lkml.org 
[lkml]   [2022]   [Mar]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH v8 5/9] hisi_acc_vfio_pci: Restrict access to VF dev BAR2 migration region
Date
> From: Shameerali Kolothum Thodi
> <shameerali.kolothum.thodi@huawei.com>
> Sent: Tuesday, March 8, 2022 4:33 PM
>
> Hi Kevin,
>
> > -----Original Message-----
> > From: Tian, Kevin [mailto:kevin.tian@intel.com]
> > Sent: 08 March 2022 06:23
> > To: Shameerali Kolothum Thodi
> <shameerali.kolothum.thodi@huawei.com>;
> > kvm@vger.kernel.org; linux-kernel@vger.kernel.org;
> > linux-crypto@vger.kernel.org
> > Cc: linux-pci@vger.kernel.org; alex.williamson@redhat.com;
> jgg@nvidia.com;
> > cohuck@redhat.com; mgurtovoy@nvidia.com; yishaih@nvidia.com;
> Linuxarm
> > <linuxarm@huawei.com>; liulongfang <liulongfang@huawei.com>;
> Zengtao (B)
> > <prime.zeng@hisilicon.com>; Jonathan Cameron
> > <jonathan.cameron@huawei.com>; Wangzhou (B)
> <wangzhou1@hisilicon.com>
> > Subject: RE: [PATCH v8 5/9] hisi_acc_vfio_pci: Restrict access to VF dev
> BAR2
> > migration region
> >
> > Hi, Shameer,
> >
> > > From: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
> > > Sent: Friday, March 4, 2022 7:01 AM
> > >
> > > HiSilicon ACC VF device BAR2 region consists of both functional
> > > register space and migration control register space. From a
> > > security point of view, it's not advisable to export the migration
> > > control region to Guest.
> > >
> > > Hence, introduce a separate struct vfio_device_ops for migration
> > > support which will override the ioctl/read/write/mmap methods to
> > > hide the migration region and limit the access only to the
> > > functional register space.
> > >
> > > This will be used in subsequent patches when we add migration
> > > support to the driver.
> >
> > As a security concern the migration control region should be always
> > disabled regardless of whether migration support is added to the
> > driver for such device... It sounds like we should first fix this security
> > hole for acc device assignment and then add the migration support
> > atop (at least organize the series in this way).
>
> By exposing the migration BAR region, there is a possibility that a malicious
> Guest can prevent migration from happening by manipulating the migration
> BAR region. I don't think there are any other security concerns now especially
> since we only support the STOP_COPY state. And the approach has been
> that
> we only restrict this if migration support is enabled. I think I can change the
> above "security concern" description to "malicious Guest can prevent
> migration"
> to make it more clear.
>

In concept migrated device state may include both the state directly
touched by the guest driver and also the one that is configured by
the PF driver. Unless there is guarantee that the state managed via
the migration control interface only touches the former (which implies
the latter managed via the PF driver) this security concern will hold
even for normal device assignment.

If the acc device has such guarantee it's worth of a clarification here.

Thanks
Kevin

\
 
 \ /
  Last update: 2022-03-08 11:09    [W:0.248 / U:0.040 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site