view docs/src/user.tex @ 0:7d21f7218375

Exact replica of unstable on 051908 + README-this
author Mukesh Rathor
date Mon May 19 15:34:57 2008 -0700 (2008-05-19)
children 5c0bf00e371d
line source
1 \documentclass[11pt,twoside,final,openright]{report}
2 \usepackage{a4,graphicx,html,parskip,setspace,times,xspace,url}
3 \setstretch{1.15}
5 \renewcommand{\ttdefault}{pcr}
7 \def\Xend{{Xend}\xspace}
8 \def\xend{{xend}\xspace}
10 \latexhtml{\renewcommand{\path}[1]{{\small {\tt #1}}}}{\renewcommand{\path}[1]{{\tt #1}}}
13 \begin{document}
16 \pagestyle{empty}
17 \begin{center}
18 \vspace*{\fill}
19 \includegraphics{figs/xenlogo.eps}
20 \vfill
21 \vfill
22 \vfill
23 \begin{tabular}{l}
24 {\Huge \bf Users' Manual} \\[4mm]
25 {\huge Xen v3.0} \\[80mm]
26 \end{tabular}
27 \end{center}
29 {\bf DISCLAIMER: This documentation is always under active development
30 and as such there may be mistakes and omissions --- watch out for
31 these and please report any you find to the developers' mailing list,
32 The latest version is always available
33 on-line. Contributions of material, suggestions and corrections are
34 welcome.}
36 \vfill
37 \clearpage
41 \pagestyle{empty}
43 \vspace*{\fill}
45 Xen is Copyright \copyright 2002-2005, University of Cambridge, UK, XenSource
46 Inc., IBM Corp., Hewlett-Packard Co., Intel Corp., AMD Inc., and others. All
47 rights reserved.
49 Xen is an open-source project. Most portions of Xen are licensed for copying
50 under the terms of the GNU General Public License, version 2. Other portions
51 are licensed under the terms of the GNU Lesser General Public License, the
52 Zope Public License 2.0, or under ``BSD-style'' licenses. Please refer to the
53 COPYING file for details.
55 Xen includes software by Christopher Clark. This software is covered by the
56 following licence:
58 \begin{quote}
59 Copyright (c) 2002, Christopher Clark. All rights reserved.
61 Redistribution and use in source and binary forms, with or without
62 modification, are permitted provided that the following conditions are met:
64 \begin{itemize}
65 \item Redistributions of source code must retain the above copyright notice,
66 this list of conditions and the following disclaimer.
68 \item Redistributions in binary form must reproduce the above copyright
69 notice, this list of conditions and the following disclaimer in the
70 documentation and/or other materials provided with the distribution.
72 \item Neither the name of the original author; nor the names of any
73 contributors may be used to endorse or promote products derived from this
74 software without specific prior written permission.
75 \end{itemize}
87 \end{quote}
89 \cleardoublepage
93 \pagestyle{plain}
94 \pagenumbering{roman}
95 { \parskip 0pt plus 1pt
96 \tableofcontents }
97 \cleardoublepage
101 \pagenumbering{arabic}
102 \raggedbottom
103 \widowpenalty=10000
104 \clubpenalty=10000
105 \parindent=0pt
106 \parskip=5pt
107 \renewcommand{\topfraction}{.8}
108 \renewcommand{\bottomfraction}{.8}
109 \renewcommand{\textfraction}{.2}
110 \renewcommand{\floatpagefraction}{.8}
111 \setstretch{1.1}
114 %% Chapter Introduction moved to introduction.tex
115 \chapter{Introduction}
118 Xen is an open-source \emph{para-virtualizing} virtual machine monitor
119 (VMM), or ``hypervisor'', for the x86 processor architecture. Xen can
120 securely execute multiple virtual machines on a single physical system
121 with close-to-native performance. Xen facilitates enterprise-grade
122 functionality, including:
124 \begin{itemize}
125 \item Virtual machines with performance close to native hardware.
126 \item Live migration of running virtual machines between physical hosts.
127 \item Up to 32\footnote{IA64 supports up to 64 virtual CPUs per guest virtual machine} virtual CPUs per guest virtual machine, with VCPU hotplug.
128 \item x86/32, x86/32 with PAE, x86/64, IA64 and Power platform support.
129 \item Intel and AMD Virtualization Technology for unmodified guest operating systems (including Microsoft Windows).
130 \item Excellent hardware support (supports almost all Linux device
131 drivers).
132 \end{itemize}
135 \section{Usage Scenarios}
137 Usage scenarios for Xen include:
139 \begin{description}
140 \item [Server Consolidation.] Move multiple servers onto a single
141 physical host with performance and fault isolation provided at the
142 virtual machine boundaries.
143 \item [Hardware Independence.] Allow legacy applications and operating
144 systems to exploit new hardware.
145 \item [Multiple OS configurations.] Run multiple operating systems
146 simultaneously, for development or testing purposes.
147 \item [Kernel Development.] Test and debug kernel modifications in a
148 sand-boxed virtual machine --- no need for a separate test machine.
149 \item [Cluster Computing.] Management at VM granularity provides more
150 flexibility than separately managing each physical host, but better
151 control and isolation than single-system image solutions,
152 particularly by using live migration for load balancing.
153 \item [Hardware support for custom OSes.] Allow development of new
154 OSes while benefiting from the wide-ranging hardware support of
155 existing OSes such as Linux.
156 \end{description}
159 \section{Operating System Support}
161 Para-virtualization permits very high performance virtualization, even
162 on architectures like x86 that are traditionally very hard to
163 virtualize.
165 This approach requires operating systems to be \emph{ported} to run on
166 Xen. Porting an OS to run on Xen is similar to supporting a new
167 hardware platform, however the process is simplified because the
168 para-virtual machine architecture is very similar to the underlying
169 native hardware. Even though operating system kernels must explicitly
170 support Xen, a key feature is that user space applications and
171 libraries \emph{do not} require modification.
173 With hardware CPU virtualization as provided by Intel VT and AMD
174 SVM technology, the ability to run an unmodified guest OS kernel
175 is available. No porting of the OS is required, although some
176 additional driver support is necessary within Xen itself. Unlike
177 traditional full virtualization hypervisors, which suffer a tremendous
178 performance overhead, the combination of Xen and VT or Xen and
179 Pacifica technology complement one another to offer superb performance
180 for para-virtualized guest operating systems and full support for
181 unmodified guests running natively on the processor.
183 Paravirtualized Xen support is available for increasingly many
184 operating systems: currently, mature Linux support is available and
185 included in the standard distribution. Other OS ports---including
186 NetBSD, FreeBSD and Solaris x86 v10---are nearing completion.
189 \section{Hardware Support}
191 Xen currently runs on the x86 architecture, requiring a ``P6'' or
192 newer processor (e.g.\ Pentium Pro, Celeron, Pentium~II, Pentium~III,
193 Pentium~IV, Xeon, AMD~Athlon, AMD~Duron). Multiprocessor machines are
194 supported, and there is support for HyperThreading (SMT). In
195 addition, ports to IA64 and Power architectures are supported.
197 The default 32-bit Xen supports for Intel's Physical Addressing Extensions (PAE), which enable x86/32 machines to address up to 64 GB of physical memory.
198 It also supports non-PAE 32-bit Xen up to 4GB of memory.
199 Xen also supports x86/64 platforms such as Intel EM64T and AMD Opteron
200 which can currently address up to 1TB of physical memory.
202 Xen offloads most of the hardware support issues to the guest OS
203 running in the \emph{Domain~0} management virtual machine. Xen itself
204 contains only the code required to detect and start secondary
205 processors, set up interrupt routing, and perform PCI bus
206 enumeration. Device drivers run within a privileged guest OS rather
207 than within Xen itself. This approach provides compatibility with the
208 majority of device hardware supported by Linux. The default XenLinux
209 build contains support for most server-class network and disk
210 hardware, but you can add support for other hardware by configuring
211 your XenLinux kernel in the normal way.
214 \section{Structure of a Xen-Based System}
216 A Xen system has multiple layers, the lowest and most privileged of
217 which is Xen itself.
219 Xen may host multiple \emph{guest} operating systems, each of which is
220 executed within a secure virtual machine. In Xen terminology, a
221 \emph{domain}. Domains are scheduled by Xen to make effective use of the
222 available physical CPUs. Each guest OS manages its own applications.
223 This management includes the responsibility of scheduling each
224 application within the time allotted to the VM by Xen.
226 The first domain, \emph{domain~0}, is created automatically when the
227 system boots and has special management privileges. Domain~0 builds
228 other domains and manages their virtual devices. It also performs
229 administrative tasks such as suspending, resuming and migrating other
230 virtual machines.
232 Within domain~0, a process called \emph{xend} runs to manage the system.
233 \Xend\ is responsible for managing virtual machines and providing access
234 to their consoles. Commands are issued to \xend\ over an HTTP interface,
235 via a command-line tool.
238 \section{History}
240 Xen was originally developed by the Systems Research Group at the
241 University of Cambridge Computer Laboratory as part of the XenoServers
242 project, funded by the UK-EPSRC\@.
244 XenoServers aim to provide a ``public infrastructure for global
245 distributed computing''. Xen plays a key part in that, allowing one to
246 efficiently partition a single machine to enable multiple independent
247 clients to run their operating systems and applications in an
248 environment. This environment provides protection, resource isolation
249 and accounting. The project web page contains further information along
250 with pointers to papers and technical reports:
251 \path{}
253 Xen has grown into a fully-fledged project in its own right, enabling us
254 to investigate interesting research issues regarding the best techniques
255 for virtualizing resources such as the CPU, memory, disk and network.
256 Project contributors now include XenSource, Intel, IBM, HP, AMD, Novell,
257 RedHat.
259 Xen was first described in a paper presented at SOSP in
260 2003\footnote{\tt
261}, and the first
262 public release (1.0) was made that October. Since then, Xen has
263 significantly matured and is now used in production scenarios on many
264 sites.
266 \section{What's New}
268 Xen 3.0.0 offers:
270 \begin{itemize}
271 \item Support for up to 32-way SMP guest operating systems
272 \item Intel (Physical Addressing Extensions) PAE to support 32-bit
273 servers with more than 4GB physical memory
274 \item x86/64 support (Intel EM64T, AMD Opteron)
275 \item Intel VT-x support to enable the running of unmodified guest
276 operating systems (Windows XP/2003, Legacy Linux)
277 \item Enhanced control tools
278 \item Improved ACPI support
279 \item AGP/DRM graphics
280 \end{itemize}
283 Xen 3.0 features greatly enhanced hardware support, configuration
284 flexibility, usability and a larger complement of supported operating
285 systems. This latest release takes Xen a step closer to being the
286 definitive open source solution for virtualization.
290 \part{Installation}
292 %% Chapter Basic Installation
293 \chapter{Basic Installation}
295 The Xen distribution includes three main components: Xen itself, ports
296 of Linux and NetBSD to run on Xen, and the userspace tools required to
297 manage a Xen-based system. This chapter describes how to install the
298 Xen~3.0 distribution from source. Alternatively, there may be pre-built
299 packages available as part of your operating system distribution.
302 \section{Prerequisites}
303 \label{sec:prerequisites}
305 The following is a full list of prerequisites. Items marked `$\dag$' are
306 required by the \xend\ control tools, and hence required if you want to
307 run more than one virtual machine; items marked `$*$' are only required
308 if you wish to build from source.
309 \begin{itemize}
310 \item A working Linux distribution using the GRUB bootloader and running
311 on a P6-class or newer CPU\@.
312 \item [$\dag$] The \path{iproute2} package.
313 \item [$\dag$] The Linux bridge-utils\footnote{Available from {\tt
314}} (e.g., \path{/sbin/brctl})
315 \item [$\dag$] The Linux hotplug system\footnote{Available from {\tt
316}} (e.g.,
317 \path{/sbin/hotplug} and related scripts). On newer distributions,
318 this is included alongside the Linux udev system\footnote{See {\tt
320 \item [$*$] Build tools (gcc v3.2.x or v3.3.x, binutils, GNU make).
321 \item [$*$] Development installation of zlib (e.g.,\ zlib-dev).
322 \item [$*$] Development installation of Python v2.2 or later (e.g.,\
323 python-dev).
324 \item [$*$] \LaTeX\ and transfig are required to build the
325 documentation.
326 \end{itemize}
328 Once you have satisfied these prerequisites, you can now install either
329 a binary or source distribution of Xen.
331 \section{Installing from Binary Tarball}
333 Pre-built tarballs are available for download from the XenSource downloads
334 page:
335 \begin{quote} {\tt}
336 \end{quote}
338 Once you've downloaded the tarball, simply unpack and install:
339 \begin{verbatim}
340 # tar zxvf xen-3.0-install.tgz
341 # cd xen-3.0-install
342 # sh ./
343 \end{verbatim}
345 Once you've installed the binaries you need to configure your system as
346 described in Section~\ref{s:configure}.
348 \section{Installing from RPMs}
349 Pre-built RPMs are available for download from the XenSource downloads
350 page:
351 \begin{quote} {\tt}
352 \end{quote}
354 Once you've downloaded the RPMs, you typically install them via the
355 RPM commands:
357 \verb|# rpm -iv rpmname|
359 See the instructions and the Release Notes for each RPM set referenced at:
360 \begin{quote}
361 {\tt}.
362 \end{quote}
364 \section{Installing from Source}
366 This section describes how to obtain, build and install Xen from source.
368 \subsection{Obtaining the Source}
370 The Xen source tree is available as either a compressed source tarball
371 or as a clone of our master Mercurial repository.
373 \begin{description}
374 \item[Obtaining the Source Tarball]\mbox{} \\
375 Stable versions and daily snapshots of the Xen source tree are
376 available from the Xen download page:
377 \begin{quote} {\tt \tt}
378 \end{quote}
379 \item[Obtaining the source via Mercurial]\mbox{} \\
380 The source tree may also be obtained via the public Mercurial
381 repository at:
382 \begin{quote}{\tt}
383 \end{quote} See the instructions and the Getting Started Guide
384 referenced at:
385 \begin{quote}
386 {\tt}
387 \end{quote}
388 \end{description}
390 % \section{The distribution}
391 %
392 % The Xen source code repository is structured as follows:
393 %
394 % \begin{description}
395 % \item[\path{tools/}] Xen node controller daemon (Xend), command line
396 % tools, control libraries
397 % \item[\path{xen/}] The Xen VMM.
398 % \item[\path{buildconfigs/}] Build configuration files
399 % \item[\path{linux-*-xen-sparse/}] Xen support for Linux.
400 % \item[\path{patches/}] Experimental patches for Linux.
401 % \item[\path{docs/}] Various documentation files for users and
402 % developers.
403 % \item[\path{extras/}] Bonus extras.
404 % \end{description}
406 \subsection{Building from Source}
408 The top-level Xen Makefile includes a target ``world'' that will do the
409 following:
411 \begin{itemize}
412 \item Build Xen.
413 \item Build the control tools, including \xend.
414 \item Download (if necessary) and unpack the Linux 2.6 source code, and
415 patch it for use with Xen.
416 \item Build a Linux kernel to use in domain~0 and a smaller unprivileged
417 kernel, which can be used for unprivileged virtual machines.
418 \end{itemize}
420 After the build has completed you should have a top-level directory
421 called \path{dist/} in which all resulting targets will be placed. Of
422 particular interest are the two XenLinux kernel images, one with a
423 ``-xen0'' extension which contains hardware device drivers and drivers
424 for Xen's virtual devices, and one with a ``-xenU'' extension that
425 just contains the virtual ones. These are found in
426 \path{dist/install/boot/} along with the image for Xen itself and the
427 configuration files used during the build.
429 %The NetBSD port can be built using:
430 %\begin{quote}
431 %\begin{verbatim}
432 %# make netbsd20
433 %\end{verbatim}\end{quote}
434 %NetBSD port is built using a snapshot of the netbsd-2-0 cvs branch.
435 %The snapshot is downloaded as part of the build process if it is not
436 %yet present in the \path{NETBSD\_SRC\_PATH} search path. The build
437 %process also downloads a toolchain which includes all of the tools
438 %necessary to build the NetBSD kernel under Linux.
440 To customize the set of kernels built you need to edit the top-level
441 Makefile. Look for the line:
442 \begin{quote}
443 \begin{verbatim}
444 KERNELS ?= linux-2.6-xen0 linux-2.6-xenU
445 \end{verbatim}
446 \end{quote}
448 You can edit this line to include any set of operating system kernels
449 which have configurations in the top-level \path{buildconfigs/}
450 directory.
452 %% Inspect the Makefile if you want to see what goes on during a
453 %% build. Building Xen and the tools is straightforward, but XenLinux
454 %% is more complicated. The makefile needs a `pristine' Linux kernel
455 %% tree to which it will then add the Xen architecture files. You can
456 %% tell the makefile the location of the appropriate Linux compressed
457 %% tar file by
458 %% setting the LINUX\_SRC environment variable, e.g. \\
459 %% \verb!# LINUX_SRC=/tmp/linux-2.6.11.tar.bz2 make world! \\ or by
460 %% placing the tar file somewhere in the search path of {\tt
461 %% LINUX\_SRC\_PATH} which defaults to `{\tt .:..}'. If the
462 %% makefile can't find a suitable kernel tar file it attempts to
463 %% download it from (this won't work if you're behind a
464 %% firewall).
466 %% After untaring the pristine kernel tree, the makefile uses the {\tt
467 %% mkbuildtree} script to add the Xen patches to the kernel.
469 %% \framebox{\parbox{5in}{
470 %% {\bf Distro specific:} \\
471 %% {\it Gentoo} --- if not using udev (most installations,
472 %% currently), you'll need to enable devfs and devfs mount at boot
473 %% time in the xen0 config. }}
475 \subsection{Custom Kernels}
477 % If you have an SMP machine you may wish to give the {\tt '-j4'}
478 % argument to make to get a parallel build.
480 If you wish to build a customized XenLinux kernel (e.g.\ to support
481 additional devices or enable distribution-required features), you can
482 use the standard Linux configuration mechanisms, specifying that the
483 architecture being built for is \path{xen}, e.g:
484 \begin{quote}
485 \begin{verbatim}
486 # cd linux-2.6.12-xen0
487 # make ARCH=xen xconfig
488 # cd ..
489 # make
490 \end{verbatim}
491 \end{quote}
493 You can also copy an existing Linux configuration (\path{.config}) into
494 e.g.\ \path{linux-2.6.12-xen0} and execute:
495 \begin{quote}
496 \begin{verbatim}
497 # make ARCH=xen oldconfig
498 \end{verbatim}
499 \end{quote}
501 You may be prompted with some Xen-specific options. We advise accepting
502 the defaults for these options.
504 Note that the only difference between the two types of Linux kernels
505 that are built is the configuration file used for each. The ``U''
506 suffixed (unprivileged) versions don't contain any of the physical
507 hardware device drivers, leading to a 30\% reduction in size; hence you
508 may prefer these for your non-privileged domains. The ``0'' suffixed
509 privileged versions can be used to boot the system, as well as in driver
510 domains and unprivileged domains.
512 \subsection{Installing Generated Binaries}
514 The files produced by the build process are stored under the
515 \path{dist/install/} directory. To install them in their default
516 locations, do:
517 \begin{quote}
518 \begin{verbatim}
519 # make install
520 \end{verbatim}
521 \end{quote}
523 Alternatively, users with special installation requirements may wish to
524 install them manually by copying the files to their appropriate
525 destinations.
527 %% Files in \path{install/boot/} include:
528 %% \begin{itemize}
529 %% \item \path{install/boot/xen-3.0.gz} Link to the Xen 'kernel'
530 %% \item \path{install/boot/vmlinuz-2.6-xen0} Link to domain 0
531 %% XenLinux kernel
532 %% \item \path{install/boot/vmlinuz-2.6-xenU} Link to unprivileged
533 %% XenLinux kernel
534 %% \end{itemize}
536 The \path{dist/install/boot} directory will also contain the config
537 files used for building the XenLinux kernels, and also versions of Xen
538 and XenLinux kernels that contain debug symbols such as
539 (\path{xen-syms-3.0.0} and \path{vmlinux-syms-}) which are
540 essential for interpreting crash dumps. Retain these files as the
541 developers may wish to see them if you post on the mailing list.
544 \section{Configuration}
545 \label{s:configure}
547 Once you have built and installed the Xen distribution, it is simple to
548 prepare the machine for booting and running Xen.
550 \subsection{GRUB Configuration}
552 An entry should be added to \path{grub.conf} (often found under
553 \path{/boot/} or \path{/boot/grub/}) to allow Xen / XenLinux to boot.
554 This file is sometimes called \path{menu.lst}, depending on your
555 distribution. The entry should look something like the following:
557 %% KMSelf Thu Dec 1 19:06:13 PST 2005 262144 is useful for RHEL/RH and
558 %% related Dom0s.
559 {\small
560 \begin{verbatim}
561 title Xen 3.0 / XenLinux 2.6
562 kernel /boot/xen-3.0.gz dom0_mem=262144
563 module /boot/vmlinuz-2.6-xen0 root=/dev/sda4 ro console=tty0
564 \end{verbatim}
565 }
567 The kernel line tells GRUB where to find Xen itself and what boot
568 parameters should be passed to it (in this case, setting the domain~0
569 memory allocation in kilobytes and the settings for the serial port).
570 For more details on the various Xen boot parameters see
571 Section~\ref{s:xboot}.
573 The module line of the configuration describes the location of the
574 XenLinux kernel that Xen should start and the parameters that should be
575 passed to it. These are standard Linux parameters, identifying the root
576 device and specifying it be initially mounted read only and instructing
577 that console output be sent to the screen. Some distributions such as
578 SuSE do not require the \path{ro} parameter.
580 %% \framebox{\parbox{5in}{
581 %% {\bf Distro specific:} \\
582 %% {\it SuSE} --- Omit the {\tt ro} option from the XenLinux
583 %% kernel command line, since the partition won't be remounted rw
584 %% during boot. }}
586 To use an initrd, add another \path{module} line to the configuration,
587 like: {\small
588 \begin{verbatim}
589 module /boot/my_initrd.gz
590 \end{verbatim}
591 }
593 %% KMSelf Thu Dec 1 19:05:30 PST 2005 Other configs as an appendix?
595 When installing a new kernel, it is recommended that you do not delete
596 existing menu options from \path{menu.lst}, as you may wish to boot your
597 old Linux kernel in future, particularly if you have problems.
599 \subsection{Serial Console (optional)}
601 Serial console access allows you to manage, monitor, and interact with
602 your system over a serial console. This can allow access from another
603 nearby system via a null-modem (``LapLink'') cable or remotely via a serial
604 concentrator.
606 You system's BIOS, bootloader (GRUB), Xen, Linux, and login access must
607 each be individually configured for serial console access. It is
608 \emph{not} strictly necessary to have each component fully functional,
609 but it can be quite useful.
611 For general information on serial console configuration under Linux,
612 refer to the ``Remote Serial Console HOWTO'' at The Linux Documentation
613 Project: \url{}
615 \subsubsection{Serial Console BIOS configuration}
617 Enabling system serial console output neither enables nor disables
618 serial capabilities in GRUB, Xen, or Linux, but may make remote
619 management of your system more convenient by displaying POST and other
620 boot messages over serial port and allowing remote BIOS configuration.
622 Refer to your hardware vendor's documentation for capabilities and
623 procedures to enable BIOS serial redirection.
626 \subsubsection{Serial Console GRUB configuration}
628 Enabling GRUB serial console output neither enables nor disables Xen or
629 Linux serial capabilities, but may made remote management of your system
630 more convenient by displaying GRUB prompts, menus, and actions over
631 serial port and allowing remote GRUB management.
633 Adding the following two lines to your GRUB configuration file,
634 typically either \path{/boot/grub/menu.lst} or \path{/boot/grub/grub.conf}
635 depending on your distro, will enable GRUB serial output.
637 \begin{quote}
638 {\small \begin{verbatim}
639 serial --unit=0 --speed=115200 --word=8 --parity=no --stop=1
640 terminal --timeout=10 serial console
641 \end{verbatim}}
642 \end{quote}
644 Note that when both the serial port and the local monitor and keyboard
645 are enabled, the text ``\emph{Press any key to continue}'' will appear
646 at both. Pressing a key on one device will cause GRUB to display to
647 that device. The other device will see no output. If no key is
648 pressed before the timeout period expires, the system will boot to the
649 default GRUB boot entry.
651 Please refer to the GRUB documentation for further information.
654 \subsubsection{Serial Console Xen configuration}
656 Enabling Xen serial console output neither enables nor disables Linux
657 kernel output or logging in to Linux over serial port. It does however
658 allow you to monitor and log the Xen boot process via serial console and
659 can be very useful in debugging.
661 %% kernel /boot/xen-2.0.gz dom0_mem=131072 console=com1,vga com1=115200,8n1
662 %% module /boot/vmlinuz-2.6-xen0 root=/dev/sda4 ro
664 In order to configure Xen serial console output, it is necessary to
665 add a boot option to your GRUB config; e.g.\ replace the previous
666 example kernel line with:
667 \begin{quote} {\small \begin{verbatim}
668 kernel /boot/xen.gz dom0_mem=131072 com1=115200,8n1 console=com1,vga
669 \end{verbatim}}
670 \end{quote}
672 This configures Xen to output on COM1 at 115,200 baud, 8 data bits, no
673 parity and 1 stop bit. Modify these parameters for your environment.
674 See Section~\ref{s:xboot} for an explanation of all boot parameters.
676 One can also configure XenLinux to share the serial console; to achieve
677 this append ``\path{console=ttyS0}'' to your module line.
680 \subsubsection{Serial Console Linux configuration}
682 Enabling Linux serial console output at boot neither enables nor
683 disables logging in to Linux over serial port. It does however allow
684 you to monitor and log the Linux boot process via serial console and can be
685 very useful in debugging.
687 To enable Linux output at boot time, add the parameter
688 \path{console=ttyS0} (or ttyS1, ttyS2, etc.) to your kernel GRUB line.
689 Under Xen, this might be:
690 \begin{quote}
691 {\footnotesize \begin{verbatim}
692 module /vmlinuz-2.6-xen0 ro root=/dev/VolGroup00/LogVol00 \
693 console=ttyS0, 115200
694 \end{verbatim}}
695 \end{quote}
696 to enable output over ttyS0 at 115200 baud.
700 \subsubsection{Serial Console Login configuration}
702 Logging in to Linux via serial console, under Xen or otherwise, requires
703 specifying a login prompt be started on the serial port. To permit root
704 logins over serial console, the serial port must be added to
705 \path{/etc/securetty}.
707 \newpage
708 To automatically start a login prompt over the serial port,
709 add the line: \begin{quote} {\small {\tt c:2345:respawn:/sbin/mingetty
710 ttyS0}} \end{quote} to \path{/etc/inittab}. Run \path{init q} to force
711 a reload of your inttab and start getty.
713 To enable root logins, add \path{ttyS0} to \path{/etc/securetty} if not
714 already present.
716 Your distribution may use an alternate getty; options include getty,
717 mgetty and agetty. Consult your distribution's documentation
718 for further information.
721 \subsection{TLS Libraries}
723 Users of the XenLinux 2.6 kernel should disable Thread Local Storage
724 (TLS) (e.g.\ by doing a \path{mv /lib/tls /lib/tls.disabled}) before
725 attempting to boot a XenLinux kernel\footnote{If you boot without first
726 disabling TLS, you will get a warning message during the boot process.
727 In this case, simply perform the rename after the machine is up and
728 then run \path{/sbin/ldconfig} to make it take effect.}. You can
729 always reenable TLS by restoring the directory to its original location
730 (i.e.\ \path{mv /lib/tls.disabled /lib/tls}).
