File:  [NetBSD Developer Wiki] / wikisrc / ports / xen / howto.mdwn
Revision 1.105: download - view: text, annotated - select for diffs
Tue Dec 20 03:57:26 2016 UTC (4 years, 4 months ago) by gdt
Branches: MAIN
CVS tags: HEAD
Update recommendations

    1: Introduction
    2: ============
    4: [![[Xen
    5: screenshot]](](
    7: Xen is a hypervisor (or virtual machine monitor) for x86 hardware
    8: (i686-class or higher), which supports running multiple guest
    9: operating systems on a single physical machine.  Xen is a Type 1 or
   10: bare-metal hypervisor; one uses the Xen kernel to control the CPU,
   11: memory and console, a dom0 operating system which mediates access to
   12: other hardware (e.g., disks, network, USB), and one or more domU
   13: operating systems which operate in an unprivileged virtualized
   14: environment.  IO requests from the domU systems are forwarded by the
   15: hypervisor (Xen) to the dom0 to be fulfilled.
   17: Xen supports two styles of guests.  The original is Para-Virtualized
   18: (PV) which means that the guest OS does not attempt to access hardware
   19: directly, but instead makes hypercalls to the hypervisor.  This is
   20: analogous to a user-space program making system calls.  (The dom0
   21: operating system uses PV calls for some functions, such as updating
   22: memory mapping page tables, but has direct hardware access for disk
   23: and network.)   PV guests must be specifically coded for Xen.
   25: The more recent style is HVM, which means that the guest does not have
   26: code for Xen and need not be aware that it is running under Xen.
   27: Attempts to access hardware registers are trapped and emulated.  This
   28: style is less efficient but can run unmodified guests.
   30: Generally any amd64 machine will work with Xen and PV guests.  In
   31: theory i386 computers without amd64 support can be used for Xen <=
   32: 4.2, but we have no recent reports of this working (this is a hint).
   33: For HVM guests, the VT or VMX cpu feature (Intel) or SVM/HVM/VT
   34: (amd64) is needed; "cpuctl identify 0" will show this.  TODO: Clean up
   35: and check the above features.
   37: At boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   38: The dom0 can start one or more domUs.  (Booting is explained in detail
   39: in the dom0 section.)
   41: NetBSD supports Xen in that it can serve as dom0, be used as a domU,
   42: and that Xen kernels and tools are available in pkgsrc.  This HOWTO
   43: attempts to address both the case of running a NetBSD dom0 on hardware
   44: and running domUs under it (NetBSD and other), and also running NetBSD
   45: as a domU in a VPS.
   47: Some versions of Xen support "PCI passthrough", which means that
   48: specific PCI devices can be made available to a specific domU instead
   49: of the dom0.  This can be useful to let a domU run X11, or access some
   50: network interface or other peripheral.
   52: NetBSD 6 and earlier supported Xen 2; support was removed fro NetBSD
   53: 7.  Xen 2 has been removed from pkgsrc.
   55: Prerequisites
   56: -------------
   58: Installing NetBSD/Xen is not extremely difficult, but it is more
   59: complex than a normal installation of NetBSD.
   60: In general, this HOWTO is occasionally overly restrictive about how
   61: things must be done, guiding the reader to stay on the established
   62: path when there are no known good reasons to stray.
   64: This HOWTO presumes a basic familiarity with the Xen system
   65: architecture.  This HOWTO presumes familiarity with installing NetBSD
   66: on i386/amd64 hardware and installing software from pkgsrc.
   67: See also the [Xen website](
   69: Versions of Xen and NetBSD
   70: ==========================
   72: Most of the installation concepts and instructions are independent
   73: of Xen version and NetBSD version.  This section gives advice on
   74: which version to choose.  Versions not in pkgsrc and older unsupported
   75: versions of NetBSD are intentionally ignored.
   77: Xen
   78: ---
   80: In NetBSD, xen is provided in pkgsrc, via matching pairs of packages
   81: xenkernel and xentools.  We will refer only to the kernel versions,
   82: but note that both packages must be installed together and must have
   83: matching versions.
   85: xenkernel3 provides Xen 3.1.  It is no longer maintained by Xen, and the last applied security patch was in
   86: 2011. Thus, it should not be used.  It supports PCI passthrough,
   87: which is why people use it anyway. Xen 3.1 supports i386, both PAE and
   88: non-PAE.
   90: xenkernel33 provides Xen 3.3.  It is no longer maintained by Xen, and
   91: the last applied security patch was in 2012.  Thus, it should not be
   92: used.  Xen 3.3 supports i386, but only in PAE mode.  There are no good
   93: reasons to run this version.
   95: xenkernel41 provides Xen 4.1.  It is no longer maintained by Xen, but
   96: as of 2016-11 received backported security patches.  Xen 4.1 supports
   97: i386, but only in PAE mode.  There are no good reasons to run this
   98: version.
  100: xenkernel42 provides Xen 4.2.  It is no longer maintained by Xen, but
  101: as of 2016-11 received backported security patches.  Xen 4.2 supports
  102: i386, but only in PAE mode.  The only reason to run this is if you
  103: need to use xm instead of xl, or if you need to run an i386 dom0
  104: (because your hardware is i386 only).
  106: xenkernel45 provides Xen 4.5.  It is no longer maintained by Xen, but
  107: as of 2016-11 it received security patches.  Xen 4.5 requires an amd64
  108: dom0, but domUs can be amd64 or i386 PAE.  TODO: It is either a
  109: conservative choice or somewhat old.
