Diff for /wikisrc/ports/xen/howto.mdwn between versions 1.2 and 1.136

version 1.2, 2013/10/31 12:34:39 version 1.136, 2016/12/21 17:03:03
Line 1 Line 1
 NetBSD/xen Howto  Introduction
   Xen is a hypervisor (or virtual machine monitor) for x86 hardware
   (i686-class or higher), which supports running multiple guest
   operating systems on a single physical machine.  Xen is a Type 1 or
   bare-metal hypervisor; one uses the Xen kernel to control the CPU,
   memory and console, a dom0 operating system which mediates access to
   other hardware (e.g., disks, network, USB), and one or more domU
   operating systems which operate in an unprivileged virtualized
   environment.  IO requests from the domU systems are forwarded by the
   hypervisor (Xen) to the dom0 to be fulfilled.
   Xen supports two styles of guests.  The original is Para-Virtualized
   (PV) which means that the guest OS does not attempt to access hardware
   directly, but instead makes hypercalls to the hypervisor.  This is
   analogous to a user-space program making system calls.  (The dom0
   operating system uses PV calls for some functions, such as updating
   memory mapping page tables, but has direct hardware access for disk
   and network.)   PV guests must be specifically coded for Xen.
   The more recent style is HVM, which means that the guest does not have
   code for Xen and need not be aware that it is running under Xen.
   Attempts to access hardware registers are trapped and emulated.  This
   style is less efficient but can run unmodified guests.
   Generally any machine that runs NetBSD/amd64 will work with Xen and PV
   guests.  In theory i386 computers (without x86_64/amd64 support) can
   be used for Xen <= 4.2, but we have no recent reports of this working
   (this is a hint).  For HVM guests, hardware support is needed, but it
   is common on recent machines.  For Intel CPUs, one needs the VT-x
   extension, shown in "cpuctl identify 0" as VMX.  For AMD CPUs, one
   needs the AMD-V extensions, shown in "cpuctl identify 0" as SVM.
   There are further features for IOMMU virtualization, Intel's VT-d and
   AMD's AMD-Vi.  TODO: Explain whether Xen on NetBSD makes use of these
   features.  TODO: Review by someone who really understands this.
   Note that a FreeBSD dom0 requires VT-x and VT-d (or equivalent); this
   is because the FreeBSD dom0 does not run in PV mode.
   At boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   The dom0 can start one or more domUs.  (Booting is explained in detail
   in the dom0 section.)
   NetBSD supports Xen in that it can serve as dom0, be used as a domU,
   and that Xen kernels and tools are available in pkgsrc.  This HOWTO
   attempts to address both the case of running a NetBSD dom0 on hardware
   and running domUs under it (NetBSD and other), and also running NetBSD
   as a domU in a VPS.
   Xen 3.1 in pkgsrc supports "PCI passthrough", which means that
   specific PCI devices can be made available to a specific domU instead
   of the dom0.  This can be useful to let a domU run X11, or access some
   network interface or other peripheral.
   NetBSD 6 and earlier supported Xen 2; support was removed from NetBSD
   7.  Xen 2 has been removed from pkgsrc.
   Installing NetBSD/Xen is not extremely difficult, but it is more
   complex than a normal installation of NetBSD.
   In general, this HOWTO is occasionally overly restrictive about how
   things must be done, guiding the reader to stay on the established
   path when there are no known good reasons to stray.
   This HOWTO presumes a basic familiarity with the Xen system
   architecture, with installing NetBSD on i386/amd64 hardware, and with
   installing software from pkgsrc.  See also the [Xen
   Versions of Xen and NetBSD
   Most of the installation concepts and instructions are independent
   of Xen version and NetBSD version.  This section gives advice on
   which version to choose.  Versions not in pkgsrc and older unsupported
   versions of NetBSD are intentionally ignored.
   The term "amd64" is used to refer to both the NetBSD port and to the
   hardware architecture on which it runs.  (Such hardware is made by
   both Intel and AMD, and in 2016 a normal PC has this CPU
   In NetBSD, Xen is provided in pkgsrc, via matching pairs of packages
   xenkernel and xentools.  We will refer only to the kernel versions,
   but note that both packages must be installed together and must have
   matching versions.
   xenkernel3 provides Xen 3.1.  It is no longer maintained by Xen, and
   the last applied security patch was in 2011. Thus, it should not be
   used.  It supports PCI passthrough, which is why people use it anyway.
   Xen 3.1 runs on i386 (both non-PAE and PAE) and amd64 hardware.
   xenkernel33 provides Xen 3.3.  It is no longer maintained by Xen, and
   the last applied security patch was in 2012.  Thus, it should not be
   used.  Xen 3.3 runs on i386 PAE and amd64 hardware.  There are no good
   reasons to run this version.
   xenkernel41 provides Xen 4.1.  It is no longer maintained by Xen, but
   as of 2016-12 received backported security patches.  Xen 4.1 runs on
   i386 PAE and amd64 hardware.  There are no good reasons to run this
   xenkernel42 provides Xen 4.2.  It is no longer maintained by Xen, but
   as of 2016-12 received backported security patches.  Xen 4.2 runs on
   i386 PAE and amd64 hardware.  The only reason to run this is if you
   need to use xm instead of xl, or if you need to run on hardware that
   supports i386 but not amd64.  (This might also be useful if you need
   an i386 dom0, if it turns out that an amd64 Xen kernel and an i386
   dom0 is problematic.)
   xenkernel45 provides Xen 4.5.  As of 2016-12, security patches were
   released by Xen and applied to pkgsrc.  Xen 4.5 runs on amd64 hardware
   only.  While slightly old, 4.5 has been tested and run by others, so
   it is the conservative choice.
   xenkernel46 provides Xen 4.6.  It is new to pkgsrc as of 2016-05.  As
   of 2016-12, security patches were released by Xen and applied to
   pkgsrc.  Xen 4.6 runs on amd64 hardware only For new installations,
   4.6 is probably the appropriate choice and it will likely soon be the
   standard approach.  (If using Ubuntu guests, be sure to have the
   xentools46 from December, 2016).
   Xen 4.7 (released 2016-06) and 4.8 (released 2016-12) are not yet in
   See also the [Xen Security Advisory page](http://xenbits.xen.org/xsa/).
   Note that NetBSD support is called XEN3.  It works with Xen 3 and Xen
   4 because the hypercall interface has been stable.
   Xen command program
   Early Xen used a program called xm to manipulate the system from the
   dom0.  Starting in 4.1, a replacement program with similar behavior
   called xl is provided, but it does not work well in 4.1.  In 4.2, both
   xm and xl work fine.  4.4 is the last version that has xm.
   You must make a global choice to use xm or xl, because it affects not
   only which command you use, but the command used by rc.d scripts
   (specifically xendomains) and which daemons should be run.  The
   xentools packages provide xm for 3.1, 3.3 and 4.1 and xl for 4.2 and up.
   In 4.2, you can choose to use xm by simply changing the ctl_command
   variable and setting xend=YES in rc.conf.
   With xl, virtual devices are configured in parallel, which can cause
   problems if they are written assuming serial operation (e.g., updating
   firewall rules without explicit locking).  There is now locking for
   the provided scripts, which works for normal casses (e.g, file-backed
   xbd, where a vnd must be allocated).  But, as of 201612, it has not
   been adequately tested for a complex custom setup with a large number
   of interfaces.
   The netbsd-6, netbsd-7, and -current branches are all reasonable
   choices, with more or less the same considerations for non-Xen use.
   Therefore, netbsd-7 is recommended as the stable version of the most
   recent release for production use.  In addition, netbsd-7 and -current
   have a important scheduler fix (in November of 2015) affecting
   contention between dom0 and domUs; see
   https://releng.netbsd.org/cgi-bin/req-7.cgi?show=1040 for a
   description.  For those wanting to learn Xen or without production
   stability concerns, netbsd-7 is still likely most appropriate, but
   -current is also a reasonable choice.  (Xen runs ok on netbsd-5, but
   the xentools packages are likely difficult to build, and netbsd-5 is
   not supported.)
   As of NetBSD 6, a NetBSD domU will support multiple vcpus.  There is
   no SMP support for NetBSD as dom0.  (The dom0 itself doesn't really
   need SMP for dom0 functions; the lack of support is really a problem
   when using a dom0 as a normal computer.)
   Xen itself can run on i386 (Xen < 4.2) or amd64 hardware (all Xen
   versions).  (Practically, almost any computer where one would want to
   run Xen today supports amd64.)
   Xen, the dom0 system, and each domU system can be either i386 or
   amd64.  When building a xenkernel package, one obtains an i386 Xen
   kernel on an i386 host, and an amd64 Xen kernel on an amd64 host.  If
   the Xen kernel is i386, then the dom0 kernel and all domU kernels must
   be i386.  With an amd64 Xen kernel, an amd64 dom0 kernel is known to
   work, and an i386 dom0 kernel should in theory work.  An amd64
   Xen/dom0 is known to support both i386 and amd64 domUs.
   i386 dom0 and domU kernels must be PAE (except for an i386 Xen 3.1
   kernel, where one can use non-PAE for dom0 and all domUs); PAE kernels
   are included in the NetBSD default build.  (Note that emacs (at least)
   fails if run on i386 with PAE when built without, and vice versa,
   presumably due to bugs in the undump code.)
   Because of the above, the standard approach is to use an amd64 Xen
   kernel and NetBSD/amd64 for the dom0.  For domUs, NetBSD/i386 (with
   the PAE kernel) and NetBSD/amd64 are in widespread use, and there is
   little to no Xen-specific reason to prefer one over the other.
   Note that to use an i386 dom0 with Xen 4.5 or higher, one must build
   (or obtain from pre-built packages) an amd64 Xen kernel and install
   that on the system.  (One must also use a PAE i386 kernel, but this is
   also required with an i386 Xen kernel.).  Almost no one in the
   NetBSD/Xen community does this, and the standard, well-tested,
   approach is to use an amd64 dom0.
