Message-ID: <CAKMK7uFzxWF7V=7vkeNC-8shsPZRgdz9fYTsn0ayENv2BpnFEg@mail.gmail.com>
Date:   Mon, 5 Oct 2020 20:16:33 +0200
From:   Daniel Vetter <daniel.vetter@...ll.ch>
To:     Jason Gunthorpe <jgg@...pe.ca>
Cc:     DRI Development <dri-devel@...ts.freedesktop.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Daniel Vetter <daniel.vetter@...el.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        John Hubbard <jhubbard@...dia.com>,
        Jérôme Glisse <jglisse@...hat.com>,
        Jan Kara <jack@...e.cz>,
        Dan Williams <dan.j.williams@...el.com>,
        Linux MM <linux-mm@...ck.org>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Pawel Osciak <pawel@...iak.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Kyungmin Park <kyungmin.park@...sung.com>,
        Tomasz Figa <tfiga@...omium.org>,
        Inki Dae <inki.dae@...sung.com>,
        Joonyoung Shim <jy0922.shim@...sung.com>,
        Seung-Woo Kim <sw0312.kim@...sung.com>,
        linux-samsung-soc <linux-samsung-soc@...r.kernel.org>,
        "open list:DMA BUFFER SHARING FRAMEWORK" 
        <linux-media@...r.kernel.org>, Oded Gabbay <oded.gabbay@...il.com>
Subject: Re: [PATCH 2/2] mm/frame-vec: use FOLL_LONGTERM

On Mon, Oct 5, 2020 at 7:28 PM Jason Gunthorpe <jgg@...pe.ca> wrote:
>
> On Sun, Oct 04, 2020 at 06:09:29PM +0200, Daniel Vetter wrote:
> > On Sun, Oct 4, 2020 at 2:51 PM Jason Gunthorpe <jgg@...pe.ca> wrote:
> > >
> > > On Sat, Oct 03, 2020 at 11:40:22AM +0200, Daniel Vetter wrote:
> > >
> > > > > That leaves the only interesting places as vb2_dc_get_userptr() and
> > > > > vb2_vmalloc_get_userptr() which both completely fail to follow the
> > > > > REQUIRED behavior in the function's comment about checking PTEs. It
> > > > > just DMA maps them. Badly broken.
> > > > >
> > > > > Guessing this hackery is for some embedded P2P DMA transfer?
> > > >
> > > > Yeah, see also the follow_pfn trickery in
> > > > videobuf_dma_contig_user_get(), I think this is fully intentional and
> > > > userspace abi we can't break :-/
> > >
> > > We don't need to break uABI, it just needs to work properly in the
> > > kernel:
> > >
> > >   vma = find_vma_intersection()
> > >   dma_buf = dma_buf_get_from_vma(vma)
> > >   sg = dma_buf_p2p_dma_map(dma_buf)
> > >   [.. do dma ..]
> > >   dma_buf_unmap(sg)
> > >   dma_buf_put(dma_buf)
> > >
> > > It is as we discussed before, dma buf needs to be discoverable from a
> > > VMA, at least for users doing this kind of stuff.
> >
> > I'm not a big fan of magic behaviour like this, there's more to
> > dma-buf buffer sharing than just "how do I get at the backing
> > storage". Thus far we've done everything rather explicitly. Plus with
> > exynos and habanalabs converted there's only v4l left over, and that
> > has a proper dma-buf import path already.
>
> Well, any VA approach like this has to access some backing refcount
> via the VMA. Not really any way to avoid something like that
>
> > > A VM flag doesn't help - we need to introduce some kind of lifetime,
> > > and that has to be derived from the VMA. It needs data not just a flag
> >
> > I don't want to make it work, I just want to make it fail. Rough idea
> > I have in mind is to add a follow_pfn_longterm, for all callers which
> > aren't either synchronized through mmap_sem or an mmu_notifier.
>
> follow_pfn() doesn't work outside the pagetable locks or mmu notifier
> protection. Can't be fixed.
>
> We only have a few users:
>
> arch/s390/pci/pci_mmio.c:       ret = follow_pfn(vma, user_addr, pfn);
> drivers/media/v4l2-core/videobuf-dma-contig.c:          ret = follow_pfn(vma, user_address, &this_pfn);
> drivers/vfio/vfio_iommu_type1.c:        ret = follow_pfn(vma, vaddr, pfn);
> drivers/vfio/vfio_iommu_type1.c:                ret = follow_pfn(vma, vaddr, pfn);
> mm/frame_vector.c:                      err = follow_pfn(vma, start, &nums[ret]);
> virt/kvm/kvm_main.c:    r = follow_pfn(vma, addr, &pfn);
> virt/kvm/kvm_main.c:            r = follow_pfn(vma, addr, &pfn);
>
> VFIO is broken like media, but I saw patches fixing the vfio cases
> using the VMA and a vfio specific refcount.
>
> media & frame_vector we are talking about here.
>
> kvm is some similar hack added for P2P DMA, see commit
> add6a0cd1c5ba51b201e1361b05a5df817083618. It might be protected by notifiers..

Yeah my thinking is that kvm (and I think also vfio, also seems to
have mmu notifier nearby) are ok because of the mmu notiifer. Assuming
that one works correctly.

> s390 looks broken too, needs to hold the page table locks.

Hm yeah I guess that looks fairly reasonable to fix too.

> So, the answer really is that s390 and media need fixing, and this API
> should go away (or become kvm specific)

I'm still not clear how you want fo fix this, since your vma->dma_buf
idea is kinda a decade long plan and so just not going to happen:
- v4l used this mostly (afaik the lore at least) for buffer sharing
with v4l itself, and also a bit with fbdev. Neither even has any
dma-buf exporter code as-is.
- like I said, there's no central dma-buf instance, it was fairly
intentionally create as an all-to-all abstraction. Which means you
either have to roll out a vm_ops->gimme_the_dmabuf or, even more work,
refactor all the dma-buf exporters to go through the same things
- even where we have dma-buf, most mmaps of buffer objects aren't a
dma-buf. Those  are only set up when userspace explicitly asks for
one, so we'd also need to change the mmap code of all drivers involved
to make sure the dma-buf is always created when we do any kind of
mmap.

I don't see that as a realistic thing to ever happen, and meanwhile we
can't leave the gap open for a few years.

> > If this really breaks anyone's use-case we can add a tainting kernel
> > option which re-enables this (we've done something similar for
> > phys_addr_t based buffer sharing in fbdev, entirely unfixable since
> > the other driver has to just blindly trust that what userspace
> > passes around is legit). This here isn't unfixable, but if v4l
> > people want to keep it without a big "security hole here" sticker,
> > they should do the work, not me :-)
>
> This seems fairly reasonable..
>
> So after frame_vec is purged and we have the one caller in media, move
> all this stuff to media and taint the kernel if it goes down the
> follow_pfn path

Yeah I think moving frame_vec back to media sounds like a good idea,
it should stop new users like habanalbas/exynos from popping up at
least. It's follow_pfn that freaks me out more.
-Daniel

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

Powered by blists - more mailing lists