  111: xenkernel45 provides Xen 4.6.  It is new to pkgsrc in 2016-05.  It is
  112: no longer maintained by Xen, but as of 2016-11 it received security
  113: patches.  Xen 4.6 requires an amd64 dom0, but domUs can be amd64 or
  114: i386 PAE.  TODO: It is either a somewhat aggressive choice or the
  115: standard choice
  117: See also the [Xen Security Advisory page](
  119: Ideally newer versions of Xen will be added to pkgsrc.
  121: Note that NetBSD support is called XEN3.  It works with Xen 3 and Xen
  122: 4 because the hypercall interface has been stable.
  124: Xen command program
  125: -------------------
  127: Early Xen used a program called xm to manipulate the system from the
  128: dom0.  Starting in 4.1, a replacement program with similar behavior
  129: called xl is provided, but it does not work well in 4.1.  In 4.2, both
  130: xm and xl work fine.  4.4 is the last version that has xm.  You must
  131: choose one or the other, because it affects which daemons you run.
  133: NetBSD
  134: ------
  136: The netbsd-6, netbsd-7, and -current branches are all reasonable
  137: choices, with more or less the same considerations for non-Xen use.
  138: Therefore, netbsd-7 is recommended as the stable version of the most
  139: recent release for production use.  For those wanting to learn Xen or
  140: without production stability concerns, netbsd-7 is still likely most
  141: appropriate.  Xen runs fine on netbsd-5, but the xentools packages are
  142: likely difficult to build.
  144: As of NetBSD 6, a NetBSD domU will support multiple vcpus.  There is
  145: no SMP support for NetBSD as dom0.  (The dom0 itself doesn't really
  146: need SMP for dom0 functions; the lack of support is really a problem
  147: when using a dom0 as a normal computer.)
  149: Architecture
  150: ------------
  152: Xen itself can run on i386 (some versions) or amd64 machines (all
  153: versions).  (Practically, almost any computer where one would want to
  154: run Xen today supports amd64.)
  156: Xen, the dom0 kernel, and each domU kernel can be either i386 or
  157: amd64.  When building a xenkernel package, one obtains i386 on an i386
  158: host, and amd64 on an amd64 host.  If the xen kernel is i386, then the
  159: dom0 kernel and all domU kernels must be i386.  With an amd64 xen
  160: kernel, an amd64 dom0 kernel is known to work, and an i386PAE dom0
  161: kernel should in theory work.  An amd64 xen/dom0 is known to support
  162: both i386PAE and amd64 domUs.
  164: i386 dom0 and domU kernels must be PAE (except for Xen 3.1); these are
  165: built by default.  (Note that emacs (at least) fails if run on i386
  166: with PAE when built without, and vice versa, presumably due to bugs in
  167: the undump code.)
  169: Because of the above, the standard approach is to use amd64 for the
  170: dom0.
  172: Xen 4.2 is the last version to support i386 as a host.  TODO: Clarify
  173: if this is about the CPU, the xen kernel, or the dom0 kernel having to
  174: be amd64.
  177: Stability
  178: ---------
  180: Mostly, NetBSD as a dom0 or domU is quite stable.
  181: However, there are some open PRs indicating problems.
  183:  - [PR 48125](
  184:  - [PR 47720](
  186: Note also that there are issues with sparse vnd(4) instances, but
  187: these are not about Xen -- they just are noticed with sparse vnd(4)
  188: instances in support of virtual disks in a dom0.
  190: Recommendation
  191: --------------
  193: Therefore, this HOWTO recommends running xenkernel45 or xenkernel46,
  194: xl, the NetBSD 7 stable branch, and to use an amd64 kernel as the
  195: dom0.  Either the i386PAE or amd64 version of NetBSD may be used as
  196: domUs.
  198: Build problems
  199: --------------
  201: Ideally, all versions of Xen in pkgsrc would build on all versions of
  202: NetBSD on both i386 and amd64.  However, that isn't the case.  Besides
  203: aging code and aging compilers, qemu (included in xentools for HVM
  204: support) is difficult to build.  The following are known to work or FAIL:
  206:         xenkernel3 netbsd-5 amd64
  207:         xentools3 netbsd-5 amd64
  208:         xentools3=hvm netbsd-5 amd64 ????
  209:         xenkernel33 netbsd-5 amd64
  210:         xentools33 netbsd-5 amd64
  211:         xenkernel41 netbsd-5 amd64
  212:         xentools41 netbsd-5 amd64
  213:         xenkernel42 netbsd-5 amd64
  214:         xentools42 netbsd-5 amd64
  216:         xenkernel3 netbsd-6 i386 FAIL
  217:         xentools3 netbsd-6 i386
  218:         xentools3-hvm netbsd-6 i386 FAIL (dependencies fail)
  219:         xenkernel33 netbsd-6 i386
  220:         xentools33 netbsd-6 i386
  221:         xenkernel41 netbsd-6 i386
  222:         xentools41 netbsd-6 i386
  223:         xenkernel42 netbsd-6 i386
  224:         xentools42 netbsd-6 i386 *MIXED
  226: 	(all 3 and 33 seem to FAIL)
  227:         xenkernel41 netbsd-7 i386
  228:         xentools41 netbsd-7 i386
  229:         xenkernel42 netbsd-7 i386
  230:         xentools42 netbsd-7 i386 ??FAIL
  232: (*On netbsd-6 i386, there is a xentools42 in the 2014Q3 official builds,
  233: but it does not build for gdt.)