   A [posting on
   explained that PV system call overhead was higher on amd64, and thus
   there is some notion that i386 guests are faster.  It goes on to
   caution that the total situation is complex and not entirely
   understood. On top of that caution, the post is about Linux, not
   NetBSD.  TODO: Include link to benchmarks, if someone posts them.
   Mostly, NetBSD as a dom0 or domU is quite stable.
   However, there are some open PRs indicating problems.
    - [PR 48125](http://gnats.netbsd.org/48125)
    - [PR 47720](http://gnats.netbsd.org/47720)
   Note also that there are issues with sparse vnd(4) instances, but
   these are not about Xen -- they just are noticed with sparse vnd(4)
   instances in support of virtual disks in a dom0.
   Therefore, this HOWTO recommends running xenkernel45 or xenkernel46,
   xl, the NetBSD 7 stable branch, and to use an amd64 kernel as the
   dom0.  Either the i386PAE or amd64 version of NetBSD may be used as
   Because bugs are fixed quite often, and because of Xen security
   advisories, it is good to stay up to date with NetBSD (tracking a
   stable branch), with the Xen kernel (tracking a Xen version via
   pkgsrc), and with the Xen tools.  Specifically, NetBSD (-7 and
   -current) got an important fix affecting dom0/domU timesharing in
   November, 2015, and xentools46 got a fix to enable Ubuntu guests to
   boot in December, 2016.
   Ideally, all versions of Xen in pkgsrc would build on all supported
   versions of NetBSD/amd64, to the point where this section would be
   silly.  However, that has not always been the case.  Besides aging
   code and aging compilers, qemu (included in xentools for HVM support)
   is difficult to build.  Note that there is intentionally no data for
   4.5+ up for i386, and often omits xentools info if the corresponding
   kernel fails.
   The following table gives status, with the date last checked
   (generally on the most recent quarterly branch).  The first code is
   "builds" if it builds ok, and "FAIL" for a failure to build.  The
   second code/date only appears for xenkernel* and is "works" if it runs
   ok as a dom0 and can support a domU, and "FAIL" if it won't boot or
   run a domU.
           xenkernel3 netbsd-6 i386 FAIL 201612
           xenkernel33 netbsd-6 i386 FAIL 201612
           xenkernel41 netbsd-6 i386 builds 201612
           xenkernel42 netbsd-6 i386 builds 201612
           xentools3 netbsd-6 i386 FAIL 201612
           xentools33 netbsd-6 i386 FAIL 201612
           xentools41 netbsd-6 i386 builds 201612
           xentools42 netbsd-6 i386 FAIL 201612
           xenkernel3 netbsd-7 i386 FAIL 201412
           xenkernel33 netbsd-7 i386 FAIL 201412
           xenkernel41 netbsd-7 i386 builds 201412
           xenkernel42 netbsd-7 i386 builds 201412
           xentools41 netbsd-7 i386 builds 201412
           xentools42 netbsd-7 i386 ??FAIL 201412
           xenkernel3 netbsd-6 amd64 FAIL 201612
           xenkernel33 netbsd-6 amd64 FAIL 201612
           xenkernel41 netbsd-6 amd64 builds 201612 works 201612
           xenkernel42 netbsd-6 amd64 builds 201612 works 201612
           xenkernel45 netbsd-6 amd64 builds 201612
           xenkernel46 netbsd-6 amd64 builds 201612
           xentools41 netbsd-6 amd64 builds 201612
           xentools42 netbsd-6 amd64 builds 201612
           xentools45 netbsd-6 amd64 builds 201612
           xentools46 netbsd-6 amd64 FAIL 201612
           xenkernel3 netbsd-7 amd64 builds 201612
           xenkernel33 netbsd-7 amd64 builds 201612
           xenkernel41 netbsd-7 amd64 builds 201612
           xenkernel42 netbsd-7 amd64 builds 201612
           xenkernel45 netbsd-7 amd64 builds 201612
           xenkernel46 netbsd-7 amd64 builds 201612
           xentools3 netbsd-7 amd64 builds 201612
           xentools3-hvm netbsd-7 amd64 builds 201612
           xentools33 netbsd-7 amd64 FAIL 201612
           xentools41 netbsd-7 amd64 builds 201612
           xentools42 netbsd-7 amd64 builds 201612
           xentools45 netbsd-7 amd64 builds 201612
           xentools46 netbsd-7 amd64 builds 201612
   NetBSD as a dom0
 ================  ================
 [![BSD  NetBSD can be used as a dom0 and works very well.  The following
 daemon](../../images/BSD-daemon.jpg)](../../about/disclaimer.html#bsd-daemon)  sections address installation, updating NetBSD, and updating Xen.
   Note that it doesn't make sense to talk about installing a dom0 OS
   without also installing Xen itself.  We first address installing
   NetBSD, which is not yet a dom0, and then adding Xen, pivoting the
   NetBSD install to a dom0 install by just changing the kernel and boot
   For experimenting with Xen, a machine with as little as 1G of RAM and
   100G of disk can work.  For running many domUs in productions, far
   more will be needed; e.g. 4-8G and 1T of disk is reasonable for a
   half-dozen domUs of 512M and 32G each.  Basically, the RAM and disk
   have to be bigger than the sum of the RAM/disk needs of the dom0 and
   all the domUs.
   Styles of dom0 operation
   There are two basic ways to use Xen.  The traditional method is for
   the dom0 to do absolutely nothing other than providing support to some
   number of domUs.  Such a system was probably installed for the sole
   purpose of hosting domUs, and sits in a server room on a UPS.
   The other way is to put Xen under a normal-usage computer, so that the
   dom0 is what the computer would have been without Xen, perhaps a
   desktop or laptop.  Then, one can run domUs at will.  Purists will
   deride this as less secure than the previous approach, and for a
   computer whose purpose is to run domUs, they are right.  But Xen and a
   dom0 (without domUs) is not meaningfully less secure than the same
   things running without Xen.  One can boot Xen or boot regular NetBSD
   alternately with little problems, simply refraining from starting the
   Xen daemons when not running Xen.
   Note that NetBSD as dom0 does not support multiple CPUs.  This will
   limit the performance of the Xen/dom0 workstation approach.  In theory
   the only issue is that the "backend drivers" are not yet MPSAFE:
   Installation of NetBSD
   [install NetBSD/amd64](/guide/inst/)
   just as you would if you were not using Xen.
   However, the partitioning approach is very important.
   If you want to use RAIDframe for the dom0, there are no special issues
   for Xen.  Typically one provides RAID storage for the dom0, and the
   domU systems are unaware of RAID.  The 2nd-stage loader bootxx_* skips
   over a RAID1 header to find /boot from a file system within a RAID
   partition; this is no different when booting Xen.
   There are 4 styles of providing backing storage for the virtual disks
   used by domUs: raw partitions, LVM, file-backed vnd(4), and SAN.
   With raw partitions, one has a disklabel (or gpt) partition sized for
   each virtual disk to be used by the domU.  (If you are able to predict
   how domU usage will evolve, please add an explanation to the HOWTO.
   Seriously, needs tend to change over time.)
   One can use [lvm(8)](/guide/lvm/) to create logical devices to use
   for domU disks.  This is almost as efficient as raw disk partitions
   and more flexible.  Hence raw disk partitions should typically not
   be used.
   One can use files in the dom0 file system, typically created by dd'ing
   /dev/zero to create a specific size.  This is somewhat less efficient,
   but very convenient, as one can cp the files for backup, or move them
   between dom0 hosts.
   Finally, in theory one can place the files backing the domU disks in a
   SAN.  (This is an invitation for someone who has done this to add a
   HOWTO page.)
   Installation of Xen
   In the dom0, install sysutils/xenkernel42 and sysutils/xentools42 from
   pkgsrc (or another matching pair).  See [the pkgsrc
   documentation](http://www.NetBSD.org/docs/pkgsrc/) for help with
   pkgsrc.  Ensure that your packages are recent; the HOWTO does not
   contemplate old builds.
   For Xen 3.1, support for HVM guests is in sysutils/xentool3-hvm.  More
   recent versions have HVM support integrated in the main xentools
   package.  It is entirely reasonable to run only PV guests.
   Next you need to install the selected Xen kernel itself, which is
   installed by pkgsrc as "/usr/pkg/xen*-kernel/xen.gz".  Copy it to /.
   For debugging, one may copy xen-debug.gz; this is conceptually similar
   to DIAGNOSTIC and DEBUG in NetBSD.  xen-debug.gz is basically only
   useful with a serial console.  Then, place a NetBSD XEN3_DOM0 kernel
   in /, copied from releasedir/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
   of a NetBSD build.  If using i386, use
   releasedir/i386/binary/kernel/netbsd-XEN3PAE_DOM0.gz.  (If using Xen
   3.1 and i386, you may use XEN3_DOM0 with the non-PAE Xen.  But you
   should not use Xen 3.1.)  Both xen and the NetBSD kernel may be (and
   typically are) left compressed.
   In a dom0, kernfs is mandatory for xend to communicate with the
   kernel, so ensure that /kern is in fstab.  (A standard NetBSD install
   should already mount /kern.)
   Because you already installed NetBSD, you have a working boot setup
   with an MBR bootblock, either bootxx_ffsv1 or bootxx_ffsv2 at the
   beginning of your root file system, have /boot, and likely also
   /boot.cfg.  (If not, fix before continuing!)
