On Thu, 2024-06-06 at 11:27 -0600, Alex Williamson wrote:
> On Mon, 3 Jun 2024 17:50:13 +0200
> Christian Borntraeger <borntraeger@xxxxxxxxxxxxx> wrote:
>
> > Am 29.05.24 um 13:36 schrieb Niklas Schnelle:
> > > With the introduction of memory I/O (MIO) instructions enbaled in commit
> > > 71ba41c9b1d9 ("s390/pci: provide support for MIO instructions") s390
> > > gained support for direct user-space access to mapped PCI resources.
> > > Even without those however user-space can access mapped PCI resources
> > > via the s390 specific MMIO syscalls. There is thus nothing fundamentally
> > > preventing s390 from supporting VFIO_PCI_MMAP allowing user-space drivers
> > > to access PCI resources without going through the pread() interface.
> > > To actually enable VFIO_PCI_MMAP a few issues need fixing however.
> > >
> > > Firstly the s390 MMIO syscalls do not cause a page fault when
> > > follow_pte() fails due to the page not being present. This breaks
> > > vfio-pci's mmap() handling which lazily maps on first access.
> > >
> > > Secondly on s390 there is a virtual PCI device called ISM which has
> > > a few oddities. For one it claims to have a 256 TiB PCI BAR (not a typo)
> > > which leads to any attempt to mmap() it fail with the following message:
> > >
> > > vmap allocation for size 281474976714752 failed: use vmalloc=<size> to increase size
> > >
> > > Even if one tried to map this BAR only partially the mapping would not
> > > be usable on systems with MIO support enabled. So just block mapping
> > > BARs which don't fit between IOREMAP_START and IOREMAP_END.
> > >
> > > Note:
> > > For your convenience the code is also available in the tagged
> > > b4/vfio_pci_mmap branch on my git.kernel.org site below:
> > > https: //git.kernel.org/pub/scm/linux/kernel/git/niks/linux.git/
> >
> >
> > I guess its now mostly a question of who picks those patches? Alex?
> >
> > Any patch suitable for stable?
>
> Nothing here looks like stable material to me. 1/ only becomes an
> issue when mmap of MMIO is allowed on s390 (ie. 3/), 2/ is generic, but
> only really targets a device found on s390, and finally 3/ is
> essentially enabling a new feature.
I trust your judgement and was unsure too. I think for the
s390_mmio_read/write syscalls the only existing users out there are via
rdma-core, so unless Jason tells us that he thinks they could also be
affected by the lack of page fault handling I see no problem in going
upstream only.
>
> If we expect any conflicts with 1/ in the next merge window I can take
> a branch for it and apply 2/ and 3/ through the vfio tree, otherwise I
> can bring them all through the vfio tree if the s390 folks agree.
> Thanks,
>
> Alex
>
I also agree with this going via the vfio tree. I don't forsee
conflicts with 1.
Thanks,
Niklas