732 The reason for this is that the current TLS implementation uses
733 segmentation in a way that is not permissible under Xen. If TLS is not
734 disabled, an emulation mode is used within Xen which reduces performance
735 substantially. To ensure full performance you should install a
736 `Xen-friendly' (nosegneg) version of the library.
739 \section{Booting Xen}
741 It should now be possible to restart the system and use Xen. Reboot and
742 choose the new Xen option when the Grub screen appears.
744 What follows should look much like a conventional Linux boot. The first
745 portion of the output comes from Xen itself, supplying low level
746 information about itself and the underlying hardware. The last portion
747 of the output comes from XenLinux.
749 You may see some error messages during the XenLinux boot. These are not
750 necessarily anything to worry about---they may result from kernel
751 configuration differences between your XenLinux kernel and the one you
752 usually use.
754 When the boot completes, you should be able to log into your system as
755 usual. If you are unable to log in, you should still be able to reboot
756 with your normal Linux kernel by selecting it at the GRUB prompt.
759 % Booting Xen
760 \chapter{Booting a Xen System}
762 Booting the system into Xen will bring you up into the privileged
763 management domain, Domain0. At that point you are ready to create
764 guest domains and ``boot'' them using the \texttt{xm create} command.
766 \section{Booting Domain0}
768 After installation and configuration is complete, reboot the system
769 and and choose the new Xen option when the Grub screen appears.
771 What follows should look much like a conventional Linux boot. The
772 first portion of the output comes from Xen itself, supplying low level
773 information about itself and the underlying hardware. The last
774 portion of the output comes from XenLinux.
776 %% KMSelf Wed Nov 30 18:09:37 PST 2005: We should specify what these are.
778 When the boot completes, you should be able to log into your system as
779 usual. If you are unable to log in, you should still be able to
780 reboot with your normal Linux kernel by selecting it at the GRUB prompt.
782 The first step in creating a new domain is to prepare a root
783 filesystem for it to boot. Typically, this might be stored in a normal
784 partition, an LVM or other volume manager partition, a disk file or on
785 an NFS server. A simple way to do this is simply to boot from your
786 standard OS install CD and install the distribution into another
787 partition on your hard drive.
789 To start the \xend\ control daemon, type
790 \begin{quote}
791 \verb!# xend start!
792 \end{quote}
794 If you wish the daemon to start automatically, see the instructions in
795 Section~\ref{s:xend}. Once the daemon is running, you can use the
796 \path{xm} tool to monitor and maintain the domains running on your
797 system. This chapter provides only a brief tutorial. We provide full
798 details of the \path{xm} tool in the next chapter.
800 % \section{From the web interface}
801 %
802 % Boot the Xen machine and start Xensv (see Chapter~\ref{cha:xensv}
803 % for more details) using the command: \\
804 % \verb_# xensv start_ \\
805 % This will also start Xend (see Chapter~\ref{cha:xend} for more
806 % information).
807 %
808 % The domain management interface will then be available at {\tt
809 % http://your\_machine:8080/}. This provides a user friendly wizard
810 % for starting domains and functions for managing running domains.
811 %
812 % \section{From the command line}
813 \section{Booting Guest Domains}
815 \subsection{Creating a Domain Configuration File}
817 Before you can start an additional domain, you must create a
818 configuration file. We provide two example files which you can use as
819 a starting point:
820 \begin{itemize}
821 \item \path{/etc/xen/xmexample1} is a simple template configuration
822 file for describing a single VM\@.
823 \item \path{/etc/xen/xmexample2} file is a template description that
824 is intended to be reused for multiple virtual machines. Setting the
825 value of the \path{vmid} variable on the \path{xm} command line
826 fills in parts of this template.
827 \end{itemize}
829 There are also a number of other examples which you may find useful.
830 Copy one of these files and edit it as appropriate. Typical values
831 you may wish to edit include:
833 \begin{quote}
834 \begin{description}
835 \item[kernel] Set this to the path of the kernel you compiled for use
836 with Xen (e.g.\ \path{kernel = ``/boot/vmlinuz-2.6-xenU''})
837 \item[memory] Set this to the size of the domain's memory in megabytes
838 (e.g.\ \path{memory = 64})
839 \item[disk] Set the first entry in this list to calculate the offset
840 of the domain's root partition, based on the domain ID\@. Set the
841 second to the location of \path{/usr} if you are sharing it between
842 domains (e.g.\ \path{disk = ['phy:your\_hard\_drive\%d,sda1,w' \%
843 (base\_partition\_number + vmid),
844 'phy:your\_usr\_partition,sda6,r' ]}
845 \item[dhcp] Uncomment the dhcp variable, so that the domain will
846 receive its IP address from a DHCP server (e.g.\ \path{dhcp=``dhcp''})
847 \end{description}
848 \end{quote}
850 You may also want to edit the {\bf vif} variable in order to choose
851 the MAC address of the virtual ethernet interface yourself. For
852 example:
854 \begin{quote}
855 \verb_vif = ['mac=00:16:3E:F6:BB:B3']_
856 \end{quote}
857 If you do not set this variable, \xend\ will automatically generate a
858 random MAC address from the range 00:16:3E:xx:xx:xx, assigned by IEEE to
859 XenSource as an OUI (organizationally unique identifier). XenSource
860 Inc. gives permission for anyone to use addresses randomly allocated
861 from this range for use by their Xen domains.
863 For a list of IEEE OUI assignments, see
864 \url{}
867 \subsection{Booting the Guest Domain}
869 The \path{xm} tool provides a variety of commands for managing
870 domains. Use the \path{create} command to start new domains. Assuming
871 you've created a configuration file \path{myvmconf} based around
872 \path{/etc/xen/xmexample2}, to start a domain with virtual machine
873 ID~1 you should type:
875 \begin{quote}
876 \begin{verbatim}
877 # xm create -c myvmconf vmid=1
878 \end{verbatim}
879 \end{quote}
881 The \path{-c} switch causes \path{xm} to turn into the domain's
882 console after creation. The \path{vmid=1} sets the \path{vmid}
883 variable used in the \path{myvmconf} file.
885 You should see the console boot messages from the new domain appearing
886 in the terminal in which you typed the command, culminating in a login
887 prompt.
890 \section{Starting / Stopping Domains Automatically}
892 It is possible to have certain domains start automatically at boot
893 time and to have dom0 wait for all running domains to shutdown before
894 it shuts down the system.
896 To specify a domain is to start at boot-time, place its configuration
897 file (or a link to it) under \path{/etc/xen/auto/}.
899 A Sys-V style init script for Red Hat and LSB-compliant systems is
900 provided and will be automatically copied to \path{/etc/init.d/}
901 during install. You can then enable it in the appropriate way for
902 your distribution.
904 For instance, on Red Hat:
906 \begin{quote}
907 \verb_# chkconfig --add xendomains_
908 \end{quote}
910 By default, this will start the boot-time domains in runlevels 3, 4
911 and 5.
913 You can also use the \path{service} command to run this script
914 manually, e.g:
916 \begin{quote}
917 \verb_# service xendomains start_
919 Starts all the domains with config files under /etc/xen/auto/.
920 \end{quote}
922 \begin{quote}
923 \verb_# service xendomains stop_
925 Shuts down all running Xen domains.
926 \end{quote}
930 \part{Configuration and Management}
932 %% Chapter Domain Management Tools and Daemons
933 \chapter{Domain Management Tools}
935 This chapter summarizes the management software and tools available.
938 \section{\Xend\ }
939 \label{s:xend}
942 The \Xend\ node control daemon performs system management functions
943 related to virtual machines. It forms a central point of control of
944 virtualized resources, and must be running in order to start and manage
945 virtual machines. \Xend\ must be run as root because it needs access to
946 privileged system management functions.
948 An initialization script named \texttt{/etc/init.d/xend} is provided to
949 start \Xend\ at boot time. Use the tool appropriate (i.e. chkconfig) for
950 your Linux distribution to specify the runlevels at which this script
951 should be executed, or manually create symbolic links in the correct
952 runlevel directories.
954 \Xend\ can be started on the command line as well, and supports the
955 following set of parameters:
957 \begin{tabular}{ll}
958 \verb!# xend start! & start \xend, if not already running \\
959 \verb!# xend stop! & stop \xend\ if already running \\
960 \verb!# xend restart! & restart \xend\ if running, otherwise start it \\
961 % \verb!# xend trace_start! & start \xend, with very detailed debug logging \\
962 \verb!# xend status! & indicates \xend\ status by its return code
963 \end{tabular}
965 A SysV init script called {\tt xend} is provided to start \xend\ at
966 boot time. {\tt make install} installs this script in
967 \path{/etc/init.d}. To enable it, you have to make symbolic links in
968 the appropriate runlevel directories or use the {\tt chkconfig} tool,
969 where available. Once \xend\ is running, administration can be done
970 using the \texttt{xm} tool.
972 \subsection{Logging}
974 As \xend\ runs, events will be logged to \path{/var/log/xen/xend.log} and
975 (less frequently) to \path{/var/log/xen/xend-debug.log}. These, along with
976 the standard syslog files, are useful when troubleshooting problems.
978 \subsection{Configuring \Xend\ }
980 \Xend\ is written in Python. At startup, it reads its configuration
981 information from the file \path{/etc/xen/xend-config.sxp}. The Xen
982 installation places an example \texttt{xend-config.sxp} file in the
983 \texttt{/etc/xen} subdirectory which should work for most installations.
985 See the example configuration file \texttt{xend-debug.sxp} and the
986 section 5 man page \texttt{xend-config.sxp} for a full list of
987 parameters and more detailed information. Some of the most important
988 parameters are discussed below.
990 An HTTP interface and a Unix domain socket API are available to
991 communicate with \Xend. This allows remote users to pass commands to the
992 daemon. By default, \Xend does not start an HTTP server. It does start a
993 Unix domain socket management server, as the low level utility
994 \texttt{xm} requires it. For support of cross-machine migration, \Xend\
995 can start a relocation server. This support is not enabled by default
996 for security reasons.
998 Note: the example \texttt{xend} configuration file modifies the defaults and
999 starts up \Xend\ as an HTTP server as well as a relocation server.
1001 From the file:
1003 \begin{verbatim}
1004 #(xend-http-server no)
1005 (xend-http-server yes)
1006 #(xend-unix-server yes)
1007 #(xend-relocation-server no)
1008 (xend-relocation-server yes)
1009 \end{verbatim}
1011 Comment or uncomment lines in that file to disable or enable features
1012 that you require.
1014 Connections from remote hosts are disabled by default:
1016 \begin{verbatim}
1017 # Address xend should listen on for HTTP connections, if xend-http-server is
1018 # set.
1019 # Specifying 'localhost' prevents remote connections.
1020 # Specifying the empty string '' (the default) allows all connections.
1021 #(xend-address '')
1022 (xend-address localhost)
1023 \end{verbatim}
1025 It is recommended that if migration support is not needed, the
1026 \texttt{xend-relocation-server} parameter value be changed to
1027 ``\texttt{no}'' or commented out.
1029 \section{Xm}
1030 \label{s:xm}
1032 The xm tool is the primary tool for managing Xen from the console. The
1033 general format of an xm command line is:
1035 \begin{verbatim}
1036 # xm command [switches] [arguments] [variables]
1037 \end{verbatim}
1039 The available \emph{switches} and \emph{arguments} are dependent on the
1040 \emph{command} chosen. The \emph{variables} may be set using
1041 declarations of the form {\tt variable=value} and command line
1042 declarations override any of the values in the configuration file being
1043 used, including the standard variables described above and any custom
1044 variables (for instance, the \path{xmdefconfig} file uses a {\tt vmid}
1045 variable).
1047 For online help for the commands available, type:
1049 \begin{quote}
1050 \begin{verbatim}
1051 # xm help
1052 \end{verbatim}
1053 \end{quote}
1055 This will list the most commonly used commands. The full list can be obtained
1056 using \verb_xm help --long_. You can also type \path{xm help $<$command$>$}
1057 for more information on a given command.
1059 \subsection{Basic Management Commands}
1061 One useful command is \verb_# xm list_ which lists all domains running in rows
1062 of the following format:
1063 \begin{center} {\tt name domid memory vcpus state cputime}
1064 \end{center}
1066 The meaning of each field is as follows:
1067 \begin{quote}
1068 \begin{description}
1069 \item[name] The descriptive name of the virtual machine.
1070 \item[domid] The number of the domain ID this virtual machine is
1071 running in.
1072 \item[memory] Memory size in megabytes.
1073 \item[vcpus] The number of virtual CPUs this domain has.
1074 \item[state] Domain state consists of 5 fields:
1075 \begin{description}
1076 \item[r] running
1077 \item[b] blocked
1078 \item[p] paused
1079 \item[s] shutdown
1080 \item[c] crashed
1081 \end{description}
1082 \item[cputime] How much CPU time (in seconds) the domain has used so
1083 far.
1084 \end{description}
1085 \end{quote}
1087 The \path{xm list} command also supports a long output format when the
1088 \path{-l} switch is used. This outputs the full details of the
1089 running domains in \xend's SXP configuration format.
1091 If you want to know how long your domains have been running for, then
1092 you can use the \verb_# xm uptime_ command.
1095 You can get access to the console of a particular domain using
1096 the \verb_# xm console_ command (e.g.\ \verb_# xm console myVM_).
1098 \subsection{Domain Scheduling Management Commands}
1100 The credit CPU scheduler automatically load balances guest VCPUs
1101 across all available physical CPUs on an SMP host. The user need
1102 not manually pin VCPUs to load balance the system. However, she
1103 can restrict which CPUs a particular VCPU may run on using
1104 the \path{xm vcpu-pin} command.
1106 Each guest domain is assigned a \path{weight} and a \path{cap}.
1108 A domain with a weight of 512 will get twice as much CPU as a
1109 domain with a weight of 256 on a contended host. Legal weights
1110 range from 1 to 65535 and the default is 256.
1112 The cap optionally fixes the maximum amount of CPU a guest will
1113 be able to consume, even if the host system has idle CPU cycles.
1114 The cap is expressed in percentage of one physical CPU: 100 is
1115 1 physical CPU, 50 is half a CPU, 400 is 4 CPUs, etc... The
1116 default, 0, means there is no upper cap.
1118 When you are running with the credit scheduler, you can check and
1119 modify your domains' weights and caps using the \path{xm sched-credit}
1120 command:
1122 \begin{tabular}{ll}
1123 \verb!xm sched-credit -d <domain>! & lists weight and cap \\
1124 \verb!xm sched-credit -d <domain> -w <weight>! & sets the weight \\
1125 \verb!xm sched-credit -d <domain> -c <cap>! & sets the cap
1126 \end{tabular}
1130 %% Chapter Domain Configuration
1131 \chapter{Domain Configuration}
1132 \label{cha:config}
1134 The following contains the syntax of the domain configuration files
1135 and description of how to further specify networking, driver domain
1136 and general scheduling behavior.
1139 \section{Configuration Files}
1140 \label{s:cfiles}
1142 Xen configuration files contain the following standard variables.
1143 Unless otherwise stated, configuration items should be enclosed in
1144 quotes: see the configuration scripts in \path{/etc/xen/}
1145 for concrete examples.
1147 \begin{description}
1148 \item[kernel] Path to the kernel image.
1149 \item[ramdisk] Path to a ramdisk image (optional).
1150 % \item[builder] The name of the domain build function (e.g.
1151 % {\tt'linux'} or {\tt'netbsd'}.
1152 \item[memory] Memory size in megabytes.
1153 \item[vcpus] The number of virtual CPUs.
1154 \item[console] Port to export the domain console on (default 9600 +
1155 domain ID).
1156 \item[vif] Network interface configuration. This may simply contain
1157 an empty string for each desired interface, or may override various
1158 settings, e.g.\
1159 \begin{verbatim}
1160 vif = [ 'mac=00:16:3E:00:00:11, bridge=xen-br0',
1161 'bridge=xen-br1' ]
1162 \end{verbatim}
1163 to assign a MAC address and bridge to the first interface and assign
1164 a different bridge to the second interface, leaving \xend\ to choose
1165 the MAC address. The settings that may be overridden in this way are
1166 type, mac, bridge, ip, script, backend, and vifname.
1167 \item[disk] List of block devices to export to the domain e.g.
1168 \verb_disk = [ 'phy:hda1,sda1,r' ]_
1169 exports physical device \path{/dev/hda1} to the domain as
1170 \path{/dev/sda1} with read-only access. Exporting a disk read-write
1171 which is currently mounted is dangerous -- if you are \emph{certain}
1172 you wish to do this, you can specify \path{w!} as the mode.
1173 \item[dhcp] Set to {\tt `dhcp'} if you want to use DHCP to configure
1174 networking.
1175 \item[netmask] Manually configured IP netmask.
1176 \item[gateway] Manually configured IP gateway.
1177 \item[hostname] Set the hostname for the virtual machine.
1178 \item[root] Specify the root device parameter on the kernel command
1179 line.
1180 \item[nfs\_server] IP address for the NFS server (if any).
1181 \item[nfs\_root] Path of the root filesystem on the NFS server (if
1182 any).
1183 \item[extra] Extra string to append to the kernel command line (if
1184 any)
1185 \end{description}
1187 Additional fields are documented in the example configuration files
1188 (e.g. to configure virtual TPM functionality).
1190 For additional flexibility, it is also possible to include Python
1191 scripting commands in configuration files. An example of this is the
1192 \path{xmexample2} file, which uses Python code to handle the
1193 \path{vmid} variable.
1196 %\part{Advanced Topics}
1199 \section{Network Configuration}
1201 For many users, the default installation should work ``out of the
1202 box''. More complicated network setups, for instance with multiple
1203 Ethernet interfaces and/or existing bridging setups will require some
1204 special configuration.
1206 The purpose of this section is to describe the mechanisms provided by
1207 \xend\ to allow a flexible configuration for Xen's virtual networking.
1209 \subsection{Xen virtual network topology}
1211 Each domain network interface is connected to a virtual network
1212 interface in dom0 by a point to point link (effectively a ``virtual
1213 crossover cable''). These devices are named {\tt
1214 vif$<$domid$>$.$<$vifid$>$} (e.g.\ {\tt vif1.0} for the first
1215 interface in domain~1, {\tt vif3.1} for the second interface in
1216 domain~3).
1218 Traffic on these virtual interfaces is handled in domain~0 using
1219 standard Linux mechanisms for bridging, routing, rate limiting, etc.
1220 Xend calls on two shell scripts to perform initial configuration of
1221 the network and configuration of new virtual interfaces. By default,
1222 these scripts configure a single bridge for all the virtual
1223 interfaces. Arbitrary routing / bridging configurations can be
1224 configured by customizing the scripts, as described in the following
1225 section.
1227 \subsection{Xen networking scripts}
1229 Xen's virtual networking is configured by two shell scripts (by
1230 default \path{network-bridge} and \path{vif-bridge}). These are called
1231 automatically by \xend\ when certain events occur, with arguments to
1232 the scripts providing further contextual information. These scripts
1233 are found by default in \path{/etc/xen/scripts}. The names and
1234 locations of the scripts can be configured in
1235 \path{/etc/xen/xend-config.sxp}.
1237 \begin{description}
1238 \item[network-bridge:] This script is called whenever \xend\ is started or
1239 stopped to respectively initialize or tear down the Xen virtual
1240 network. In the default configuration initialization creates the
1241 bridge `xen-br0' and moves eth0 onto that bridge, modifying the
1242 routing accordingly. When \xend\ exits, it deletes the Xen bridge
1243 and removes eth0, restoring the normal IP and routing configuration.
1245 %% In configurations where the bridge already exists, this script
1246 %% could be replaced with a link to \path{/bin/true} (for instance).
1248 \item[vif-bridge:] This script is called for every domain virtual
1249 interface and can configure firewalling rules and add the vif to the
1250 appropriate bridge. By default, this adds and removes VIFs on the
1251 default Xen bridge.
1252 \end{description}
1254 Other example scripts are available (\path{network-route} and
1255 \path{vif-route}, \path{network-nat} and \path{vif-nat}).
1256 For more complex network setups (e.g.\ where routing is required or
1257 integrate with existing bridges) these scripts may be replaced with
1258 customized variants for your site's preferred configuration.
1260 \section{Driver Domain Configuration}
1261 \label{s:ddconf}
1263 \subsection{PCI}
1264 \label{ss:pcidd}
1266 Individual PCI devices can be assigned to a given domain (a PCI driver domain)
1267 to allow that domain direct access to the PCI hardware.
1269 While PCI Driver Domains can increase the stability and security of a system
1270 by addressing a number of security concerns, there are some security issues
1271 that remain that you can read about in Section~\ref{s:ddsecurity}.
1273 \subsubsection{Compile-Time Setup}
1274 To use this functionality, ensure
1275 that the PCI Backend is compiled in to a privileged domain (e.g. domain 0)
1276 and that the domains which will be assigned PCI devices have the PCI Frontend
1277 compiled in. In XenLinux, the PCI Backend is available under the Xen
1278 configuration section while the PCI Frontend is under the
1279 architecture-specific "Bus Options" section. You may compile both the backend
1280 and the frontend into the same kernel; they will not affect each other.
1282 \subsubsection{PCI Backend Configuration - Binding at Boot}
1283 The PCI devices you wish to assign to unprivileged domains must be "hidden"
1284 from your backend domain (usually domain 0) so that it does not load a driver
1285 for them. Use the \path{pciback.hide} kernel parameter which is specified on
1286 the kernel command-line and is configurable through GRUB (see
1287 Section~\ref{s:configure}). Note that devices are not really hidden from the
1288 backend domain. The PCI Backend appears to the Linux kernel as a regular PCI
1289 device driver. The PCI Backend ensures that no other device driver loads
1290 for the devices by binding itself as the device driver for those devices.
1291 PCI devices are identified by hexadecimal slot/function numbers (on Linux,
1292 use \path{lspci} to determine slot/function numbers of your devices) and
1293 can be specified with or without the PCI domain: \\
1294 \centerline{ {\tt ({\em bus}:{\em slot}.{\em func})} example {\tt (02:1d.3)}} \\
1295 \centerline{ {\tt ({\em domain}:{\em bus}:{\em slot}.{\em func})} example {\tt (0000:02:1d.3)}} \\
1297 An example kernel command-line which hides two PCI devices might be: \\
1298 \centerline{ {\tt root=/dev/sda4 ro console=tty0 pciback.hide=(02:01.f)(0000:04:1d.0) } } \\
1300 \subsubsection{PCI Backend Configuration - Late Binding}
1301 PCI devices can also be bound to the PCI Backend after boot through the manual
1302 binding/unbinding facilities provided by the Linux kernel in sysfs (allowing
1303 for a Xen user to give PCI devices to driver domains that were not specified
1304 on the kernel command-line). There are several attributes with the PCI
1305 Backend's sysfs directory (\path{/sys/bus/pci/drivers/pciback}) that can be
1306 used to bind/unbind devices:
1308 \begin{description}
1309 \item[slots] lists all of the PCI slots that the PCI Backend will try to seize
1310 (or "hide" from Domain 0). A PCI slot must appear in this list before it can
1311 be bound to the PCI Backend through the \path{bind} attribute.
1312 \item[new\_slot] write the name of a slot here (in 0000:00:00.0 format) to
1313 have the PCI Backend seize the device in this slot.
1314 \item[remove\_slot] write the name of a slot here (same format as
1315 \path{new\_slot}) to have the PCI Backend no longer try to seize devices in
1316 this slot. Note that this does not unbind the driver from a device it has
1317 already seized.
1318 \item[bind] write the name of a slot here (in 0000:00:00.0 format) to have
1319 the Linux kernel attempt to bind the device in that slot to the PCI Backend
1320 driver.
1321 \item[unbind] write the name of a skit here (same format as \path{bind}) to have
1322 the Linux kernel unbind the device from the PCI Backend. DO NOT unbind a
1323 device while it is currently given to a PCI driver domain!
1324 \end{description}
1326 Some examples:
1328 Bind a device to the PCI Backend which is not bound to any other driver.
1329 \begin{verbatim}
1330 # # Add a new slot to the PCI Backend's list
1331 # echo -n 0000:01:04.d > /sys/bus/pci/drivers/pciback/new_slot
1332 # # Now that the backend is watching for the slot, bind to it
1333 # echo -n 0000:01:04.d > /sys/bus/pci/drivers/pciback/bind
1334 \end{verbatim}
1336 Unbind a device from its driver and bind to the PCI Backend.
1337 \begin{verbatim}
1338 # # Unbind a PCI network card from its network driver
1339 # echo -n 0000:05:02.0 > /sys/bus/pci/drivers/3c905/unbind
1340 # # And now bind it to the PCI Backend
1341 # echo -n 0000:05:02.0 > /sys/bus/pci/drivers/pciback/new_slot
1342 # echo -n 0000:05:02.0 > /sys/bus/pci/drivers/pciback/bind
1343 \end{verbatim}
1345 Note that the "-n" option in the example is important as it causes echo to not
1346 output a new-line.
1348 \subsubsection{PCI Backend Configuration - User-space Quirks}
1349 Quirky devices (such as the Broadcom Tigon 3) may need write access to their
1350 configuration space registers. Xen can be instructed to allow specified PCI
1351 devices write access to specific configuration space registers. The policy may
1352 be found in:
1354 \centerline{ \path{/etc/xen/xend-pci-quirks.sxp} }
1356 The policy file is heavily commented and is intended to provide enough
1357 documentation for developers to extend it.
1359 \subsubsection{PCI Backend Configuration - Permissive Flag}
1360 If the user-space quirks approach doesn't meet your needs you may want to enable
1361 the permissive flag for that device. To do so, first get the PCI domain, bus,
1362 slot, and function information from dom0 via \path{lspci}. Then augment the
1363 user-space policy for permissive devices. The permissive policy can be found
1364 in:
1366 \centerline{ \path{/etc/xen/xend-pci-permissive.sxp} }
1368 Currently, the only way to reset the permissive flag is to unbind the device
1369 from the PCI Backend driver.
1371 \subsubsection{PCI Backend - Checking Status}
1372 There two important sysfs nodes that provide a mechanism to view specifics on
1373 quirks and permissive devices:
1374 \begin{description}
1375 \item \path{/sys/bus/drivers/pciback/permissive} \\
1376 Use \path{cat} on this file to view a list of permissive slots.
1377 \item \path{/sys/bus/drivers/pciback/quirks} \\
1378 Use \path{cat} on this file view a hierarchical view of devices bound to the
1379 PCI backend, their PCI vendor/device ID, and any quirks that are associated with
1380 that particular slot.
1381 \end{description}
1383 You may notice that every device bound to the PCI backend has 17 quirks standard
1384 "quirks" regardless of \path{xend-pci-quirks.sxp}. These default entries are
1385 necessary to support interactions between the PCI bus manager and the device bound
1386 to it. Even non-quirky devices should have these standard entries.
1388 In this case, preference was given to accuracy over aesthetics by choosing to
1389 show the standard quirks in the quirks list rather than hide them from the
1390 inquiring user
1392 \subsubsection{PCI Frontend Configuration}
1393 To configure a domU to receive a PCI device:
1395 \begin{description}
1396 \item[Command-line:]
1397 Use the {\em pci} command-line flag. For multiple devices, use the option
1398 multiple times. \\
1399 \centerline{ {\tt xm create netcard-dd pci=01:00.0 pci=02:03.0 }} \\
1401 \item[Flat Format configuration file:]
1402 Specify all of your PCI devices in a python list named {\em pci}. \\
1403 \centerline{ {\tt pci=['01:00.0','02:03.0'] }} \\
1405 \item[SXP Format configuration file:]
1406 Use a single PCI device section for all of your devices (specify the numbers
1407 in hexadecimal with the preceding '0x'). Note that {\em domain} here refers
1408 to the PCI domain, not a virtual machine within Xen.