   Add a line to to /boot.cfg to boot Xen.  See boot.cfg(5) for an
   example.  The basic line is
           menu=Xen:load /netbsd-XEN3_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=256M
   which specifies that the dom0 should have 256M, leaving the rest to be
   allocated for domUs.  To use a serial console, use
           menu=Xen:load /netbsd-XEN3_DOM0.gz console=com0;multiboot /xen.gz dom0_mem=256M console=com1 com1=9600,8n1
   which will use the first serial port for Xen (which counts starting
   from 1), forcing speed/parity, and also for NetBSD (which counts
   starting at 0).  In an attempt to add performance, one can also add
           dom0_max_vcpus=1 dom0_vcpus_pin
   to force only one vcpu to be provided (since NetBSD dom0 can't use
   more) and to pin that vcpu to a physical CPU.  TODO: benchmark this.
   Xen has [many boot
   and other than dom0 memory and max_vcpus, they are generally not
   As with non-Xen systems, you should have a line to boot /netbsd (a
   kernel that works without Xen).  Consider a line to boot /netbsd.ok (a
   fallback version of the non-Xen kernel, updated manually when you are
   sure /netbsd is ok).  Consider also a line to boot fallback versions
   of Xen and the dom0 kernel, but note that non-Xen NetBSD can be used
   to resolve Xen booting issues.
   Probably you want a default=N line to choose Xen in the absence of
   Now, reboot so that you are running a DOM0 kernel under Xen, rather
   than GENERIC without Xen.
   Using grub (historic)
   Before NetBSD's native bootloader could support Xen, the use of
   grub was recommended.  If necessary, see the
   [old grub information](/ports/xen/howto-grub).
   The [HowTo on Installing into
   explains how to set up booting a dom0 with Xen using grub with
   NetBSD's RAIDframe.  (This is obsolete with the use of NetBSD's native
   boot.  Now, just create a system with RAID-1, and alter /boot.cfg as
   described above.)
   Configuring Xen
   Xen logs will be in /var/log/xen.
   Now, you have a system that will boot Xen and the dom0 kernel, but not
   do anything else special.  Make sure that you have rebooted into Xen.
   There will be no domUs, and none can be started because you still have
   to configure the dom0 daemons.
   The daemons which should be run vary with Xen version and with whether
   one is using xm or xl.  The Xen 3.1, 3.3 and 4.1 packages use xm.  Xen
   4.2 and up packages use xl.  To use xm with 4.2, edit xendomains to
   use xm instead.
   For 3.1 and 3.3, you should enable xend and xenbackendd:
   For 4.1 and up, you should enable xencommons.  Not enabling xencommons
   will result in a hang; it is necessary to hit ^C on the console to let
   the machine finish booting.  If you are using xm (default in 4.1, or
   if you changed xendomains in 4.2), you should also enable xend:
           xend=YES # only if using xm, and only installed <= 4.2
   TODO: Recommend for/against xen-watchdog.
   After you have configured the daemons and either started them (in the
   order given) or rebooted, use xm or xl to inspect Xen's boot messages,
   available resources, and running domains.  An example with xl follows:
           # xl dmesg
           [xen's boot info]
           # xl info
           [available memory, etc.]
           # xl list
           Name              Id  Mem(MB)  CPU  State  Time(s)  Console
           Domain-0           0       64    0  r----     58.1
   ### Issues with xencommons
   xencommons starts xenstored, which stores data on behalf of dom0 and
   domUs.  It does not currently work to stop and start xenstored.
   Certainly all domUs should be shutdown first, following the sort order
   of the rc.d scripts.  However, the dom0 sets up state with xenstored,
   and is not notified when xenstored exits, leading to not recreating
   the state when the new xenstored starts.  Until there's a mechanism to
   make this work, one should not expect to be able to restart xenstored
   (and thus xencommons).  There is currently no reason to expect that
   this will get fixed any time soon.
   ### No-longer needed advice about devices
   The installation of NetBSD should already have created devices for xen
   (xencons, xenevt, xsd_kva), but if they are not present, create them:
           cd /dev && sh MAKEDEV xen
   anita (for testing NetBSD)
   With the setup so far (assuming 4.2/xl), one should be able to run
   anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
   root, because anita must create a domU):
           anita --vmm=xl test file:///usr/obj/i386/
   Alternatively, one can use --vmm=xm to use xm-based domU creation
   instead (and must, on Xen <= 4.1).   TODO: confirm that anita xl really works.
   Xen-specific NetBSD issues
   There are (at least) two additional things different about NetBSD as a
   dom0 kernel compared to hardware.
   One is that the module ABI is different because some of the #defines
   change, so one must build modules for Xen.  As of netbsd-7, the build
   system does this automatically.  TODO: check this.  (Before building
   Xen modules was added, it was awkward to use modules to the point
   where it was considered that it did not work.)
   The other difference is that XEN3_DOM0 does not have exactly the same
   options as GENERIC.  While it is debatable whether or not this is a
   bug, users should be aware of this and can simply add missing config
   items if desired.
   Updating NetBSD in a dom0
   This is just like updating NetBSD on bare hardware, assuming the new
   version supports the version of Xen you are running.  Generally, one
   replaces the kernel and reboots, and then overlays userland binaries
   and adjusts /etc.
 Table Of Contents  Note that one must update both the non-Xen kernel typically used for
 -----------------  rescue purposes and the DOM0 kernel used with Xen.
 -   [Introduction](#introduction)  Converting from grub to /boot
 -   [Installing NetBSD as privileged domain (Dom0)](#netbsd-dom0)  -----------------------------
 -   [Creating an unprivileged NetBSD domain (DomU)](#netbsd-domU)  
 -   [Creating an unprivileged Linux domain (DomU)](#linux-domU)  
 -   [Creating an unprivileged Solaris domain (DomU)](#solaris-domU)  
 -   [Using PCI devices in guest domains](#pci-pass-through)  
 -   [Links and further information](#links-and-more)  
 * * * * *  These instructions were [TODO: will be] used to convert a system from
   grub to /boot.  The system was originally installed in February of
   2006 with a RAID1 setup and grub to boot Xen 2, and has been updated
   over time.  Before these commands, it was running NetBSD 6 i386, Xen
   4.1 and grub, much like the message linked earlier in the grub
           # Install MBR bootblocks on both disks. 
           fdisk -i /dev/rwd0d
           fdisk -i /dev/rwd1d
           # Install NetBSD primary boot loader (/ is FFSv1) into RAID1 components.
           installboot -v /dev/rwd0d /usr/mdec/bootxx_ffsv1
           installboot -v /dev/rwd1d /usr/mdec/bootxx_ffsv1
           # Install secondary boot loader
           cp -p /usr/mdec/boot /
           # Create boot.cfg following earlier guidance:
           menu=Xen:load /netbsd-XEN3PAE_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=256M
           menu=Xen.ok:load /netbsd-XEN3PAE_DOM0.ok.gz console=pc;multiboot /xen.ok.gz dom0_mem=256M
           menu=GENERIC single-user:boot -s
           menu=GENERIC.ok:boot netbsd.ok
           menu=GENERIC.ok single-user:boot netbsd.ok -s
           menu=Drop to boot prompt:prompt
   TODO: actually do this and fix it if necessary.
   Upgrading Xen versions
   Minor version upgrades are trivial.  Just rebuild/replace the
   xenkernel version and copy the new xen.gz to / (where /boot.cfg
   references it), and reboot.
   Major version upgrades are conceptually not difficult, but can run
   into all the issues found when installing Xen.  Assuming migration
   from 4.1 to 4.2, remove the xenkernel41 and xentools41 packages and
   install the xenkernel42 and xentools42 packages.  Copy the 4.2 xen.gz
   to /.
   Ensure that the contents of /etc/rc.d/xen* are correct.  Specifically,
   they must match the package you just installed and not be left over
   from some previous installation.
   Enable the correct set of daemons; see the configuring section above.
   (Upgrading from 3.x to 4.x without doing this will result in a hang.)
   Ensure that the domU config files are valid for the new version.
   Specifically, for 4.x remove autorestart=True, and ensure that disks
   are specified with numbers as the second argument, as the examples
   above show, and not NetBSD device names.
   Hardware known to work
   Arguably, this section is misplaced, and there should be a page of
   hardware that runs NetBSD/amd64 well, with the mostly-well-founded
   assumption that NetBSD/xen runs fine on any modern hardware that
   NetBSD/amd64 runs well on.  Until then, we give motherboard/CPU (and
   sometimes RAM) pairs/triples to aid those choosing a motherboard.
   Note that Xen systems usually do not run X, so a listing here does not
   imply that X works at all.
           Supermicro X9SRL-F, Xeon E5-1650 v2, 96 GiB ECC
           Supermicro ??, Atom C2758 (8 core), 32 GiB ECC
           ASUS M5A78L-M/USB3 AM3+ microATX, AMD Piledriver X8 4000MHz, 16 GiB ECC
   Older hardware:
           Intel D915GEV, Pentium4 CPU 3.40GHz, 4GB 533MHz Synchronous DDR2
           INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E6850  @ 3.00GHz"
           INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E8400  @ 3.00GHz"
   Running Xen under qemu
   The astute reader will note that this section is somewhat twisted.
   However, it can be useful to run Xen under qemu either because the
   version of NetBSD as a dom0 does not run on the hardware in use, or to
   generate automated test cases involving Xen.
   In 2015-01, the following combination was reported to mostly work:
           host OS: NetBSD/amd64 6.1.4
           qemu: 2.2.0 from pkgsrc
           Xen kernel: xenkernel42-4.2.5nb1 from pkgsrc
           dom0 kernel: NetBSD/amd64 6.1.5
           Xen tools: xentools42-4.2.5 from pkgsrc
   See [PR 47720](http://gnats.netbsd.org/47720) for a problem with dom0
   Unprivileged domains (domU)
   This section describes general concepts about domUs.  It does not
   address specific domU operating systems or how to install them.  The
   config files for domUs are typically in /usr/pkg/etc/xen, and are
   typically named so that the file name, domU name and the domU's host
   name match.
