NAK/cmnt: [SRU Bionic 0/1] CVE-2021-38198

Thadeu Lima de Souza Cascardo cascardo at canonical.com
Thu Sep 23 17:04:11 UTC 2021


On Thu, Sep 23, 2021 at 06:29:14PM +0200, Kleber Souza wrote:
> On 22.09.21 20:47, Thadeu Lima de Souza Cascardo wrote:
> > [Impact]
> >   It was discovered that the KVM hypervisor implementation in the Linux
> >   kernel did not properly compute the access permissions for shadow pages in
> >   some situations. A local attacker could use this to cause a denial of
> >   service.
> > 
> > [Backport]
> > I picked up the backport that was done for 4.19.y, as there was a small
> > conflict in one of the hunks and a file rename.
> > 
> > [Test case]
> > kvm-unit-tests access test will check for this bug. It fails before the
> > fix and passes after the fix.
> > 
> > [Potential regression]
> > KVM guest page faults could be erronously handled, causing issues on
> > KVM guests.
> > 
> > Lai Jiangshan (1):
> >    KVM: X86: MMU: Use the correct inherited permissions to get shadow
> >      page
> > 
> >   Documentation/virtual/kvm/mmu.txt |  4 ++--
> >   arch/x86/kvm/paging_tmpl.h        | 14 +++++++++-----
> >   2 files changed, 11 insertions(+), 7 deletions(-)
> > 
> 
> Already applied to bionic:linux as part of LP#1944600 (Bionic update:
> upstream stable patchset 2021-09-22).
> 
> 
> Thanks,
> Kleber

Race condition detected! I found that this was already part of stable updates
for Focal, but it was not there for Bionic, which is explained by the patchset
date here.  :-)

I see the resulting code is the same, so I am fine with keeping the stable
update in this case.

Thanks.
Cascardo.



More information about the kernel-team mailing list