  235: NetBSD as a dom0
  236: ================
  238: NetBSD can be used as a dom0 and works very well.  The following
  239: sections address installation, updating NetBSD, and updating Xen.
  240: Note that it doesn't make sense to talk about installing a dom0 OS
  241: without also installing Xen itself.  We first address installing
  242: NetBSD, which is not yet a dom0, and then adding Xen, pivoting the
  243: NetBSD install to a dom0 install by just changing the kernel and boot
  244: configuration.
  246: For experimenting with Xen, a machine with as little as 1G of RAM and
  247: 100G of disk can work.  For running many domUs in productions, far
  248: more will be needed.
  250: Styles of dom0 operation
  251: ------------------------
  253: There are two basic ways to use Xen.  The traditional method is for
  254: the dom0 to do absolutely nothing other than providing support to some
  255: number of domUs.  Such a system was probably installed for the sole
  256: purpose of hosting domUs, and sits in a server room on a UPS.
  258: The other way is to put Xen under a normal-usage computer, so that the
  259: dom0 is what the computer would have been without Xen, perhaps a
  260: desktop or laptop.  Then, one can run domUs at will.  Purists will
  261: deride this as less secure than the previous approach, and for a
  262: computer whose purpose is to run domUs, they are right.  But Xen and a
  263: dom0 (without domUs) is not meaningfully less secure than the same
  264: things running without Xen.  One can boot Xen or boot regular NetBSD
  265: alternately with little problems, simply refraining from starting the
  266: Xen daemons when not running Xen.
  268: Note that NetBSD as dom0 does not support multiple CPUs.  This will
  269: limit the performance of the Xen/dom0 workstation approach.  In theory
  270: the only issue is that the "backend drivers" are not yet MPSAFE:
  273: Installation of NetBSD
  274: ----------------------
  276: First,
  277: [install NetBSD/amd64](/guide/inst/)
  278: just as you would if you were not using Xen.
  279: However, the partitioning approach is very important.
  281: If you want to use RAIDframe for the dom0, there are no special issues
  282: for Xen.  Typically one provides RAID storage for the dom0, and the
  283: domU systems are unaware of RAID.  The 2nd-stage loader bootxx_* skips
  284: over a RAID1 header to find /boot from a filesystem within a RAID
  285: partition; this is no different when booting Xen.
  287: There are 4 styles of providing backing storage for the virtual disks
  288: used by domUs: raw partitions, LVM, file-backed vnd(4), and SAN.
  290: With raw partitions, one has a disklabel (or gpt) partition sized for
  291: each virtual disk to be used by the domU.  (If you are able to predict
  292: how domU usage will evolve, please add an explanation to the HOWTO.
  293: Seriously, needs tend to change over time.)
  295: One can use [lvm(8)](/guide/lvm/) to create logical devices to use
  296: for domU disks.  This is almost as efficient as raw disk partitions
  297: and more flexible.  Hence raw disk partitions should typically not
  298: be used.
  300: One can use files in the dom0 filesystem, typically created by dd'ing
  301: /dev/zero to create a specific size.  This is somewhat less efficient,
  302: but very convenient, as one can cp the files for backup, or move them
  303: between dom0 hosts.
  305: Finally, in theory one can place the files backing the domU disks in a
  306: SAN.  (This is an invitation for someone who has done this to add a
  307: HOWTO page.)
  309: Installation of Xen
  310: -------------------
  312: In the dom0, install sysutils/xenkernel42 and sysutils/xentools42 from
  313: pkgsrc (or another matching pair).
  314: See [the pkgsrc
  315: documentation]( for help with pkgsrc.
  317: For Xen 3.1, support for HVM guests is in sysutils/xentool3-hvm.  More
  318: recent versions have HVM support integrated in the main xentools
  319: package.  It is entirely reasonable to run only PV guests.
  321: Next you need to install the selected Xen kernel itself, which is
  322: installed by pkgsrc as "/usr/pkg/xen*-kernel/xen.gz".  Copy it to /.
  323: For debugging, one may copy xen-debug.gz; this is conceptually similar
  324: to DIAGNOSTIC and DEBUG in NetBSD.  xen-debug.gz is basically only
  325: useful with a serial console.  Then, place a NetBSD XEN3_DOM0 kernel
  326: in /, copied from releasedir/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
  327: of a NetBSD build.  If using i386, use
  328: releasedir/i386/binary/kernel/netbsd-XEN3PAE_DOM0.gz.  (If using Xen
  329: 3.1 and i386, you may use XEN3_DOM0 with the non-PAE Xen.  But you
  330: should not use Xen 3.1.)  Both xen and the NetBSD kernel may be (and
  331: typically are) left compressed.
  333: In a dom0 kernel, kernfs is mandatory for xend to comunicate with the
  334: kernel, so ensure that /kern is in fstab.  TODO: Say this is default,
  335: or file a PR and give a reference.
  337: Because you already installed NetBSD, you have a working boot setup
  338: with an MBR bootblock, either bootxx_ffsv1 or bootxx_ffsv2 at the
  339: beginning of your root filesystem, /boot present, and likely
  340: /boot.cfg.  (If not, fix before continuing!)