   The domU is provided with CPU and memory by Xen, configured by the
   dom0.  The domU is provided with disk and network by the dom0,
   mediated by Xen, and configured in the dom0.
   Entropy in domUs can be an issue; physical disks and network are on
   the dom0.  NetBSD's /dev/random system works, but is often challenged.
   Config files
   There is no good order to present config files and the concepts
   surrounding what is being configured.  We first show an example config
   file, and then in the various sections give details.
   See (at least in xentools41) /usr/pkg/share/examples/xen/xmexample*,
   for a large number of well-commented examples, mostly for running
   The following is an example minimal domain configuration file
   "/usr/pkg/etc/xen/foo".  It is (with only a name change) an actual
   known working config file on Xen 4.1 (NetBSD 5 amd64 dom0 and NetBSD 5
   i386 domU).  The domU serves as a network file server.
           # -*- mode: python; -*-
           kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
           memory = 1024
           vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
           disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
                    'file:/n0/xen/foo-wd1,0x1,w' ]
   The domain will have the same name as the file.  The kernel has the
   host/domU name in it, so that on the dom0 one can update the various
   domUs independently.  The vif line causes an interface to be provided,
   with a specific mac address (do not reuse MAC addresses!), in bridge
   mode.  Two disks are provided, and they are both writable; the bits
   are stored in files and Xen attaches them to a vnd(4) device in the
   dom0 on domain creation.  The system treats xbd0 as the boot device
   without needing explicit configuration.
   By default xm looks for domain config files in /usr/pkg/etc/xen.  Note
   that "xm create" takes the name of a config file, while other commands
   take the name of a domain.  To create the domain, connect to the
   console, create the domain while attaching the console, shutdown the
   domain, and see if it has finished stopping, do (or xl with Xen >=
           xm create foo
           xm console foo
           xm create -c foo
           xm shutdown foo
           xm list
   Typing ^] will exit the console session.  Shutting down a domain is
   equivalent to pushing the power button; a NetBSD domU will receive a
   power-press event and do a clean shutdown.  Shutting down the dom0
   will trigger controlled shutdowns of all configured domUs.
   domU kernels
   On a physical computer, the BIOS reads sector 0, and a chain of boot
   loaders finds and loads a kernel.  Normally this comes from the root
   file system.  With Xen domUs, the process is totally different.  The
   normal path is for the domU kernel to be a file in the dom0's
   file system.  At the request of the dom0, Xen loads that kernel into a
   new domU instance and starts execution.  While domU kernels can be
   anyplace, reasonable places to store domU kernels on the dom0 are in /
   (so they are near the dom0 kernel), in /usr/pkg/etc/xen (near the
   config files), or in /u0/xen (where the vdisks are).
   Note that loading the domU kernel from the dom0 implies that boot
   blocks, /boot, /boot.cfg, and so on are all ignored in the domU.
   See the VPS section near the end for discussion of alternate ways to
   obtain domU kernels.
   CPU and memory
   A domain is provided with some number of vcpus, less than the number
   of CPUs seen by the hypervisor.  (For a dom0, this is controlled by
   the boot argument "dom0_max_vcpus=1".)  For a domU, it is controlled
   from the config file by the "vcpus = N" directive.
   A domain is provided with memory; this is controlled in the config
   file by "memory = N" (in megabytes).  In the straightforward case, the
   sum of the the memory allocated to the dom0 and all domUs must be less
   than the available memory.
   Xen also provides a "balloon" driver, which can be used to let domains
   use more memory temporarily.  TODO: Explain better, and explain how
   well it works with NetBSD.
   Virtual disks
   With the file/vnd style, typically one creates a directory,
   e.g. /u0/xen, on a disk large enough to hold virtual disks for all
   domUs.  Then, for each domU disk, one writes zeros to a file that then
   serves to hold the virtual disk's bits; a suggested name is foo-xbd0
   for the first virtual disk for the domU called foo.  Writing zeros to
   the file serves two purposes.  One is that preallocating the contents
   improves performance.  The other is that vnd on sparse files has
   failed to work.  TODO: give working/notworking NetBSD versions for
   sparse vnd and gnats reference.  Note that the use of file/vnd for Xen
   is not really different than creating a file-backed virtual disk for
   some other purpose, except that xentools handles the vnconfig
   commands.  To create an empty 4G virtual disk, simply do
           dd if=/dev/zero of=foo-xbd0 bs=1m count=4096
   Do not use qemu-img-xen, because this will create sparse file.  There
   have been recent (2015) reports of sparse vnd(4) devices causing
   lockups, but there is apparently no PR.
   With the lvm style, one creates logical devices.  They are then used
   similarly to vnds.  TODO: Add an example with lvm.
   In domU config files, the disks are defined as a sequence of 3-tuples.
   The first element is "method:/path/to/disk".  Common methods are
   "file:" for file-backed vnd. and "phy:" for something that is already
   a (TODO: character or block) device.
   The second element is an artifact of how virtual disks are passed to
   Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
   are given a device name to associate with the disk, and values like
   "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
   as xbd0, the second as xbd1, and so on.  However, xm/xl demand a
   second argument.  The name given is converted to a major/minor by
   calling stat(2) on the name in /dev and this is passed to the domU.
   In the general case, the dom0 and domU can be different operating
   systems, and it is an unwarranted assumption that they have consistent
   numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
   as both dom0 and domU, using values of 0x0 for the first disk and 0x1
   for the second works fine and avoids this issue.  For a GNU/Linux
   guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
   The third element is "w" for writable disks, and "r" for read-only
   Note that NetBSD by default creates only vnd[0123].  If you need more
   than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
   Note that NetBSD by default creates only xbd[0123].  If you need more
   virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
   Virtual Networking
   Xen provides virtual Ethernets, each of which connects the dom0 and a
   domU.  For each virtual network, there is an interface "xvifN.M" in
   the dom0, and in domU index N, a matching interface xennetM (NetBSD
   name).  The interfaces behave as if there is an Ethernet with two
   adapters connected.  From this primitive, one can construct various
   configurations.  We focus on two common and useful cases for which
   there are existing scripts: bridging and NAT.
   With bridging (in the example above), the domU perceives itself to be
   on the same network as the dom0.  For server virtualization, this is
   usually best.  Bridging is accomplished by creating a bridge(4) device
   and adding the dom0's physical interface and the various xvifN.0
   interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
   config file.  The bridge must be set up already in the dom0; an
   example /etc/ifconfig.bridge0 is:
           !brconfig bridge0 add wm0
   With NAT, the domU perceives itself to be behind a NAT running on the
   dom0.  This is often appropriate when running Xen on a workstation.
   TODO: NAT appears to be configured by "vif = [ '' ]".
   The MAC address specified is the one used for the interface in the new
   domain.  The interface in dom0 will use this address XOR'd with
   00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
   Sizing domains
   Modern x86 hardware has vast amounts of resources.  However, many
   virtual servers can function just fine on far less.  A system with
   256M of RAM and a 4G disk can be a reasonable choice.  Note that it is
   far easier to adjust virtual resources than physical ones.  For
   memory, it's just a config file edit and a reboot.  For disk, one can
   create a new file and vnconfig it (or lvm), and then dump/restore,
   just like updating physical disks, but without having to be there and
   without those pesky connectors.
   Starting domains automatically
   To start domains foo at bar at boot and shut them down cleanly on dom0
   shutdown, in rc.conf add:
           xendomains="foo bar"
   Note that earlier versions of the xentools41 xendomains rc.d script
   used xl, when one should use xm with 4.1.
   Creating specific unprivileged domains (domU)
   Creating domUs is almost entirely independent of operating system.  We
   have already presented the basics of config files.  Note that you must
   have already completed the dom0 setup so that "xl list" (or "xm list")
   Creating an unprivileged NetBSD domain (domU)
   See the earlier config file, and adjust memory.  Decide on how much
   storage you will provide, and prepare it (file or lvm).
   While the kernel will be obtained from the dom0 file system, the same
   file should be present in the domU as /netbsd so that tools like
   savecore(8) can work.   (This is helpful but not necessary.)
   The kernel must be specifically for Xen and for use as a domU.  The
   i386 and amd64 provide the following kernels:
           i386 XEN3_DOMU
           i386 XEN3PAE_DOMU
           amd64 XEN3_DOMU
   Unless using Xen 3.1 (and you shouldn't) with i386-mode Xen, you must
   use the PAE version of the i386 kernel.
   This will boot NetBSD, but this is not that useful if the disk is
   empty.  One approach is to unpack sets onto the disk outside of xen
   (by mounting it, just as you would prepare a physical disk for a
   system you can't run the installer on).
   A second approach is to run an INSTALL kernel, which has a miniroot
   and can load sets from the network.  To do this, copy the INSTALL
   kernel to / and change the kernel line in the config file to:
 ### Introduction          kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
 [![[Xen  Then, start the domain as "xl create -c configname".
   Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
   line should be used in the config file.
