render / rpms / libvirt

Forked from rpms/libvirt 9 months ago
Clone
b971b8
From 026160bc76bab26772c2a80bd50ae55119e3b60b Mon Sep 17 00:00:00 2001
b971b8
Message-Id: <026160bc76bab26772c2a80bd50ae55119e3b60b@dist-git>
b971b8
From: Viktor Mihajlovski <mihajlov@linux.ibm.com>
b971b8
Date: Wed, 24 Jun 2020 13:16:23 +0200
b971b8
Subject: [PATCH] docs: Describe protected virtualization guest setup
b971b8
MIME-Version: 1.0
b971b8
Content-Type: text/plain; charset=UTF-8
b971b8
Content-Transfer-Encoding: 8bit
b971b8
b971b8
Protected virtualization/IBM Secure Execution for Linux protects
b971b8
guest memory and state from the host.
b971b8
b971b8
Add some basic information about technology and a brief guide
b971b8
on setting up secure guests with libvirt.
b971b8
b971b8
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.ibm.com>
b971b8
Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
b971b8
Reviewed-by: Paulo de Rezende Pinatti <ppinatti@linux.ibm.com>
b971b8
Reviewed-by: Erik Skultety <eskultet@redhat.com>
b971b8
(cherry picked from commit f0d0cd6179709461b026f24569a688065e90d766)
b971b8
b971b8
https://bugzilla.redhat.com/show_bug.cgi?id=1848997
b971b8
https://bugzilla.redhat.com/show_bug.cgi?id=1850351
b971b8
b971b8
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
b971b8
Message-Id: <cf598d01d242bb56e64e14bfc32cece69341d949.1592996194.git.jdenemar@redhat.com>
b971b8
Reviewed-by: Ján Tomko <jtomko@redhat.com>
b971b8
---
b971b8
 docs/kbase.html.in                 |   3 +
b971b8
 docs/kbase/s390_protected_virt.rst | 189 +++++++++++++++++++++++++++++
b971b8
 2 files changed, 192 insertions(+)
b971b8
 create mode 100644 docs/kbase/s390_protected_virt.rst
b971b8
b971b8
diff --git a/docs/kbase.html.in b/docs/kbase.html.in
b971b8
index f2975960f6..05773db16d 100644
b971b8
--- a/docs/kbase.html.in
b971b8
+++ b/docs/kbase.html.in
b971b8
@@ -14,6 +14,9 @@
b971b8
         
Secure usage
b971b8
         
Secure usage of the libvirt APIs
b971b8
 
b971b8
+        
Protected virtualization on s390
b971b8
+        
Running secure s390 guests with IBM Secure Execution
b971b8
+
b971b8
         