  342: Add a line to to /boot.cfg to boot Xen.  See boot.cfg(5) for an
  343: example.  The basic line is
  345:         menu=Xen:load /netbsd-XEN3_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=256M
  347: which specifies that the dom0 should have 256M, leaving the rest to be
  348: allocated for domUs.  To use a serial console, use
  350:         menu=Xen:load /netbsd-XEN3_DOM0.gz console=com0;multiboot /xen.gz dom0_mem=256M console=com1 com1=9600,8n1
  352: which will use the first serial port for Xen (which counts starting
  353: from 1), forcing speed/parity, and also for NetBSD (which counts
  354: starting at 0).  In an attempt to add performance, one can also add
  356:         dom0_max_vcpus=1 dom0_vcpus_pin
  358: to force only one vcpu to be provided (since NetBSD dom0 can't use
  359: more) and to pin that vcpu to a physical cpu.  TODO: benchmark this.
  361: Xen has [many boot
  362: options](,
  363: and other tham dom0 memory and max_vcpus, they are generally not
  364: necessary.
  366: As with non-Xen systems, you should have a line to boot /netbsd (a
  367: kernel that works without Xen) and fallback versions of the non-Xen
  368: kernel, Xen, and the dom0 kernel.
  370: Now, reboot so that you are running a DOM0 kernel under Xen, rather
  371: than GENERIC without Xen.
  373: Using grub (historic)
  374: ---------------------
  376: Before NetBSD's native bootloader could support Xen, the use of
  377: grub was recommended.  If necessary, see the
  378: [old grub information](/ports/xen/howto-grub/).
  380: The [HowTo on Installing into
  381: RAID-1](
  382: explains how to set up booting a dom0 with Xen using grub with
  383: NetBSD's RAIDframe.  (This is obsolete with the use of NetBSD's native
  384: boot.)
  386: Configuring Xen
  387: ---------------
  389: Xen logs will be in /var/log/xen.
  391: Now, you have a system that will boot Xen and the dom0 kernel, but not
  392: do anything else special.  Make sure that you have rebooted into Xen.
  393: There will be no domUs, and none can be started because you still have
  394: to configure the dom0 daemons.
  396: The daemons which should be run vary with Xen version and with whether
  397: one is using xm or xl.  The Xen 3.1 and 3.3 packages use xm.  Xen 4.1
  398: and higher packages use xl.  While is is possible to use xm with some
  399: 4.x versions (TODO: 4.1 and 4.2?), the pkgsrc-provided rc.d scripts do
  400: not support this as of 2014-12-26, and thus the HOWTO does not support
  401: it either.  (Make sure your packages are reasonably recent.)
  403: For "xm" (3.1 and 3.3), you should enable xend and xenbackendd (but
  404: note that you should be using 4.x):
  406:         xend=YES
  407:         xenbackendd=YES
  409: For "xl" (4.x), you should enabled xend and xencommons (xenstored).
  410: Trying to boot 4.x without xencommons=YES will result in a hang; it is
  411: necessary to hig ^C on the console to let the machine finish booting.
  412: TODO: explain why xend is installed by the package.
  414:         xencommons=YES
  416: The installation of NetBSD should already have created devices for xen
  417: (xencons, xenevt), but if they are not present, create them:
  419:         cd /dev && sh MAKEDEV xen
  421: TODO: Recommend for/against xen-watchdog.
  423: After you have configured the daemons and either started them (in the
  424: order given) or rebooted, use xm or xl to inspect Xen's boot messages,
  425: available resources, and running domains.  An example with xl follows:
  427:         # xl dmesg
  428: 	[xen's boot info]
  429:         # xl info
  430: 	[available memory, etc.]
  431:         # xl list
  432:         Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  433:         Domain-0           0       64    0  r----     58.1
  435: ### Issues with xencommons
  437: xencommons starts xenstored, which stores data on behalf of dom0 and
  438: domUs.  It does not currently work to stop and start xenstored.
  439: Certainly all domUs should be shutdown first, following the sort order
  440: of the rc.d scripts.  However, the dom0 sets up state with xenstored,
  441: and is not notified when xenstored exits, leading to not recreating
  442: the state when the new xenstored starts.  Until there's a mechanism to
  443: make this work, one should not expect to be able to restart xenstored
  444: (and thus xencommons).  There is currently no reason to expect that
  445: this will get fixed any time soon.
  447: anita (for testing NetBSD)
  448: --------------------------
  450: With the setup so far (assuming 4.2/xl), one should be able to run
  451: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
  452: root, because anita must create a domU):
  454:         anita --vmm=xl test file:///usr/obj/i386/
  456: Alternatively, one can use --vmm=xm to use xm-based domU creation
  457: instead (and must, on Xen <= 4.1).   TODO: confirm that anita xl really works.
  459: Xen-specific NetBSD issues
  460: --------------------------
  462: There are (at least) two additional things different about NetBSD as a
  463: dom0 kernel compared to hardware.
  465: One is that modules are not usable in DOM0 kernels, so one must
  466: compile in what's needed.  It's not really that modules cannot work,
  467: but that modules must be built for XEN3_DOM0 because some of the
  468: defines change and the normal module builds don't do this.  Basically,
  469: enabling Xen changes the kernel ABI, and the module build system
  470: doesn't cope with this.
  472: The other difference is that XEN3_DOM0 does not have exactly the same
  473: options as GENERIC.  While it is debatable whether or not this is a
  474: bug, users should be aware of this and can simply add missing config
  475: items if desired.
  477: Updating NetBSD in a dom0
  478: -------------------------
  480: This is just like updating NetBSD on bare hardware, assuming the new
  481: version supports the version of Xen you are running.  Generally, one
  482: replaces the kernel and reboots, and then overlays userland binaries
  483: and adjusts /etc.