 Xen is a virtual machine monitor for x86 hardware (requires i686-class      disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
 CPUs), which supports running multiple guest operating systems on a  
 single machine. Guest OSes (also called <E2><80><9C>domains<E2><80><9D>) require a modified  
 kernel which supports Xen hypercalls in replacement to access to the  
 physical hardware. At boot, the Xen kernel (also known as the Xen  
 hypervisor) is loaded (via the bootloader) along with the guest kernel  
 for the first domain (called *domain0*). The Xen kernel has to be loaded  
 using the multiboot protocol. You would use the NetBSD boot loader for  
 this, or alternatively the **grub** boot loader (**grub** has some  
 limitations, detailed below). *domain0* has special privileges to access  
 the physical hardware (PCI and ISA devices), administrate other domains  
 and provide virtual devices (disks and network) to other domains that  
 lack those privileges. For more details, see  
 NetBSD can be used for both *domain0 (Dom0)* and further, unprivileged  
 (DomU) domains. (Actually there can be multiple privileged domains  
 accessing different parts of the hardware, all providing virtual devices  
 to unprivileged domains. We will only talk about the case of a single  
 privileged domain, *domain0*). *domain0* will see physical devices much  
 like a regular i386 or amd64 kernel, and will own the physical console  
 (VGA or serial). Unprivileged domains will only see a character-only  
 virtual console, virtual disks (`xbd`{.code}) and virtual network  
 interfaces (`xennet`{.code}) provided by a privileged domain (usually  
 *domain0*). xbd devices are connected to a block device (i.e., a  
 partition of a disk, raid, ccd, ... device) in the privileged domain.  
 xennet devices are connected to virtual devices in the privileged  
 domain, named xvif\<domain number\>.\<if number for this domain\>, e.g.,  
 xvif1.0. Both xennet and xvif devices are seen as regular Ethernet  
 devices (they can be seen as a crossover cable between 2 PCs) and can be  
 assigned addresses (and be routed or NATed, filtered using IPF, etc ...)  
 or be added as part of a bridge.  
 * * * * *  
 ### Installing NetBSD as privileged domain (Dom0)  
 First do a NetBSD/i386 or NetBSD/amd64  
 [installation](../../docs/guide/en/chap-inst.html) of the 5.1 release  
 (or newer) as you usually do on x86 hardware. The binary releases are  
 available from  
 Binary snapshots for current and the stable branches are available on  
 [daily autobuilds](http://nyftp.NetBSD.org/pub/NetBSD-daily/). If you  
 plan to use the **grub** boot loader, when partitioning the disk you  
 have to make the root partition smaller than 512Mb, and formatted as  
 FFSv1 with 8k block/1k fragments. If the partition is larger than this,  
 uses FFSv2 or has different block/fragment sizes, grub may fail to load  
 some files. Also keep in mind that you'll probably want to provide  
 virtual disks to other domains, so reserve some partitions for these  
 virtual disks. Alternatively, you can create large files in the file  
 system, map them to vnd(4) devices and export theses vnd devices to  
 other domains.  
 Next step is to install the Xen packages via pkgsrc or from binary  
 packages. See [the pkgsrc  
 documentation](http://www.NetBSD.org/docs/pkgsrc/) if you are unfamiliar  
 with pkgsrc and/or handling of binary packages. Xen 3.1, 3.3, 4.1 and  
 4.2 are available. 3.1 supports PCI pass-through while other versions do  
 not. You'll need either  
 for Xen 3.1,  
 for Xen 3.3,  
 for Xen 4.1. or  
 for Xen 4.2. You'll also need  
 if you plan do use the grub boot loader. If using Xen 3.1, you may also  
 want to install  
 which contains the utilities to run unmodified guests OSes using the  
 *HVM* support (for later versions this is included in  
 Note that your CPU needs to support this. Intel CPUs must have the 'VT'  
 instruction, AMD CPUs the 'SVM' instruction. You can easily find out if  
 your CPU support HVM by using NetBSD's cpuctl command:  
 # cpuctl identify 0  
 cpu0: Intel Core 2 (Merom) (686-class), id 0x6f6  
 cpu0: features 0xbfebfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR>  
 cpu0: features 0xbfebfbff<PGE,MCA,CMOV,PAT,PSE36,CFLUSH,DS,ACPI,MMX>  
 cpu0: features 0xbfebfbff<FXSR,SSE,SSE2,SS,HTT,TM,SBF>  
 cpu0: features2 0x4e33d<SSE3,DTES64,MONITOR,DS-CPL,VMX,TM2,SSSE3,CX16,xTPR,PDCM,DCA>  
 cpu0: features3 0x20100800<SYSCALL/SYSRET,XD,EM64T>  
 cpu0: "Intel(R) Xeon(R) CPU            5130  @ 2.00GHz"  
 cpu0: I-cache 32KB 64B/line 8-way, D-cache 32KB 64B/line 8-way  
 cpu0: L2 cache 4MB 64B/line 16-way  
 cpu0: ITLB 128 4KB entries 4-way  
 cpu0: DTLB 256 4KB entries 4-way, 32 4MB entries 4-way  
 cpu0: Initial APIC ID 0  
 cpu0: Cluster/Package ID 0  
 cpu0: Core ID 0  
 cpu0: family 06 model 0f extfamily 00 extmodel 00  
 Depending on your CPU, the feature you are looking for is called HVM,  
 SVM or VMX.  
 Next you need to copy the selected Xen kernel itself. pkgsrc installed  
 them under `/usr/pkg/xen*-kernel/`{.filename}. The file you're looking  
 for is `xen.gz`{.filename}. Copy it to your root file system.  
 `xen-debug.gz`{.filename} is a kernel with more consistency checks and  
 more details printed on the serial console. It is useful for debugging  
 crashing guests if you use a serial console. It is not useful with a VGA  
 You'll then need a NetBSD/Xen kernel for *domain0* on your root file  
 system. The XEN3PAE\_DOM0 kernel or XEN3\_DOM0 provided as part of the  
 i386 or amd64 binaries is suitable for this, but you may want to  
 customize it. Keep your native kernel around, as it can be useful for  
 recovery. *Note:* the *domain0* kernel must support KERNFS and  
 `/kern`{.filename} must be mounted because *xend* needs access to  
 Next you need to get a bootloader to load the `xen.gz`{.filename}  
 kernel, and the NetBSD *domain0* kernel as a module. This can be  
 **grub** or NetBSD's boot loader. Below is a detailled example for grub,  
 see the boot.cfg(5) manual page for an example using the latter.  
 This is also where you'll specify the memory allocated to *domain0*, the  
 console to use, etc ...  
 Here is a commented `/grub/menu.lst`{.filename} file:  
 ~~~ {.programlisting}  
 #Grub config file for NetBSD/xen. Copy as /grub/menu.lst and run  
 # grub-install /dev/rwd0d (assuming your boot device is wd0).  
 # The default entry to load will be the first one  
 # boot the default entry after 10s if the user didn't hit keyboard  
 # Configure serial port to use as console. Ignore if you'll use VGA only  
 serial --unit=0 --speed=115200 --word=8 --parity=no --stop=1  
 # Let the user select which console to use (serial or VGA), default  
 # to serial after 10s  
 terminal --timeout=10 serial console  
 # An entry for NetBSD/xen, using /netbsd as the domain0 kernel, and serial  
 # console. Domain0 will have 64MB RAM allocated.  
 # Assume NetBSD is installed in the first MBR partition.  
 title Xen 3 / NetBSD (hda0, serial)  
   kernel (hd0,a)/xen.gz dom0_mem=65536 com1=115200,8n1  
   module (hd0,a)/netbsd bootdev=wd0a ro console=ttyS0  
 # Same as above, but using VGA console  
 # We can use console=tty0 (Linux syntax) or console=pc (NetBSD syntax)  
 title Xen 3 / NetBSD (hda0, vga)  
   kernel (hd0,a)/xen.gz dom0_mem=65536  
   module (hd0,a)/netbsd bootdev=wd0a ro console=tty0  
 # NetBSD/xen using a backup domain0 kernel (in case you installed a  
 # nonworking kernel as /netbsd  
 title Xen 3 / NetBSD (hda0, backup, serial)  
   kernel (hd0,a)/xen.gz dom0_mem=65536 com1=115200,8n1  
   module (hd0,a)/netbsd.backup bootdev=wd0a ro console=ttyS0  
 title Xen 3 / NetBSD (hda0, backup, VGA)  
   kernel (hd0,a)/xen.gz dom0_mem=65536  
   module (hd0,a)/netbsd.backup bootdev=wd0a ro console=tty0  
 #Load a regular NetBSD/i386 kernel. Can be useful if you end up with a  
 #nonworking /xen.gz  
 title NetBSD 5.1  
   root (hd0,a)  
   kernel --type=netbsd /netbsd-GENERIC  
 #Load the NetBSD bootloader, letting it load the NetBSD/i386 kernel.  
 #May be better than the above, as grub can't pass all required infos  
 #to the NetBSD/i386 kernel (e.g. console, root device, ...)  
 title NetBSD chain  
   root        (hd0,0)  
   chainloader +1  
 ## end of grub config file.  
 Install grub with the following command:  
 ~~~ {.programlisting}  
 # grub --no-floppy  
 grub> root (hd0,a)  
  Filesystem type is ffs, partition type 0xa9  
 grub> setup (hd0)  
  Checking if "/boot/grub/stage1" exists... no  
  Checking if "/grub/stage1" exists... yes  
  Checking if "/grub/stage2" exists... yes  
  Checking if "/grub/ffs_stage1_5" exists... yes  
  Running "embed /grub/ffs_stage1_5 (hd0)"...  14 sectors are embedded.  
  Running "install /grub/stage1 (hd0) (hd0)1+14 p (hd0,0,a)/grub/stage2 /grub/menu.lst"...  
 * * * * *  
 ### Creating an unprivileged NetBSD domain (DomU)  
 Once you have *domain0* running, you need to start the xen tool daemon  
 (**/usr/pkg/share/examples/rc.d/xend start**) and the xen backend daemon  
 (**/usr/pkg/share/examples/rc.d/xenbackendd start** for Xen3\*,  
 **/usr/pkg/share/examples/rc.d/xencommons start** for Xen4.\*). Make  
 sure that `/dev/xencons`{.filename} and `/dev/xenevt`{.filename} exist  
 before starting **xend**. You can create them with this command:  
 ~~~ {.programlisting}  
 # cd /dev && sh MAKEDEV xen  
 xend will write logs to `/var/log/xend.log`{.filename} and  
 `/var/log/xend-debug.log`{.filename}. You can then control xen with the  
 xm tool. 'xm list' will show something like:  
 ~~~ {.programlisting}  
 # xm list  
 Name              Id  Mem(MB)  CPU  State  Time(s)  Console  
 Domain-0           0       64    0  r----     58.1  
 'xm create' allows you to create a new domain. It uses a config file in  
 PKG\_SYSCONFDIR for its parameters. By default, this file will be in  
 `/usr/pkg/etc/xen/`{.filename}. On creation, a kernel has to be  
 specified, which will be executed in the new domain (this kernel is in  
 the *domain0* file system, not on the new domain virtual disk; but  
 please note, you should install the same kernel into *domainU* as  
 `/netbsd`{.filename} in order to make your system tools, like  
 work). A suitable kernel is provided as part of the i386 and amd64  
 binary sets: XEN3\_DOMU.  