1409 {\small
1410 \begin{verbatim}
1411 (device (pci
1412 (dev (domain 0x0)(bus 0x3)(slot 0x1a)(func 0x1)
1413 (dev (domain 0x0)(bus 0x1)(slot 0x5)(func 0x0)
1415 \end{verbatim}
1417 \end{description}
1419 %% There are two possible types of privileges: IO privileges and
1420 %% administration privileges.
1422 \section{Support for virtual Trusted Platform Module (vTPM)}
1423 \label{ss:vtpm}
1425 Paravirtualized domains can be given access to a virtualized version
1426 of a TPM. This enables applications in these domains to use the services
1427 of the TPM device for example through a TSS stack
1428 \footnote{Trousers TSS stack:}.
1429 The Xen source repository provides the necessary software components to
1430 enable virtual TPM access. Support is provided through several
1431 different pieces. First, a TPM emulator has been modified to provide TPM's
1432 functionality for the virtual TPM subsystem. Second, a virtual TPM Manager
1433 coordinates the virtual TPMs efforts, manages their creation, and provides
1434 protected key storage using the TPM. Third, a device driver pair providing
1435 a TPM front- and backend is available for XenLinux to deliver TPM commands
1436 from the domain to the virtual TPM manager, which dispatches it to a
1437 software TPM. Since the TPM Manager relies on a HW TPM for protected key
1438 storage, therefore this subsystem requires a Linux-supported hardware TPM.
1439 For development purposes, a TPM emulator is available for use on non-TPM
1440 enabled platforms.
1442 \subsubsection{Compile-Time Setup}
1443 To enable access to the virtual TPM, the virtual TPM backend driver must
1444 be compiled for a privileged domain (e.g. domain 0). Using the XenLinux
1445 configuration, the necessary driver can be selected in the Xen configuration
1446 section. Unless the driver has been compiled into the kernel, its module
1447 must be activated using the following command:
1449 \begin{verbatim}
1450 modprobe tpmbk
1451 \end{verbatim}
1453 Similarly, the TPM frontend driver must be compiled for the kernel trying
1454 to use TPM functionality. Its driver can be selected in the kernel
1455 configuration section Device Driver / Character Devices / TPM Devices.
1456 Along with that the TPM driver for the built-in TPM must be selected.
1457 If the virtual TPM driver has been compiled as module, it
1458 must be activated using the following command:
1460 \begin{verbatim}
1461 modprobe tpm_xenu
1462 \end{verbatim}
1464 Furthermore, it is necessary to build the virtual TPM manager and software
1465 TPM by making changes to entries in Xen build configuration files.
1466 The following entry in the file in the Xen root source
1467 directory must be made:
1469 \begin{verbatim}
1470 VTPM_TOOLS ?= y
1471 \end{verbatim}
1473 After a build of the Xen tree and a reboot of the machine, the TPM backend
1474 drive must be loaded. Once loaded, the virtual TPM manager daemon
1475 must be started before TPM-enabled guest domains may be launched.
1476 To enable being the destination of a virtual TPM Migration, the virtual TPM
1477 migration daemon must also be loaded.
1479 \begin{verbatim}
1480 vtpm_managerd
1481 \end{verbatim}
1482 \begin{verbatim}
1483 vtpm_migratord
1484 \end{verbatim}
1486 Once the VTPM manager is running, the VTPM can be accessed by loading the
1487 front end driver in a guest domain.
1489 \subsubsection{Development and Testing TPM Emulator}
1490 For development and testing on non-TPM enabled platforms, a TPM emulator
1491 can be used in replacement of a platform TPM. First, the entry in the file
1492 tools/vtpm/ must look as follows:
1494 \begin{verbatim}
1496 \end{verbatim}
1498 Second, the entry in the file tool/vtpm\_manager/ must be uncommented
1499 as follows:
1501 \begin{verbatim}
1502 # TCS talks to fifo's rather than /dev/tpm. TPM Emulator assumed on fifos
1504 \end{verbatim}
1506 Before starting the virtual TPM Manager, start the emulator by executing
1507 the following in dom0:
1509 \begin{verbatim}
1510 tpm_emulator clear
1511 \end{verbatim}
1513 \subsubsection{vTPM Frontend Configuration}
1514 To provide TPM functionality to a user domain, a line must be added to
1515 the virtual TPM configuration file using the following format:
1517 \begin{verbatim}
1518 vtpm = ['instance=<instance number>, backend=<domain id>']
1519 \end{verbatim}
1521 The { \it instance number} reflects the preferred virtual TPM instance
1522 to associate with the domain. If the selected instance is
1523 already associated with another domain, the system will automatically
1524 select the next available instance. An instance number greater than
1525 zero must be provided. It is possible to omit the instance
1526 parameter from the configuration file.
1528 The {\it domain id} provides the ID of the domain where the
1529 virtual TPM backend driver and virtual TPM are running in. It should
1530 currently always be set to '0'.
1533 Examples for valid vtpm entries in the configuration file are
1535 \begin{verbatim}
1536 vtpm = ['instance=1, backend=0']
1537 \end{verbatim}
1538 and
1539 \begin{verbatim}
1540 vtpm = ['backend=0'].
1541 \end{verbatim}
1543 \subsubsection{Using the virtual TPM}
1545 Access to TPM functionality is provided by the virtual TPM frontend driver.
1546 Similar to existing hardware TPM drivers, this driver provides basic TPM
1547 status information through the {\it sysfs} filesystem. In a Xen user domain
1548 the sysfs entries can be found in /sys/devices/xen/vtpm-0.
1550 Commands can be sent to the virtual TPM instance using the character
1551 device /dev/tpm0 (major 10, minor 224).
1553 % Chapter Storage and FileSytem Management
1554 \chapter{Storage and File System Management}
1556 Storage can be made available to virtual machines in a number of
1557 different ways. This chapter covers some possible configurations.
1559 The most straightforward method is to export a physical block device (a
1560 hard drive or partition) from dom0 directly to the guest domain as a
1561 virtual block device (VBD).
1563 Storage may also be exported from a filesystem image or a partitioned
1564 filesystem image as a \emph{file-backed VBD}.
1566 Finally, standard network storage protocols such as NBD, iSCSI, NFS,
1567 etc., can be used to provide storage to virtual machines.
1570 \section{Exporting Physical Devices as VBDs}
1571 \label{s:exporting-physical-devices-as-vbds}
1573 One of the simplest configurations is to directly export individual
1574 partitions from domain~0 to other domains. To achieve this use the
1575 \path{phy:} specifier in your domain configuration file. For example a
1576 line like
1577 \begin{quote}
1578 \verb_disk = ['phy:hda3,sda1,w']_
1579 \end{quote}
1580 specifies that the partition \path{/dev/hda3} in domain~0 should be
1581 exported read-write to the new domain as \path{/dev/sda1}; one could
1582 equally well export it as \path{/dev/hda} or \path{/dev/sdb5} should
1583 one wish.
1585 In addition to local disks and partitions, it is possible to export
1586 any device that Linux considers to be ``a disk'' in the same manner.
1587 For example, if you have iSCSI disks or GNBD volumes imported into
1588 domain~0 you can export these to other domains using the \path{phy:}
1589 disk syntax. E.g.:
1590 \begin{quote}
1591 \verb_disk = ['phy:vg/lvm1,sda2,w']_
1592 \end{quote}
1594 \begin{center}
1595 \framebox{\bf Warning: Block device sharing}
1596 \end{center}
1597 \begin{quote}
1598 Block devices should typically only be shared between domains in a
1599 read-only fashion otherwise the Linux kernel's file systems will get
1600 very confused as the file system structure may change underneath
1601 them (having the same ext3 partition mounted \path{rw} twice is a
1602 sure fire way to cause irreparable damage)! \Xend\ will attempt to
1603 prevent you from doing this by checking that the device is not
1604 mounted read-write in domain~0, and hasn't already been exported
1605 read-write to another domain. If you want read-write sharing,
1606 export the directory to other domains via NFS from domain~0 (or use
1607 a cluster file system such as GFS or ocfs2).
1608 \end{quote}
1611 \section{Using File-backed VBDs}
1613 It is also possible to use a file in Domain~0 as the primary storage
1614 for a virtual machine. As well as being convenient, this also has the
1615 advantage that the virtual block device will be \emph{sparse} ---
1616 space will only really be allocated as parts of the file are used. So
1617 if a virtual machine uses only half of its disk space then the file
1618 really takes up half of the size allocated.
1620 For example, to create a 2GB sparse file-backed virtual block device
1621 (actually only consumes no disk space at all):
1622 \begin{quote}
1623 \verb_# dd if=/dev/zero of=vm1disk bs=1k seek=2048k count=0_
1624 \end{quote}
1626 Make a file system in the disk file:
1627 \begin{quote}
1628 \verb_# mkfs -t ext3 vm1disk_
1629 \end{quote}
1631 (when the tool asks for confirmation, answer `y')
1633 Populate the file system e.g.\ by copying from the current root:
1634 \begin{quote}
1635 \begin{verbatim}
1636 # mount -o loop vm1disk /mnt
1637 # cp -ax /{root,dev,var,etc,usr,bin,sbin,lib} /mnt
1638 # mkdir /mnt/{proc,sys,home,tmp}
1639 \end{verbatim}
1640 \end{quote}
1642 Tailor the file system by editing \path{/etc/fstab},
1643 \path{/etc/hostname}, etc.\ Don't forget to edit the files in the
1644 mounted file system, instead of your domain~0 filesystem, e.g.\ you
1645 would edit \path{/mnt/etc/fstab} instead of \path{/etc/fstab}. For
1646 this example put \path{/dev/sda1} to root in fstab.
1648 Now unmount (this is important!):
1649 \begin{quote}
1650 \verb_# umount /mnt_
1651 \end{quote}
1653 In the configuration file set:
1654 \begin{quote}
1655 \verb_disk = ['tap:aio:/full/path/to/vm1disk,sda1,w']_
1656 \end{quote}
1658 As the virtual machine writes to its `disk', the sparse file will be
1659 filled in and consume more space up to the original 2GB.
1661 {\em{Note:}} Users that have worked with file-backed VBDs on Xen in previous
1662 versions will be interested to know that this support is now provided through
1663 the blktap driver instead of the loopback driver. This change results in
1664 file-based block devices that are higher-performance, more scalable, and which
1665 provide better safety properties for VBD data. All that is required to update
1666 your existing file-backed VM configurations is to change VBD configuration
1667 lines from:
1668 \begin{quote}
1669 \verb_disk = ['file:/full/path/to/vm1disk,sda1,w']_
1670 \end{quote}
1671 to:
1672 \begin{quote}
1673 \verb_disk = ['tap:aio:/full/path/to/vm1disk,sda1,w']_
1674 \end{quote}
1677 \subsection{Loopback-mounted file-backed VBDs (deprecated)}
1679 {\em{{\bf{Note:}} Loopback mounted VBDs have now been replaced with
1680 blktap-based support for raw image files, as described above. This
1681 section remains to detail a configuration that was used by older Xen
1682 versions.}}
1684 Raw image file-backed VBDs may also be attached to VMs using the
1685 Linux loopback driver. The only required change to the raw file
1686 instructions above are to specify the configuration entry as:
1687 \begin{quote}
1688 \verb_disk = ['file:/full/path/to/vm1disk,sda1,w']_
1689 \end{quote}
1691 {\bf Note that loopback file-backed VBDs may not be appropriate for backing
1692 I/O-intensive domains.} This approach is known to experience
1693 substantial slowdowns under heavy I/O workloads, due to the I/O
1694 handling by the loopback block device used to support file-backed VBDs
1695 in dom0. Loopback support remains for old Xen installations, and users
1696 are strongly encouraged to use the blktap-based file support (using
1697 ``{\tt{tap:aio}}'' as described above).
1699 Additionally, Linux supports a maximum of eight loopback file-backed
1700 VBDs across all domains by default. This limit can be statically
1701 increased by using the \emph{max\_loop} module parameter if
1702 CONFIG\_BLK\_DEV\_LOOP is compiled as a module in the dom0 kernel, or
1703 by using the \emph{max\_loop=n} boot option if CONFIG\_BLK\_DEV\_LOOP
1704 is compiled directly into the dom0 kernel. Again, users are encouraged
1705 to use the blktap-based file support described above which scales to much
1706 larger number of active VBDs.
1709 \section{Using LVM-backed VBDs}
1710 \label{s:using-lvm-backed-vbds}
1712 A particularly appealing solution is to use LVM volumes as backing for
1713 domain file-systems since this allows dynamic growing/shrinking of
1714 volumes as well as snapshot and other features.
1716 To initialize a partition to support LVM volumes:
1717 \begin{quote}
1718 \begin{verbatim}
1719 # pvcreate /dev/sda10
1720 \end{verbatim}
1721 \end{quote}
1723 Create a volume group named `vg' on the physical partition:
1724 \begin{quote}
1725 \begin{verbatim}
1726 # vgcreate vg /dev/sda10
1727 \end{verbatim}
1728 \end{quote}
1730 Create a logical volume of size 4GB named `myvmdisk1':
1731 \begin{quote}
1732 \begin{verbatim}
1733 # lvcreate -L4096M -n myvmdisk1 vg
1734 \end{verbatim}
1735 \end{quote}
1737 You should now see that you have a \path{/dev/vg/myvmdisk1} Make a
1738 filesystem, mount it and populate it, e.g.:
1739 \begin{quote}
1740 \begin{verbatim}
1741 # mkfs -t ext3 /dev/vg/myvmdisk1
1742 # mount /dev/vg/myvmdisk1 /mnt
1743 # cp -ax / /mnt
1744 # umount /mnt
1745 \end{verbatim}
1746 \end{quote}
1748 Now configure your VM with the following disk configuration:
1749 \begin{quote}
1750 \begin{verbatim}
1751 disk = [ 'phy:vg/myvmdisk1,sda1,w' ]
1752 \end{verbatim}
1753 \end{quote}
1755 LVM enables you to grow the size of logical volumes, but you'll need
1756 to resize the corresponding file system to make use of the new space.
1757 Some file systems (e.g.\ ext3) now support online resize. See the LVM
1758 manuals for more details.
1760 You can also use LVM for creating copy-on-write (CoW) clones of LVM
1761 volumes (known as writable persistent snapshots in LVM terminology).
1762 This facility is new in Linux 2.6.8, so isn't as stable as one might
1763 hope. In particular, using lots of CoW LVM disks consumes a lot of
1764 dom0 memory, and error conditions such as running out of disk space
1765 are not handled well. Hopefully this will improve in future.
1767 To create two copy-on-write clones of the above file system you would
1768 use the following commands:
1770 \begin{quote}
1771 \begin{verbatim}
1772 # lvcreate -s -L1024M -n myclonedisk1 /dev/vg/myvmdisk1
1773 # lvcreate -s -L1024M -n myclonedisk2 /dev/vg/myvmdisk1
1774 \end{verbatim}
1775 \end{quote}
1777 Each of these can grow to have 1GB of differences from the master
1778 volume. You can grow the amount of space for storing the differences
1779 using the lvextend command, e.g.:
1780 \begin{quote}
1781 \begin{verbatim}
1782 # lvextend +100M /dev/vg/myclonedisk1
1783 \end{verbatim}
1784 \end{quote}
1786 Don't let the `differences volume' ever fill up otherwise LVM gets
1787 rather confused. It may be possible to automate the growing process by
1788 using \path{dmsetup wait} to spot the volume getting full and then
1789 issue an \path{lvextend}.
1791 In principle, it is possible to continue writing to the volume that
1792 has been cloned (the changes will not be visible to the clones), but
1793 we wouldn't recommend this: have the cloned volume as a `pristine'
1794 file system install that isn't mounted directly by any of the virtual
1795 machines.
1798 \section{Using NFS Root}
1800 First, populate a root filesystem in a directory on the server
1801 machine. This can be on a distinct physical machine, or simply run
1802 within a virtual machine on the same node.
1804 Now configure the NFS server to export this filesystem over the
1805 network by adding a line to \path{/etc/exports}, for instance:
1807 \begin{quote}
1808 \begin{small}
1809 \begin{verbatim}
1810 /export/vm1root (rw,sync,no_root_squash)
1811 \end{verbatim}
1812 \end{small}
1813 \end{quote}
1815 Finally, configure the domain to use NFS root. In addition to the
1816 normal variables, you should make sure to set the following values in
1817 the domain's configuration file:
1819 \begin{quote}
1820 \begin{small}
1821 \begin{verbatim}
1822 root = '/dev/nfs'
1823 nfs_server = '' # substitute IP address of server
1824 nfs_root = '/path/to/root' # path to root FS on the server
1825 \end{verbatim}
1826 \end{small}
1827 \end{quote}
1829 The domain will need network access at boot time, so either statically
1830 configure an IP address using the config variables \path{ip},
1831 \path{netmask}, \path{gateway}, \path{hostname}; or enable DHCP
1832 (\path{dhcp='dhcp'}).
1834 Note that the Linux NFS root implementation is known to have stability
1835 problems under high load (this is not a Xen-specific problem), so this
1836 configuration may not be appropriate for critical servers.
1839 \chapter{CPU Management}
1841 %% KMS Something sage about CPU / processor management.
1843 Xen allows a domain's virtual CPU(s) to be associated with one or more
1844 host CPUs. This can be used to allocate real resources among one or
1845 more guests, or to make optimal use of processor resources when
1846 utilizing dual-core, hyperthreading, or other advanced CPU technologies.
1848 Xen enumerates physical CPUs in a `depth first' fashion. For a system
1849 with both hyperthreading and multiple cores, this would be all the
1850 hyperthreads on a given core, then all the cores on a given socket,
1851 and then all sockets. I.e. if you had a two socket, dual core,
1852 hyperthreaded Xeon the CPU order would be:
1855 \begin{center}
1856 \begin{tabular}{l|l|l|l|l|l|l|r}
1857 \multicolumn{4}{c|}{socket0} & \multicolumn{4}{c}{socket1} \\ \hline
1858 \multicolumn{2}{c|}{core0} & \multicolumn{2}{c|}{core1} &
1859 \multicolumn{2}{c|}{core0} & \multicolumn{2}{c}{core1} \\ \hline
1860 ht0 & ht1 & ht0 & ht1 & ht0 & ht1 & ht0 & ht1 \\
1861 \#0 & \#1 & \#2 & \#3 & \#4 & \#5 & \#6 & \#7 \\
1862 \end{tabular}
1863 \end{center}
1866 Having multiple vcpus belonging to the same domain mapped to the same
1867 physical CPU is very likely to lead to poor performance. It's better to
1868 use `vcpus-set' to hot-unplug one of the vcpus and ensure the others are
1869 pinned on different CPUs.
1871 If you are running IO intensive tasks, its typically better to dedicate
1872 either a hyperthread or whole core to running domain 0, and hence pin
1873 other domains so that they can't use CPU 0. If your workload is mostly
1874 compute intensive, you may want to pin vcpus such that all physical CPU
1875 threads are available for guest domains.
1877 \chapter{Migrating Domains}
1879 \section{Domain Save and Restore}
1881 The administrator of a Xen system may suspend a virtual machine's
1882 current state into a disk file in domain~0, allowing it to be resumed at
1883 a later time.
1885 For example you can suspend a domain called ``VM1'' to disk using the
1886 command:
1887 \begin{verbatim}
1888 # xm save VM1 VM1.chk
1889 \end{verbatim}
1891 This will stop the domain named ``VM1'' and save its current state
1892 into a file called \path{VM1.chk}.
1894 To resume execution of this domain, use the \path{xm restore} command:
1895 \begin{verbatim}
1896 # xm restore VM1.chk
1897 \end{verbatim}
1899 This will restore the state of the domain and resume its execution.
1900 The domain will carry on as before and the console may be reconnected
1901 using the \path{xm console} command, as described earlier.
1903 \section{Migration and Live Migration}
1905 Migration is used to transfer a domain between physical hosts. There
1906 are two varieties: regular and live migration. The former moves a
1907 virtual machine from one host to another by pausing it, copying its
1908 memory contents, and then resuming it on the destination. The latter
1909 performs the same logical functionality but without needing to pause
1910 the domain for the duration. In general when performing live migration
1911 the domain continues its usual activities and---from the user's
1912 perspective---the migration should be imperceptible.
1914 To perform a live migration, both hosts must be running Xen / \xend\ and
1915 the destination host must have sufficient resources (e.g.\ memory
1916 capacity) to accommodate the domain after the move. Furthermore we
1917 currently require both source and destination machines to be on the same
1918 L2 subnet.
1920 Currently, there is no support for providing automatic remote access
1921 to filesystems stored on local disk when a domain is migrated.
1922 Administrators should choose an appropriate storage solution (i.e.\
1923 SAN, NAS, etc.) to ensure that domain filesystems are also available
1924 on their destination node. GNBD is a good method for exporting a
1925 volume from one machine to another. iSCSI can do a similar job, but is
1926 more complex to set up.
1928 When a domain migrates, it's MAC and IP address move with it, thus it is
1929 only possible to migrate VMs within the same layer-2 network and IP
1930 subnet. If the destination node is on a different subnet, the
1931 administrator would need to manually configure a suitable etherip or IP
1932 tunnel in the domain~0 of the remote node.
1934 A domain may be migrated using the \path{xm migrate} command. To live
1935 migrate a domain to another machine, we would use the command:
1937 \begin{verbatim}
1938 # xm migrate --live mydomain
1939 \end{verbatim}
1941 Without the \path{--live} flag, \xend\ simply stops the domain and
1942 copies the memory image over to the new node and restarts it. Since
1943 domains can have large allocations this can be quite time consuming,
1944 even on a Gigabit network. With the \path{--live} flag \xend\ attempts
1945 to keep the domain running while the migration is in progress, resulting
1946 in typical down times of just 60--300ms.
1948 For now it will be necessary to reconnect to the domain's console on the
1949 new machine using the \path{xm console} command. If a migrated domain
1950 has any open network connections then they will be preserved, so SSH
1951 connections do not have this limitation.
1954 %% Chapter Securing Xen
1955 \chapter{Securing Xen}
1957 This chapter describes how to secure a Xen system. It describes a number
1958 of scenarios and provides a corresponding set of best practices. It
1959 begins with a section devoted to understanding the security implications
1960 of a Xen system.
1963 \section{Xen Security Considerations}
1965 When deploying a Xen system, one must be sure to secure the management
1966 domain (Domain-0) as much as possible. If the management domain is
1967 compromised, all other domains are also vulnerable. The following are a
1968 set of best practices for Domain-0:
1970 \begin{enumerate}
1971 \item \textbf{Run the smallest number of necessary services.} The less
1972 things that are present in a management partition, the better.
1973 Remember, a service running as root in the management domain has full
1974 access to all other domains on the system.
1975 \item \textbf{Use a firewall to restrict the traffic to the management
1976 domain.} A firewall with default-reject rules will help prevent
1977 attacks on the management domain.
1978 \item \textbf{Do not allow users to access Domain-0.} The Linux kernel
1979 has been known to have local-user root exploits. If you allow normal
1980 users to access Domain-0 (even as unprivileged users) you run the risk
1981 of a kernel exploit making all of your domains vulnerable.
1982 \end{enumerate}
1984 \section{Driver Domain Security Considerations}
1985 \label{s:ddsecurity}
1987 Driver domains address a range of security problems that exist regarding
1988 the use of device drivers and hardware. On many operating systems in common
1989 use today, device drivers run within the kernel with the same privileges as
1990 the kernel. Few or no mechanisms exist to protect the integrity of the kernel
1991 from a misbehaving (read "buggy") or malicious device driver. Driver
1992 domains exist to aid in isolating a device driver within its own virtual
1993 machine where it cannot affect the stability and integrity of other
1994 domains. If a driver crashes, the driver domain can be restarted rather than
1995 have the entire machine crash (and restart) with it. Drivers written by
1996 unknown or untrusted third-parties can be confined to an isolated space.
1997 Driver domains thus address a number of security and stability issues with
1998 device drivers.
2000 However, due to limitations in current hardware, a number of security
2001 concerns remain that need to be considered when setting up driver domains (it
2002 should be noted that the following list is not intended to be exhaustive).
2004 \begin{enumerate}
2005 \item \textbf{Without an IOMMU, a hardware device can DMA to memory regions
2006 outside of its controlling domain.} Architectures which do not have an
2007 IOMMU (e.g. most x86-based platforms) to restrict DMA usage by hardware
2008 are vulnerable. A hardware device which can perform arbitrary memory reads
2009 and writes can read/write outside of the memory of its controlling domain.
2010 A malicious or misbehaving domain could use a hardware device it controls
2011 to send data overwriting memory in another domain or to read arbitrary
2012 regions of memory in another domain.
2013 \item \textbf{Shared buses are vulnerable to sniffing.} Devices that share
2014 a data bus can sniff (and possible spoof) each others' data. Device A that
2015 is assigned to Domain A could eavesdrop on data being transmitted by
2016 Domain B to Device B and then relay that data back to Domain A.
2017 \item \textbf{Devices which share interrupt lines can either prevent the
2018 reception of that interrupt by the driver domain or can trigger the
2019 interrupt service routine of that guest needlessly.} A devices which shares
2020 a level-triggered interrupt (e.g. PCI devices) with another device can
2021 raise an interrupt and never clear it. This effectively blocks other devices
2022 which share that interrupt line from notifying their controlling driver
2023 domains that they need to be serviced. A device which shares an
2024 any type of interrupt line can trigger its interrupt continually which
2025 forces execution time to be spent (in multiple guests) in the interrupt
2026 service routine (potentially denying time to other processes within that
2027 guest). System architectures which allow each device to have its own
2028 interrupt line (e.g. PCI's Message Signaled Interrupts) are less
2029 vulnerable to this denial-of-service problem.
2030 \item \textbf{Devices may share the use of I/O memory address space.} Xen can
2031 only restrict access to a device's physical I/O resources at a certain
2032 granularity. For interrupt lines and I/O port address space, that
2033 granularity is very fine (per interrupt line and per I/O port). However,
2034 Xen can only restrict access to I/O memory address space on a page size
2035 basis. If more than one device shares use of a page in I/O memory address
2036 space, the domains to which those devices are assigned will be able to
2037 access the I/O memory address space of each other's devices.
2038 \end{enumerate}
2041 \section{Security Scenarios}
2044 \subsection{The Isolated Management Network}
2046 In this scenario, each node has two network cards in the cluster. One
2047 network card is connected to the outside world and one network card is a
2048 physically isolated management network specifically for Xen instances to
2049 use.
2051 As long as all of the management partitions are trusted equally, this is
2052 the most secure scenario. No additional configuration is needed other
2053 than forcing Xend to bind to the management interface for relocation.
2056 \subsection{A Subnet Behind a Firewall}
2058 In this scenario, each node has only one network card but the entire
2059 cluster sits behind a firewall. This firewall should do at least the
2060 following:
2062 \begin{enumerate}
2063 \item Prevent IP spoofing from outside of the subnet.
2064 \item Prevent access to the relocation port of any of the nodes in the
2065 cluster except from within the cluster.