  485: Note that one must update both the non-Xen kernel typically used for
  486: rescue purposes and the DOM0 kernel used with Xen.
  488: Converting from grub to /boot
  489: -----------------------------
  491: These instructions were [TODO: will be] used to convert a system from
  492: grub to /boot.  The system was originally installed in February of
  493: 2006 with a RAID1 setup and grub to boot Xen 2, and has been updated
  494: over time.  Before these commands, it was running NetBSD 6 i386, Xen
  495: 4.1 and grub, much like the message linked earlier in the grub
  496: section.
  498:         # Install mbr bootblocks on both disks. 
  499:         fdisk -i /dev/rwd0d
  500:         fdisk -i /dev/rwd1d
  501:         # Install NetBSD primary boot loader (/ is FFSv1) into RAID1 components.
  502:         installboot -v /dev/rwd0d /usr/mdec/bootxx_ffsv1
  503:         installboot -v /dev/rwd1d /usr/mdec/bootxx_ffsv1
  504:         # Install secondary boot loader
  505:         cp -p /usr/mdec/boot /
  506:         # Create boog.cfg following earlier guidance:
  507:         menu=Xen:load /netbsd-XEN3PAE_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=256M
  508:         menu=Xen.ok:load /netbsd-XEN3PAE_DOM0.ok.gz console=pc;multiboot /xen.ok.gz dom0_mem=256M
  509:         menu=GENERIC:boot
  510:         menu=GENERIC single-user:boot -s
  511:         menu=GENERIC.ok:boot netbsd.ok
  512:         menu=GENERIC.ok single-user:boot netbsd.ok -s
  513:         menu=Drop to boot prompt:prompt
  514:         default=1
  515:         timeout=30
  517: TODO: actually do this and fix it if necessary.
  519: Upgrading Xen versions
  520: ---------------------
  522: Updating Xen is conceptually not difficult, but can run into all the
  523: issues found when installing Xen.  Assuming migration from 4.1 to 4.2,
  524: remove the xenkernel41 and xentools41 packages and install the
  525: xenkernel42 and xentools42 packages.  Copy the 4.2 xen.gz to /.
  527: Ensure that the contents of /etc/rc.d/xen* are correct.  Specifically,
  528: they must match the package you just installed and not be left over
  529: from some previous installation.
  531: Enable the correct set of daemons; see the configuring section above.
  532: (Upgrading from 3.x to 4.x without doing this will result in a hang.)
  534: Ensure that the domU config files are valid for the new version.
  535: Specifically: remove autorestart=True, and ensure that disks are
  536: specified with numbers as the second argument, as the examples above
  537: show, and not NetBSD device names.
  539: Hardware known to work
  540: ----------------------
  542: Arguably, this section is misplaced, and there should be a page of
  543: hardware that runs NetBSD/amd64 well, with the mostly-well-founded
  544: assumption that NetBSD/xen runs fine on any modern hardware that
  545: NetBSD/amd64 runs well on.  Until then, we give motherboard/CPU/RAM
  546: triples to aid those choosing a motherboard.  Note that Xen systems
  547: usually do not run X, so a listing here does not imply that X works at
  548: all.
  550:         Supermicro X9SRL-F, Xeon E5-1650 v2, 96 GiB ECC
  551:         Supermicro ??, Atom C2758 (8 core), 32 GiB ECC
  552:         ASUS M5A78L-M/USB3 AM3+ microATX, AMD Piledriver X8 4000MHz, 16 GiB ECC
  554: Older hardware:
  556:         Intel D915GEV, Pentium4 CPU 3.40GHz, 4GB 533MHz Synchronous DDR2
  558: Running Xen under qemu
  559: ----------------------
  561: The astute reader will note that this section is somewhat twisted.
  562: However, it can be useful to run Xen under qemu either because the
  563: version of NetBSD as a dom0 does not run on the hardware in use, or to
  564: generate automated test cases involving Xen.
  566: In 2015-01, the following combination was reported to mostly work:
  568:         host OS: NetBSD/amd64 6.1.4
  569:         qemu: 2.2.0 from pkgsrc
  570:         Xen kernel: xenkernel42-4.2.5nb1 from pkgsrc
  571:         dom0 kernel: NetBSD/amd64 6.1.5
  572:         Xen tools: xentools42-4.2.5 from pkgsrc
  574: See [PR 47720]( for a problem with dom0
  575: shutdown.
  577: Unprivileged domains (domU)
  578: ===========================
  580: This section describes general concepts about domUs.  It does not
  581: address specific domU operating systems or how to install them.  The
  582: config files for domUs are typically in /usr/pkg/etc/xen, and are
  583: typically named so that the file name, domU name and the domU's host
  584: name match.
  586: The domU is provided with cpu and memory by Xen, configured by the
  587: dom0.  The domU is provided with disk and network by the dom0,
  588: mediated by Xen, and configured in the dom0.
  590: Entropy in domUs can be an issue; physical disks and network are on
  591: the dom0.  NetBSD's /dev/random system works, but is often challenged.
  593: Config files
  594: ------------
  596: There is no good order to present config files and the concepts
  597: surrounding what is being configured.  We first show an example config
  598: file, and then in the various sections give details.
  600: See (at least in xentools41) /usr/pkg/share/examples/xen/xmexample*,
  601: for a large number of well-commented examples, mostly for running
  602: GNU/Linux.