 Here is an /usr/pkg/etc/xen/nbsd example config file:  
 ~~~ {.programlisting}  
 #  -*- mode: python; -*-  
 # Python defaults setup for 'xm create'.  
 # Edit this file to reflect the configuration of your system.  
 # Kernel image file. This kernel will be loaded in the new domain.  
 kernel = "/home/bouyer/netbsd-XEN3_DOMU"  
 #kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"  
 # Memory allocation (in megabytes) for the new domain.  
 memory = 128  
 # A handy name for your new domain. This will appear in 'xm list',  
 # and you can use this as parameters for xm in place of the domain  
 # number. All domains must have different names.  
 name = "nbsd"  
 # The number of virtual CPUs this domain has.  
 vcpus = 1  
 # Define network interfaces for the new domain.  
 # Number of network interfaces (must be at least 1). Default is 1.  
 nics = 1  
 # Define MAC and/or bridge for the network interfaces.  
 # The MAC address specified in ``mac'' is the one used for the interface  
 # in the new domain. The interface in domain0 will use this address XOR'd  
 # with 00:00:00:01:00:00 (i.e. aa:00:00:51:02:f0 in our example). Random  
 # MACs are assigned if not given.  
 # ``bridge'' is a required parameter, which will be passed to the  
 # vif-script called by xend(8) when a new domain is created to configure  
 # the new xvif interface in domain0.  
 # In this example, the xvif is added to bridge0, which should have been  
 # set up prior to the new domain being created -- either in the  
 # ``network'' script or using a /etc/ifconfig.bridge0 file.  
 vif = [ 'mac=aa:00:00:50:02:f0, bridge=bridge0' ]  
 # Define the disk devices you want the domain to have access to, and  
 # what you want them accessible as.  
 # Each disk entry is of the form:  
 #       phy:DEV,VDEV,MODE  
 # where DEV is the device, VDEV is the device name the domain will see,  
 # and MODE is r for read-only, w for read-write.  You can also create  
 # file-backed domains using disk entries of the form:  
 #       file:PATH,VDEV,MODE  
 # where PATH is the path to the file used as the virtual disk, and VDEV  
 # and MODE have the same meaning as for ``phy'' devices.  
 # VDEV doesn't really matter for a NetBSD guest OS (it's just used as an index),  
 # but it does for Linux.  
 # Worse, the device has to exist in /dev/ of domain0, because xm will  
 # try to stat() it. This means that in order to load a Linux guest OS  
 # from a NetBSD domain0, you'll have to create /dev/hda1, /dev/hda2, ...  
 # on domain0, with the major/minor from Linux :(  
 # Alternatively it's possible to specify the device number in hex,  
 # e.g. 0x301 for /dev/hda1, 0x302 for /dev/hda2, etc ...  
 disk = [ 'phy:/dev/wd0e,0x1,w' ]  
 #disk = [ 'file:/var/xen/nbsd-disk,0x01,w' ]  
 #disk = [ 'file:/var/xen/nbsd-disk,0x301,w' ]  
 # Set the kernel command line for the new domain.  
 # Set root device. This one does matter for NetBSD  
 root = "xbd0"  
 # extra parameters passed to the kernel  
 # this is where you can set boot flags like -s, -a, etc ...  
 #extra = ""  
 # Set according to whether you want the domain restarted when it exits.  
 # The default is False.  
 #autorestart = True  
 # end of nbsd config file ====================================================  
 When a new domain is created, xen calls the  
 `/usr/pkg/etc/xen/vif-bridge`{.filename} script for each virtual network  
 interface created in *domain0*. This can be used to automatically  
 configure the xvif?.? interfaces in *domain0*. In our example, these  
 will be bridged with the bridge0 device in *domain0*, but the bridge has  
 to exist first. To do this, create the file  
 `/etc/ifconfig.bridge0`{.filename} and make it look like this:  
 ~~~ {.programlisting}  
 !brconfig $int add ex0 up  
 (replace `ex0`{.literal} with the name of your physical interface). Then  
 bridge0 will be created on boot. See the  
 man page for details.  
 So, here is a suitable `/usr/pkg/etc/xen/vif-bridge`{.filename} for  
 xvif?.? (a working vif-bridge is also provided with xentools20)  
 ~~~ {.programlisting}  
 # $NetBSD: howto.mdwn,v 1.1 2013/10/31 12:20:57 mspo Exp $  
 # /usr/pkg/etc/xen/vif-bridge  
 # Script for configuring a vif in bridged mode with a dom0 interface.  
 # The xend(8) daemon calls a vif script when bringing a vif up or down.  
 # The script name to use is defined in /usr/pkg/etc/xen/xend-config.sxp  
 # in the ``vif-script'' field.  
 # Usage: vif-bridge up|down [var=value ...]  
 # Actions:  
 #    up         Adds the vif interface to the bridge.  
 #    down       Removes the vif interface from the bridge.  
 # Variables:  
 #    domain     name of the domain the interface is on (required).  
 #    vifq       vif interface name (required).  
 #    mac        vif MAC address (required).  
 #    bridge     bridge to add the vif to (required).  
 # Example invocation:  
 # vif-bridge up domain=VM1 vif=xvif1.0 mac="ee:14:01:d0:ec:af" bridge=bridge0  
 # Exit if anything goes wrong  
 set -e  
 echo "vif-bridge $*"  
 # Operation name.  
 OP=$1; shift  
 # Pull variables in args into environment  
 for arg ; do export "${arg}" ; done  
 # Required parameters. Fail if not set.  
 # Optional parameters. Set defaults.  
 ip=${ip:-''}   # default to null (do nothing)  
 # Are we going up or down?  
 case $OP in  
 up)     brcmd='add' ;;  
 down)   brcmd='delete' ;;  
         echo 'Invalid command: ' $OP  
         echo 'Valid commands are: up, down'  
         exit 1  
 # Don't do anything if the bridge is "null".  
 if [ "${bridge}" = "null" ] ; then  
 # Don't do anything if the bridge doesn't exist.  
 if ! ifconfig -l | grep "${bridge}" >/dev/null; then  
 # Add/remove vif to/from bridge.  
 ifconfig x${vif} $OP  
 brconfig ${bridge} ${brcmd} x${vif}  
 Now, running  
 ~~~ {.programlisting}  
 xm create -c /usr/pkg/etc/xen/nbsd  
 should create a domain and load a NetBSD kernel in it. (Note:  
 `-c`{.code} causes xm to connect to the domain's console once created.)  
 The kernel will try to find its root file system on xbd0 (i.e., wd0e)  
 which hasn't been created yet. wd0e will be seen as a disk device in the  
 new domain, so it will be 'sub-partitioned'. We could attach a ccd to  
 wd0e in *domain0* and partition it, newfs and extract the NetBSD/i386 or  
 amd64 tarballs there, but there's an easier way: load the  
 `netbsd-INSTALL_XEN3_DOMU`{.filename} kernel provided in the NetBSD  
 binary sets. Like other install kernels, it contains a ramdisk with  
 sysinst, so you can install NetBSD using sysinst on your new domain.  
 If you want to install NetBSD/Xen with a CDROM image, the following line  
 should be used in the `/usr/pkg/etc/xen/nbsd`{.filename} file:  
 ~~~ {.programlisting}  
 disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]  
 After booting the domain, the option to install via CDROM may be  After booting the domain, the option to install via CDROM may be
 selected. The CDROM device should be changed to **xbd1d**.  selected.  The CDROM device should be changed to `xbd1d`.
 Once done installing, **halt -p** the new domain (don't reboot or halt,  Once done installing, "halt -p" the new domain (don't reboot or halt,
 it would reload the INSTALL\_XEN3\_DOMU kernel even if you changed the  it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
 config file), switch the config file back to the XEN3\_DOMU kernel, and  config file), switch the config file back to the XEN3_DOMU kernel,
 start the new domain again. Now it should be able to use **root on  and start the new domain again. Now it should be able to use "root on
 xbd0a** and you should have a second, functional NetBSD system on your  xbd0a" and you should have a, functional NetBSD domU.
 xen installation.  
   TODO: check if this is still accurate.
 When the new domain is booting you'll see some warnings about *wscons*  When the new domain is booting you'll see some warnings about *wscons*
 and the pseudo-terminals. These can be fixed by editing the files  and the pseudo-terminals. These can be fixed by editing the files
 `/etc/ttys`{.filename} and `/etc/wscons.conf`{.filename}. You must  `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
 disable all terminals in `/etc/ttys`{.filename}, except *console*, like  `/etc/ttys`, except *console*, like this:
 ~~~ {.programlisting}  
 console "/usr/libexec/getty Pc"         vt100   on secure  
 ttyE0   "/usr/libexec/getty Pc"         vt220   off secure  
 ttyE1   "/usr/libexec/getty Pc"         vt220   off secure  
 ttyE2   "/usr/libexec/getty Pc"         vt220   off secure  
 ttyE3   "/usr/libexec/getty Pc"         vt220   off secure  
 Finally, all screens must be commented out from      console "/usr/libexec/getty Pc"         vt100   on secure
 `/etc/wscons.conf`{.filename}.      ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
       ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
       ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
       ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
 It is also desirable to add  Finally, all screens must be commented out from `/etc/wscons.conf`.
