2020-03-19 14:19:20 +01:00
|
|
|
Protected Virtualization on s390x
|
|
|
|
=================================
|
|
|
|
|
|
|
|
The memory and most of the registers of Protected Virtual Machines
|
|
|
|
(PVMs) are encrypted or inaccessible to the hypervisor, effectively
|
|
|
|
prohibiting VM introspection when the VM is running. At rest, PVMs are
|
|
|
|
encrypted and can only be decrypted by the firmware, represented by an
|
|
|
|
entity called Ultravisor, of specific IBM Z machines.
|
|
|
|
|
|
|
|
|
|
|
|
Prerequisites
|
|
|
|
-------------
|
|
|
|
|
|
|
|
To run PVMs, a machine with the Protected Virtualization feature, as
|
|
|
|
indicated by the Ultravisor Call facility (stfle bit 158), is
|
|
|
|
required. The Ultravisor needs to be initialized at boot by setting
|
2021-07-26 16:23:34 +02:00
|
|
|
``prot_virt=1`` on the host's kernel command line.
|
2020-03-19 14:19:20 +01:00
|
|
|
|
|
|
|
Running PVMs requires using the KVM hypervisor.
|
|
|
|
|
2021-07-26 16:23:34 +02:00
|
|
|
If those requirements are met, the capability ``KVM_CAP_S390_PROTECTED``
|
2020-03-19 14:19:20 +01:00
|
|
|
will indicate that KVM can support PVMs on that LPAR.
|
|
|
|
|
|
|
|
|
s390: Recognize confidential-guest-support option
At least some s390 cpu models support "Protected Virtualization" (PV),
a mechanism to protect guests from eavesdropping by a compromised
hypervisor.
This is similar in function to other mechanisms like AMD's SEV and
POWER's PEF, which are controlled by the "confidential-guest-support"
machine option. s390 is a slightly special case, because we already
supported PV, simply by using a CPU model with the required feature
(S390_FEAT_UNPACK).
To integrate this with the option used by other platforms, we
implement the following compromise:
- When the confidential-guest-support option is set, s390 will
recognize it, verify that the CPU can support PV (failing if not)
and set virtio default options necessary for encrypted or protected
guests, as on other platforms. i.e. if confidential-guest-support
is set, we will either create a guest capable of entering PV mode,
or fail outright.
- If confidential-guest-support is not set, guests might still be
able to enter PV mode, if the CPU has the right model. This may be
a little surprising, but shouldn't actually be harmful.
To start a guest supporting Protected Virtualization using the new
option use the command line arguments:
-object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
2020-07-23 06:36:45 +02:00
|
|
|
Running a Protected Virtual Machine
|
|
|
|
-----------------------------------
|
2020-03-19 14:19:20 +01:00
|
|
|
|
s390: Recognize confidential-guest-support option
At least some s390 cpu models support "Protected Virtualization" (PV),
a mechanism to protect guests from eavesdropping by a compromised
hypervisor.
This is similar in function to other mechanisms like AMD's SEV and
POWER's PEF, which are controlled by the "confidential-guest-support"
machine option. s390 is a slightly special case, because we already
supported PV, simply by using a CPU model with the required feature
(S390_FEAT_UNPACK).
To integrate this with the option used by other platforms, we
implement the following compromise:
- When the confidential-guest-support option is set, s390 will
recognize it, verify that the CPU can support PV (failing if not)
and set virtio default options necessary for encrypted or protected
guests, as on other platforms. i.e. if confidential-guest-support
is set, we will either create a guest capable of entering PV mode,
or fail outright.
- If confidential-guest-support is not set, guests might still be
able to enter PV mode, if the CPU has the right model. This may be
a little surprising, but shouldn't actually be harmful.
To start a guest supporting Protected Virtualization using the new
option use the command line arguments:
-object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
2020-07-23 06:36:45 +02:00
|
|
|
To run a PVM you will need to select a CPU model which includes the
|
2021-07-26 16:23:34 +02:00
|
|
|
``Unpack facility`` (stfle bit 161 represented by the feature
|
|
|
|
``unpack``/``S390_FEAT_UNPACK``), and add these options to the command line::
|
s390: Recognize confidential-guest-support option
At least some s390 cpu models support "Protected Virtualization" (PV),
a mechanism to protect guests from eavesdropping by a compromised
hypervisor.
