lkml.org 
[lkml]   [2020]   [Mar]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH v3 0/7] vfio/pci: SR-IOV support
Date
> From: Alex Williamson <alex.williamson@redhat.com>
> Sent: Thursday, March 12, 2020 5:58 AM
>
> Only minor tweaks since v2, GET and SET on VFIO_DEVICE_FEATURE are
> enforced mutually exclusive except with the PROBE option as suggested
> by Connie, the modinfo text has been expanded for the opt-in to enable
> SR-IOV support in the vfio-pci driver per discussion with Kevin.
>
> I have not incorporated runtime warnings attempting to detect misuse
> of SR-IOV or imposed a session lifetime of a VF token, both of which
> were significant portions of the discussion of the v2 series. Both of
> these also seem to impose a usage model or make assumptions about VF
> resource usage or configuration requirements that don't seem necessary
> except for the sake of generating a warning or requiring an otherwise
> unnecessary and implicit token reinitialization. If there are new
> thoughts around these or other discussion points, please raise them.
>
> Series overview (same as provided with v1):
>
> The synopsis of this series is that we have an ongoing desire to drive
> PCIe SR-IOV PFs from userspace with VFIO. There's an immediate need
> for this with DPDK drivers and potentially interesting future use
> cases in virtualization. We've been reluctant to add this support
> previously due to the dependency and trust relationship between the
> VF device and PF driver. Minimally the PF driver can induce a denial
> of service to the VF, but depending on the specific implementation,
> the PF driver might also be responsible for moving data between VFs
> or have direct access to the state of the VF, including data or state
> otherwise private to the VF or VF driver.
>
> To help resolve these concerns, we introduce a VF token into the VFIO
> PCI ABI, which acts as a shared secret key between drivers. The
> userspace PF driver is required to set the VF token to a known value
> and userspace VF drivers are required to provide the token to access
> the VF device. If a PF driver is restarted with VF drivers in use, it
> must also provide the current token in order to prevent a rogue
> untrusted PF driver from replacing a known driver. The degree to
> which this new token is considered secret is left to the userspace
> drivers, the kernel intentionally provides no means to retrieve the
> current token.
>
> Note that the above token is only required for this new model where
> both the PF and VF devices are usable through vfio-pci. Existing
> models of VFIO drivers where the PF is used without SR-IOV enabled
> or the VF is bound to a userspace driver with an in-kernel, host PF
> driver are unaffected.
>
> The latter configuration above also highlights a new inverted scenario
> that is now possible, a userspace PF driver with in-kernel VF drivers.
> I believe this is a scenario that should be allowed, but should not be
> enabled by default. This series includes code to set a default
> driver_override for VFs sourced from a vfio-pci user owned PF, such
> that the VFs are also bound to vfio-pci. This model is compatible
> with tools like driverctl and allows the system administrator to
> decide if other bindings should be enabled. The VF token interface
> above exists only between vfio-pci PF and VF drivers, once a VF is
> bound to another driver, the administrator has effectively pronounced
> the device as trusted. The vfio-pci driver will note alternate
> binding in dmesg for logging and debugging purposes.
>
> Please review, comment, and test. The example QEMU implementation
> provided with the RFC is still current for this version. Thanks,
>
> Alex

The whole series looks good to me:
Reviewed-by: Kevin Tian <kevin.tian@intel.com>

and confirm one understanding here, since it is not discussed anywhere. For
VM live migration with assigned VF device, it is not necessary to migrate the
VF token itself and actually we don't allow userspace to retrieve it. Instead,
Qemu just follows whatever token requirement on the dest to open the new
VF: could be same or different token as/from src, or even no token if PF
driver runs in kernel on dest. I suppose either combination could work, correct?

Thanks
Kevin

>
> RFC:
> https://lore.kernel.org/lkml/158085337582.9445.17682266437583505502.stg
> it@gimli.home/
> v1:
> https://lore.kernel.org/lkml/158145472604.16827.15751375540102298130.st
> git@gimli.home/
> v2:
> https://lore.kernel.org/lkml/158213716959.17090.8399427017403507114.stg
> it@gimli.home/
>
> ---
>
> Alex Williamson (7):
> vfio: Include optional device match in vfio_device_ops callbacks
> vfio/pci: Implement match ops
> vfio/pci: Introduce VF token
> vfio: Introduce VFIO_DEVICE_FEATURE ioctl and first user
> vfio/pci: Add sriov_configure support
> vfio/pci: Remove dev_fmt definition
> vfio/pci: Cleanup .probe() exit paths
>
>
> drivers/vfio/pci/vfio_pci.c | 390
> +++++++++++++++++++++++++++++++++--
> drivers/vfio/pci/vfio_pci_private.h | 10 +
> drivers/vfio/vfio.c | 20 +-
> include/linux/vfio.h | 4
> include/uapi/linux/vfio.h | 37 +++
> 5 files changed, 433 insertions(+), 28 deletions(-)

\
 
 \ /
  Last update: 2020-03-19 07:33    [W:0.082 / U:2.664 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site