2066 \end{enumerate}
2068 The following iptables rules can be used on each node to prevent
2069 migrations to that node from outside the subnet assuming the main
2070 firewall does not do this for you:
2072 \begin{verbatim}
2073 # this command disables all access to the Xen relocation
2074 # port:
2075 iptables -A INPUT -p tcp --destination-port 8002 -j REJECT
2077 # this command enables Xen relocations only from the specific
2078 # subnet:
2079 iptables -I INPUT -p tcp -{}-source \
2080 --destination-port 8002 -j ACCEPT
2081 \end{verbatim}
2083 \subsection{Nodes on an Untrusted Subnet}
2085 Migration on an untrusted subnet is not safe in current versions of Xen.
2086 It may be possible to perform migrations through a secure tunnel via an
2087 VPN or SSH. The only safe option in the absence of a secure tunnel is to
2088 disable migration completely. The easiest way to do this is with
2089 iptables:
2091 \begin{verbatim}
2092 # this command disables all access to the Xen relocation port
2093 iptables -A INPUT -p tcp -{}-destination-port 8002 -j REJECT
2094 \end{verbatim}
2096 %% Chapter Xen Mandatory Access Control Framework
2097 \chapter{sHype/Xen Access Control}
2098 The Xen mandatory access control framework is an implementation of the
2099 sHype Hypervisor Security Architecture
2100 (\_shype). It permits or denies communication
2101 and resource access of domains based on a security policy. The
2102 mandatory access controls are enforced in addition to the Xen core
2103 controls, such as memory protection. They are designed to remain
2104 transparent during normal operation of domains (policy-conform
2105 behavior) but to intervene when domains move outside their intended
2106 sharing behavior. This chapter will describe how the sHype access
2107 controls in Xen can be configured to prevent viruses from spilling
2108 over from one into another workload type and secrets from leaking from
2109 one workload type to another. sHype/Xen depends on the correct
2110 behavior of Domain-0 (cf previous chapter).
2112 Benefits of configuring sHype/ACM in Xen include:
2113 \begin{itemize}
2114 \item robust workload and resource protection effective against rogue
2115 user domains
2116 \item simple, platform- and operating system-independent security
2117 policies (ideal for heterogeneous distributed environments)
2118 \item safety net with minimal performance overhead in case operating
2119 system security is missing, does not scale, or fails
2120 \end{itemize}
2122 These benefits are very valuable because today's operating systems
2123 become increasingly complex and often have no or insufficient
2124 mandatory access controls. (Discretionary access controls, supported
2125 by most operating systems, are not effective against viruses or
2126 misbehaving programs.) Where mandatory access control exists (e.g.,
2127 SELinux), they usually deploy platform-specific, complex, and difficult
2128 to understand security policies. Multi-tier applications in business
2129 environments typically require different operating systems
2130 (e.g., AIX, Windows, Linux) in different tiers. Related distributed
2131 transactions and workloads cannot be easily protected on the OS level.
2132 The Xen access control framework steps in to offer a coarse-grained
2133 but very robust and consistent security layer and safety net across
2134 different platforms and operating systems.
2136 To control sharing between domains, Xen mediates all inter-domain
2137 communication (shared memory, events) as well as the access of domains
2138 to resources such as storage disks. Thus, Xen can confine distributed
2139 workloads (domain payloads) by permitting sharing among domains
2140 running the same type of workload and denying sharing between pairs of
2141 domains that run different workload types. We assume that--from a Xen
2142 perspective--only one workload type is running per user domain. To
2143 enable Xen to associate domains and resources with workload types,
2144 security labels including the workload types are attached to domains
2145 and resources. These labels and the hypervisor sHype controls cannot
2146 be manipulated or bypassed by user domains and are effective even
2147 against compromised or rogue domains.
2149 \section{Overview}
2150 This section gives an overview of how workloads can be protected using
2151 the sHype mandatory access control framework in Xen.
2152 Figure~\ref{fig:acmoverview} shows the necessary steps in activating
2153 the Xen workload protection. These steps are described in detail in
2154 Section~\ref{section:acmexample}.
2156 \begin{figure}
2157 \centering
2158 \includegraphics[width=13cm]{figs/acm_overview.eps}
2159 \caption{Overview of activating sHype workload protection in Xen.
2160 Section numbers point to representative examples.}
2161 \label{fig:acmoverview}
2162 \end{figure}
2164 First, the sHype/ACM access control must be enabled in the Xen
2165 distribution and the distribution must be built and installed (cf
2166 Subsection~\ref{subsection:acmexampleconfigure}). Before we can
2167 enforce security, a Xen security policy must be created (cf
2168 Subsection~\ref{subsection:acmexamplecreate}) and deployed (cf
2169 Subsection~\ref{subsection:acmexampleinstall}). This policy defines
2170 the workload types differentiated during access control. It also
2171 defines the rules that compare workload types of domains and resources
2172 to decide about access requests. Workload types are represented by
2173 security labels that can be securely associated to domains and resources (cf
2174 Subsections~\ref{subsection:acmexamplelabeldomains}
2175 and~\ref{subsection:acmexamplelabelresources}). The functioning of
2176 the active sHype/Xen workload protection is demonstrated using simple
2177 resource assignment, and domain creation tests in
2178 Subsection~\ref{subsection:acmexampletest}.
2179 Section~\ref{section:acmpolicy} describes the syntax and semantics of
2180 the sHype/Xen security policy in detail and introduces briefly the
2181 tools that are available to help you create your own sHype security policies.
2183 The next section describes all the necessary steps to create, deploy,
2184 and test a simple workload protection policy. It is meant to enable
2185 Xen users and developers to quickly try out the sHype/Xen workload
2186 protection. Those readers who are interested in learning more about
2187 how the sHype access control in Xen works and how it is configured
2188 using the XML security policy should read Section~\ref{section:acmpolicy}
2189 as well. Section~\ref{section:acmlimitations} concludes this chapter with
2190 current limitations of the sHype implementation for Xen.
2192 \section{Xen Workload Protection Step-by-Step}
2193 \label{section:acmexample}
2195 You are about to configure and deploy the Xen sHype workload protection
2196 by following 5 simple steps:
2197 \begin{itemize}
2198 \item configure and install sHype/Xen
2199 \item create a simple workload protection security policy
2200 \item deploy the sHype/Xen security policy
2201 \item associate domains and resources with workload labels,
2202 \item test the workload protection
2203 \end{itemize}
2204 The essential commands to create and deploy an sHype/Xen security
2205 policy are numbered throughout the following sections. If you want a
2206 quick-guide or return at a later time to go quickly through this
2207 demonstration, simply look for the numbered commands and apply them in
2208 order.
2210 \subsection{Configuring/Building sHype Support into Xen}
2211 \label{subsection:acmexampleconfigure}
2212 First, we need to configure the access control module in Xen and
2213 install the ACM-enabled Xen hypervisor. This step installs security
2214 tools and compiles sHype/ACM controls into the Xen hypervisor.
2216 To enable sHype/ACM in Xen, please edit the file in the top
2217 Xen directory.
2219 \begin{verbatim}
2220 (1) In
2221 Change: XSM_ENABLE ?= n
2222 To: XSM_ENABLE ?= y
2224 Change: ACM_SECURITY ?= n
2225 To: ACM_SECURITY ?= y
2226 \end{verbatim}
2228 Then install the security-enabled Xen environment as follows:
2230 \begin{verbatim}
2231 (2) # make world
2232 # make install
2233 \end{verbatim}
2235 Reboot into the security-enabled Xen hypervisor.
2237 \begin{verbatim}
2238 (3) # reboot
2239 \end{verbatim}
2241 Xen will boot into the default security policy. After reboot,
2242 you can explore the simple DEFAULT policy.
2243 \begin{scriptsize}
2244 \begin{verbatim}
2245 # xm getpolicy
2246 Supported security subsystems : ACM
2247 Policy name : DEFAULT
2248 Policy type : ACM
2249 Version of XML policy : 1.0
2250 Policy configuration : loaded
2252 # xm labels
2253 SystemManagement
2255 # xm list --label
2256 Name ID Mem VCPUs State Time(s) Label
2257 Domain-0 0 941 1 r----- 38.1 ACM:DEFAULT:SystemManagement
2258 \end{verbatim}
2259 \end{scriptsize}
2261 In this state, no domains can be started.
2262 Now, a policy can be created and loaded into the hypervisor.
2264 \subsection{Creating A WLP Policy in 3 Simple Steps with ezPolicy}
2265 \label{subsection:acmexamplecreate}
2267 We will use the ezPolicy tool to quickly create a policy that protects
2268 workloads. You will need both the Python and wxPython packages to run
2269 this tool. To run the tool in Domain-0, you can download the wxPython
2270 package from or use the command \verb|yum install wxPython|
2271 in Redhat/Fedora. To run the tool on MS Windows, you also need to download
2272 the Python package from After these packages are installed,
2273 start the ezPolicy tool with the following command:
2275 \begin{verbatim}
2276 (4) # xensec_ezpolicy
2277 \end{verbatim}
2279 Figure~\ref{fig:acmezpolicy} shows a screen-shot of the tool. The
2280 following steps illustrate how you can create the workload definition
2281 shown in Figure~\ref{fig:acmezpolicy}. You can use \verb|<CTRL>-h| to
2282 pop up a help window at any time. The indicators (a), (b), and (c) in
2283 Figure~\ref{fig:acmezpolicy} show the buttons that are used during the
2284 3 steps of creating a policy:
2285 \begin{enumerate}
2286 \item defining workloads
2287 \item defining run-time conflicts
2288 \item translating the workload definition into an sHype/Xen access
2289 control policy
2290 \end{enumerate}
2292 \paragraph{Defining workloads.} Workloads are defined for each
2293 organization and department that you enter in the left panel.
2295 To ease the transition from an unlabeled to a fully labeled workload-protection
2296 environment, we have added support to sHype/Xen to run unlabeled domains accessing
2297 unlabeled resources in addition to labeled domains accessing labeled resources.
2299 Support for running unlabeled domains on sHype/Xen is enabled by adding the
2300 predefined workload type and label \verb|__UNLABELED__| to the security
2301 policy. (This is a double underscore
2302 followed by the string ''\verb|UNLABELED|'' followed by a double underscore.)
2303 The ezPolicy tool automatically adds this organization-level workload type
2304 to a new workload definition (cf Figure~\ref{fig:acmezpolicy}). It can simply be
2305 deleted from the workload definition if no such support is desired. If unlabeled domains
2306 are supported in the policy, then any domain or resource that has no label will implicitly
2307 inherit this label when access control decisions are made. In effect, unlabeled
2308 domains and resources define a new workload type \verb|__UNLABELED__|, which is
2309 confined from any other labeled workload.
2311 Please use now the ``New Org'' button to add the organization workload types
2312 ``A-Bank'', ``B-Bank'', and ``AutoCorp''.
2314 You can refine an organization to differentiate between multiple
2315 department workloads by right-clicking the organization and selecting
2316 \verb|Add Department| (or selecting an organization and pressing
2317 \verb|<CRTL>-a|). Create department workloads ``SecurityUnderwriting'',
2318 and ``MarketAnalysis'' for the ``A-Bank''. The resulting layout of the
2319 tool should be similar to the left panel shown in
2320 Figure~\ref{fig:acmezpolicy}.
2322 \begin{figure}[htb]
2323 \centering
2324 \includegraphics[width=13cm]{figs/acm_ezpolicy_gui.eps}
2325 \caption{Final layout including workload definition and Run-time Exclusion rules.}
2326 \label{fig:acmezpolicy}
2327 \end{figure}
2329 \paragraph{Defining run-time conflicts.} Workloads that shall be
2330 prohibited from running concurrently on the same hypervisor platform
2331 are grouped into ``Run-time Exclusion rules'' on the right panel of
2332 the window. Cautious users should include the \verb|__UNLABELED__|
2333 workload type in all run-time exclusion rules because any workload
2334 could run inside unlabeled domains.
2336 To prevent A-Bank and B-Bank workloads (including their
2337 departmental workloads) from running simultaneously on the same
2338 hypervisor system, select the organization ``A-Bank'' and, while
2339 pressing the \verb|<CTRL>|-key, select the organization ``B-Bank''.
2340 Being cautious, we also prevent unlabeled workloads from running with
2341 any of those workloads by pressing the \verb|<CTRL>|-key and selecting
2342 ``\_\_UNLABELED\_\_''. Now press the button named ``Create run-time exclusion
2343 rule from selection''. A popup window will ask for the name for this run-time
2344 exclusion rule (enter a name or just hit \verb|<ENTER>|). A rule will
2345 appear on the right panel. The name is used as reference only and does
2346 not affect access control decisions.
2348 Please repeat this process to create another run-time exclusion rule
2349 for the department workloads ``A-Bank.SecurityUnderwriting'',
2350 ``A-Bank.MarketAnalysis''. Also add the ``\_\_UNLABELED\_\_''
2351 workload type to this conflict set.
2353 The resulting layout of your window should be similar to
2354 Figure~\ref{fig:acmezpolicy}. Save this workload definition by
2355 selecting ``Save Workload Definition as ...'' in the ``File'' menu.
2356 This workload definition can be later refined if required.
2358 \paragraph{Translating the workload definition into an sHype/Xen access
2359 control policy.} To translate the workload definition into a access
2360 control policy understood by Xen, please select the ``Save as Xen ACM
2361 Security Policy'' in the ``File'' menu. Enter the following policy
2362 name in the popup window: \verb|mytest|. If you are running ezPolicy in
2363 Domain-0, the resulting policy file mytest\_security-policy.xml will
2364 automatically be placed into the right directory (/etc/xen/acm-security/policies/).
2365 If you run the tool on another system, then you need to copy the
2366 resulting policy file into Domain-0 before continuing. See
2367 Section~\ref{subsection:acmnaming} for naming conventions of security
2368 policies.
2370 \begin{scriptsize}
2371 \textbf{Note:} The support for \verb|__UNLABELED__| domains and
2372 resources is meant to help transitioning from an uncontrolled
2373 environment to a workload-protected environment by starting with
2374 unlabeled domains and resources and then step-by-step labeling domains
2375 and resources. Once all workloads are labeled, the \verb|__UNLABELED__|
2376 type can simply be removed from the Domain-0 label or from the policy
2377 through a policy update. Section~\ref{subsection:acmpolicymanagement} will
2378 show how unlabeled domains can be disabled by updating the
2379 \verb|mytest| policy at run-time.
2380 \end{scriptsize}
2382 \subsection{Deploying a WLP Policy}
2383 \label{subsection:acmexampleinstall}
2384 To deploy the workload protection policy we created in
2385 Section~\ref{subsection:acmexamplecreate}, we create a policy
2386 representation (mytest.bin), load it into the Xen
2387 hypervisor, and configure Xen to also load this policy during
2388 reboot.
2390 The following command translates the source policy representation
2391 into a format that can be loaded into Xen with sHype/ACM support,
2392 activates the policy, and configures this policy for future boot
2393 cycles into the boot sequence. Please refer to the \verb|xm|
2394 man page for further details:
2396 \begin{verbatim}
2397 (5) # xm setpolicy ACM mytest
2398 Successfully set the new policy.
2399 Supported security subsystems : ACM
2400 Policy name : mytest
2401 Policy type : ACM
2402 Version of XML policy : 1.0
2403 Policy configuration : loaded, activated for boot
2404 \end{verbatim}
2406 Alternatively, if installing the policy fails (e.g., because it cannot
2407 identify the Xen boot entry), you can manually install the policy in 3
2408 steps a-c.
2410 (\textit{Alternatively to 5 - step a}) Manually copy the policy binary
2411 file into the boot directory:
2413 \begin{scriptsize}
2414 \begin{verbatim}
2415 # cp /etc/xen/acm-security/policies/mytest.bin /boot/mytest.bin
2416 \end{verbatim}
2417 \end{scriptsize}
2419 (\textit{Alternatively to 5 - step b}) Manually add a module line to your
2420 Xen boot entry so that grub loads this policy file during startup:
2422 \begin{scriptsize}
2423 \begin{verbatim}
2424 title XEN Devel with
2425 kernel /xen.gz
2426 module /vmlinuz- root=/dev/sda3 ro console=tty0
2427 module /initrd-
2428 module /mytest.bin
2429 \end{verbatim}
2430 \end{scriptsize}
2432 (\textit{Alternatively to 5 - step c}) Reboot. Xen will choose the
2433 bootstrap label defined in the policy as Domain-0 label during reboot.
2434 After reboot, you can re-label Domain-0 at run-time,
2435 cf Section~\ref{subsection:acmlabeldom0}.
2437 Assuming that command (5) succeeded or you followed the alternative
2438 instructions above, you should see the new policy and label appear
2439 when listing domains:
2441 \begin{scriptsize}
2442 \begin{verbatim}
2443 # xm list --label
2444 Name ID Mem VCPUs State Time(s) Label
2445 Domain-0 0 941 1 r----- 81.5 ACM:mytest:SystemManagement
2446 \end{verbatim}
2447 \end{scriptsize}
2449 If the security label at the end of the line says ``INACTIVE'' then the
2450 security is not enabled. Verify the previous steps. Note: Domain-0 is
2451 assigned a default label (see \verb|bootstrap| policy attribute
2452 explained in Section~\ref{section:acmpolicy}). All other domains must
2453 be explicitly labeled, which we describe in detail below.
2455 \subsection{Labeling Unmanaged User Domains}
2456 \label{subsection:acmexamplelabeldomains}
2458 Unmanaged domains are started in Xen by using a configuration
2459 file. Please refer to Section~\ref{subsection:acmlabelmanageddomains}
2460 if you are using managed domains.
2462 The following configuration file defines \verb|domain1|:
2464 \begin{scriptsize}
2465 \begin{verbatim}
2466 # cat domain1.xm
2467 kernel= "/boot/vmlinuz-"
2468 memory = 128
2469 name = "domain1"
2470 vif = ['']
2471 dhcp = "dhcp"
2472 disk = ['file:/home/xen/dom_fc5/fedora.fc5.img,sda1,w', \
2473 'file:/home/xen/dom_fc5/fedora.fc5.swap,sda2,w']
2474 root = "/dev/sda1 ro xencons=tty"
2475 \end{verbatim}
2476 \end{scriptsize}
2478 Every domain must be associated with a security label before it can start
2479 on sHype/Xen. Otherwise, sHype/Xen would not be able to enforce the policy
2480 consistently. Our \verb|mytest| policy is configured so that Xen
2481 assigns a default label \verb|__UNLABELED__| to domains and resources that
2482 have no label and supports them in a controlled manner. Since neither the domain,
2483 nor the resources are (yet) labeled, this domain can start under the \verb|mytest|
2484 policy:
2486 \begin{scriptsize}
2487 \begin{verbatim}
2488 # xm create domain1.xm
2489 Using config file "./domain1.xm".
2490 Started domain domain1
2492 # xm list --label
2493 Name ID Mem VCPUs State Time(s) Label
2494 domain1 1 128 1 -b---- 0.7 ACM:mytest:__UNLABELED__
2495 Domain-0 0 875 1 r----- 84.6 ACM:mytest:SystemManagement
2496 \end{verbatim}
2497 \end{scriptsize}
2499 Please shutdown domain1 so that we can move it into the protection
2500 domain of workload \verb|A-Bank|.
2502 \begin{scriptsize}
2503 \begin{verbatim}
2504 # xm shutdown domain1
2505 (wait some seconds until the domain has shut down)
2507 #xm list --label
2508 Name ID Mem VCPUs State Time(s) Label
2509 Domain-0 0 875 1 r----- 86.4 ACM:mytest:SystemManagement
2510 \end{verbatim}
2511 \end{scriptsize}
2513 We assume that the processing in domain1 contributes to the \verb|A-Bank| workload.
2514 We explore now how to transition this domain into the ``A-Bank'' workload-protection.
2515 The following command prints all domain labels available in the active policy:
2517 \begin{scriptsize}
2518 \begin{verbatim}
2519 # xm labels
2520 A-Bank
2521 A-Bank.MarketAnalysis
2522 A-Bank.SecurityUnderwriting
2523 AutoCorp
2524 B-Bank
2525 SystemManagement
2526 __UNLABELED__
2527 \end{verbatim}
2528 \end{scriptsize}
2530 Now label \verb|domain1| with the A-Bank label and another \verb|domain2|
2531 with the B-Bank label. Please refer to the xm man page for
2532 further information.
2534 \begin{verbatim}
2535 (6) # xm addlabel A-Bank dom domain1.xm
2536 # xm addlabel B-Bank dom domain2.xm
2537 \end{verbatim}
2539 Let us try to start the domain again:
2541 \begin{scriptsize}
2542 \begin{verbatim}
2543 # xm create domain1.xm
2544 Using config file "./domain1.xm".
2545 Error: VM's access to block device 'file:/home/xen/dom_fc5/fedora.fc5.img' denied
2546 \end{verbatim}
2547 \end{scriptsize}
2549 This error indicates that \verb|domain1|, if started, would not be able to
2550 access its image and swap files because they are not labeled. This
2551 makes sense because to confine workloads, access of domains to
2552 resources must be controlled. Otherwise, domains that are not allowed
2553 to communicate or run simultaneously could share data through storage
2554 resources.
2556 \subsection{Labeling Resources}
2557 \label{subsection:acmexamplelabelresources}
2558 You can use the \verb|xm labels type=res| command to list available
2559 resource labels. Let us assign the A-Bank resource label to the
2560 \verb|domain1| image file representing \verb|/dev/sda1| and to its swap file:
2562 \begin{verbatim}
2563 (7) # xm addlabel A-Bank res \
2564 file:/home/xen/dom_fc5/fedora.fc5.img
2566 # xm addlabel A-Bank res \
2567 file:/home/xen/dom_fc5/fedora.fc5.swap
2568 \end{verbatim}
2570 The following command lists all labeled resources on the system, e.g.,
2571 to lookup or verify the labeling:
2573 \begin{scriptsize}
2574 \begin{verbatim}
2575 # xm resources
2576 file:/home/xen/dom_fc5/fedora.fc5.swap
2577 type: ACM
2578 policy: mytest
2579 label: A-Bank
2580 file:/home/xen/dom_fc5/fedora.fc5.img
2581 type: ACM
2582 policy: mytest
2583 label: A-Bank
2584 \end{verbatim}
2585 \end{scriptsize}
2587 Starting \verb|domain1| will now succeed:
2589 \begin{scriptsize}
2590 \begin{verbatim}
2591 # xm create domain1.xm
2592 Using config file "./domain1.xm".
2593 Started domain domain1
2595 # xm list --label
2596 Name ID Mem VCPUs State Time(s) Label
2597 domain1 3 128 1 -b---- 0.8 ACM:mytest:A-Bank
2598 Domain-0 0 875 1 r----- 90.9 ACM:mytest:SystemManagement
2599 \end{verbatim}
2600 \end{scriptsize}
2602 Currently, if a labeled resource is moved to another location, the
2603 label must first be manually removed, and after the move re-attached
2604 using the xm commands \verb|rmlabel| and \verb|addlabel|
2605 respectively. Please see Section~\ref{section:acmlimitations} for
2606 further details.
2608 \begin{verbatim}
2609 (8) Label the resources of domain2 as B-Bank
2610 but please do not start this domain yet.
2611 \end{verbatim}
2613 \subsection{Testing The Xen Workload Protection}
2614 \label{subsection:acmexampletest}
2616 We are about to demonstrate the sHype/Xen workload protection by verifying
2617 \begin{itemize}
2618 \item that user domains with conflicting workloads cannot run
2619 simultaneously
2620 \item that user domains cannot access resources of workloads other than the
2621 one they are associated with
2622 \item that user domains cannot exchange network packets if they are not
2623 associated with the same workload type (not yet supported in Xen)
2624 \end{itemize}
2626 \paragraph{Test 1: Run-time exclusion rules.} We assume that \verb|domain1|
2627 with the A-Bank label is still running. While \verb|domain1| is running,
2628 the run-time exclusion set of our policy implies that \verb|domain2| cannot
2629 start because the label of \verb|domain1| includes the CHWALL type A-Bank
2630 and the label of \verb|domain2| includes the CHWALL type B-Bank. The
2631 run-time exclusion rule of our policy enforces that A-Bank and
2632 B-Bank cannot run at the same time on the same hypervisor platform.
2633 Once domain1 is stopped, saved, or migrated to another platform,
2634 \verb|domain2| can start. Once \verb|domain2| is started, however,
2635 \verb|domain1| can no longer start or resume on this system. When creating the
2636 Chinese Wall types for the workload labels, the ezPolicy tool policy
2637 translation component ensures that department workloads inherit all the
2638 organization types (and with it any organization exclusions).
2640 \begin{scriptsize}
2641 \begin{verbatim}
2642 # xm list --label
2643 Name ID Mem VCPUs State Time(s) Label
2644 domain1 3 128 1 -b---- 0.8 ACM:mytest:A-Bank
2645 Domain-0 0 875 1 r----- 90.9 ACM:mytest:SystemManagement
2647 # xm create domain2.xm
2648 Using config file "./domain2.xm".
2649 Error: 'Domain in conflict set with running domains'
2651 # xm shutdown domain1
2652 (wait some seconds until domain 1 is shut down)
2654 # xm list --label
2655 Name ID Mem VCPUs State Time(s) Label
2656 Domain-0 0 873 1 r----- 95.3 ACM:mytest:SystemManagement
2658 # xm create domain2.xm
2659 Using config file "./domain2.xm".
2660 Started domain domain2
2662 # xm list --label
2663 Name ID Mem VCPUs State Time(s) Label
2664 domain2 5 164 1 -b---- 0.3 ACM:mytest:B-Bank
2665 Domain-0 0 839 1 r----- 96.4 ACM:mytest:SystemManagement
2667 # xm create domain1.xm
2668 Using config file "domain1.xm".
2669 Error: 'Domain in conflict with running domains'
2671 # xm shutdown domain2
2672 # xm list --label
2673 Name ID Mem VCPUs State Time(s) Label
2674 Domain-0 0 839 1 r----- 97.8 ACM:mytest:SystemManagement
2675 \end{verbatim}
2676 \end{scriptsize}
2678 You can verify that domains with AutoCorp label can run together with
2679 domains labeled A-Bank or B-Bank.
2681 \paragraph{Test2: Resource access.} In this test, we will re-label the
2682 swap file for \verb|domain1| with the \verb|B-Bank| resource label. In a
2683 real environment, the swap file must be sanitized (scrubbed/zeroed) before
2684 it is reassigned to prevent data leaks from the A-Bank to the B-Bank workload
2685 through the swap file.
2687 We expect that \verb|domain1| will no longer start because it cannot access
2688 this resource. This test checks the sharing abilities of domains, which are
2689 defined by the Simple Type Enforcement Policy component.
2691 \begin{scriptsize}
2692 \begin{verbatim}
2693 # xm rmlabel res file:/home/xen/dom_fc5/fedora.fc5.swap
2695 # xm addlabel B-Bank res file:/home/xen/dom_fc5/fedora.fc5.swap
2697 # xm resources
2698 file:/home/xen/dom_fc5/fedora.fc5.swap
2699 type: ACM
2700 policy: mytest
2701 label: B-Bank
2702 file:/home/xen/dom_fc5/fedora.fc5.img
2703 type: ACM
2704 policy: mytest
2705 label: A-Bank
2707 # xm create domain1.xm
2708 Using config file "./domain1.xm".
2709 Error:
2710 VM's access to block device 'file:/home/xen/dom_fc5/fedora.fc5.swap' denied
2711 \end{verbatim}
2712 \end{scriptsize}
2714 The resource authorization checks are performed before the domain is actually started
2715 so that failures during the startup are prevented. A domain is only started if all
2716 the resources specified in its configuration are accessible.