  604: The following is an example minimal domain configuration file
  605: "/usr/pkg/etc/xen/foo".  It is (with only a name change) an actual
  606: known working config file on Xen 4.1 (NetBSD 5 amd64 dom0 and NetBSD 5
  607: i386 domU).  The domU serves as a network file server.
  609:         # -*- mode: python; -*-
  611:         kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
  612:         memory = 1024
  613:         vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
  614:         disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
  615:                  'file:/n0/xen/foo-wd1,0x1,w' ]
  617: The domain will have the same name as the file.  The kernel has the
  618: host/domU name in it, so that on the dom0 one can update the various
  619: domUs independently.  The vif line causes an interface to be provided,
  620: with a specific mac address (do not reuse MAC addresses!), in bridge
  621: mode.  Two disks are provided, and they are both writable; the bits
  622: are stored in files and Xen attaches them to a vnd(4) device in the
  623: dom0 on domain creation.  The system treates xbd0 as the boot device
  624: without needing explicit configuration.
  626: By default xm looks for domain config files in /usr/pkg/etc/xen.  Note
  627: that "xm create" takes the name of a config file, while other commands
  628: take the name of a domain.  To create the domain, connect to the
  629: console, create the domain while attaching the console, shutdown the
  630: domain, and see if it has finished stopping, do (or xl with Xen >=
  631: 4.2):
  633:         xm create foo
  634:         xm console foo
  635:         xm create -c foo
  636:         xm shutdown foo
  637:         xm list
  639: Typing ^] will exit the console session.  Shutting down a domain is
  640: equivalent to pushing the power button; a NetBSD domU will receive a
  641: power-press event and do a clean shutdown.  Shutting down the dom0
  642: will trigger controlled shutdowns of all configured domUs.
  644: domU kernels
  645: ------------
  647: On a physical computer, the BIOS reads sector 0, and a chain of boot
  648: loaders finds and loads a kernel.  Normally this comes from the root
  649: filesystem.  With Xen domUs, the process is totally different.  The
  650: normal path is for the domU kernel to be a file in the dom0's
  651: filesystem.  At the request of the dom0, Xen loads that kernel into a
  652: new domU instance and starts execution.  While domU kernels can be
  653: anyplace, reasonable places to store domU kernels on the dom0 are in /
  654: (so they are near the dom0 kernel), in /usr/pkg/etc/xen (near the
  655: config files), or in /u0/xen (where the vdisks are).
  657: Note that loading the domU kernel from the dom0 implies that boot
  658: blocks, /boot, /boot.cfg, and so on are all ignored in the domU.
  659: See the VPS section near the end for discussion of alternate ways to
  660: obtain domU kernels.
  662: CPU and memory
  663: --------------
  665: A domain is provided with some number of vcpus, less than the number
  666: of cpus seen by the hypervisor.  (For a dom0, this is controlled by
  667: the boot argument "dom0_max_vcpus=1".)  For a domU, it is controlled
  668: from the config file by the "vcpus = N" directive.
  670: A domain is provided with memory; this is controlled in the config
  671: file by "memory = N" (in megabytes).  In the straightforward case, the
  672: sum of the the memory allocated to the dom0 and all domUs must be less
  673: than the available memory.
  675: Xen also provides a "balloon" driver, which can be used to let domains
  676: use more memory temporarily.  TODO: Explain better, and explain how
  677: well it works with NetBSD.
  679: Virtual disks
  680: -------------
  682: With the file/vnd style, typically one creates a directory,
  683: e.g. /u0/xen, on a disk large enough to hold virtual disks for all
  684: domUs.  Then, for each domU disk, one writes zeros to a file that then
  685: serves to hold the virtual disk's bits; a suggested name is foo-xbd0
  686: for the first virtual disk for the domU called foo.  Writing zeros to
  687: the file serves two purposes.  One is that preallocating the contents
  688: improves performance.  The other is that vnd on sparse files has
  689: failed to work.  TODO: give working/notworking NetBSD versions for
  690: sparse vnd.  Note that the use of file/vnd for Xen is not really
  691: different than creating a file-backed virtual disk for some other
  692: purpose, except that xentools handles the vnconfig commands.  To
  693: create an empty 4G virtual disk, simply do
  695:         dd if=/dev/zero of=foo-xbd0 bs=1m count=4096
  697: Do not use qemu-img-xen, because this will create sparse file.  There
  698: have been recent (2015) reports of sparse vnd(4) devices causing
  699: lockups, but there is apparently no PR.
  701: With the lvm style, one creates logical devices.  They are then used
  702: similarly to vnds.  TODO: Add an example with lvm.
  704: In domU config files, the disks are defined as a sequence of 3-tuples.
  705: The first element is "method:/path/to/disk".  Common methods are
  706: "file:" for file-backed vnd. and "phy:" for something that is already
  707: a (TODO: character or block) device.
  709: The second element is an artifact of how virtual disks are passed to
  710: Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
  711: are given a device name to associate with the disk, and values like
  712: "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
  713: as xbd0, the second as xbd1, and so on.  However, xm/xl demand a
  714: second argument.  The name given is converted to a major/minor by
  715: calling stat(2) on the name in /dev and this is passed to the domU.
  716: In the general case, the dom0 and domU can be different operating
  717: systems, and it is an unwarranted assumption that they have consistent
  718: numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
  719: as both dom0 and domU, using values of 0x0 for the first disk and 0x1
  720: for the second works fine and avoids this issue.  For a GNU/Linux
  721: guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
  722: /dev/hda1.
  724: The third element is "w" for writable disks, and "r" for read-only
  725: disks.