This is similar in function to other mechanisms like AMD's SEV and
POWER's PEF, which are controlled by the "confidential-guest-support"
machine option. s390 is a slightly special case, because we already
supported PV, simply by using a CPU model with the required feature
(S390_FEAT_UNPACK).
To integrate this with the option used by other platforms, we
implement the following compromise:
- When the confidential-guest-support option is set, s390 will
recognize it, verify that the CPU can support PV (failing if not)
and set virtio default options necessary for encrypted or protected
guests, as on other platforms. i.e. if confidential-guest-support
is set, we will either create a guest capable of entering PV mode,
or fail outright.
- If confidential-guest-support is not set, guests might still be
able to enter PV mode, if the CPU has the right model. This may be
a little surprising, but shouldn't actually be harmful.
To start a guest supporting Protected Virtualization using the new
option use the command line arguments:
-object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
2020-07-23 06:36:45 +02:00
|
|
|
|
|
|
|
-object s390-pv-guest,id=pv0 \
|
|
|
|
-machine confidential-guest-support=pv0
|
|
|
|
|
|
|
|
Adding these options will:
|
|
|
|
|
2021-07-26 16:23:34 +02:00
|
|
|
* Ensure the ``unpack`` facility is available
|
s390: Recognize confidential-guest-support option
At least some s390 cpu models support "Protected Virtualization" (PV),
a mechanism to protect guests from eavesdropping by a compromised
hypervisor.
This is similar in function to other mechanisms like AMD's SEV and
POWER's PEF, which are controlled by the "confidential-guest-support"
machine option. s390 is a slightly special case, because we already
supported PV, simply by using a CPU model with the required feature
(S390_FEAT_UNPACK).
To integrate this with the option used by other platforms, we
implement the following compromise:
- When the confidential-guest-support option is set, s390 will
recognize it, verify that the CPU can support PV (failing if not)
and set virtio default options necessary for encrypted or protected
guests, as on other platforms. i.e. if confidential-guest-support
is set, we will either create a guest capable of entering PV mode,
or fail outright.
- If confidential-guest-support is not set, guests might still be
able to enter PV mode, if the CPU has the right model. This may be
a little surprising, but shouldn't actually be harmful.
To start a guest supporting Protected Virtualization using the new
option use the command line arguments:
-object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
2020-07-23 06:36:45 +02:00
|
|
|
* Enable the IOMMU by default for all I/O devices
|
|
|
|
* Initialize the PV mechanism
|
2020-03-19 14:19:20 +01:00
|
|
|
|
|
|
|
Passthrough (vfio) devices are currently not supported.
|
|
|
|
|
|
|
|
Host huge page backings are not supported. However guests can use huge
|
|
|
|
pages as indicated by its facilities.
|
|
|
|
|
|
|
|
|
|
|
|
Boot Process
|
|
|
|
------------
|
|
|
|
|
|
|
|
A secure guest image can either be loaded from disk or supplied on the
|
|
|
|
QEMU command line. Booting from disk is done by the unmodified
|
|
|
|
s390-ccw BIOS. I.e., the bootmap is interpreted, multiple components
|
|
|
|
are read into memory and control is transferred to one of the
|
|
|
|
components (zipl stage3). Stage3 does some fixups and then transfers
|
|
|
|
control to some program residing in guest memory, which is normally
|
|
|
|
the OS kernel. The secure image has another component prepended
|
|
|
|
(stage3a) that uses the new diag308 subcodes 8 and 10 to trigger the
|
|
|
|
transition into secure mode.
|
|
|
|
|
|
|
|
Booting from the image supplied on the QEMU command line requires that
|
|
|
|
the file passed via -kernel has the same memory layout as would result
|
|
|
|
from the disk boot. This memory layout includes the encrypted
|
|
|
|
components (kernel, initrd, cmdline), the stage3a loader and
|
|
|
|
metadata. In case this boot method is used, the command line
|
|
|
|
options -initrd and -cmdline are ineffective. The preparation of a PVM
|
2021-07-26 16:23:34 +02:00
|
|
|
image is done via the ``genprotimg`` tool from the s390-tools
|
2020-03-19 14:19:20 +01:00
|
|
|
collection.
|