2718 \paragraph{Test 3: Communication.} In this test we would verify that
2719 two domains with labels A-Bank and B-Bank cannot exchange network packets
2720 by using the 'ping' connectivity test. It is also related to the STE
2721 policy. {\bf Note:} sHype/Xen does control direct communication between
2722 domains. However, domains associated with different workloads can
2723 currently still communicate through the Domain-0 virtual network. We
2724 are working on the sHype/ACM controls for local and remote network
2725 traffic through Domain-0. Please monitor the xen-devel mailing list
2726 for updated information.
2729 \subsection{Labeling Domain-0 --or-- Restricting System Authorization}
2730 \label{subsection:acmlabeldom0}
2731 The major use case for explicitly labeling or relabeling Domain-0 is to restrict
2732 or extend which workload types can run on a virtualized Xen system. This enables
2733 flexible partitioning of the physical infrastructure as well as the workloads
2734 running on it in a multi-platform environment.
2736 In case no Domain-0 label is explicitly stated, we automatically assigned Domain-0
2737 the \verb|SystemManagement| label, which includes all STE (workload) types that
2738 are known to the policy. In effect, the Domain-0 label authorizes the Xen system
2739 to run only those workload types, whose STE types are included in the Domain-0
2740 label. Hence, choosing the \verb|SystemManagement| label for Domain-0 permits any
2741 labeled domain to run. Resetting the label for Domain-0 at boot or run-time to
2742 a label with a subset of the known STE workload types restricts which user domains
2743 can run on this system. If Domain-0 is relabeled at run-time, then the new label
2744 must at least include all STE types of those domains that are currently running.
2745 The operation fails otherwise. This requirement ensures that the system remains
2746 in a valid security configuration after re-labelling.
2748 Restricting the Domain-0 authorization through the label creates a flexible
2749 policy-driven way to strongly partition the physical infrastructure and the
2750 workloads running on it. This partitioning will be automatically enforced during
2751 migration, start, or resume of domains and simplifies the security management
2752 considerably. Strongly competing workloads can be forced to run on separate physical
2753 infrastructure and become less depend on the domain isolation capabilities
2754 of the hypervisor.
2756 First, we relabel the swap image back to A-Bank and then start up domain1:
2757 \begin{scriptsize}
2758 \begin{verbatim}
2759 # xm rmlabel res file:/home/xen/dom_fc5/fedora.fc5.swap
2761 # xm addlabel A-Bank res file:/home/xen/dom_fc5/fedora.fc5.swap
2763 # xm create domain1.xm
2764 Using config file "./domain1.xm".
2765 Started domain domain1
2767 # xm list --label
2768 Name ID Mem VCPUs State Time(s) Label
2769 domain1 7 128 1 -b---- 0.7 ACM:mytest:A-Bank
2770 Domain-0 0 839 1 r----- 103.1 ACM:mytest:SystemManagement
2771 \end{verbatim}
2772 \end{scriptsize}
2774 The following command will restrict the Xen system to only run STE types
2775 included in the A-Bank label.
2777 \begin{scriptsize}
2778 \begin{verbatim}
2779 # xm addlabel A-Bank mgt Domain-0
2780 Successfully set the label of domain 'Domain-0' to 'A-Bank'.
2782 # xm list --label
2783 Name ID Mem VCPUs State Time(s) Label
2784 Domain-0 0 839 1 r----- 103.7 ACM:mytest:A-Bank
2785 domain1 7 128 1 -b---- 0.7 ACM:mytest:A-Bank
2787 \end{verbatim}
2788 \end{scriptsize}
2790 In our example policy in Figure~\ref{fig:acmxmlfileb}, this means that
2791 only \verb|A-Bank| domains and workloads (types) can run after the
2792 successful completion of this command because the \verb|A-Bank| label
2793 includes only a single STE type, namely \verb|A-Bank|. This command
2794 fails if any running domain has an STE type in its label that is not
2795 included in the A-Bank label.
2797 If we now label a domain3 with AutoCorp, it cannot start because Domain-0 is
2798 no longer authorized to run the workload type \verb|AutoCorp|.
2799 \begin{scriptsize}
2800 \begin{verbatim}
2801 # xm addlabel AutoCorp dom domain3.xm
2802 (remember to label its resources, too)
2804 # xm create domain3.xm
2805 Using config file "./domain3.xm".
2806 Error: VM is not authorized to run.
2808 # xm list --label
2809 Name ID Mem VCPUs State Time(s) Label
2810 Domain-0 0 839 1 r----- 104.7 ACM:mytest:A-Bank
2811 domain1 7 128 1 -b---- 0.7 ACM:mytest:A-Bank
2812 \end{verbatim}
2813 \end{scriptsize}
2815 At this point, unlabeled domains cannot start either. Let domain4.xm
2816 describe an unlabeled domain, then trying to start domain4
2817 will fail:
2818 \begin{scriptsize}
2819 \begin{verbatim}
2820 # xm getlabel dom domain4.xm
2821 Error: 'Domain not labeled'
2823 # xm create domain4.xm
2824 Using config file "./domain4.xm".
2825 Error: VM is not authorized to run.
2826 \end{verbatim}
2827 \end{scriptsize}
2829 Relabeling Domain-0 with the SystemManagement label will enable domain3 to start.
2830 \begin{scriptsize}
2831 \begin{verbatim}
2832 # xm addlabel SystemManagement mgt Domain-0
2833 Successfully set the label of domain 'Domain-0' to 'SystemManagement'.
2835 # xm list --label
2836 Name ID Mem VCPUs State Time(s) Label
2837 domain1 7 128 1 -b---- 0.8 ACM:mytest:A-Bank
2838 Domain-0 0 839 1 r----- 106.6 ACM:mytest:SystemManagement
2840 # xm create domain3.xm
2841 Using config file "./domain3.xm".
2842 Started domain domain3
2844 # xm list --label
2845 Name ID Mem VCPUs State Time(s) Label
2846 domain1 7 128 1 -b---- 0.8 ACM:mytest:A-Bank
2847 domain3 8 164 1 -b---- 0.3 ACM:mytest:AutoCorp
2848 Domain-0 0 711 1 r----- 107.6 ACM:mytest:SystemManagement
2849 \end{verbatim}
2850 \end{scriptsize}
2853 \subsection{Labeling Managed User Domains}
2854 \label{subsection:acmlabelmanageddomains}
2856 Xend has been extended with functionality to manage domains along with their
2857 configuration information. Such domains are configured and started via Xen-API
2858 calls. Since managed domains do not have an associated xm configuration file,
2859 the existing \verb|addlabel| command, which adds the security label into a
2860 domain's configuration file, will not work for such managed domains.
2862 Therefore, we have extended the \verb|xm addlabel| and \verb|xm rmlabel|
2863 subcommands to enable adding security labels to and removing security
2864 labels from managed domain configurations. The following example shows how
2865 the \verb|A-Bank| label can be assigned to the xend-managed
2866 domain configuration of \verb|domain1|. Removing labels from managed user
2867 domain configurations works similarly.
2869 Below, we show a dormant configuration of the managed domain1
2870 with ID \verb|"-1"| and state \verb|"-----"| before labeling:
2871 \begin{scriptsize}
2872 \begin{verbatim}
2873 # xm list --label
2874 Name ID Mem VCPUs State Time(s) Label
2875 domain1 -1 128 1 ------ 0.0 ACM:mytest:__UNLABELED__
2876 Domain-0 0 711 1 r----- 128.4 ACM:mytest:SystemManagement
2877 \end{verbatim}
2878 \end{scriptsize}
2880 Now we label the managed domain:
2881 \begin{scriptsize}
2882 \begin{verbatim}
2883 # xm addlabel A-Bank mgt domain1
2884 Successfully set the label of the dormant domain 'domain1' to 'A-Bank'.
2885 \end{verbatim}
2886 \end{scriptsize}
2888 After labeling, you can see that the security label is part of the
2889 domain configuration:
2890 \begin{scriptsize}
2891 \begin{verbatim}
2892 # xm list --label
2893 Name ID Mem VCPUs State Time(s) Label
2894 domain1 -1 128 1 ------ 0.0 ACM:mytest:A-Bank
2895 Domain-0 0 711 1 r----- 129.7 ACM:mytest:SystemManagement
2896 \end{verbatim}
2897 \end{scriptsize}
2899 This command extension does not support relabeling of individual running user domains
2900 for several reasons. For one, because of the difficulty to revoke resources
2901 in cases where a running domain's new label does not permit access to resources
2902 that were accessible under the old label. Another reason is that changing the
2903 label of a single domain of a workload is rarely a good choice and will affect
2904 the workload isolation properties of the overall workload.
2906 However, the name and contents of the label associated with running domains can
2907 be indirectly changed through a global policy change, which will update the whole
2908 workload consistently (domains and resources), cf.
2909 Section~\ref{subsection:acmpolicymanagement}.
2911 \section{Xen Access Control Policy}
2912 \label{section:acmpolicy}
2914 This section describes the sHype/Xen access control policy in detail.
2915 It gives enough information to enable the reader to write custom
2916 access control policies and to use the available Xen policy tools. The
2917 policy language is expressive enough to specify most symmetric access
2918 relationships between domains and resources efficiently.
2920 The Xen access control policy consists of two policy components. The
2921 first component, called Simple Type Enforcement (STE) policy, controls
2922 the sharing between running domains, i.e., communication or access to
2923 shared resources. The second component, called Chinese Wall (CHWALL)
2924 policy, controls which domains can run simultaneously on the same
2925 virtualized platform. The CHWALL and STE policy components complement
2926 each other. The XML policy file includes all information
2927 needed by Xen to enforce those policies.
2929 Figures~\ref{fig:acmxmlfilea} and \ref{fig:acmxmlfileb} show the fully
2930 functional but very simple example Xen security policy that is created
2931 by ezPolicy as shown in Figure~\ref{fig:acmezpolicy}. The policy can
2932 distinguish the 6 workload types shown in lines 11-17 in
2933 Fig.~\ref{fig:acmxmlfilea}. The whole XML Security Policy consists of
2934 four parts:
2935 \begin{enumerate}
2936 \item Policy header including the policy name
2937 \item Simple Type Enforcement block
2938 \item Chinese Wall Policy block
2939 \item Label definition block
2940 \end{enumerate}
2942 \begin{figure}
2943 \begin{scriptsize}
2944 \begin{verbatim}
2945 01 <?xml version="1.0" ?>
2946 02 <!-- Auto-generated by ezPolicy -->
2947 03 <SecurityPolicyDefinition ...">
2948 04 <PolicyHeader>
2949 05 <PolicyName>mytest</PolicyName>
2950 06 <Date>Mon Nov 19 22:51:56 2007</Date>
2951 07 <Version>1.0</Version>
2952 08 </PolicyHeader>
2953 09 <SimpleTypeEnforcement>
2954 10 <SimpleTypeEnforcementTypes>
2955 11 <Type>SystemManagement</Type>
2956 12 <Type>__UNLABELED__</Type>
2957 13 <Type>A-Bank</Type>
2958 14 <Type>A-Bank.SecurityUnderwriting</Type>
2959 15 <Type>A-Bank.MarketAnalysis</Type>
2960 16 <Type>B-Bank</Type>
2961 17 <Type>AutoCorp</Type>
2962 18 </SimpleTypeEnforcementTypes>
2963 19 </SimpleTypeEnforcement>
2964 20 <ChineseWall priority="PrimaryPolicyComponent">
2965 21 <ChineseWallTypes>
2966 22 <Type>SystemManagement</Type>
2967 23 <Type>__UNLABELED__</Type>
2968 24 <Type>A-Bank</Type>
2969 25 <Type>A-Bank.SecurityUnderwriting</Type>
2970 26 <Type>A-Bank.MarketAnalysis</Type>
2971 27 <Type>B-Bank</Type>
2972 28 <Type>AutoCorp</Type>
2973 29 </ChineseWallTypes>
2974 30 <ConflictSets>
2975 31 <Conflict name="RER">
2976 32 <Type>A-Bank</Type>
2977 33 <Type>B-Bank</Type>
2978 34 <Type>__UNLABELED__</Type>
2979 35 </Conflict>
2980 36 <Conflict name="RER">
2981 37 <Type>A-Bank.MarketAnalysis</Type>
2982 38 <Type>A-Bank.SecurityUnderwriting</Type>
2983 39 <Type>__UNLABELED__</Type>
2984 40 </Conflict>
2985 41 </ConflictSets>
2986 42 </ChineseWall>
2987 \end{verbatim}
2988 \end{scriptsize}
2989 \caption{Example XML security policy file -- Part I: Types and Rules Definition.}
2990 \label{fig:acmxmlfilea}
2991 \end{figure}
2993 \subsection{Policy Header and Policy Name}
2994 \label{subsection:acmnaming}
2995 Lines 1-2 (cf Figure~\ref{fig:acmxmlfilea}) include the usual XML
2996 header. The security policy definition starts in Line 3 and refers to
2997 the policy schema. The XML-Schema definition for the Xen policy can be
2998 found in the file
2999 \textit{/etc/xen/acm-security/policies/security-policy.xsd}. Examples
3000 for security policies can be found in the example subdirectory. The
3001 acm-security directory is only installed if ACM security is configured
3002 during installation (cf Section~\ref{subsection:acmexampleconfigure}).
3004 The \verb|Policy Header| spans lines 4-8. It includes a date field and
3005 defines the policy name \verb|mytest| as well
3006 as the version of the XML. It can also include optional fields that are
3007 not shown and are for future use (see schema definition).
3009 The policy name serves two purposes: First, it provides a unique name
3010 for the security policy. This name is also exported by the Xen
3011 hypervisor to the Xen management tools in order to ensure that both
3012 the Xen hypervisor and Domain-0 enforce the same policy.
3013 We plan to extend the policy name with a
3014 digital fingerprint of the policy contents to better protect this
3015 correlation. Second, it implicitly points the xm tools to the
3016 location where the XML policy file is stored on the Xen system.
3017 Replacing the colons in the policy name by slashes yields the local
3018 path to the policy file starting from the global policy directory
3019 \verb|/etc/xen/acm-security/policies|. The last part of the policy
3020 name is the prefix for the XML policy file name, completed by
3021 \verb|-security_policy.xml|. Our example policy with the name
3022 \verb|mytest| can be found in the XML policy file named
3023 \verb|mytest-security_policy.xml| that is stored under the global
3024 policy directory. Another, preinstalled example policy named
3025 \verb|example.test| can be found in the \verb|test-security_policy.xml|
3026 under \verb|/etc/xen/acm-security/policies/example|.
3028 \subsection{Simple Type Enforcement Policy Component}
3030 The Simple Type Enforcement (STE) policy controls which domains can
3031 communicate or share resources. This way, Xen can enforce confinement
3032 of workload types by confining the domains running those workload
3033 types and their resources. The mandatory access control framework
3034 enforces its policy when
3035 domains access intended communication or cooperation means (shared
3036 memory, events, shared resources such as block devices). It builds on
3037 top of the core hypervisor isolation, which restricts the ways of
3038 inter-communication to those intended means. STE does not protect or
3039 intend to protect from covert channels in the hypervisor or hardware;
3040 this is an orthogonal problem that can be mitigated by using the
3041 Run-time Exclusion rules described above or by fixing the problem leading
3042 to those covert channels in the core hypervisor or hardware platform.
3044 Xen controls sharing between domains on the resource and domain level
3045 because this is the abstraction the hypervisor and its management
3046 understand naturally. While this is coarse-grained, it is also very
3047 reliable and robust and it requires minimal changes to implement
3048 mandatory access controls in the hypervisor. It enables platform- and
3049 operating system-independent policies as part of a layered security
3050 approach.
3052 Lines 11-17 (cf Figure~\ref{fig:acmxmlfilea}) define the Simple Type
3053 Enforcement policy component. Essentially, they define the workload
3054 type names \verb|SystemManagement|, \verb|A-Bank|,
3055 \verb|AutoCorp| etc. that are available in the STE policy component. The
3056 policy rules are implicit: Xen permits two domains to communicate with
3057 each other if and only if their security labels have at least one STE type in
3058 common. Similarly, Xen permits a user domain to access a
3059 resource if and only if the labels of the domain and the resource
3060 have at least one STE workload type in common.
3062 \subsection{Chinese Wall Policy Component}
3064 The Chinese Wall security policy interpretation of sHype enables users
3065 to prevent certain workloads from running simultaneously on the same
3066 hypervisor platform. Run-time Exclusion rules (RER), also called
3067 Conflict Sets or Anti-Collocation rules, define a set of workload types
3068 that are not permitted to run simultaneously on the same virtualized
3069 platform. Of all the workloads specified in a Run-time
3070 Exclusion rule, at most one type can run on the same hypervisor
3071 platform at a time. Run-time Exclusion Rules implement a less
3072 rigorous variant of the original Chinese Wall security component. They
3073 do not implement the *-property of the policy, which would require to
3074 restrict also types that are not part of an exclusion rule once they
3075 are running together with a type in an exclusion rule
3076 ( provides more information
3077 on the original Chinese Wall policy).
3079 Xen considers the \verb|ChineseWallTypes| part of the label for the
3080 enforcement of the Run-time Exclusion rules. It is illegal to define
3081 labels including conflicting Chinese Wall types.
3083 Lines 20-41 (cf Figure~\ref{fig:acmxmlfilea}) define the Chinese Wall
3084 policy component. Lines 22-28 define the known Chinese Wall types,
3085 which coincide here with the STE types defined above. This usually
3086 holds if the criteria for sharing among domains and sharing of the
3087 hardware platform are the same. Lines 30-41 define one Run-time
3088 Exclusion rules, the first of which is depicted below:
3090 \begin{scriptsize}
3091 \begin{verbatim}
3092 31 <Conflict name="RER">
3093 32 <Type>A-Bank</Type>
3094 33 <Type>B-Bank</Type>
3095 34 <Type>__UNLABELED__</Type>
3096 35 </Conflict>
3097 \end{verbatim}
3098 \end{scriptsize}
3100 Based on this rule, Xen enforces that only one of the types
3101 \verb|A-Bank|, \verb|B-Bank|, or \verb|__UNLABELED__| will run
3102 on a single hypervisor platform at a time. For example, once a domain assigned a
3103 \verb|A-Bank| workload type is started, domains with the
3104 \verb|B-Bank| type or unlabeled domains will be denied to start.
3105 When the former domain stops and no other domains with the \verb|A-Bank|
3106 type are running, then domains with the \verb|B-Bank| type or unlabeled domains
3107 can start.
3109 Xen maintains reference counts on each running workload type to keep
3110 track of which workload types are running. Every time a domain starts
3111 or resumes, the reference count on those Chinese Wall types that are
3112 referenced in the domain's label are incremented. Every time a domain
3113 is destroyed or saved, the reference counts of its Chinese Wall types
3114 are decremented. sHype in Xen fully supports migration and live-migration,
3115 which is subject to access control the same way as saving a domain on
3116 the source platform and resuming it on the destination platform.
3118 Here are some reasons why users might want to restrict workloads or domains
3119 from sharing the system hardware simultaneously:
3121 \begin{itemize}
3122 \item Imperfect resource management or control might enable a compromised
3123 user domain to starve other domains and the workload running in them.
3124 \item Redundant user domains might run the same workload to increase
3125 availability; such domains should not run on the same hardware to
3126 avoid single points of failure.
3127 \item Imperfect Xen core domain isolation might enable two rogue
3128 domains running different workload types to use unintended and
3129 unknown ways (covert channels) to exchange some bits of information.
3130 This way, they bypass the policed Xen access control mechanisms. Such
3131 imperfections cannot be completely eliminated and are a result of
3132 trade-offs between security and other design requirements. For a
3133 simple example of a covert channel see
3134 Such covert channels
3135 exist also between workloads running on different platforms if they
3136 are connected through networks. The Xen Chinese Wall policy provides
3137 an approximated ``air-gap'' between selected workload types.
3138 \end{itemize}
3140 \subsection{Security Labels}
3142 To enable Xen to associate domains with workload types running in
3143 them, each domain is assigned a security label that includes the
3144 workload types of the domain.
3146 \begin{figure}[htb]
3147 \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}l|l}
3148 \begin{minipage}{0.475\textwidth}
3149 \begin{tiny}
3150 \begin{verbatim}
3151 <SecurityLabelTemplate>
3152 <SubjectLabels bootstrap="SystemManagement">
3153 <VirtualMachineLabel>
3154 <Name>SystemManagement</Name>
3155 <SimpleTypeEnforcementTypes>
3156 <Type>SystemManagement</Type>
3157 <Type>__UNLABELED__</Type>
3158 <Type>A-Bank</Type>
3159 <Type>A-Bank.SecurityUnderwriting</Type>
3160 <Type>A-Bank.MarketAnalysis</Type>
3161 <Type>B-Bank</Type>
3162 <Type>AutoCorp</Type>
3163 </SimpleTypeEnforcementTypes>
3164 <ChineseWallTypes>
3165 <Type>SystemManagement</Type>
3166 </ChineseWallTypes>
3167 </VirtualMachineLabel>
3168 <VirtualMachineLabel>
3169 <Name>__UNLABELED__</Name>
3170 <SimpleTypeEnforcementTypes>
3171 <Type>__UNLABELED__</Type>
3172 </SimpleTypeEnforcementTypes>
3173 <ChineseWallTypes>
3174 <Type>__UNLABELED__</Type>
3175 </ChineseWallTypes>
3176 </VirtualMachineLabel>
3177 <VirtualMachineLabel>
3178 <Name>A-Bank</Name>
3179 <SimpleTypeEnforcementTypes>
3180 <Type>A-Bank</Type>
3181 </SimpleTypeEnforcementTypes>
3182 <ChineseWallTypes>
3183 <Type>A-Bank</Type>
3184 </ChineseWallTypes>
3185 </VirtualMachineLabel>
3186 <VirtualMachineLabel>
3187 <Name>A-Bank.SecurityUnderwriting</Name>
3188 <SimpleTypeEnforcementTypes>
3189 <Type>A-Bank.SecurityUnderwriting</Type>
3190 </SimpleTypeEnforcementTypes>
3191 <ChineseWallTypes>
3192 <Type>A-Bank</Type>
3193 <Type>A-Bank.SecurityUnderwriting</Type>
3194 </ChineseWallTypes>
3195 </VirtualMachineLabel>
3196 <VirtualMachineLabel>
3197 <Name>A-Bank.MarketAnalysis</Name>
3198 <SimpleTypeEnforcementTypes>
3199 <Type>A-Bank.MarketAnalysis</Type>
3200 </SimpleTypeEnforcementTypes>
3201 <ChineseWallTypes>
3202 <Type>A-Bank</Type>
3203 <Type>A-Bank.MarketAnalysis</Type>
3204 </ChineseWallTypes>
3205 </VirtualMachineLabel>
3206 <VirtualMachineLabel>
3207 <Name>B-Bank</Name>
3208 <SimpleTypeEnforcementTypes>
3209 <Type>B-Bank</Type>
3210 </SimpleTypeEnforcementTypes>
3211 <ChineseWallTypes>
3212 <Type>B-Bank</Type>
3213 </ChineseWallTypes>
3214 </VirtualMachineLabel>
3215 \end{verbatim}
3216 \end{tiny}
3217 \end{minipage} &
3218 \begin{minipage}{0.475\textwidth}
3219 \begin{tiny}
3220 \begin{verbatim}
3221 <VirtualMachineLabel>
3222 <Name>AutoCorp</Name>
3223 <SimpleTypeEnforcementTypes>
3224 <Type>AutoCorp</Type>
3225 </SimpleTypeEnforcementTypes>
3226 <ChineseWallTypes>
3227 <Type>AutoCorp</Type>
3228 </ChineseWallTypes>
3229 </VirtualMachineLabel>
3230 </SubjectLabels>
3231 <ObjectLabels>
3232 <ResourceLabel>
3233 <Name>SystemManagement</Name>
3234 <SimpleTypeEnforcementTypes>
3235 <Type>SystemManagement</Type>
3236 </SimpleTypeEnforcementTypes>
3237 </ResourceLabel>
3238 <ResourceLabel>
3239 <Name>__UNLABELED__</Name>
3240 <SimpleTypeEnforcementTypes>
3241 <Type>__UNLABELED__</Type>
3242 </SimpleTypeEnforcementTypes>
3243 </ResourceLabel>
3244 <ResourceLabel>
3245 <Name>A-Bank</Name>
3246 <SimpleTypeEnforcementTypes>
3247 <Type>A-Bank</Type>
3248 </SimpleTypeEnforcementTypes>
3249 </ResourceLabel>
3250 <ResourceLabel>
3251 <Name>A-Bank.SecurityUnderwriting</Name>
3252 <SimpleTypeEnforcementTypes>
3253 <Type>A-Bank.SecurityUnderwriting</Type>
3254 </SimpleTypeEnforcementTypes>
3255 </ResourceLabel>
3256 <ResourceLabel>
3257 <Name>A-Bank.MarketAnalysis</Name>
3258 <SimpleTypeEnforcementTypes>
3259 <Type>A-Bank.MarketAnalysis</Type>
3260 </SimpleTypeEnforcementTypes>
3261 </ResourceLabel>
3262 <ResourceLabel>
3263 <Name>B-Bank</Name>
3264 <SimpleTypeEnforcementTypes>
3265 <Type>B-Bank</Type>
3266 </SimpleTypeEnforcementTypes>
3267 </ResourceLabel>
3268 <ResourceLabel>
3269 <Name>AutoCorp</Name>
3270 <SimpleTypeEnforcementTypes>
3271 <Type>AutoCorp</Type>
3272 </SimpleTypeEnforcementTypes>
3273 </ResourceLabel>
3274 </ObjectLabels>
3275 </SecurityLabelTemplate>
3276 </SecurityPolicyDefinition>
3285 \end{verbatim}
3286 \end{tiny}
3287 \end{minipage}
3288 \end{tabular*}
3289 \caption{Example XML security policy file -- Part II: Label Definition.}
3290 \label{fig:acmxmlfileb}
3291 \end{figure}
3292 % DO NOT MODIFY WHITESPACE ABOVE, it balances the columns
3293 The \verb|SecurityLabelTemplate| (cf Figure~\ref{fig:acmxmlfileb}) defines
3294 the security labels that can be associated with domains and resources when
3295 this policy is active (use the \verb|xm labels type=any| command described in
3296 Section~\ref{subsection:acmexamplelabeldomains} to list all available labels).
3298 The domain labels include
3299 Chinese Wall types while resource labels do not include Chinese Wall types.
3300 The \verb|SubjectLabels| policy section defines the labels that can be
3301 assigned to domains. The VM label
3302 \verb|A-Bank.SecurityUnderwriting| in Figure~\ref{fig:acmxmlfileb})
3303 associates the domain that carries it with the workload STE type
3304 \verb|A-Bank.SecurityUnderwriting| and with the CHWALL types \verb|A-Bank|
3305 and \verb|A-Bank.SecurityUnderwriting|. The ezPolicy tool
3306 assumes that any department workload will inherit any conflict set that
3307 is specified for its organization, i.e., if \verb|B-Bank| is running, not
3308 only \verb|A-Bank| but also all its departmental workloads are prevented
3309 from running by this first run-time exclusion set. The separation of STE
3310 and CHWALL types in the label definition ensures that
3311 all departmental workloads are isolated from each other and from their generic
3312 organization workloads, while they are sharing CHWALL types to
3313 simplify the formulation of run-time exclusion sets.
3315 The \verb|bootstrap| attribute of the \verb|<SubjectLabels>| XML node
3316 in our example policy shown in Figure~\ref{fig:acmxmlfileb} names
3317 the label \verb|SystemManagement| as the label that Xen will assign
3318 to Domain-0 at boot time (if this policy is installed as boot policy). The
3319 label of Domain-0 can be persistently changed at run-time with the
3320 \verb|addlabel| command, which adds an overriding option to the grub.conf
3321 boot entry (cf Section~\ref{subsection:acmlabeldom0}).