  727: Virtual Networking
  728: ------------------
  730: Xen provides virtual ethernets, each of which connects the dom0 and a
  731: domU.  For each virtual network, there is an interface "xvifN.M" in
  732: the dom0, and in domU index N, a matching interface xennetM (NetBSD
  733: name).  The interfaces behave as if there is an Ethernet with two
  734: adaptors connected.  From this primitive, one can construct various
  735: configurations.  We focus on two common and useful cases for which
  736: there are existing scripts: bridging and NAT.
  738: With bridging (in the example above), the domU perceives itself to be
  739: on the same network as the dom0.  For server virtualization, this is
  740: usually best.  Bridging is accomplished by creating a bridge(4) device
  741: and adding the dom0's physical interface and the various xvifN.0
  742: interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
  743: config file.  The bridge must be set up already in the dom0; an
  744: example /etc/ifconfig.bridge0 is:
  746:         create
  747:         up
  748:         !brconfig bridge0 add wm0
  750: With NAT, the domU perceives itself to be behind a NAT running on the
  751: dom0.  This is often appropriate when running Xen on a workstation.
  752: TODO: NAT appears to be configured by "vif = [ '' ]".
  754: The MAC address specified is the one used for the interface in the new
  755: domain.  The interface in dom0 will use this address XOR'd with
  756: 00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
  758: Sizing domains
  759: --------------
  761: Modern x86 hardware has vast amounts of resources.  However, many
  762: virtual servers can function just fine on far less.  A system with
  763: 256M of RAM and a 4G disk can be a reasonable choice.  Note that it is
  764: far easier to adjust virtual resources than physical ones.  For
  765: memory, it's just a config file edit and a reboot.  For disk, one can
  766: create a new file and vnconfig it (or lvm), and then dump/restore,
  767: just like updating physical disks, but without having to be there and
  768: without those pesky connectors.
  770: Starting domains automatically
  771: ------------------------------
  773: To start domains foo at bar at boot and shut them down cleanly on dom0
  774: shutdown, in rc.conf add:
  776:         xendomains="foo bar"
  778: Note that earlier versions of the xentools41 xendomains rc.d scripth
  779: usd xl, when one should use xm with 4.1.
  781: Creating specific unprivileged domains (domU)
  782: =============================================
  784: Creating domUs is almost entirely independent of operating system.  We
  785: have already presented the basics of config files.  Note that you must
  786: have already completed the dom0 setup so that "xl list" (or "xm list")
  787: works.
  789: Creating an unprivileged NetBSD domain (domU)
  790: ---------------------------------------------
  792: See the earlier config file, and adjust memory.  Decide on how much
  793: storage you will provide, and prepare it (file or lvm).
  795: While the kernel will be obtained from the dom0 filesystem, the same
  796: file should be present in the domU as /netbsd so that tools like
  797: savecore(8) can work.   (This is helpful but not necessary.)
  799: The kernel must be specifically for Xen and for use as a domU.  The
  800: i386 and amd64 provide the following kernels:
  802:         i386 XEN3_DOMU
  803:         i386 XEN3PAE_DOMU
  804:         amd64 XEN3_DOMU
  806: Unless using Xen 3.1 (and you shouldn't) with i386-mode Xen, you must
  807: use the PAE version of the i386 kernel.
  809: This will boot NetBSD, but this is not that useful if the disk is
  810: empty.  One approach is to unpack sets onto the disk outside of xen
  811: (by mounting it, just as you would prepare a physical disk for a
  812: system you can't run the installer on).
  814: A second approach is to run an INSTALL kernel, which has a miniroot
  815: and can load sets from the network.  To do this, copy the INSTALL
  816: kernel to / and change the kernel line in the config file to:
  818:         kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
  820: Then, start the domain as "xl create -c configname".
  822: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
  823: line should be used in the config file.
  825:     disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
  827: After booting the domain, the option to install via CDROM may be
  828: selected.  The CDROM device should be changed to `xbd1d`.
  830: Once done installing, "halt -p" the new domain (don't reboot or halt,
  831: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
  832: config file), switch the config file back to the XEN3_DOMU kernel,
  833: and start the new domain again. Now it should be able to use "root on
  834: xbd0a" and you should have a, functional NetBSD domU.
  836: TODO: check if this is still accurate.
  837: When the new domain is booting you'll see some warnings about *wscons*
  838: and the pseudo-terminals. These can be fixed by editing the files
  839: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
  840: `/etc/ttys`, except *console*, like this:
  842:     console "/usr/libexec/getty Pc"         vt100   on secure
  843:     ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
  844:     ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
  845:     ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
  846:     ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
  848: Finally, all screens must be commented out from `/etc/wscons.conf`.
  850: It is also desirable to add
  852:         powerd=YES
  854: in rc.conf. This way, the domain will be properly shut down if
  855: `xm shutdown -R` or `xm shutdown -H` is used on the dom0.
  857: It is not strictly necessary to have a kernel (as /netbsd) in the domU
  858: filesystem.  However, various programs (e.g. netstat) will use that
  859: kernel to look up symbols to read from kernel virtual memory.  If
  860: /netbsd is not the running kernel, those lookups will fail.  (This is
  861: not really a Xen-specific issue, but because the domU kernel is
  862: obtained from the dom0, it is far more likely to be out of sync or
  863: missing with Xen.)
  865: Creating an unprivileged Linux domain (domU)
  866: --------------------------------------------
  868: Creating unprivileged Linux domains isn't much different from
  869: unprivileged NetBSD domains, but there are some details to know.
