeps working with videobuf1 with my patch series. So
depending upon which videobuf1 implementations these 8 drivers use,
you might not even have any real breakage there.

> We have moved media drivers to staging in the past when there wasn't
> enough maintenance effort, we could do so here too.

I'm not breaking the world with this, it's really very minimal
use-case. At least as far as I'm understanding the entire media
subsystem here.
-Daniel

> > > - roll this out everywhere we still need it.
> > >
> > > Realistically this just isn't going to happen. And anything else just
> > > reimplements half of dma-buf, which is kinda pointless (you need
> > > minimally refcounting and some way to get at a promise of a permanent
> > > sg list for dma. Plus probably the vmap for kernel cpu access.
> > >
> > > > > Please let address the issue on this way, instead of broken an
> > > > > userspace API that it is there since 1991.
> > > >
> > > > It has happened before :( It took 4 years for RDMA to undo the uAPI
> > > > breakage caused by a security fix for something that was a 15 years
> > > > old bug.
> > >
> > > Yeah we have a bunch of these on the drm side too. Some of them are
> > > really just "you have to upgrade userspace", and there's no real fix
> > > for the security nightmare without that.
> >
> > I think we need to phase out such userspace indeed. The Kconfig symbol
> > allows enabling the unsafe functionality for anyone who still needs
> > it, so I think it's not entirely a breakage.
>
> --
> Regards,
>
> Laurent Pinchart



-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

  reply index

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-09  7:59 [PATCH v2 00/17] follow_pfn and other iomap races Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 01/17] drm/exynos: Stop using frame_vector helpers Daniel Vetter
2020-10-16  7:42   ` John Hubbard
2020-10-09  7:59 ` [PATCH v2 02/17] drm/exynos: Use FOLL_LONGTERM for g2d cmdlists Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 03/17] misc/habana: Stop using frame_vector helpers Daniel Vetter
2020-10-10 20:26   ` Oded Gabbay
2020-10-10 21:32     ` Daniel Vetter
2020-10-10 21:41       ` Daniel Vetter
2020-10-10 21:47         ` Oded Gabbay
2020-10-16  7:45   ` John Hubbard
2020-10-09  7:59 ` [PATCH v2 04/17] misc/habana: Use FOLL_LONGTERM for userptr Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 05/17] mm/frame-vector: Use FOLL_LONGTERM Daniel Vetter
2020-10-16  7:54   ` John Hubbard
2020-10-16  8:03     ` Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 06/17] media: videobuf2: Move frame_vector into media subsystem Daniel Vetter
2020-10-09 10:14   ` Mauro Carvalho Chehab
2020-10-09 16:57     ` Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 07/17] mm: Close race in generic_access_phys Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 08/17] s390/pci: Remove races against pte updates Daniel Vetter
2020-10-12 14:03   ` Niklas Schnelle
2020-10-12 14:19     ` Daniel Vetter
2020-10-12 14:39       ` Niklas Schnelle
2020-10-21  7:55       ` Niklas Schnelle
2020-10-22  7:39         ` Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 09/17] mm: Add unsafe_follow_pfn Daniel Vetter
2020-10-09 10:34   ` Mauro Carvalho Chehab
2020-10-09 12:21     ` Jason Gunthorpe
2020-10-09 12:37       ` Mauro Carvalho Chehab
2020-10-09 12:39         ` Mauro Carvalho Chehab
2020-10-09 12:48         ` Jason Gunthorpe
2020-10-09 17:52           ` Daniel Vetter
2020-10-09 18:01             ` Jason Gunthorpe
2020-10-09 19:31               ` Daniel Vetter
2020-10-10  9:21             ` Mauro Carvalho Chehab
2020-10-10 10:53               ` Daniel Vetter
2020-10-10 11:39                 ` Mauro Carvalho Chehab
2020-10-10 11:56                   ` Daniel Vetter
2020-10-10 17:22             ` Tomasz Figa
2020-10-10 21:35               ` Laurent Pinchart
2020-10-10 21:50                 ` Daniel Vetter [this message]
2020-10-11  6:27                   ` Mauro Carvalho Chehab
2020-10-11  6:36                     ` Mauro Carvalho Chehab
2020-10-10 21:11             ` Laurent Pinchart
2020-10-12 10:46           ` Marek Szyprowski
2020-10-12 13:49             ` Daniel Vetter
2020-10-10 17:30         ` Tomasz Figa
2020-10-09  7:59 ` [PATCH v2 10/17] media/videbuf1|2: Mark follow_pfn usage as unsafe Daniel Vetter
2020-10-10  9:24   ` Mauro Carvalho Chehab
2020-10-09  7:59 ` [PATCH v2 11/17] vfio/type1: Mark follow_pfn " Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 12/17] PCI: Obey iomem restrictions for procfs mmap Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 13/17] /dev/mem: Only set filp->f_mapping Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 14/17] resource: Move devmem revoke code to resource framework Daniel Vetter
2020-10-09 10:59   ` Greg Kroah-Hartman
2020-10-09 12:31   ` Jason Gunthorpe
2020-10-09 14:24     ` Daniel Vetter
2020-10-09 14:32       ` Jason Gunthorpe
2020-10-09 18:28         ` Dan Williams
2020-10-15  0:09           ` Jason Gunthorpe
2020-10-15  7:52             ` Daniel Vetter
2020-10-15  7:55               ` Daniel Vetter
2020-10-15 15:29             ` Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 15/17] sysfs: Support zapping of binary attr mmaps Daniel Vetter
2020-10-09 10:58   ` Greg Kroah-Hartman
2020-10-09  7:59 ` [PATCH v2 16/17] PCI: Revoke mappings like devmem Daniel Vetter
2020-10-09  7:59 ` [PATCH v2 17/17] drm/i915: Properly request PCI BARs Daniel Vetter
2020-10-09  9:47   ` Ville Syrjälä
2020-10-09 10:01     ` Daniel Vetter
2020-10-09 10:41       ` Ville Syrjälä
2020-10-09 14:18         ` Daniel Vetter

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=CAKMK7uGhq+BiaJ5jD+bkO4VOaCPuUZ_empA3Ojr1AsvwNef6QQ@mail.gmail.com \
    --to=daniel.vetter@ffwll.ch \
    --cc=akpm@linux-foundation.org \
    --cc=dan.j.williams@intel.com \
    --cc=daniel.vetter@intel.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=jack@suse.cz \
    --cc=jgg@ziepe.ca \
    --cc=jglisse@redhat.com \
    --cc=jhubbard@nvidia.com \
    --cc=keescook@chromium.org \
    --cc=kvm@vger.kernel.org \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=linux-samsung-soc@vger.kernel.org \
    --cc=mchehab+huawei@kernel.org \
    --cc=tfiga@chromium.org \
    --cc=torvalds@linux-foundation.org \
    /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

KVM Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/kvm/0 kvm/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 kvm kvm/ https://lore.kernel.org/kvm \
		kvm@vger.kernel.org
	public-inbox-index kvm

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.kvm


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git