3322 All user domains are assigned labels according to their domain configuration
3323 (see Section~\ref{subsection:acmexamplelabeldomains} for examples of
3324 how to label domains).
3326 The \verb|ObjectLabels| depicted in Figure~\ref{fig:acmxmlfileb} can be
3327 assigned to resources when this policy is active.
3329 In general, user domains should be assigned labels that have only a
3330 single SimpleTypeEnforcement workload type. This way, workloads remain
3331 confined even if user domains become rogue. Any domain that is
3332 assigned a label with multiple STE types must be trusted to keep
3333 information belonging to the different STE types separate (confined).
3334 For example, Domain-0 is assigned the bootstrap label
3335 \verb|SystemManagement|, which includes all existing STE types.
3336 Therefore, Domain-0 must take care not to enable unauthorized
3337 information flow (eg. through block devices or virtual networking)
3338 between domains or resources that are assigned different STE types.
3340 Security administrators simply use the name of a label (specified in
3341 the \verb|<Name>| field) to associate a label with a domain (cf.
3342 Section~\ref{subsection:acmexamplelabeldomains}). The types inside the
3343 label are used by the Xen access control enforcement. While the name
3344 can be arbitrarily chosen (as long as it is unique), it is advisable
3345 to choose the label name in accordance to the security types included.
3346 Similarly, the STE and CHWALL types should be named according to the
3347 workloads they represent. While the XML representation of the label
3348 in the above example seems unnecessary flexible, labels in general
3349 must be able to include multiple types.
3351 We assume in the following example, that \verb|A-Bank.SecurityUnderwriting| and
3352 \verb|A-Bank.MarketAnalysis| workloads use virtual disks that are provided
3353 by a virtual I/O domain hosting a physical storage device and carrying
3354 the following label:
3356 \begin{scriptsize}
3357 \begin{verbatim}
3358 <VirtualMachineLabel>
3359 <Name>VIOServer</Name>
3360 <SimpleTypeEnforcementTypes>
3361 <Type>A-Bank</Type>
3362 <Type>A-Bank.SecurityUnderwriting</Type>
3363 <Type>A-Bank.MarketAnalysis</Type>
3364 <Type>VIOServer</Type>
3365 </SimpleTypeEnforcementTypes>
3366 <ChineseWallTypes>
3367 <Type>VIOServer</Type>
3368 </ChineseWallTypes>
3369 </VirtualMachineLabel>
3370 \end{verbatim}
3371 \end{scriptsize}
3373 This Virtual I/O domain (VIO) exports its virtualized disks by
3374 communicating to all domains labeled with the
3375 \verb|A-Bank.SecurityUnderwriting|, the \verb|A-Bank|, or the
3376 \verb|A-Bank.MarketAnalysis| label. This requires the
3377 VIO domain to carry those STE types. In addition, this label includes a
3378 new \verb|VIOServer| type that can be used to restrict direct access to the
3379 physical storage resource to the VIODomain.
3381 In this example, the confinement of these A-Bank workloads depends on the
3382 VIO domain that must keep the data of those different workloads separate.
3383 The virtual disks are labeled as well to keep track of their assignments
3384 to workload types (see Section~\ref{subsection:acmexamplelabelresources}
3385 for labeling resources) and enforcement functions inside the VIO
3386 domain must ensure that the labels of the domain mounting a virtual
3387 disk and the virtual disk label share a common STE type. The VIO label
3388 carrying its own VIOServer CHWALL type introduces the flexibility to
3389 permit the trusted VIO server to run together with \verb|A-Bank.SecurityUnderwriting|
3390 or \verb|A-Bank.MarketAnalysis| workloads.
3392 Alternatively, a system that has two hard-drives does not need a VIO
3393 domain but can directly assign one hardware storage device to each of
3394 the workloads if the platform offers an IO-MMU, cf
3395 Section~\ref{s:ddsecurity}. Sharing hardware through virtualized devices
3396 is a trade-off between the amount of trusted code (size of the trusted
3397 computing base) and the amount of acceptable over-provisioning. This
3398 holds both for peripherals and for system platforms.
3401 \subsection{Managing sHype/Xen Security Policies at Run-time}
3402 \label{subsection:acmpolicymanagement}
3404 \subsubsection{Removing the sHype/Xen Security Policy}
3405 When resetting the policy, no labeled domains can be running.
3406 Please stop or shutdown all running labeled domains. Then you can reset
3407 the policy to the default policy using the \verb|resetpolicy| command:
3409 \begin{scriptsize}
3410 \begin{verbatim}
3411 # xm getpolicy
3412 Supported security subsystems : ACM
3413 Policy name : mytest
3414 Policy type : ACM
3415 Version of XML policy : 1.0
3416 Policy configuration : loaded, activated for boot
3418 # xm resetpolicy
3419 Successfully reset the system's policy.
3421 # xm getpolicy
3422 Supported security subsystems : ACM
3423 Policy name : DEFAULT
3424 Policy type : ACM
3425 Version of XML policy : 1.0
3426 Policy configuration : loaded
3428 # xm resources
3429 file:/home/xen/dom_fc5/fedora.fc5.swap
3430 type: INV_ACM
3431 policy: mytest
3432 label: A-Bank
3433 file:/home/xen/dom_fc5/fedora.fc5.img
3434 type: INV_ACM
3435 policy: mytest
3436 label: A-Bank
3437 \end{verbatim}
3438 \end{scriptsize}
3440 As the \verb|xm resources| output shows, all resource labels have
3441 invalidated type information but their semantics remain associated
3442 with the resources so that they can later on either be relabeled
3443 with semantically equivalent labels or sanitized and reused
3444 (storage resources).
3446 At this point, the system is in the same initial state as after
3447 configuring XSM and sHype/ACM and rebooting the system without
3448 a specific policy. No user domains can run.
3450 \subsubsection{Changing to a Different sHype/Xen Security Policy}
3451 The easiest way to change to a different, unrelated policy is to reset the system
3452 policy and then set the new policy. Please consider that the existing
3453 domain and resource labels become invalid at this point. Please refer
3454 to the next section for an example of how to seamlessly update an
3455 active policy at run-time without invalidating labels.
3457 \begin{scriptsize}
3458 \begin{verbatim}
3459 # xm resetpolicy
3460 Successfully reset the system's policy.
3462 # xm setpolicy ACM example.test
3463 Successfully set the new policy.
3464 Supported security subsystems : ACM
3465 Policy name : example.test
3466 Policy type : ACM
3467 Version of XML policy : 1.0
3468 Policy configuration : loaded, activated for boot
3470 # xm labels
3471 CocaCola
3472 PepsiCo
3473 SystemManagement
3474 VIO
3475 # xm list --label
3476 Name ID Mem VCPUs State Time(s) Label
3477 Domain-0 0 873 1 r----- 56.3 ACM:example.test:SystemManagement
3479 # xm resetpolicy
3480 Successfully reset the system's policy.
3482 # xm getpolicy
3483 Supported security subsystems : ACM
3484 Policy name : DEFAULT
3485 Policy type : ACM
3486 Version of XML policy : 1.0
3487 Policy configuration : loaded
3489 # xm list --label
3490 Name ID Mem VCPUs State Time(s) Label
3491 Domain-0 0 873 1 r----- 57.2 ACM:DEFAULT:SystemManagement
3493 # xm setpolicy ACM mytest
3494 Successfully set the new policy.
3495 Supported security subsystems : ACM
3496 Policy name : mytest
3497 Policy type : ACM
3498 Version of XML policy : 1.0
3499 Policy configuration : loaded, activated for boot
3501 # xm labels
3502 A-Bank
3503 A-Bank.MarketAnalysis
3504 A-Bank.SecurityUnderwriting
3505 AutoCorp
3506 B-Bank
3507 SystemManagement
3508 __UNLABELED__
3510 # xm list --label
3511 Name ID Mem VCPUs State Time(s) Label
3512 Domain-0 0 873 1 r----- 58.0 ACM:mytest:SystemManagement
3513 \end{verbatim}
3514 \end{scriptsize}
3516 The described way of changing policies by resetting the existing
3517 policy is useful for testing different policies. For real deployment
3518 environments, a policy update as described in the following section
3519 is more appropriate and can be applied seamlessly at run-time while
3520 user domains are running.
3522 \subsubsection{Update an sHype/Xen Security Policy at Run-time}
3524 Once an ACM security policy is activated (loaded into the Xen
3525 hypervisor), the policy may be updated at run-time without the
3526 need to re-boot the system. The XML update-policy contains several
3527 additional information fields that are required to safely link the
3528 new policy contents to the old policy and ensure a consistent
3529 transformation of the system security state from the old to the
3530 new policy. Those additional fields are required for policies that
3531 are updating an existing policy at run-time.
3533 The major benefit of policy updates is the ability to add, delete,
3534 or rename workload types, labels, and conflict sets (run-time
3535 exclusion rules) to accommodate changes in the managed virtual
3536 environment without the need to reboot the Xen system. When a
3537 new policy renames labels of the current policy, the labels
3538 attached to resources and domains are automatically updated
3539 during a successful policy update.
3541 We have manually crafted an update policy for the \verb|mytest|
3542 security policy and stored it in the file mytest\_update-security\_policy.xml
3543 in the policies directory. We will discuss this policy in detail before
3544 using it to update a running sHype/Xen system. The following figures contain
3545 the whole contents of the update policy file.
3547 Figure~\ref{fig:acmupdateheader} shows the policy
3548 header of an update-policy and the new \verb|FromPolicy| XML
3549 node. For the policy update to succeed, the policy name and the
3550 policy version fields of the \verb|FromPolicy| XML node must
3551 exactly match those of the currently enforced policy. This
3552 ensures a controlled update path of the policy.
3554 \begin{figure}[htb]
3555 \begin{scriptsize}
3556 \begin{verbatim}
3557 <?xml version="1.0" encoding="UTF-8"?>
3558 <!-- Auto-generated by ezPolicy -->
3559 <SecurityPolicyDefinition xmlns=""
3560 xmlns:xsi=""
3561 xsi:schemaLocation=" ../../security_policy.xsd ">
3562 <PolicyHeader>
3563 <PolicyName>mytest</PolicyName>
3564 <Date>Tue Nov 27 21:53:45 2007</Date>
3565 <Version>1.1</Version>
3566 <FromPolicy>
3567 <PolicyName>mytest</PolicyName>
3568 <Version>1.0</Version>
3569 </FromPolicy>
3570 </PolicyHeader>
3571 \end{verbatim}
3572 \end{scriptsize}
3573 \caption{XML security policy update -- Part I: Updated Policy Header.}
3574 \label{fig:acmupdateheader}
3575 \end{figure}
3577 The version number of the new policy, which is shown in the
3578 node following the \verb|Date| node, must be a logical increment
3579 to the current policy's version. Therefore at least the minor
3580 number of the policy version must be incremented. This ensures
3581 that a policy update is applied only to exactly the policy for
3582 which this update was created and minimizes unforseen side-effects
3583 of policy updates.
3585 \paragraph{Types and Conflic Sets}
3586 The type names and the assignment of types to labels or conflict
3587 sets (run-time exclusion rules) can
3588 simply be changed consistently throughout the policy. Types,
3589 as opposed to labels, are not directly associated or referenced
3590 outside the policy so they do not need to carry their history
3591 in a ``From'' field. The figure below shows the update for the
3592 types and conflict sets. The \verb|__UNLABELED__| type is removed
3593 to disable support for running unlabeled domains. Additionally,
3594 we have renamed the two \verb|A-Bank| department types with
3595 abbreviated names \verb|A-Bank.SU| and \verb|A-Bank.MA|. You
3596 can also see how those type names are
3597 consistently changed within the conflict set definition.
3599 \begin{figure}[htb]
3600 \begin{scriptsize}
3601 \begin{verbatim}
3602 <SimpleTypeEnforcement>
3603 <SimpleTypeEnforcementTypes>
3604 <Type>SystemManagement</Type>
3605 <Type>A-Bank</Type>
3606 <Type>A-Bank.SU</Type>
3607 <Type>A-Bank.MA</Type>
3608 <Type>B-Bank</Type>
3609 <Type>AutoCorp</Type>
3610 </SimpleTypeEnforcementTypes>
3611 </SimpleTypeEnforcement>
3613 <ChineseWall priority="PrimaryPolicyComponent">
3614 <ChineseWallTypes>
3615 <Type>SystemManagement</Type>
3616 <Type>A-Bank</Type>
3617 <Type>A-Bank.SU</Type>
3618 <Type>A-Bank.MA</Type>
3619 <Type>B-Bank</Type>
3620 <Type>AutoCorp</Type>
3621 </ChineseWallTypes>
3623 <ConflictSets>
3624 <Conflict name="RER">
3625 <Type>A-Bank</Type>
3626 <Type>B-Bank</Type>
3627 </Conflict>
3628 <Conflict name="RER">
3629 <Type>A-Bank.MA</Type>
3630 <Type>A-Bank.SU</Type>
3631 </Conflict>
3632 </ConflictSets>
3633 </ChineseWall>
3634 \end{verbatim}
3635 \end{scriptsize}
3636 \caption{XML security policy update -- Part II: Updated Types and Conflict Sets.}
3637 \label{fig:acmupdatetypesnrules}
3638 \end{figure}
3640 In the same way, new types can be introduced and new conflict sets
3641 can be defined by simply adding the types or conflict sets to the
3642 update policy.
3644 \paragraph{Labels} Virtual machine and resource labels of an existing policy can be
3645 deleted through a policy update simply by omitting them in the
3646 update-policy. However, if a currently running virtual machine
3647 or a currently used resource is labeled with a label not stated
3648 in the update-policy, then the policy update is rejected. This
3649 ensures that a policy update leaves the system in a consistent
3650 security state.
3652 A policy update also enables the renaming of virtual machine and
3653 resource labels. Linking the old label name with the new label
3654 name is achieved through the \verb|from| attribute in the
3655 \verb|VirtualMachineLabel| or \verb|ResourceLabel| nodes in the
3656 update-policy. Figure~\ref{fig:acmupdatelabels} shown how subject
3657 and resource labels
3658 are updated from their old name \verb|A-Bank.SecurityUnterwriting|
3659 to their new name \verb|A-Bank.SU| using the \verb|from| attribute.
3661 \begin{figure}[htb]
3662 \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}l|l}
3663 \begin{minipage}{0.475\textwidth}
3664 \begin{tiny}
3665 \begin{verbatim}
3666 <SecurityLabelTemplate>
3667 <SubjectLabels bootstrap="SystemManagement">
3668 <VirtualMachineLabel>
3669 <Name>SystemManagement</Name>
3670 <SimpleTypeEnforcementTypes>
3671 <Type>SystemManagement</Type>
3672 <Type>A-Bank</Type>
3673 <Type>A-Bank.SU</Type>
3674 <Type>A-Bank.MA</Type>
3675 <Type>B-Bank</Type>
3676 <Type>AutoCorp</Type>
3677 </SimpleTypeEnforcementTypes>
3678 <ChineseWallTypes>
3679 <Type>SystemManagement</Type>
3680 </ChineseWallTypes>
3681 </VirtualMachineLabel>
3682 <VirtualMachineLabel>
3683 <Name>A-Bank-WL</Name>
3684 <SimpleTypeEnforcementTypes>
3685 <Type>SystemManagement</Type>
3686 <Type>A-Bank</Type>
3687 <Type>A-Bank.SU</Type>
3688 <Type>A-Bank.MA</Type>
3689 </SimpleTypeEnforcementTypes>
3690 <ChineseWallTypes>
3691 <Type>SystemManagement</Type>
3692 </ChineseWallTypes>
3693 </VirtualMachineLabel>
3694 <VirtualMachineLabel>
3695 <Name>A-Bank</Name>
3696 <SimpleTypeEnforcementTypes>
3697 <Type>A-Bank</Type>
3698 </SimpleTypeEnforcementTypes>
3699 <ChineseWallTypes>
3700 <Type>A-Bank</Type>
3701 </ChineseWallTypes>
3702 </VirtualMachineLabel>
3703 <VirtualMachineLabel>
3704 <Name from="A-Bank.SecurityUnderwriting">
3705 A-Bank.SU</Name>
3706 <SimpleTypeEnforcementTypes>
3707 <Type>A-Bank.SU</Type>
3708 </SimpleTypeEnforcementTypes>
3709 <ChineseWallTypes>
3710 <Type>A-Bank</Type>
3711 <Type>A-Bank.SU</Type>
3712 </ChineseWallTypes>
3713 </VirtualMachineLabel>
3714 <VirtualMachineLabel>
3715 <Name from="A-Bank.MarketAnalysis">
3716 A-Bank.MA</Name>
3717 <SimpleTypeEnforcementTypes>
3718 <Type>A-Bank.MA</Type>
3719 </SimpleTypeEnforcementTypes>
3720 <ChineseWallTypes>
3721 <Type>A-Bank</Type>
3722 <Type>A-Bank.MA</Type>
3723 </ChineseWallTypes>
3724 </VirtualMachineLabel>
3725 \end{verbatim}
3726 \end{tiny}
3727 \end{minipage} &
3728 \begin{minipage}{0.475\textwidth}
3729 \begin{tiny}
3730 \begin{verbatim}
3731 <VirtualMachineLabel>
3732 <Name>B-Bank</Name>
3733 <SimpleTypeEnforcementTypes>
3734 <Type>B-Bank</Type>
3735 </SimpleTypeEnforcementTypes>
3736 <ChineseWallTypes>
3737 <Type>B-Bank</Type>
3738 </ChineseWallTypes>
3739 </VirtualMachineLabel>
3740 <VirtualMachineLabel>
3741 <Name>AutoCorp</Name>
3742 <SimpleTypeEnforcementTypes>
3743 <Type>AutoCorp</Type>
3744 </SimpleTypeEnforcementTypes>
3745 <ChineseWallTypes>
3746 <Type>AutoCorp</Type>
3747 </ChineseWallTypes>
3748 </VirtualMachineLabel>
3749 </SubjectLabels>
3751 <ObjectLabels>
3752 <ResourceLabel>
3753 <Name>SystemManagement</Name>
3754 <SimpleTypeEnforcementTypes>
3755 <Type>SystemManagement</Type>
3756 </SimpleTypeEnforcementTypes>
3757 </ResourceLabel>
3758 <ResourceLabel>
3759 <Name>A-Bank</Name>
3760 <SimpleTypeEnforcementTypes>
3761 <Type>A-Bank</Type>
3762 </SimpleTypeEnforcementTypes>
3763 </ResourceLabel>
3764 <ResourceLabel>
3765 <Name from="A-Bank.SecurityUnderwriting">
3766 A-Bank.SU</Name>
3767 <SimpleTypeEnforcementTypes>
3768 <Type>A-Bank.SU</Type>
3769 </SimpleTypeEnforcementTypes>
3770 </ResourceLabel>
3771 <ResourceLabel>
3772 <Name from="A-Bank.MarketAnalysis">
3773 A-Bank.MA</Name>
3774 <SimpleTypeEnforcementTypes>
3775 <Type>A-Bank.MA</Type>
3776 </SimpleTypeEnforcementTypes>
3777 </ResourceLabel>
3778 <ResourceLabel>
3779 <Name>B-Bank</Name>
3780 <SimpleTypeEnforcementTypes>
3781 <Type>B-Bank</Type>
3782 </SimpleTypeEnforcementTypes>
3783 </ResourceLabel>
3784 <ResourceLabel>
3785 <Name>AutoCorp</Name>
3786 <SimpleTypeEnforcementTypes>
3787 <Type>AutoCorp</Type>
3788 </SimpleTypeEnforcementTypes>
3789 </ResourceLabel>
3790 </ObjectLabels>
3791 </SecurityLabelTemplate>
3792 </SecurityPolicyDefinition>
3793 \end{verbatim}
3794 \end{tiny}
3795 \end{minipage}
3796 \end{tabular*}
3797 \caption{XML security policy update -- Part III: Updated Label Definition.}
3798 \label{fig:acmupdatelabels}
3799 \end{figure}
3800 % DO NOT MODIFY WHITESPACE ABOVE, it balances the columns
3802 The updated label definition also includes a new label \verb|A-Bank-WL|
3803 that includes all STE types related to A-Bank. Its CHWALL type
3804 is \verb|SystemManagement|. This indicates that this label is designed
3805 as Domain-0 label. A Xen system can be restricted to only run A-Bank
3806 related workloads by relabeling Domain-0 with the \verb|A-Bank-WL| label.
3808 We assume that the update-policy shown in
3809 Figures~\ref{fig:acmupdateheader}, \ref{fig:acmupdatetypesnrules}
3810 and \ref{fig:acmupdatelabels}
3811 is stored in the XML file mytest\_update-security\_policy.xml located
3812 in the ACM policy directory. See Section~\ref{subsection:acmnaming}
3813 for information about policy names and locations.
3815 The following \verb|xm setpolicy| command updates the active ACM
3816 security policy at run-time.
3818 \begin{scriptsize}
3819 \begin{verbatim}
3820 # xm list --label
3821 Name ID Mem VCPUs State Time(s) Label
3822 domain1 2 128 1 -b---- 0.6 ACM:mytest:A-Bank
3823 domain4 3 164 1 -b---- 0.3 ACM:mytest:A-Bank.SecurityUnderwriting
3824 Domain-0 0 711 1 r----- 71.8 ACM:mytest:SystemManagement
3826 # xm resources
3827 file:/home/xen/dom_fc5/fedora.fc5.swap
3828 type: ACM
3829 policy: mytest
3830 label: A-Bank
3831 file:/home/xen/dom_fc5/fedora.fc5.img
3832 type: ACM
3833 policy: mytest
3834 label: A-Bank
3836 # xm setpolicy ACM mytest_update
3837 Successfully set the new policy.
3838 Supported security subsystems : ACM
3839 Policy name : mytest
3840 Policy type : ACM
3841 Version of XML policy : 1.1
3842 Policy configuration : loaded, activated for boot
3844 # xm list --label
3845 Name ID Mem VCPUs State Time(s) Label
3846 domain1 2 128 1 -b---- 0.7 ACM:mytest:A-Bank
3847 domain4 3 164 1 -b---- 0.3 ACM:mytest:A-Bank.SU
3848 Domain-0 0 711 1 r----- 72.8 ACM:mytest:SystemManagement
3850 # xm labels
3851 A-Bank
3852 A-Bank-WL
3853 A-Bank.MA
3854 A-Bank.SU
3855 AutoCorp
3856 B-Bank
3858 # xm resources
3859 file:/home/xen/dom_fc5/fedora.fc5.swap
3860 type: ACM
3861 policy: mytest
3862 label: A-Bank
3863 file:/home/xen/dom_fc5/fedora.fc5.img
3864 type: ACM
3865 policy: mytest
3866 label: A-Bank
3867 \end{verbatim}
3868 \end{scriptsize}
3870 After successful completion of this command, \verb|xm list --label|
3871 shows that the labels of running domains changed to their new names.
3872 \verb|xm labels| shows that new labels \verb|A-Bank.SU| and \verb|A-Bank.AM|
3873 are now available in the policy. The resource labels remain valid after
3874 the successful update as \verb|xm resources| confirms.
3876 The \verb|setpolicy| command fails if the new policy is inconsistent
3877 with the current one or the policy is inconsistent internally (e.g., types
3878 are renamed in the type definition but not in the label definition part of
3879 the policy). In this case, the old policy remains active.
3881 After relabeling Domain-0 with the new \verb|A-Bank-WL| label, we can no
3882 longer run domains labeled \verb|B-Bank| or \verb|AutoCorp| since their
3883 STE types are not a subset of the new Domain-0 label.
3885 \begin{scriptsize}
3886 \begin{verbatim}
3887 # xm addlabel A-Bank-WL mgt Domain-0
3888 Successfully set the label of domain 'Domain-0' to 'A-Bank-WL'.
3890 # xm list --label
3891 Name ID Mem VCPUs State Time(s) Label
3892 domain1 2 128 1 -b---- 0.8 ACM:mytest:A-Bank
3893 Domain-0 0 711 1 r----- 74.5 ACM:mytest:A-Bank-WL
3894 domain4 3 164 1 -b---- 0.3 ACM:mytest:A-Bank.SU
3896 # xm getlabel dom domain3.xm
3897 policytype=ACM,policy=mytest,label=AutoCorp
3899 # xm create domain3.xm
3900 Using config file "./domain3.xm".
3901 Error: VM is not authorized to run.
3903 # xm addlabel SystemManagement mgt Domain-0
3904 Successfully set the label of domain 'Domain-0' to 'SystemManagement'.
3906 # xm list --label
3907 Name ID Mem VCPUs State Time(s) Label
3908 domain1 2 128 1 -b---- 0.8 ACM:mytest:A-Bank
3909 domain4 3 164 1 -b---- 0.3 ACM:mytest:A-Bank.SU
3910 Domain-0 0 709 1 r----- 76.4 ACM:mytest:SystemManagement
3912 # xm create domain3.xm
3913 Using config file "./domain3.xm".
3914 Started domain domain3
3916 # xm list --label
3917 Name ID Mem VCPUs State Time(s) Label
3918 domain1 2 128 1 -b---- 0.8 ACM:mytest:A-Bank
3919 domain4 3 164 1 -b---- 0.3 ACM:mytest:A-Bank.SU
3920 domain3 4 164 1 -b---- 0.3 ACM:mytest:AutoCorp
3921 Domain-0 0 547 1 r----- 77.5 ACM:mytest:SystemManagement
3922 \end{verbatim}
3923 \end{scriptsize}
3925 In the same manner, you can add new labels to support new workloads and
3926 add, delete, or rename workload types (STE and/or CHWALL types) simply
3927 by changing the composition of labels. Another use case is to add new
3928 workload types to the current Domain-0 label to enable them to run.
3929 Conflict sets (run-time exclusion rules) can be simply omitted or added.
3930 The policy and label changes become active at once and new workloads
3931 can be run in protected mode without rebooting the Xen system.
3933 In all these cases, if any running user domain would--under the new policy--not
3934 be allowed to run or would not be allowed to access any of the resources
3935 it currently uses, then the policy update is rejected. In this case, you
3936 can stop domains that conflict with the new policy and update the policy
3937 afterwards. The old policy remains active until a policy update succeeds
3938 or Xen is re-booted into a new policy.
3940 \subsection{Tools For Creating sHype/Xen Security Policies}
3941 To create a security policy for Xen, you can use one of the following
3942 tools:
3943 \begin{itemize}
3944 \item \verb|ezPolicy| GUI tool -- start writing policies
3945 \item \verb|xensec_gen| tool -- refine policies created with \verb|ezPolicy|
3946 \item text or XML editor
3947 \end{itemize}
3949 We use the \verb|ezPolicy| tool in
3950 Section~\ref{subsection:acmexamplecreate} to quickly create a workload
3951 protection policy. If desired, the resulting XML policy file can be
3952 loaded into the \verb|xensec_gen| tool to refine it. It can also be
3953 directly edited using an XML editor. Any XML policy file is verified
3954 against the security policy schema when it is translated (see
3955 Subsection~\ref{subsection:acmexampleinstall}).
3957 \section{Current Limitations}
3958 \label{section:acmlimitations}
3960 The sHype/ACM configuration for Xen is work in progress. There is
3961 ongoing work for protecting virtualized resources and planned and
3962 ongoing work for protecting access to remote resources and domains.
3963 The following sections describe limitations of some of the areas into
3964 which access control is being extended.