 ~~~ {.programlisting}  It is also desirable to add
 in rc.conf. This way, the domain will be properly shut down if **xm          powerd=YES
 shutdown -R** or **xm shutdown -H** is used on the domain0.  
 Your domain should be now ready to work, enjoy.  in rc.conf. This way, the domain will be properly shut down if
   `xm shutdown -R` or `xm shutdown -H` is used on the dom0.
 * * * * *  It is not strictly necessary to have a kernel (as /netbsd) in the domU
   file system.  However, various programs (e.g. netstat) will use that
   kernel to look up symbols to read from kernel virtual memory.  If
   /netbsd is not the running kernel, those lookups will fail.  (This is
   not really a Xen-specific issue, but because the domU kernel is
   obtained from the dom0, it is far more likely to be out of sync or
   missing with Xen.)
 ### Creating an unprivileged Linux domain (DomU)  Creating an unprivileged Linux domain (domU)
 Creating unprivileged Linux domains isn't much different from  Creating unprivileged Linux domains isn't much different from
 unprivileged NetBSD domains, but there are some details to know.  unprivileged NetBSD domains, but there are some details to know.
Line 541  unprivileged NetBSD domains, but there a Line 975  unprivileged NetBSD domains, but there a
 First, the second parameter passed to the disk declaration (the '0x1' in  First, the second parameter passed to the disk declaration (the '0x1' in
 the example below)  the example below)
 ~~~ {.programlisting}      disk = [ 'phy:/dev/wd0e,0x1,w' ]
 disk = [ 'phy:/dev/wd0e,0x1,w' ]  
 does matter to Linux. It wants a Linux device number here (e.g. 0x300  does matter to Linux. It wants a Linux device number here (e.g. 0x300
 for hda). Linux builds device numbers as: (major \<\< 8 + minor). So,  for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
 hda1 which has major 3 and minor 1 on a Linux system will have device  So, hda1 which has major 3 and minor 1 on a Linux system will have
 number 0x301. Alternatively, devices names can be used (hda, hdb, ...)  device number 0x301.  Alternatively, devices names can be used (hda,
 as xentools has a table to map these names to devices numbers. To export  hdb, ...)  as xentools has a table to map these names to devices
 a partition to a Linux guest we can use:  numbers.  To export a partition to a Linux guest we can use:
 ~~~ {.programlisting}          disk = [ 'phy:/dev/wd0e,0x300,w' ]
 disk = [ 'phy:/dev/wd0e,0x300,w' ]          root = "/dev/hda1 ro"
 root = "/dev/hda1 ro"  
 and it will appear as /dev/hda on the Linux system, and be used as root  and it will appear as /dev/hda on the Linux system, and be used as root
 partition.  partition.
 To install the Linux system on the partition to be exported to the guest  To install the Linux system on the partition to be exported to the
 domain, the following method can be used: install sysutils/e2fsprogs  guest domain, the following method can be used: install
 from pkgsrc. Use mke2fs to format the partition that will be the root  sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
 partition of your Linux domain, and mount it. Then copy the files from a  that will be the root partition of your Linux domain, and mount it.
 working Linux system, make adjustments in `/etc`{.filename} (fstab,  Then copy the files from a working Linux system, make adjustments in
 network config). It should also be possible to extract binary packages  `/etc` (fstab, network config).  It should also be possible to extract
 such as .rpm or .deb directly to the mounted partition using the  binary packages such as .rpm or .deb directly to the mounted partition
 appropriate tool, possibly running under NetBSD's Linux emulation. Once  using the appropriate tool, possibly running under NetBSD's Linux
 the filesystem has been populated, umount it. If desirable, the  emulation.  Once the file system has been populated, umount it.  If
 filesystem can be converted to ext3 using tune2fs -j. It should now be  desirable, the file system can be converted to ext3 using tune2fs -j.
 possible to boot the Linux guest domain, using one of the  It should now be possible to boot the Linux guest domain, using one of
 vmlinuz-\*-xenU kernels available in the Xen binary distribution.  the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
 To get the linux console right, you need to add:  
 ~~~ {.programlisting}  
 extra = "xencons=tty1"  
 to your configuration since not all linux distributions auto-attach a  To get the Linux console right, you need to add:
 tty to the xen console.  
 * * * * *      extra = "xencons=tty1"
 ### Creating an unprivileged Solaris domain (DomU)  to your configuration since not all Linux distributions auto-attach a
   tty to the xen console.
 Download an Opensolaris [release](http://opensolaris.org/os/downloads/)  
 or [development snapshot](http://genunix.org/) DVD image. Attach the DVD  
 image to a  
 device. Copy the kernel and ramdisk filesystem image to your dom0  
 ~~~ {.programlisting}  
 dom0# mkdir /root/solaris  
 dom0# vnconfig vnd0 osol-1002-124-x86.iso  
 dom0# mount /dev/vnd0a /mnt  
 ## for a 64-bit guest  
 dom0# cp /mnt/boot/amd64/x86.microroot /root/solaris  
 dom0# cp /mnt/platform/i86xpv/kernel/amd64/unix /root/solaris  
 ## for a 32-bit guest  
 dom0# cp /mnt/boot/x86.microroot /root/solaris  
 dom0# cp /mnt/platform/i86xpv/kernel/unix /root/solaris  
 dom0# umount /mnt  
 Keep the  
 configured. For some reason the boot process stalls unless the DVD image  
 is attached to the guest as a "phy" device. Create an initial  
 configuration file with the following contents. Substitute */dev/wd0k*  
 with an empty partition at least 8 GB large.  
 ~~~ {.programlisting}  
 memory = 640  
 name = 'solaris'  
 disk = [ 'phy:/dev/wd0k,0,w' ]  
 disk += [ 'phy:/dev/vnd0d,6:cdrom,r' ]  
 vif = [ 'bridge=bridge0' ]  
 kernel = '/root/solaris/unix'  
 ramdisk = '/root/solaris/x86.microroot'  
 # for a 64-bit guest  
 extra = '/platform/i86xpv/kernel/amd64/unix - nowin -B install_media=cdrom'  
 # for a 32-bit guest  
 #extra = '/platform/i86xpv/kernel/unix - nowin -B install_media=cdrom'  
 Start the guest.  
 ~~~ {.programlisting}  
 dom0# xm create -c solaris.cfg  
 Started domain solaris  
                       v3.3.2 chgset 'unavailable'  
 SunOS Release 5.11 Version snv_124 64-bit  
 Copyright 1983-2009 Sun Microsystems, Inc.  All rights reserved.  
 Use is subject to license terms.  
 Hostname: opensolaris  
 Remounting root read/write  
 Probing for device nodes ...  
 WARNING: emlxs: ddi_modopen drv/fct failed: err 2  
 Preparing live image for use  
 Done mounting Live image  
 Make sure the network is configured. Note that it can take a minute for  
 the xnf0 interface to appear.  
 ~~~ {.programlisting}  
 opensolaris console login: jack  
 Password: jack  
 Sun Microsystems Inc.   SunOS 5.11      snv_124 November 2008  
 jack@opensolaris:~$ pfexec sh  
 sh-3.2# ifconfig -a  
 sh-3.2# exit  
 Set a password for VNC and start the VNC server which provides the X11  
 display where the installation program runs.  
 ~~~ {.programlisting}  
 jack@opensolaris:~$ vncpasswd  
 Password: solaris  
 Verify: solaris  
 jack@opensolaris:~$ cp .Xclients .vnc/xstartup  
 jack@opensolaris:~$ vncserver :1  
 From a remote machine connect to the VNC server. Use **ifconfig xnf0**  
 on the guest to find the correct IP address to use.  
 ~~~ {.programlisting}  
 remote$ vncviewer  
 It is also possible to launch the installation on a remote X11 display.  
 ~~~ {.programlisting}  
 jack@opensolaris:~$ export DISPLAY=  
 jack@opensolaris:~$ pfexec gui-install  
 After the GUI installation is complete you will be asked to reboot.  
 Before that you need to determine the ZFS ID for the new boot filesystem  
 and update the configuration file accordingly. Return to the guest  
 ~~~ {.programlisting}  
 jack@opensolaris:~$ pfexec zdb -vvv rpool | grep bootfs  
                 bootfs = 43  
 The final configuration file should look like this. Note in particular  
 the last line.  
 ~~~ {.programlisting}  
 memory = 640  
 name = 'solaris'  
 disk = [ 'phy:/dev/wd0k,0,w' ]  
 vif = [ 'bridge=bridge0' ]  
 kernel = '/root/solaris/unix'  
 ramdisk = '/root/solaris/x86.microroot'  
 extra = '/platform/i86xpv/kernel/amd64/unix -B zfs-bootfs=rpool/43,bootpath="/xpvd/xdf@0:a"'  
 Restart the guest to verify it works correctly.  
 ~~~ {.programlisting}  
 dom0# xm destroy solaris  
 dom0# xm create -c solaris.cfg  
 Using config file "./solaris.cfg".  
 v3.3.2 chgset 'unavailable'  
 Started domain solaris  
 SunOS Release 5.11 Version snv_124 64-bit  
 Copyright 1983-2009 Sun Microsystems, Inc.  All rights reserved.  
 Use is subject to license terms.  
 WARNING: emlxs: ddi_modopen drv/fct failed: err 2  
 Hostname: osol  
 Configuring devices.  
 Loading smf(5) service descriptions: 160/160  
 svccfg import warnings. See /var/svc/log/system-manifest-import:default.log .  
 Reading ZFS config: done.  