Launch security
b971b8
         
Securely launching VMs with AMD SEV
b971b8
 
b971b8
diff --git a/docs/kbase/s390_protected_virt.rst b/docs/kbase/s390_protected_virt.rst
b971b8
new file mode 100644
b971b8
index 0000000000..f38d16d743
b971b8
--- /dev/null
b971b8
+++ b/docs/kbase/s390_protected_virt.rst
b971b8
@@ -0,0 +1,189 @@
b971b8
+================================
b971b8
+Protected Virtualization on s390
b971b8
+================================
b971b8
+
b971b8
+.. contents::
b971b8
+
b971b8
+Overview
b971b8
+========
b971b8
+
b971b8
+Protected virtualization, also known as IBM Secure Execution is a
b971b8
+hardware-based privacy protection technology for s390x (IBM Z).
b971b8
+It allows to execute virtual machines such that the host system
b971b8
+has no access to a VM's state and memory contents.
b971b8
+
b971b8
+Unlike other similar technologies, the memory of a running guest
b971b8
+is not encrypted but protected by hardware access controls, which
b971b8
+may only be manipulated by trusted system firmware, called
b971b8
+ultravisor.
b971b8
+
b971b8
+For the cases where the host needs access to guest memory (e.g. for
b971b8
+paging), it can request pages to be exported to it. The exported page
b971b8
+will be encrypted with a unique key for the running guest by the
b971b8
+ultravisor. The ultravisor also computes an integrity value for
b971b8
+the page, and stores it in a special table, together with the page
b971b8
+index and a counter. This way it can verify the integrity of
b971b8
+the page content upon re-import into the guest.
b971b8
+
b971b8
+In other cases it may be necessary for a guest to grant the host access
b971b8
+to dedicated memory regions (e.g. for I/O). The guest can request
b971b8
+that the ultravisor removes the memory protection from individual
b971b8
+pages, so that they can be shared with the host. Likewise, the
b971b8
+guest can undo the sharing.
b971b8
+
b971b8
+A secure guest will initially start in a regular non-protected VM.
b971b8
+The start-up is controlled by a small bootstrap program loaded
b971b8
+into memory together with encrypted operating system components and
b971b8
+a control structure (the PV header).
b971b8
+The operating system components (e.g. Linux kernel, initial RAM
b971b8
+file system, kernel parameters) are encrypted and integrity
b971b8
+protected. The component encryption keys and integrity values are
b971b8
+stored in the PV header.
b971b8
+The PV header is wrapped with a public key belonging to a specific
b971b8
+system (in fact it can be wrapped with multiple such keys). The
b971b8
+matching private key is only accessible by trusted hardware and
b971b8
+firmware in that specific system.
b971b8
+Consequently, such a secure guest boot image can only be run on the
b971b8
+systems it has been prepared for. Its contents can't be decrypted
b971b8
+without access to the private key and it can't be modified as
b971b8
+it is integrity protected.
b971b8
+
b971b8
+Host Requirements
b971b8
+=================
b971b8
+
b971b8
+IBM Secure Execution for Linux has some hardware and firmware
b971b8
+requirements. The system hardware must be an IBM z15 (or newer),
b971b8
+or an IBM LinuxONE III (or newer).
b971b8
+
b971b8
+It is also necessary that the IBM Secure Execution feature is
b971b8
+enabled for that system. With libvirt >= 6.5.0 you can run
b971b8
+``libvirt-host--validate`` or otherwise check for facility '158', e.g.
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   $ grep facilities /proc/cpuinfo | grep 158
b971b8
+
b971b8
+The kernel must include the protected virtualization support
b971b8
+which can be verified by checking for the presence of directory
b971b8
+``/sys/firmware/uv``. It will only be present when both the
b971b8
+hardware and the kernel support are available.
b971b8
+
b971b8
+Finally, the host operating system must donate some memory to
b971b8
+the ultravisor needed to store memory security information.
b971b8
+This is achieved by specifying the following kernel command
b971b8
+line parameter to the host boot configuration
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   prot_virt=1
b971b8
+
b971b8
+
b971b8
+Guest Requirements
b971b8
+==================
b971b8
+
b971b8
+Guest Boot
b971b8
+----------
b971b8
+
b971b8
+To start a guest in protected virtualization secure mode, the
b971b8
+boot image must have been prepared first with the program
b971b8
+``genprotimg`` using the correct public key for this host.
b971b8
+``genprotimg`` is part of the package ``s390-tools``, or
b971b8
+``s390-utils``, depending on the Linux distribution being used.
b971b8
+It can also be found at
b971b8
+`<https://github.com/ibm-s390-tools/s390-tools/tree/master/genprotimg>`_
b971b8
+
b971b8
+The guests have to be configured to use the host CPU model, which
b971b8
+must contain the ``unpack`` facility indicating ultravisor guest support.
b971b8
+
b971b8
+With the following command it's possible to check whether the host
b971b8
+CPU model satisfies the requirement
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   $ virsh domcapabilities | grep unpack
b971b8
+
b971b8
+which should return
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   <feature policy='require' name='unpack'/>
b971b8
+
b971b8
+Note that on hosts with libvirt < 6.5.0 if the check fails despite
b971b8
+the host system actually supporting protected virtualization guests,
b971b8
+this can be caused by a stale libvirt capabilities cache.
b971b8
+To recover, run the following commands
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   $ systemctl stop libvirtd
b971b8
+   $ rm /var/cache/libvirt/qemu/capabilities/*.xml
b971b8
+   $ systemctl start libvirtd
b971b8
+
b971b8
+
b971b8
+Guest I/O
b971b8
+---------
b971b8
+
b971b8
+Protected virtualization guests support I/O using virtio devices.
b971b8
+As the virtio data structures of secure guests are not accessible
b971b8
+by the host, it is necessary to use shared memory ('bounce buffers').
b971b8
+
b971b8
+To enable virtio devices to use shared buffers, it is necessary
b971b8
+to configure them with platform_iommu enabled. This can done by adding
b971b8
+``iommu='on'`` to the driver element of a virtio device definition in the
b971b8
+guest's XML, e.g.
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   <interface type='network'>
b971b8
+     <source network='default'/>
b971b8
+     <model type='virtio'/>
b971b8
+     <driver name='vhost' iommu='on'/>
b971b8
+   </interface>
b971b8
+
b971b8
+It is mandatory to define all virtio bus devices in this way to
b971b8
+prevent the host from attempting to access protected memory.
b971b8
+Ballooning will not work and is fenced by QEMU. It should be
b971b8
+disabled by specifying
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   <memballoon model='none'/>
b971b8
+
b971b8
+Finally, the guest Linux must be instructed to allocate I/O
b971b8
+buffers using memory shared between host and guest using SWIOTLB.
b971b8
+This is done by adding ``swiotlb=nnn`` to the guest's kernel command
b971b8
+line string, where ``nnn`` stands for the number of statically
b971b8
+allocated 2K entries. A commonly used value for swiotlb is 262144.
b971b8
+
b971b8
+Example guest definition
b971b8
+========================
b971b8
+
b971b8
+Minimal domain XML for a protected virtualization guest, essentially
b971b8
+it's mostly about the ``iommu`` property
b971b8
+
b971b8
+::
b971b8
+
b971b8
+   <domain type='kvm'>
b971b8
+     <name>protected</name>
b971b8
+     <memory unit='KiB'>2048000</memory>
b971b8
+     <currentMemory unit='KiB'>2048000</currentMemory>
b971b8
+     <vcpu>1</vcpu>
b971b8
+     <os>
b971b8
+       <type arch='s390x'>hvm</type>
b971b8
+     </os>
b971b8
+     <cpu mode='host-model'/>
b971b8
+     <devices>
b971b8
+       <disk type='file' device='disk'>
b971b8
+         <driver name='qemu' type='qcow2' cache='none' io='native' iommu='on'>
b971b8
+         <source file='/var/lib/libvirt/images/protected.qcow2'/>
b971b8
+         <target dev='vda' bus='virtio'/>
b971b8
+       </disk>
b971b8
+       <interface type='network'>
b971b8
+         <driver iommu='on'/>
b971b8
+         <source network='default'/>
b971b8
+         <model type='virtio'/>
b971b8
+       </interface>
b971b8
+       <console type='pty'/>
b971b8
+       <memballoon model='none'/>
b971b8
+     </devices>
b971b8
+   </domain>
b971b8
-- 
b971b8
2.27.0
b971b8