3966 \subsection{Network Traffic}
3967 Local and remote network traffic is currently not controlled.
3968 Solutions to add sHype/ACM policy enforcement to the virtual network
3969 exist but need to be discussed before they can become part of Xen.
3970 Subjecting external network traffic to the ACM security policy is work
3971 in progress. Manually setting up filters in domain 0 is required for
3972 now but does not scale well.
3974 \subsection{Resource Access and Usage Control}
3976 Enforcing the security policy across multiple hypervisor systems and
3977 on access to remote shared resources is work in progress. Extending
3978 access control to new types of resources is ongoing work (e.g. network
3979 storage).
3981 On a single Xen system, information about the association of resources
3982 and security labels is stored in
3983 \verb|/var/lib/xend/security/policies/resource_labels|. This file relates
3984 a full resource path with a security label. This association is weak
3985 and will break if resources are moved or renamed without adapting the
3986 label file. Improving the protection of label-resource relationships
3987 is ongoing work.
3989 Controlling resource usage and enforcing resource limits in general is
3990 ongoing work in the Xen community.
3992 \subsection{Domain Migration}
3994 Labels on domains are enforced during domain migration and the
3995 destination hypervisor will ensure that the domain label is valid and
3996 the domain is permitted to run (considering the Chinese Wall policy
3997 rules) before it accepts the migration. However, the network between
3998 the source and destination hypervisor as well as both hypervisors must
3999 be trusted. Architectures and prototypes exist that both protect the
4000 network connection and ensure that the hypervisors enforce access
4001 control consistently but patches are not yet available for the main
4002 stream.
4004 \subsection{Covert Channels}
4006 The sHype access control aims at system independent security policies.
4007 It builds on top of the core hypervisor isolation. Any covert channels
4008 that exist in the core hypervisor or in the hardware (e.g., shared
4009 processor cache) will be inherited. If those covert channels are not
4010 the result of trade-offs between security and other system properties,
4011 then they are most effectively minimized or eliminated where they are
4012 caused. sHype offers however some means to mitigate their impact, e.g.,
4013 run-time exclusion rules (cf Section~\ref{subsection:acmexamplecreate})
4014 or limiting the system authorization (cf Section~\ref{subsection:acmlabeldom0}).
4017 \part{Reference}
4019 %% Chapter Build and Boot Options
4020 \chapter{Build and Boot Options}
4022 This chapter describes the build- and boot-time options which may be
4023 used to tailor your Xen system.
4025 \section{Top-level Configuration Options}
4027 Top-level configuration is achieved by editing one of two
4028 files: \path{} and \path{Makefile}.
4030 The former allows the overall build target architecture to be
4031 specified. You will typically not need to modify this unless
4032 you are cross-compiling or if you wish to build a non-PAE
4033 Xen system. Additional configuration options are documented
4034 in the \path{} file.
4036 The top-level \path{Makefile} is chiefly used to customize the set of
4037 kernels built. Look for the line:
4038 \begin{quote}
4039 \begin{verbatim}
4040 KERNELS ?= linux-2.6-xen0 linux-2.6-xenU
4041 \end{verbatim}
4042 \end{quote}
4044 Allowable options here are any kernels which have a corresponding
4045 build configuration file in the \path{buildconfigs/} directory.
4049 \section{Xen Build Options}
4051 Xen provides a number of build-time options which should be set as
4052 environment variables or passed on make's command-line.
4054 \begin{description}
4055 \item[verbose=y] Enable debugging messages when Xen detects an
4056 unexpected condition. Also enables console output from all domains.
4057 \item[debug=y] Enable debug assertions. Implies {\bf verbose=y}.
4058 (Primarily useful for tracing bugs in Xen).
4059 \item[debugger=y] Enable the in-Xen debugger. This can be used to
4060 debug Xen, guest OSes, and applications.
4061 \item[perfc=y] Enable performance counters for significant events
4062 within Xen. The counts can be reset or displayed on Xen's console
4063 via console control keys.
4064 \end{description}
4067 \section{Xen Boot Options}
4068 \label{s:xboot}
4070 These options are used to configure Xen's behaviour at runtime. They
4071 should be appended to Xen's command line, either manually or by
4072 editing \path{grub.conf}.
4074 \begin{description}
4075 \item [ noreboot ] Don't reboot the machine automatically on errors.
4076 This is useful to catch debug output if you aren't catching console
4077 messages via the serial line.
4078 \item [ nosmp ] Disable SMP support. This option is implied by
4079 `ignorebiostables'.
4080 \item [ watchdog ] Enable NMI watchdog which can report certain
4081 failures.
4082 \item [ noirqbalance ] Disable software IRQ balancing and affinity.
4083 This can be used on systems such as Dell 1850/2850 that have
4084 workarounds in hardware for IRQ-routing issues.
4085 \item [ badpage=$<$page number$>$,$<$page number$>$, \ldots ] Specify
4086 a list of pages not to be allocated for use because they contain bad
4087 bytes. For example, if your memory tester says that byte 0x12345678
4088 is bad, you would place `badpage=0x12345' on Xen's command line.
4089 \item [ serial\_tx\_buffer=$<$size$>$ ] Size of serial transmit
4090 buffers. Default is 16kB.
4091 \item [ com1=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$
4092 com2=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$ ] \mbox{}\\
4093 Xen supports up to two 16550-compatible serial ports. For example:
4094 `com1=9600, 8n1, 0x408, 5' maps COM1 to a 9600-baud port, 8 data
4095 bits, no parity, 1 stop bit, I/O port base 0x408, IRQ 5. If some
4096 configuration options are standard (e.g., I/O base and IRQ), then
4097 only a prefix of the full configuration string need be specified. If
4098 the baud rate is pre-configured (e.g., by the bootloader) then you
4099 can specify `auto' in place of a numeric baud rate.
4100 \item [ console=$<$specifier list$>$ ] Specify the destination for Xen
4101 console I/O. This is a comma-separated list of, for example:
4102 \begin{description}
4103 \item[ vga ] Use VGA console (until domain 0 boots, unless {\bf
4104 vga=...keep } is specified).
4105 \item[ com1 ] Use serial port com1.
4106 \item[ com2H ] Use serial port com2. Transmitted chars will have the
4107 MSB set. Received chars must have MSB set.
4108 \item[ com2L] Use serial port com2. Transmitted chars will have the
4109 MSB cleared. Received chars must have MSB cleared.
4110 \end{description}
4111 The latter two examples allow a single port to be shared by two
4112 subsystems (e.g.\ console and debugger). Sharing is controlled by
4113 MSB of each transmitted/received character. [NB. Default for this
4114 option is `com1,vga']
4115 \item [ vga=$<$mode$>$(,keep) ] The mode is one of the following options:
4116 \begin{description}
4117 \item[ ask ] Display a vga menu allowing manual selection of video
4118 mode.
4119 \item[ current ] Use existing vga mode without modification.
4120 \item[ text-$<$mode$>$ ] Select text-mode resolution, where mode is
4121 one of 80x25, 80x28, 80x30, 80x34, 80x43, 80x50, 80x60.
4122 \item[ gfx-$<$mode$>$ ] Select VESA graphics mode
4123 $<$width$>$x$<$height$>$x$<$depth$>$ (e.g., `vga=gfx-1024x768x32').
4124 \item[ mode-$<$mode$>$ ] Specify a mode number as discovered by `vga
4125 ask'. Note that the numbers are displayed in hex and hence must be
4126 prefixed by `0x' here (e.g., `vga=mode-0x0335').
4127 \end{description}
4128 The mode may optionally be followed by `{\bf,keep}' to cause Xen to keep
4129 writing to the VGA console after domain 0 starts booting (e.g., `vga=text-80x50,keep').
4130 \item [ no-real-mode ] (x86 only) Do not execute real-mode bootstrap
4131 code when booting Xen. This option should not be used except for
4132 debugging. It will effectively disable the {\bf vga} option, which
4133 relies on real mode to set the video mode.
4134 \item [ edid=no,force ] (x86 only) Either force retrieval of monitor
4135 EDID information via VESA DDC, or disable it (edid=no). This option
4136 should not normally be required except for debugging purposes.
4137 \item [ edd=off,on,skipmbr ] (x86 only) Control retrieval of Extended
4138 Disc Data (EDD) from the BIOS during boot.
4139 \item [ console\_to\_ring ] Place guest console output into the
4140 hypervisor console ring buffer. This is disabled by default.
4141 When enabled, both hypervisor output and guest console output
4142 is available from the ring buffer. This can be useful for logging
4143 and/or remote presentation of console data.
4144 \item [ sync\_console ] Force synchronous console output. This is
4145 useful if you system fails unexpectedly before it has sent all
4146 available output to the console. In most cases Xen will
4147 automatically enter synchronous mode when an exceptional event
4148 occurs, but this option provides a manual fallback.
4149 \item [ conswitch=$<$switch-char$><$auto-switch-char$>$ ] Specify how
4150 to switch serial-console input between Xen and DOM0. The required
4151 sequence is CTRL-$<$switch-char$>$ pressed three times. Specifying
4152 the backtick character disables switching. The
4153 $<$auto-switch-char$>$ specifies whether Xen should auto-switch
4154 input to DOM0 when it boots --- if it is `x' then auto-switching is
4155 disabled. Any other value, or omitting the character, enables
4156 auto-switching. [NB. Default switch-char is `a'.]
4157 \item [ loglvl=$<$level$>/<$level$>$ ]
4158 Specify logging level. Messages of the specified severity level (and
4159 higher) will be printed to the Xen console. Valid levels are `none',
4160 `error', `warning', `info', `debug', and `all'. The second level
4161 specifier is optional: it is used to specify message severities
4162 which are to be rate limited. Default is `loglvl=warning'.
4163 \item [ guest\_loglvl=$<$level$>/<$level$>$ ] As for loglvl, but
4164 applies to messages relating to guests. Default is
4165 `guest\_loglvl=none/warning'.
4166 \item [ console\_timestamps ]
4167 Adds a timestamp prefix to each line of Xen console output.
4168 \item [ nmi=xxx ]
4169 Specify what to do with an NMI parity or I/O error. \\
4170 `nmi=fatal': Xen prints a diagnostic and then hangs. \\
4171 `nmi=dom0': Inform DOM0 of the NMI. \\
4172 `nmi=ignore': Ignore the NMI.
4173 \item [ mem=xxx ] Set the physical RAM address limit. Any RAM
4174 appearing beyond this physical address in the memory map will be
4175 ignored. This parameter may be specified with a B, K, M or G suffix,
4176 representing bytes, kilobytes, megabytes and gigabytes respectively.
4177 The default unit, if no suffix is specified, is kilobytes.
4178 \item [ dom0\_mem=$<$specifier list$>$ ] Set the amount of memory to
4179 be allocated to domain 0. This is a comma-separated list containing
4180 the following optional components:
4181 \begin{description}
4182 \item[ min:$<$min\_amt$>$ ] Minimum amount to allocate to domain 0
4183 \item[ max:$<$min\_amt$>$ ] Maximum amount to allocate to domain 0
4184 \item[ $<$amt$>$ ] Precise amount to allocate to domain 0
4185 \end{description}
4186 Each numeric parameter may be specified with a B, K, M or
4187 G suffix, representing bytes, kilobytes, megabytes and gigabytes
4188 respectively; if no suffix is specified, the parameter defaults to
4189 kilobytes. Negative values are subtracted from total available
4190 memory. If $<$amt$>$ is not specified, it defaults to all available
4191 memory less a small amount (clamped to 128MB) for uses such as DMA
4192 buffers.
4193 \item [ dom0\_vcpus\_pin ] Pins domain 0 VCPUs on their respective
4194 physical CPUS (default=false).
4195 \item [ tbuf\_size=xxx ] Set the size of the per-cpu trace buffers, in
4196 pages (default 0).
4197 \item [ sched=xxx ] Select the CPU scheduler Xen should use. The
4198 current possibilities are `credit' (default), and `sedf'.
4199 \item [ apic\_verbosity=debug,verbose ] Print more detailed
4200 information about local APIC and IOAPIC configuration.
4201 \item [ lapic ] Force use of local APIC even when left disabled by
4202 uniprocessor BIOS.
4203 \item [ nolapic ] Ignore local APIC in a uniprocessor system, even if
4204 enabled by the BIOS.
4205 \item [ apic=bigsmp,default,es7000,summit ] Specify NUMA platform.
4206 This can usually be probed automatically.
4207 \item [ dma\_bits=xxx ] Specify width of DMA
4208 addresses in bits. Default is 30 bits (addresses up to 1GB are DMAable).
4209 \item [ dma\_emergency\_pool=xxx ] Specify lower bound on size of DMA
4210 pool below which ordinary allocations will fail rather than fall
4211 back to allocating from the DMA pool.
4212 \end{description}
4214 In addition, the following options may be specified on the Xen command
4215 line. Since domain 0 shares responsibility for booting the platform,
4216 Xen will automatically propagate these options to its command line.
4217 These options are taken from Linux's command-line syntax with
4218 unchanged semantics.
4220 \begin{description}
4221 \item [ acpi=off,force,strict,ht,noirq,\ldots ] Modify how Xen (and
4222 domain 0) parses the BIOS ACPI tables.
4223 \item [ acpi\_skip\_timer\_override ] Instruct Xen (and domain~0) to
4224 ignore timer-interrupt override instructions specified by the BIOS
4225 ACPI tables.
4226 \item [ noapic ] Instruct Xen (and domain~0) to ignore any IOAPICs
4227 that are present in the system, and instead continue to use the
4228 legacy PIC.
4229 \end{description}
4232 \section{XenLinux Boot Options}
4234 In addition to the standard Linux kernel boot options, we support:
4235 \begin{description}
4236 \item[ xencons=xxx ] Specify the device node to which the Xen virtual
4237 console driver is attached. The following options are supported:
4238 \begin{center}
4239 \begin{tabular}{l}
4240 `xencons=off': disable virtual console \\
4241 `xencons=tty': attach console to /dev/tty1 (tty0 at boot-time) \\
4242 `xencons=ttyS': attach console to /dev/ttyS0 \\
4243 `xencons=xvc': attach console to /dev/xvc0
4244 \end{tabular}
4245 \end{center}
4246 The default is ttyS for dom0 and xvc for all other domains.
4247 \end{description}
4250 %% Chapter Further Support
4251 \chapter{Further Support}
4253 If you have questions that are not answered by this manual, the
4254 sources of information listed below may be of interest to you. Note
4255 that bug reports, suggestions and contributions related to the
4256 software (or the documentation) should be sent to the Xen developers'
4257 mailing list (address below).
4260 \section{Other Documentation}
4262 For developers interested in porting operating systems to Xen, the
4263 \emph{Xen Interface Manual} is distributed in the \path{docs/}
4264 directory of the Xen source distribution.
4267 \section{Online References}
4269 The official Xen web site can be found at:
4270 \begin{quote} {\tt}
4271 \end{quote}
4273 This contains links to the latest versions of all online
4274 documentation, including the latest version of the FAQ.
4276 Information regarding Xen is also available at the Xen Wiki at
4277 \begin{quote} {\tt}\end{quote}
4278 The Xen project uses Bugzilla as its bug tracking system. You'll find
4279 the Xen Bugzilla at
4282 \section{Mailing Lists}
4284 There are several mailing lists that are used to discuss Xen related
4285 topics. The most widely relevant are listed below. An official page of
4286 mailing lists and subscription information can be found at \begin{quote}
4287 {\tt} \end{quote}
4289 \begin{description}
4290 \item[] Used for development
4291 discussions and bug reports. Subscribe at: \\
4292 {\small {\tt}}
4293 \item[] Used for installation and usage
4294 discussions and requests for help. Subscribe at: \\
4295 {\small {\tt}}
4296 \item[] Used for announcements only.
4297 Subscribe at: \\
4298 {\small {\tt}}
4299 \item[] Changelog feed
4300 from the unstable and 2.0 trees - developer oriented. Subscribe at: \\
4301 {\small {\tt}}
4302 \end{description}
4306 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4308 \appendix
4310 \chapter{Unmodified (HVM) guest domains in Xen with Hardware support for Virtualization}
4312 Xen supports guest domains running unmodified guest operating systems using
4313 virtualization extensions available on recent processors. Currently processors
4314 featuring the Intel Virtualization Extension (Intel-VT) or the AMD extension
4315 (AMD-V) are supported. The technology covering both implementations is
4316 called HVM (for Hardware Virtual Machine) in Xen. More information about the
4317 virtualization extensions are available on the respective websites:
4318 {\small {\tt}}
4321 {\small {\tt\_type/white\_papers\_and\_tech\_docs/24593.pdf}}
4323 \section{Building Xen with HVM support}
4325 The following packages need to be installed in order to build Xen with HVM support. Some Linux distributions do not provide these packages by default.
4327 \begin{tabular}{lp{11.0cm}}
4328 {\bfseries Package} & {\bfseries Description} \\
4330 dev86 & The dev86 package provides an assembler and linker for real mode 80x86 instructions. You need to have this package installed in order to build the BIOS code which runs in (virtual) real mode.
4332 If the dev86 package is not available on the x86\_64 distribution, you can install the i386 version of it. The dev86 rpm package for various distributions can be found at {\scriptsize {\tt\&submit=Search}} \\
4334 SDL-devel, SDL & Simple DirectMedia Layer (SDL) is another way of virtualizing the unmodified guest console. It provides an X window for the guest console.
4336 If the SDL and SDL-devel packages are not installed by default on the build system, they can be obtained from {\scriptsize {\tt\&amp;submit=Search}}
4339 {\scriptsize {\tt\&submit=Search}} \\
4341 \end{tabular}
4343 \section{Configuration file for unmodified HVM guests}
4345 The Xen installation includes a sample configuration file, {\small {\tt /etc/xen/xmexample.hvm}}. There are comments describing all the options. In addition to the common options that are the same as those for paravirtualized guest configurations, HVM guest configurations have the following settings:
4347 \begin{tabular}{lp{11.0cm}}
4349 {\bfseries Parameter} & {\bfseries Description} \\
4351 kernel & The HVM firmware loader, {\small {\tt /usr/lib/xen/boot/hvmloader}}\\
4353 builder & The domain build function. The HVM domain uses the 'hvm' builder.\\
4355 acpi & Enable HVM guest ACPI, default=1 (enabled)\\
4357 apic & Enable HVM guest APIC, default=1 (enabled)\\
4359 pae & Enable HVM guest PAE, default=1 (enabled)\\
4361 hap & Enable hardware-assisted paging support, such as AMD-V's nested paging
4362 or Intel\textregistered VT's extended paging. If available, Xen will
4363 use hardware-assisted paging instead of shadow paging for this guest's memory
4364 management.\\
4366 vif & Optionally defines MAC address and/or bridge for the network interfaces. Random MACs are assigned if not given. {\small {\tt type=ioemu}} means ioemu is used to virtualize the HVM NIC. If no type is specified, vbd is used, as with paravirtualized guests.\\
4368 disk & Defines the disk devices you want the domain to have access to, and what you want them accessible as. If using a physical device as the HVM guest's disk, each disk entry is of the form
4370 {\small {\tt phy:UNAME,ioemu:DEV,MODE,}}
4372 where UNAME is the host device file, DEV is the device name the domain will see, and MODE is r for read-only, w for read-write. ioemu means the disk will use ioemu to virtualize the HVM disk. If not adding ioemu, it uses vbd like paravirtualized guests.
4374 If using disk image file, its form should be like
4376 {\small {\tt file:FILEPATH,ioemu:DEV,MODE}}
4378 Optical devices can be emulated by appending cdrom to the device type
4380 {\small {\tt ',hdc:cdrom,r'}}
4382 If using more than one disk, there should be a comma between each disk entry. For example:
4384 {\scriptsize {\tt disk = ['file:/var/images/image1.img,ioemu:hda,w', 'phy:hda1,hdb1,w', 'file:/var/images/install1.iso,hdc:cdrom,r']}}\\
4386 boot & Boot from floppy (a), hard disk (c) or CD-ROM (d). For example, to boot from CD-ROM and fallback to HD, the entry should be:
4388 boot='dc'\\
4390 device\_model & The device emulation tool for HVM guests. This parameter should not be changed.\\
4392 sdl & Enable SDL library for graphics, default = 0 (disabled)\\
4394 vnc & Enable VNC library for graphics, default = 1 (enabled)\\
4396 vncconsole & Enable spawning of the vncviewer (only valid when vnc=1), default = 0 (disabled)
4398 If vnc=1 and vncconsole=0, user can use vncviewer to manually connect HVM from remote. For example:
4400 {\small {\tt vncviewer domain0\_IP\_address:HVM\_domain\_id}} \\
4402 serial & Enable redirection of HVM serial output to pty device\\
4404 \end{tabular}
4406 \begin{tabular}{lp{10cm}}
4408 usb & Enable USB support without defining a specific USB device.
4409 This option defaults to 0 (disabled) unless the option usbdevice is
4410 specified in which case this option then defaults to 1 (enabled).\\
4412 usbdevice & Enable USB support and also enable support for the given
4413 device. Devices that can be specified are {\small {\tt mouse}} (a PS/2 style
4414 mouse), {\small {\tt tablet}} (an absolute pointing device) and
4415 {\small {\tt host:id1:id2}} (a physical USB device on the host machine whose
4416 ids are {\small {\tt id1}} and {\small {\tt id2}}). The advantage
4417 of {\small {\tt tablet}} is that Windows guests will automatically recognize
4418 and support this device so specifying the config line
4420 {\small
4421 \begin{verbatim}
4422 usbdevice='tablet'
4423 \end{verbatim}
4426 will create a mouse that works transparently with Windows guests under VNC.
4427 Linux doesn't recognize the USB tablet yet so Linux guests under VNC will
4428 still need the Summagraphics emulation.
4429 Details about mouse emulation are provided in section \textbf{A.4.3}.\\
4431 localtime & Set the real time clock to local time [default=0, that is, set to UTC].\\
4433 soundhw & Enable sound card support and specify the hardware to emulate. Values can be sb16, es1370 or all. Default is none.\\
4435 full-screen & Start in full screen.\\
4437 nographic & Another way to redirect serial output. If enabled, no 'sdl' or 'vnc' can work. Not recommended.\\
4439 \end{tabular}
4442 \section{Creating virtual disks from scratch}
4443 \subsection{Using physical disks}
4444 If you are using a physical disk or physical disk partition, you need to install a Linux OS on the disk first. Then the boot loader should be installed in the correct place. For example {\small {\tt dev/sda}} for booting from the whole disk, or {\small {\tt /dev/sda1}} for booting from partition 1.
4446 \subsection{Using disk image files}
4447 You need to create a large empty disk image file first; then, you need to install a Linux OS onto it. There are two methods you can choose. One is directly installing it using a HVM guest while booting from the OS installation CD-ROM. The other is copying an installed OS into it. The boot loader will also need to be installed.
4449 \subsubsection*{To create the image file:}
4450 The image size should be big enough to accommodate the entire OS. This example assumes the size is 1G (which is probably too small for most OSes).
4452 {\small {\tt \# dd if=/dev/zero of=hd.img bs=1M count=0 seek=1024}}
4454 \subsubsection*{To directly install Linux OS into an image file using a HVM guest:}
4456 Install Xen and create HVM with the original image file with booting from CD-ROM. Then it is just like a normal Linux OS installation. The HVM configuration file should have a stanza for the CD-ROM as well as a boot device specification:
4458 {\small {\tt disk=['file:/var/images/your-hd.img,hda,w', ',hdc:cdrom,r' ]
4459 boot='d'}}
4461 If this method does not succeed, you can choose the following method of copying an installed Linux OS into an image file.
4463 \subsubsection*{To copy a installed OS into an image file:}
4464 Directly installing is an easier way to make partitions and install an OS in a disk image file. But if you want to create a specific OS in your disk image, then you will most likely want to use this method.
4466 \begin{enumerate}
4467 \item {\bfseries Install a normal Linux OS on the host machine}\\
4468 You can choose any way to install Linux, such as using yum to install Red Hat Linux or YAST to install Novell SuSE Linux. The rest of this example assumes the Linux OS is installed in {\small {\tt /var/guestos/}}.
4470 \item {\bfseries Make the partition table}\\
4471 The image file will be treated as hard disk, so you should make the partition table in the image file. For example:
4473 {\scriptsize {\tt \# losetup /dev/loop0 hd.img\\
4474 \# fdisk -b 512 -C 4096 -H 16 -S 32 /dev/loop0\\
4475 press 'n' to add new partition\\
4476 press 'p' to choose primary partition\\
4477 press '1' to set partition number\\
4478 press "Enter" keys to choose default value of "First Cylinder" parameter.\\
4479 press "Enter" keys to choose default value of "Last Cylinder" parameter.\\
4480 press 'w' to write partition table and exit\\
4481 \# losetup -d /dev/loop0}}
4483 \item {\bfseries Make the file system and install grub}\\
4484 {\scriptsize {\tt \# ln -s /dev/loop0 /dev/loop\\
4485 \# losetup /dev/loop0 hd.img\\
4486 \# losetup -o 16384 /dev/loop1 hd.img\\
4487 \# mkfs.ext3 /dev/loop1\\
4488 \# mount /dev/loop1 /mnt\\
4489 \# mkdir -p /mnt/boot/grub\\
4490 \# cp /boot/grub/stage* /boot/grub/e2fs\_stage1\_5 /mnt/boot/grub\\
4491 \# umount /mnt\\
4492 \# grub\\
4493 grub> device (hd0) /dev/loop\\
4494 grub> root (hd0,0)\\
4495 grub> setup (hd0)\\
4496 grub> quit\\
4497 \# rm /dev/loop\\
4498 \# losetup -d /dev/loop0\\
4499 \# losetup -d /dev/loop1}}
4501 The {\small {\tt losetup}} option {\small {\tt -o 16384}} skips the partition table in the image file. It is the number of sectors times 512. We need {\small {\tt /dev/loop}} because grub is expecting a disk device \emph{name}, where \emph{name} represents the entire disk and \emph{name1} represents the first partition.
4503 \item {\bfseries Copy the OS files to the image}\\
4504 If you have Xen installed, you can easily use {\small {\tt lomount}} instead of {\small {\tt losetup}} and {\small {\tt mount}} when coping files to some partitions. {\small {\tt lomount}} just needs the partition information.
4506 {\scriptsize {\tt \# lomount -t ext3 -diskimage hd.img -partition 1 /mnt/guest\\
4507 \# cp -ax /var/guestos/\{root,dev,var,etc,usr,bin,sbin,lib\} /mnt/guest\\
4508 \# mkdir /mnt/guest/\{proc,sys,home,tmp\}}}
4510 \item {\bfseries Edit the {\small {\tt /etc/fstab}} of the guest image}\\
4511 The fstab should look like this:
4513 {\scriptsize {\tt \# vim /mnt/guest/etc/fstab\\
4514 /dev/hda1 / ext3 defaults 1 1\\
4515 none /dev/pts devpts gid=5,mode=620 0 0\\
4516 none /dev/shm tmpfs defaults 0 0\\
4517 none /proc proc defaults 0 0\\
4518 none /sys sysfs efaults 0 0}}
4520 \item {\bfseries umount the image file}\\
4521 {\small {\tt \# umount /mnt/guest}}
4522 \end{enumerate}
4524 Now, the guest OS image {\small {\tt hd.img}} is ready. You can also reference {\small {\tt}} for quickstart images. But make sure to install the boot loader.
4526 \section{HVM Guests}
4527 \subsection{Editing the Xen HVM config file}
4528 Make a copy of the example HVM configuration file {\small {\tt /etc/xen/xmexample.hvm}} and edit the line that reads
4530 {\small {\tt disk = [ 'file:/var/images/\emph{min-el3-i386.img},hda,w' ]}}
4532 replacing \emph{min-el3-i386.img} with the name of the guest OS image file you just made.