 Mounting ZFS filesystems: (6/6)  
 Creating new rsa public/private host key pair  
 Creating new dsa public/private host key pair  
 osol console login:  
 Using PCI devices in guest domains  
 The domain0 can give other domains access to selected PCI devices. This  
 can allow, for example, a non-privileged domain to have access to a  
 physical network interface or disk controller. However, keep in mind  
 that giving a domain access to a PCI device most likely will give the  
 domain read/write access to the whole physical memory, as PCs don't have  
 an IOMMU to restrict memory access to DMA-capable device. Also, it's not  
 possible to export ISA devices to non-domain0 domains (which means that  
 the primary VGA adapter can't be exported. A guest domain trying to  
 access the VGA registers will panic).  
 This functionality is only available in NetBSD-5.1 (and later) domain0  
 and domU. If the domain0 is NetBSD, it has to be running Xen 3.1, as  
 support has not been ported to later versions at this time.  
 For a PCI device to be exported to a domU, is has to be attached to the  
 `pciback`{.literal} driver in domain0. Devices passed to the domain0 via  
 the pciback.hide boot parameter will attach to `pciback`{.literal}  
 instead of the usual driver. The list of devices is specified as  
 `(bus:dev.func)`{.literal}, where bus and dev are 2-digit hexadecimal  
 numbers, and func a single-digit number:  
 ~~~ {.programlisting}  
 pciback devices should show up in the domain0's boot messages, and the  
 devices should be listed in the `/kern/xen/pci`{.filename} directory.  
 PCI devices to be exported to a domU are listed in the `pci`{.literal}  
 array of the domU's config file, with the format  
 ~~~ {.programlisting}  
 pci = [ '0000:00:06.0', '0000:00:0a.0' ]  
 In the domU an `xpci`{.literal} device will show up, to which one or  
 more pci busses will attach. Then the PCI drivers will attach to PCI  
 busses as usual. Note that the default NetBSD DOMU kernels do not have  
 `xpci`{.literal} or any PCI drivers built in by default; you have to  
 build your own kernel to use PCI devices in a domU. Here's a kernel  
 config example:  
 ~~~ {.programlisting}  
 include         "arch/i386/conf/XEN3_DOMU"  
 #include         "arch/i386/conf/XENU"           # in NetBSD 3.0  
 # Add support for PCI busses to the XEN3_DOMU kernel  
 xpci* at xenbus ?  
 pci* at xpci ?  
 # Now add PCI and related devices to be used by this domain  
 # USB Controller and Devices  
 # PCI USB controllers  
 uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)  
 # USB bus support  
 usb*    at uhci?  
 # USB Hubs  
 uhub*   at usb?  
 uhub*   at uhub? port ? configuration ? interface ?  
 # USB Mass Storage  
 umass*  at uhub? port ? configuration ? interface ?  
 wd*     at umass?  
 # SCSI controllers  
 ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI  
 # SCSI bus support (for both ahc and umass)  
 scsibus* at scsi?  
 # SCSI devices  
 sd*     at scsibus? target ? lun ?      # SCSI disk drives  
 cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives  
 Links and further information  Creating an unprivileged Solaris domain (domU)
 -----------------------------  ----------------------------------------------
 -   The HowTo on [Installing into  See possibly outdated
     RAID-1](http://mail-index.NetBSD.org/port-xen/2006/03/01/0010.html)  [Solaris domU instructions](/ports/xen/howto-solaris/).
     gives some hints on using Xen (grub) with NetBSD's RAIDframe  
 -   Harold Gutch wrote documentation on [setting up a Linux DomU with a  
     NetBSD Dom0](http://www.gutch.de/NetBSD/docs/xen.html)  
 -   An example of how to use NetBSD's native bootloader to load  
     NetBSD/Xen instead of Grub can be found in the i386/amd64  
   PCI passthrough: Using PCI devices in guest domains
   The dom0 can give other domains access to selected PCI
   devices. This can allow, for example, a non-privileged domain to have
   access to a physical network interface or disk controller.  However,
   keep in mind that giving a domain access to a PCI device most likely
   will give the domain read/write access to the whole physical memory,
   as PCs don't have an IOMMU to restrict memory access to DMA-capable
   device.  Also, it's not possible to export ISA devices to non-dom0
   domains, which means that the primary VGA adapter can't be exported.
   A guest domain trying to access the VGA registers will panic.
   If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
   not been ported to later versions at this time.
   For a PCI device to be exported to a domU, is has to be attached to
   the "pciback" driver in dom0.  Devices passed to the dom0 via the
   pciback.hide boot parameter will attach to "pciback" instead of the
   usual driver.  The list of devices is specified as "(bus:dev.func)",
   where bus and dev are 2-digit hexadecimal numbers, and func a
   single-digit number:
   pciback devices should show up in the dom0's boot messages, and the
   devices should be listed in the `/kern/xen/pci` directory.
   PCI devices to be exported to a domU are listed in the "pci" array of
   the domU's config file, with the format "0000:bus:dev.func".
           pci = [ '0000:00:06.0', '0000:00:0a.0' ]
   In the domU an "xpci" device will show up, to which one or more pci
   buses will attach.  Then the PCI drivers will attach to PCI buses as
   usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
   or any PCI drivers built in by default; you have to build your own
   kernel to use PCI devices in a domU.  Here's a kernel config example;
   note that only the "xpci" lines are unusual.
           include         "arch/i386/conf/XEN3_DOMU"
           # Add support for PCI buses to the XEN3_DOMU kernel
           xpci* at xenbus ?
           pci* at xpci ?
           # PCI USB controllers
           uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
           # USB bus support
           usb*    at uhci?
           # USB Hubs
           uhub*   at usb?
           uhub*   at uhub? port ? configuration ? interface ?
           # USB Mass Storage
           umass*  at uhub? port ? configuration ? interface ?
           wd*     at umass?
           # SCSI controllers
           ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
           # SCSI bus support (for both ahc and umass)
           scsibus* at scsi?
           # SCSI devices
           sd*     at scsibus? target ? lun ?      # SCSI disk drives
           cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
   NetBSD as a domU in a VPS
   The bulk of the HOWTO is about using NetBSD as a dom0 on your own
   hardware.  This section explains how to deal with Xen in a domU as a
   virtual private server where you do not control or have access to the
   dom0.  This is not intended to be an exhaustive list of VPS providers;
   only a few are mentioned that specifically support NetBSD.
   VPS operators provide varying degrees of access and mechanisms for
   configuration.  The big issue is usually how one controls which kernel
   is booted, because the kernel is nominally in the dom0 file system (to
   which VPS users do not normally have access).  A second issue is how
   to install NetBSD.
   A VPS user may want to compile a kernel for security updates, to run
   npf, run IPsec, or any other reason why someone would want to change
   their kernel.
   One approach is to have an administrative interface to upload a kernel,
   or to select from a prepopulated list.  Other approaches are pygrub
   (deprecated) and pvgrub, which are ways to have a bootloader obtain a
   kernel from the domU file system.  This is closer to a regular physical
   computer, where someone who controls a machine can replace the kernel.
   A second issue is multiple CPUs.  With NetBSD 6, domUs support
   multiple vcpus, and it is typical for VPS providers to enable multiple
   CPUs for NetBSD domUs.
   pygrub runs in the dom0 and looks into the domU file system.  This
   implies that the domU must have a kernel in a file system in a format
   known to pygrub.  As of 2014, pygrub seems to be of mostly historical
   pvgrub is a version of grub that uses PV operations instead of BIOS
   calls.  It is booted from the dom0 as the domU kernel, and then reads
   /grub/menu.lst and loads a kernel from the domU file system.
   [Panix](http://www.panix.com/) lets users use pvgrub.  Panix reports
   that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
   (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
   page](http://www.panix.com/v-colo/grub.html), which describes only
   Linux but should be updated to cover NetBSD :-).
   [prgmr.com](http://prgmr.com/) also lets users with pvgrub to boot
   their own kernel.  See then [prgmr.com NetBSD
   (which is in need of updating).
   It appears that [grub's FFS
   does not support all aspects of modern FFS, but there are also reports
   that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
   partition for the kernel with the intent that grub can understand it,
   which leads to /netbsd not being the actual kernel.  One must remember
   to update the special boot partition.
   See the [Amazon EC2 page](../amazon_ec2/).
   Using npf
   In standard kernels, npf is a module, and thus cannot be loaded in a
   DOMU kernel.
   TODO: Explain how to compile npf into a custom kernel, answering (but
   note that the problem was caused by not booting the right kernel)
   [this email to
   TODO items for improving NetBSD/xen
   * Make the NetBSD dom0 kernel work with SMP.
   * Test the Xen 4.5 packages adequately to be able to recommend them as
     the standard approach.
   * Get PCI passthrough working on Xen 4.5
   * Get pvgrub into pkgsrc, either via xentools or separately.
   * grub
     * Check/add support to pkgsrc grub2 for UFS2 and arbitrary
       fragsize/blocksize (UFS2 support may be present; the point is to
       make it so that with any UFS1/UFS2 file system setup that works
       with NetBSD grub will also work).
       See [pkg/40258](http://gnats.netbsd.org/40258).
     * Push patches upstream.
     * Get UFS2 patches into pvgrub.
   * Add support for PV ops to a version of /boot, and make it usable as
     a kernel in Xen, similar to pvgrub.
   * Solve somehow the issue with modules for GENERIC not being loadable
     in a Xen dom0 or domU kernel.
   Random pointers
   This section contains links from elsewhere not yet integrated into the
   HOWTO, and other guides.
   * http://www.lumbercartel.ca/library/xen/
   * http://pbraun.nethence.com/doc/sysutils/xen_netbsd_dom0.html
   * https://gmplib.org/~tege/xen.html

Removed from v.1.2  
changed lines
  Added in v.1.136

CVSweb for NetBSD wikisrc <wikimaster@NetBSD.org> software: FreeBSD-CVSweb