4534 \subsection{Creating HVM guests}
4535 Simply follow the usual method of creating the guest, providing the filename of your HVM configuration file:\\
4537 {\small {\tt \# xend start\\
4538 \# xm create /etc/xen/hvmguest.hvm}}
4540 In the default configuration, VNC is on and SDL is off. Therefore VNC windows will open when HVM guests are created. If you want to use SDL to create HVM guests, set {\small {\tt sdl=1}} in your HVM configuration file. You can also turn off VNC by setting {\small {\tt vnc=0}}.
4542 \subsection{Mouse issues, especially under VNC}
4543 Mouse handling when using VNC is a little problematic.
4544 The problem is that the VNC viewer provides a virtual pointer which is
4545 located at an absolute location in the VNC window and only absolute
4546 coordinates are provided.
4547 The HVM device model converts these absolute mouse coordinates
4548 into the relative motion deltas that are expected by the PS/2
4549 mouse driver running in the guest.
4550 Unfortunately,
4551 it is impossible to keep these generated mouse deltas
4552 accurate enough for the guest cursor to exactly match
4553 the VNC pointer.
4554 This can lead to situations where the guest's cursor
4555 is in the center of the screen and there's no way to
4556 move that cursor to the left
4557 (it can happen that the VNC pointer is at the left
4558 edge of the screen and,
4559 therefore,
4560 there are no longer any left mouse deltas that
4561 can be provided by the device model emulation code.)
4563 To deal with these mouse issues there are 4 different
4564 mouse emulations available from the HVM device model:
4566 \begin{description}
4567 \item[PS/2 mouse over the PS/2 port.]
4568 This is the default mouse
4569 that works perfectly well under SDL.
4570 Under VNC the guest cursor will get
4571 out of sync with the VNC pointer.
4572 When this happens you can re-synchronize
4573 the guest cursor to the VNC pointer by
4574 holding down the
4575 \textbf{left-ctl}
4576 and
4577 \textbf{left-alt}
4578 keys together.
4579 While these keys are down VNC pointer motions
4580 will not be reported to the guest so
4581 that the VNC pointer can be moved
4582 to a place where it is possible
4583 to move the guest cursor again.
4585 \item[Summagraphics mouse over the serial port.]
4586 The device model also provides emulation
4587 for a Summagraphics tablet,
4588 an absolute pointer device.
4589 This emulation is provided over the second
4590 serial port,
4591 \textbf{/dev/ttyS1}
4592 for Linux guests and
4593 \textbf{COM2}
4594 for Windows guests.
4595 Unfortunately,
4596 neither Linux nor Windows provides
4597 default support for the Summagraphics
4598 tablet so the guest will have to be
4599 manually configured for this mouse.
4601 \textbf{Linux configuration.}
4603 First,
4604 configure the GPM service to use the Summagraphics tablet.
4605 This can vary between distributions but,
4606 typically,
4607 all that needs to be done is modify the file
4608 \path{/etc/sysconfig/mouse} to contain the lines:
4610 {\small
4611 \begin{verbatim}
4612 MOUSETYPE="summa"
4614 DEVICE=/dev/ttyS1
4615 \end{verbatim}
4618 and then restart the GPM daemon.
4620 Next,
4621 modify the X11 config
4622 \path{/etc/X11/xorg.conf}
4623 to support the Summgraphics tablet by replacing
4624 the input device stanza with the following:
4626 {\small
4627 \begin{verbatim}
4628 Section "InputDevice"
4629 Identifier "Mouse0"
4630 Driver "summa"
4631 Option "Device" "/dev/ttyS1"
4632 Option "InputFashion" "Tablet"
4633 Option "Mode" "Absolute"
4634 Option "Name" "EasyPen"
4635 Option "Compatible" "True"
4636 Option "Protocol" "Auto"
4637 Option "SendCoreEvents" "on"
4638 Option "Vendor" "GENIUS"
4639 EndSection
4640 \end{verbatim}
4643 Restart X and the X cursor should now properly
4644 track the VNC pointer.
4647 \textbf{Windows configuration.}
4649 Get the file
4650 \path{}
4651 and execute that file on the guest,
4652 answering the questions as follows:
4654 \begin{enumerate}
4655 \item When the program asks for \textbf{model},
4656 scroll down and select \textbf{SummaSketch (MM Compatible)}.
4658 \item When the program asks for \textbf{COM Port} specify \textbf{com2}.
4660 \item When the programs asks for a \textbf{Cursor Type} specify
4661 \textbf{4 button cursor/puck}.
4663 \item The guest system will then reboot and,
4664 when it comes back up,
4665 the guest cursor will now properly track
4666 the VNC pointer.
4667 \end{enumerate}
4669 \item[PS/2 mouse over USB port.]
4670 This is just the same PS/2 emulation except it is
4671 provided over a USB port.
4672 This emulation is enabled by the configuration flag:
4673 {\small
4674 \begin{verbatim}
4675 usbdevice='mouse'
4676 \end{verbatim}
4679 \item[USB tablet over USB port.]
4680 The USB tablet is an absolute pointing device
4681 that has the advantage that it is automatically
4682 supported under Windows guests,
4683 although Linux guests still require some
4684 manual configuration.
4685 This mouse emulation is enabled by the
4686 configuration flag:
4687 {\small
4688 \begin{verbatim}
4689 usbdevice='tablet'
4690 \end{verbatim}
4693 \textbf{Linux configuration.}
4695 Unfortunately,
4696 there is no GPM support for the
4697 USB tablet at this point in time.
4698 If you intend to use a GPM pointing
4699 device under VNC you should
4700 configure the guest for Summagraphics
4701 emulation.
4703 Support for X11 is available by following
4704 the instructions at\\
4705 \verb+\\
4706 with one minor change.
4707 The
4708 \path{xorg.conf}
4709 given in those instructions
4710 uses the wrong values for the X \& Y minimums and maximums,
4711 use the following config stanza instead:
4713 {\small
4714 \begin{verbatim}
4715 Section "InputDevice"
4716 Identifier "Tablet"
4717 Driver "evtouch"
4718 Option "Device" "/dev/input/event2"
4719 Option "DeviceName" "touchscreen"
4720 Option "MinX" "0"
4721 Option "MinY" "0"
4722 Option "MaxX" "32256"
4723 Option "MaxY" "32256"
4724 Option "ReportingMode" "Raw"
4725 Option "Emulate3Buttons"
4726 Option "Emulate3Timeout" "50"
4727 Option "SendCoreEvents" "On"
4728 EndSection
4729 \end{verbatim}
4732 \textbf{Windows configuration.}
4734 Just enabling the USB tablet in the
4735 guest's configuration file is sufficient,
4736 Windows will automatically recognize and
4737 configure device drivers for this
4738 pointing device.
4740 \end{description}
4742 \subsection{USB Support}
4743 There is support for an emulated USB mouse,
4744 an emulated USB tablet
4745 and physical low speed USB devices
4746 (support for high speed USB 2.0 devices is
4747 still under development).
4749 \begin{description}
4750 \item[USB PS/2 style mouse.]
4751 Details on the USB mouse emulation are
4752 given in sections
4753 \textbf{A.2}
4754 and
4755 \textbf{A.4.3}.
4756 Enabling USB PS/2 style mouse emulation
4757 is just a matter of adding the line
4759 {\small
4760 \begin{verbatim}
4761 usbdevice='mouse'
4762 \end{verbatim}
4765 to the configuration file.
4766 \item[USB tablet.]
4767 Details on the USB tablet emulation are
4768 given in sections
4769 \textbf{A.2}
4770 and
4771 \textbf{A.4.3}.
4772 Enabling USB tablet emulation
4773 is just a matter of adding the line
4775 {\small
4776 \begin{verbatim}
4777 usbdevice='tablet'
4778 \end{verbatim}
4781 to the configuration file.
4782 \item[USB physical devices.]
4783 Access to a physical (low speed) USB device
4784 is enabled by adding a line of the form
4786 {\small
4787 \begin{verbatim}
4788 usbdevice='host:vid:pid'
4789 \end{verbatim}
4792 into the the configuration file.\footnote{
4793 There is an alternate
4794 way of specifying a USB device that
4795 uses the syntax
4796 \textbf{host:bus.addr}
4797 but this syntax suffers from
4798 a major problem that makes
4799 it effectively useless.
4800 The problem is that the
4801 \textbf{addr}
4802 portion of this address
4803 changes every time the USB device
4804 is plugged into the system.
4805 For this reason this addressing
4806 scheme is not recommended and
4807 will not be documented further.
4809 \textbf{vid}
4810 and
4811 \textbf{pid}
4812 are a
4813 product id and
4814 vendor id
4815 that uniquely identify
4816 the USB device.
4817 These ids can be identified
4818 in two ways:
4820 \begin{enumerate}
4821 \item Through the control window.
4822 As described in section
4823 \textbf{A.4.6}
4824 the control window
4825 is activated by pressing
4826 \textbf{ctl-alt-2}
4827 in the guest VGA window.
4828 As long as USB support is
4829 enabled in the guest by including
4830 the config file line
4831 {\small
4832 \begin{verbatim}
4833 usb=1
4834 \end{verbatim}
4836 then executing the command
4837 {\small
4838 \begin{verbatim}
4839 info usbhost
4840 \end{verbatim}
4842 in the control window
4843 will display a list of all
4844 usb devices and their ids.
4845 For example,
4846 this output:
4847 {\small
4848 \begin{verbatim}
4849 Device 1.3, speed 1.5 Mb/s
4850 Class 00: USB device 04b3:310b
4851 \end{verbatim}
4853 was created from a USB mouse with
4854 vendor id
4855 \textbf{04b3}
4856 and product id
4857 \textbf{310b}.
4858 This device could be made available
4859 to the HVM guest by including the
4860 config file entry
4861 {\small
4862 \begin{verbatim}
4863 usbdevice='host:04be:310b'
4864 \end{verbatim}
4867 It is also possible to
4868 enable access to a USB
4869 device dynamically through
4870 the control window.
4871 The control window command
4872 {\small
4873 \begin{verbatim}
4874 usb_add host:vid:pid
4875 \end{verbatim}
4877 will also allow access to a
4878 USB device with vendor id
4879 \textbf{vid}
4880 and product id
4881 \textbf{pid}.
4882 \item Through the
4883 \path{/proc} file system.
4884 The contents of the pseudo file
4885 \path{/proc/bus/usb/devices}
4886 can also be used to identify
4887 vendor and product ids.
4888 Looking at this file,
4889 the line starting with
4890 \textbf{P:}
4891 has a field
4892 \textbf{Vendor}
4893 giving the vendor id and
4894 another field
4895 \textbf{ProdID}
4896 giving the product id.
4897 The contents of
4898 \path{/proc/bus/usb/devices}
4899 for the example mouse is as
4900 follows:
4901 {\small
4902 \begin{verbatim}
4903 T: Bus=01 Lev=01 Prnt=01 Port=01 Cnt=02 Dev#= 3 Spd=1.5 MxCh= 0
4904 D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1
4905 P: Vendor=04b3 ProdID=310b Rev= 1.60
4906 C:* #Ifs= 1 Cfg#= 1 Atr=a0 MxPwr=100mA
4907 I: If#= 0 Alt= 0 #EPs= 1 Cls=03(HID ) Sub=01 Prot=02 Driver=(none)
4908 E: Ad=81(I) Atr=03(Int.) MxPS= 4 Ivl=10ms
4909 \end{verbatim}
4911 Note that the
4912 \textbf{P:}
4913 line correctly identifies the
4914 vendor id and product id
4915 for this mouse as
4916 \textbf{04b3:310b}.
4917 \end{enumerate}
4918 There is one other issue to
4919 be aware of when accessing a
4920 physical USB device from the guest.
4921 The Dom0 kernel must not have
4922 a device driver loaded for
4923 the device that the guest wishes
4924 to access.
4925 This means that the Dom0
4926 kernel must not have that
4927 device driver compiled into
4928 the kernel or,
4929 if using modules,
4930 that driver module must
4931 not be loaded.
4932 Note that this is the device
4933 specific USB driver that must
4934 not be loaded,
4935 either the
4936 \textbf{UHCI}
4937 or
4938 \textbf{OHCI}
4939 USB controller driver must
4940 still be loaded.
4942 Going back to the USB mouse
4943 as an example,
4944 if \textbf{lsmod}
4945 gives the output:
4947 {\small
4948 \begin{verbatim}
4949 Module Size Used by
4950 usbmouse 4128 0
4951 usbhid 28996 0
4952 uhci_hcd 35409 0
4953 \end{verbatim}
4956 then the USB mouse is being
4957 used by the Dom0 kernel and is
4958 not available to the guest.
4959 Executing the command
4960 \textbf{rmmod usbhid}\footnote{
4961 Turns out the
4962 \textbf{usbhid}
4963 driver is the significant
4964 one for the USB mouse,
4965 the presence or absence of
4966 the module
4967 \textbf{usbmouse}
4968 has no effect on whether or
4969 not the guest can see a USB mouse.}
4970 will remove the USB mouse
4971 driver from the Dom0 kernel
4972 and the mouse will now be
4973 accessible by the HVM guest.
4975 Be aware the the Linux USB
4976 hotplug system will reload
4977 the drivers if a USB device
4978 is removed and plugged back
4979 in.
4980 This means that just unloading
4981 the driver module might not
4982 be sufficient if the USB device
4983 is removed and added back.
4984 A more reliable technique is
4985 to first
4986 \textbf{rmmod}
4987 the driver and then rename the
4988 driver file in the
4989 \path{/lib/modules}
4990 directory,
4991 just to make sure it doesn't get
4992 reloaded.
4993 \end{description}
4995 \subsection{Destroy HVM guests}
4996 HVM guests can be destroyed in the same way as can paravirtualized guests. We recommend that you shut-down the guest using the guest OS' provided method, for Linux, type the command
4998 {\small {\tt poweroff}}
5000 in the HVM guest's console, for Windows use Start -> Shutdown first to prevent
5001 data loss. Depending on the configuration the guest will be automatically
5002 destroyed, otherwise execute the command
5004 {\small {\tt xm destroy \emph{vmx\_guest\_id} }}
5006 at the Domain0 console.
5008 \subsection{HVM window (X or VNC) Hot Key}
5009 If you are running in the X environment after creating a HVM guest, an X window is created. There are several hot keys for control of the HVM guest that can be used in the window.
5011 {\bfseries Ctrl+Alt+2} switches from guest VGA window to the control window. Typing {\small {\tt help }} shows the control commands help. For example, 'q' is the command to destroy the HVM guest.\\
5012 {\bfseries Ctrl+Alt+1} switches back to HVM guest's VGA.\\
5013 {\bfseries Ctrl+Alt+3} switches to serial port output. It captures serial output from the HVM guest. It works only if the HVM guest was configured to use the serial port. \\
5015 \chapter{Vnets - Domain Virtual Networking}
5017 Xen optionally supports virtual networking for domains using {\em vnets}.
5018 These emulate private LANs that domains can use. Domains on the same
5019 vnet can be hosted on the same machine or on separate machines, and the
5020 vnets remain connected if domains are migrated. Ethernet traffic
5021 on a vnet is tunneled inside IP packets on the physical network. A vnet is a virtual
5022 network and addressing within it need have no relation to addressing on
5023 the underlying physical network. Separate vnets, or vnets and the physical network,
5024 can be connected using domains with more than one network interface and
5025 enabling IP forwarding or bridging in the usual way.
5027 Vnet support is included in \texttt{xm} and \xend:
5028 \begin{verbatim}
5029 # xm vnet-create <config>
5030 \end{verbatim}
5031 creates a vnet using the configuration in the file \verb|<config>|.
5032 When a vnet is created its configuration is stored by \xend and the vnet persists until it is
5033 deleted using
5034 \begin{verbatim}
5035 # xm vnet-delete <vnetid>
5036 \end{verbatim}
5037 The vnets \xend knows about are listed by
5038 \begin{verbatim}
5039 # xm vnet-list
5040 \end{verbatim}
5041 More vnet management commands are available using the
5042 \texttt{vn} tool included in the vnet distribution.
5044 The format of a vnet configuration file is
5045 \begin{verbatim}
5046 (vnet (id <vnetid>)
5047 (bridge <bridge>)
5048 (vnetif <vnet interface>)
5049 (security <level>))
5050 \end{verbatim}
5051 White space is not significant. The parameters are:
5052 \begin{itemize}
5053 \item \verb|<vnetid>|: vnet id, the 128-bit vnet identifier. This can be given
5054 as 8 4-digit hex numbers separated by colons, or in short form as a single 4-digit hex number.
5055 The short form is the same as the long form with the first 7 fields zero.
5056 Vnet ids must be non-zero and id 1 is reserved.
5058 \item \verb|<bridge>|: the name of a bridge interface to create for the vnet. Domains
5059 are connected to the vnet by connecting their virtual interfaces to the bridge.
5060 Bridge names are limited to 14 characters by the kernel.
5062 \item \verb|<vnetif>|: the name of the virtual interface onto the vnet (optional). The
5063 interface encapsulates and decapsulates vnet traffic for the network and is attached
5064 to the vnet bridge. Interface names are limited to 14 characters by the kernel.
5066 \item \verb|<level>|: security level for the vnet (optional). The level may be one of
5067 \begin{itemize}
5068 \item \verb|none|: no security (default). Vnet traffic is in clear on the network.
5069 \item \verb|auth|: authentication. Vnet traffic is authenticated using IPSEC
5070 ESP with hmac96.
5071 \item \verb|conf|: confidentiality. Vnet traffic is authenticated and encrypted
5072 using IPSEC ESP with hmac96 and AES-128.
5073 \end{itemize}
5074 Authentication and confidentiality are experimental and use hard-wired keys at present.
5075 \end{itemize}
5076 When a vnet is created its configuration is stored by \xend and the vnet persists until it is
5077 deleted using \texttt{xm vnet-delete <vnetid>}. The interfaces and bridges used by vnets
5078 are visible in the output of \texttt{ifconfig} and \texttt{brctl show}.
5080 \section{Example}
5081 If the file \path{vnet97.sxp} contains
5082 \begin{verbatim}
5083 (vnet (id 97) (bridge vnet97) (vnetif vnif97)
5084 (security none))
5085 \end{verbatim}
5086 Then \texttt{xm vnet-create vnet97.sxp} will define a vnet with id 97 and no security.
5087 The bridge for the vnet is called vnet97 and the virtual interface for it is vnif97.
5088 To add an interface on a domain to this vnet set its bridge to vnet97
5089 in its configuration. In Python:
5090 \begin{verbatim}
5091 vif="bridge=vnet97"
5092 \end{verbatim}
5093 In sxp:
5094 \begin{verbatim}
5095 (dev (vif (mac aa:00:00:01:02:03) (bridge vnet97)))
5096 \end{verbatim}
5097 Once the domain is started you should see its interface in the output of \texttt{brctl show}
5098 under the ports for \texttt{vnet97}.
5100 To get best performance it is a good idea to reduce the MTU of a domain's interface
5101 onto a vnet to 1400. For example using \texttt{ifconfig eth0 mtu 1400} or putting
5102 \texttt{MTU=1400} in \texttt{ifcfg-eth0}.
5103 You may also have to change or remove cached config files for eth0 under
5104 \texttt{/etc/sysconfig/networking}. Vnets work anyway, but performance can be reduced
5105 by IP fragmentation caused by the vnet encapsulation exceeding the hardware MTU.
5107 \section{Installing vnet support}
5108 Vnets are implemented using a kernel module, which needs to be loaded before
5109 they can be used. You can either do this manually before starting \xend, using the
5110 command \texttt{vn insmod}, or configure \xend to use the \path{network-vnet}
5111 script in the xend configuration file \texttt{/etc/xend/xend-config.sxp}:
5112 \begin{verbatim}
5113 (network-script network-vnet)
5114 \end{verbatim}
5115 This script insmods the module and calls the \path{network-bridge} script.
5117 The vnet code is not compiled and installed by default.
5118 To compile the code and install on the current system
5119 use \texttt{make install} in the root of the vnet source tree,
5120 \path{tools/vnet}. It is also possible to install to an installation
5121 directory using \texttt{make dist}. See the \path{Makefile} in
5122 the source for details.
5124 The vnet module creates vnet interfaces \texttt{vnif0002},
5125 \texttt{vnif0003} and \texttt{vnif0004} by default. You can test that
5126 vnets are working by configuring IP addresses on these interfaces
5127 and trying to ping them across the network. For example, using machines
5128 hostA and hostB:
5129 \begin{verbatim}
5130 hostA# ifconfig vnif0004 up
5131 hostB# ifconfig vnif0004 up
5132 hostB# ping
5133 \end{verbatim}
5135 The vnet implementation uses IP multicast to discover vnet interfaces, so
5136 all machines hosting vnets must be reachable by multicast. Network switches
5137 are often configured not to forward multicast packets, so this often
5138 means that all machines using a vnet must be on the same LAN segment,
5139 unless you configure vnet forwarding.
5141 You can test multicast coverage by pinging the vnet multicast address:
5142 \begin{verbatim}
5143 # ping -b
5144 \end{verbatim}
5145 You should see replies from all machines with the vnet module running.
5146 You can see if vnet packets are being sent or received by dumping traffic
5147 on the vnet UDP port:
5148 \begin{verbatim}
5149 # tcpdump udp port 1798
5150 \end{verbatim}
5152 If multicast is not being forwarded between machines you can configure
5153 multicast forwarding using vn. Suppose we have machines hostA on
5154 and hostB on and that multicast is not forwarded between them.
5155 We use vn to configure each machine to forward to the other:
5156 \begin{verbatim}
5157 hostA# vn peer-add hostB
5158 hostB# vn peer-add hostA
5159 \end{verbatim}
5160 Multicast forwarding needs to be used carefully - you must avoid creating forwarding
5161 loops. Typically only one machine on a subnet needs to be configured to forward,
5162 as it will forward multicasts received from other machines on the subnet.
5164 %% Chapter Glossary of Terms moved to glossary.tex
5165 \chapter{Glossary of Terms}
5167 \begin{description}
5169 \item[Domain] A domain is the execution context that contains a
5170 running {\bf virtual machine}. The relationship between virtual
5171 machines and domains on Xen is similar to that between programs and
5172 processes in an operating system: a virtual machine is a persistent
5173 entity that resides on disk (somewhat like a program). When it is
5174 loaded for execution, it runs in a domain. Each domain has a {\bf
5175 domain ID}.
5177 \item[Domain 0] The first domain to be started on a Xen machine.
5178 Domain 0 is responsible for managing the system.
5180 \item[Domain ID] A unique identifier for a {\bf domain}, analogous to
5181 a process ID in an operating system.
5183 \item[Full virtualization] An approach to virtualization which
5184 requires no modifications to the hosted operating system, providing
5185 the illusion of a complete system of real hardware devices.
5187 \item[Hypervisor] An alternative term for {\bf VMM}, used because it
5188 means `beyond supervisor', since it is responsible for managing
5189 multiple `supervisor' kernels.
5191 \item[Live migration] A technique for moving a running virtual machine
5192 to another physical host, without stopping it or the services
5193 running on it.
5195 \item[Paravirtualization] An approach to virtualization which requires
5196 modifications to the operating system in order to run in a virtual
5197 machine. Xen uses paravirtualization but preserves binary
5198 compatibility for user space applications.
5200 \item[Shadow pagetables] A technique for hiding the layout of machine
5201 memory from a virtual machine's operating system. Used in some {\bf
5202 VMMs} to provide the illusion of contiguous physical memory, in
5203 Xen this is used during {\bf live migration}.
5205 \item[Virtual Block Device] Persistent storage available to a virtual
5206 machine, providing the abstraction of an actual block storage device.
5207 {\bf VBD}s may be actual block devices, filesystem images, or
5208 remote/network storage.
5210 \item[Virtual Machine] The environment in which a hosted operating
5211 system runs, providing the abstraction of a dedicated machine. A
5212 virtual machine may be identical to the underlying hardware (as in
5213 {\bf full virtualization}, or it may differ, as in {\bf
5214 paravirtualization}).
5216 \item[VMM] Virtual Machine Monitor - the software that allows multiple
5217 virtual machines to be multiplexed on a single physical machine.
5219 \item[Xen] Xen is a paravirtualizing virtual machine monitor,
5220 developed primarily by the Systems Research Group at the University
5221 of Cambridge Computer Laboratory.
5223 \item[XenLinux] A name for the port of the Linux kernel that
5224 runs on Xen.
5226 \end{description}
5229 \end{document}
5232 %% Other stuff without a home
5234 %% Instructions Re Python API
5236 %% Other Control Tasks using Python
5237 %% ================================
5239 %% A Python module 'Xc' is installed as part of the tools-install
5240 %% process. This can be imported, and an 'xc object' instantiated, to
5241 %% provide access to privileged command operations:
5243 %% # import Xc
5244 %% # xc =
5245 %% # dir(xc)
5246 %% # help(xc.domain_create)
5248 %% In this way you can see that the class 'xc' contains useful
5249 %% documentation for you to consult.
5251 %% A further package of useful routines (xenctl) is also installed:
5253 %% # import xenctl.utils
5254 %% # help(xenctl.utils)
5256 %% You can use these modules to write your own custom scripts or you
5257 %% can customise the scripts supplied in the Xen distribution.
5261 % Explain about AGP GART
5264 %% If you're not intending to configure the new domain with an IP
5265 %% address on your LAN, then you'll probably want to use NAT. The
5266 %% 'xen_nat_enable' installs a few useful iptables rules into domain0
5267 %% to enable NAT. [NB: We plan to support RSIP in future]
5271 %% Installing the file systems from the CD
5272 %% =======================================
5274 %% If you haven't got an existing Linux installation onto which you
5275 %% can just drop down the Xen and Xenlinux images, then the file
5276 %% systems on the CD provide a quick way of doing an install. However,
5277 %% you would be better off in the long run doing a proper install of
5278 %% your preferred distro and installing Xen onto that, rather than
5279 %% just doing the hack described below:
5281 %% Choose one or two partitions, depending on whether you want a
5282 %% separate /usr or not. Make file systems on it/them e.g.:
5283 %% mkfs -t ext3 /dev/hda3
5284 %% [or mkfs -t ext2 /dev/hda3 && tune2fs -j /dev/hda3 if using an old
5285 %% version of mkfs]
5287 %% Next, mount the file system(s) e.g.:
5288 %% mkdir /mnt/root && mount /dev/hda3 /mnt/root
5289 %% [mkdir /mnt/usr && mount /dev/hda4 /mnt/usr]
5291 %% To install the root file system, simply untar /usr/XenDemoCD/root.tar.gz:
5292 %% cd /mnt/root && tar -zxpf /usr/XenDemoCD/root.tar.gz
5294 %% You'll need to edit /mnt/root/etc/fstab to reflect your file system
5295 %% configuration. Changing the password file (etc/shadow) is probably a
5296 %% good idea too.
5298 %% To install the usr file system, copy the file system from CD on
5299 %% /usr, though leaving out the "XenDemoCD" and "boot" directories:
5300 %% cd /usr && cp -a X11R6 etc java libexec root src bin dict kerberos
5301 %% local sbin tmp doc include lib man share /mnt/usr
5303 %% If you intend to boot off these file systems (i.e. use them for
5304 %% domain 0), then you probably want to copy the /usr/boot
5305 %% directory on the cd over the top of the current symlink to /boot
5306 %% on your root filesystem (after deleting the current symlink)
5307 %% i.e.:
5308 %% cd /mnt/root ; rm boot ; cp -a /usr/boot .