  871: First, the second parameter passed to the disk declaration (the '0x1' in
  872: the example below)
  874:     disk = [ 'phy:/dev/wd0e,0x1,w' ]
  876: does matter to Linux. It wants a Linux device number here (e.g. 0x300
  877: for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
  878: So, hda1 which has major 3 and minor 1 on a Linux system will have
  879: device number 0x301.  Alternatively, devices names can be used (hda,
  880: hdb, ...)  as xentools has a table to map these names to devices
  881: numbers.  To export a partition to a Linux guest we can use:
  883:         disk = [ 'phy:/dev/wd0e,0x300,w' ]
  884:         root = "/dev/hda1 ro"
  886: and it will appear as /dev/hda on the Linux system, and be used as root
  887: partition.
  889: To install the Linux system on the partition to be exported to the
  890: guest domain, the following method can be used: install
  891: sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
  892: that will be the root partition of your Linux domain, and mount it.
  893: Then copy the files from a working Linux system, make adjustments in
  894: `/etc` (fstab, network config).  It should also be possible to extract
  895: binary packages such as .rpm or .deb directly to the mounted partition
  896: using the appropriate tool, possibly running under NetBSD's Linux
  897: emulation.  Once the filesystem has been populated, umount it.  If
  898: desirable, the filesystem can be converted to ext3 using tune2fs -j.
  899: It should now be possible to boot the Linux guest domain, using one of
  900: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
  902: To get the linux console right, you need to add:
  904:     extra = "xencons=tty1"
  906: to your configuration since not all linux distributions auto-attach a
  907: tty to the xen console.
  909: Creating an unprivileged Solaris domain (domU)
  910: ----------------------------------------------
  912: See possibly outdated
  913: [Solaris domU instructions](/ports/xen/howto-solaris/).
  916: PCI passthrough: Using PCI devices in guest domains
  917: ---------------------------------------------------
  919: The dom0 can give other domains access to selected PCI
  920: devices. This can allow, for example, a non-privileged domain to have
  921: access to a physical network interface or disk controller.  However,
  922: keep in mind that giving a domain access to a PCI device most likely
  923: will give the domain read/write access to the whole physical memory,
  924: as PCs don't have an IOMMU to restrict memory access to DMA-capable
  925: device.  Also, it's not possible to export ISA devices to non-dom0
  926: domains, which means that the primary VGA adapter can't be exported.
  927: A guest domain trying to access the VGA registers will panic.
  929: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
  930: not been ported to later versions at this time.
  932: For a PCI device to be exported to a domU, is has to be attached to
  933: the "pciback" driver in dom0.  Devices passed to the dom0 via the
  934: pciback.hide boot parameter will attach to "pciback" instead of the
  935: usual driver.  The list of devices is specified as "(bus:dev.func)",
  936: where bus and dev are 2-digit hexadecimal numbers, and func a
  937: single-digit number:
  939:         pciback.hide=(00:0a.0)(00:06.0)
  941: pciback devices should show up in the dom0's boot messages, and the
  942: devices should be listed in the `/kern/xen/pci` directory.
  944: PCI devices to be exported to a domU are listed in the "pci" array of
  945: the domU's config file, with the format "0000:bus:dev.func".
  947:         pci = [ '0000:00:06.0', '0000:00:0a.0' ]
  949: In the domU an "xpci" device will show up, to which one or more pci
  950: busses will attach.  Then the PCI drivers will attach to PCI busses as
  951: usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
  952: or any PCI drivers built in by default; you have to build your own
  953: kernel to use PCI devices in a domU.  Here's a kernel config example;
  954: note that only the "xpci" lines are unusual.
  956:         include         "arch/i386/conf/XEN3_DOMU"
  958:         # Add support for PCI busses to the XEN3_DOMU kernel
  959:         xpci* at xenbus ?
  960:         pci* at xpci ?
  962:         # PCI USB controllers
  963:         uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
  965:         # USB bus support
  966:         usb*    at uhci?
  968:         # USB Hubs
  969:         uhub*   at usb?
  970:         uhub*   at uhub? port ? configuration ? interface ?
  972:         # USB Mass Storage
  973:         umass*  at uhub? port ? configuration ? interface ?
  974:         wd*     at umass?
  975:         # SCSI controllers
  976:         ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
  978:         # SCSI bus support (for both ahc and umass)
  979:         scsibus* at scsi?
  981:         # SCSI devices
  982:         sd*     at scsibus? target ? lun ?      # SCSI disk drives
  983:         cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
  986: NetBSD as a domU in a VPS
  987: =========================
  989: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
  990: hardware.  This section explains how to deal with Xen in a domU as a
  991: virtual private server where you do not control or have access to the
  992: dom0.  This is not intended to be an exhaustive list of VPS providers;
  993: only a few are mentioned that specifically support NetBSD.
  995: VPS operators provide varying degrees of access and mechanisms for
  996: configuration.  The big issue is usually how one controls which kernel
  997: is booted, because the kernel is nominally in the dom0 filesystem (to
  998: which VPS users do not normally have acesss).  A second issue is how
  999: to install NetBSD.
 1000: A VPS user may want to compile a kernel for security updates, to run
 1001: npf, run IPsec, or any other reason why someone would want to change
 1002: their kernel.
 1004: One approach is to have an adminstrative interface to upload a kernel,
 1005: or to select from a prepopulated list.  Other approaches are pygrub
 1006: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
 1007: kernel from the domU filesystem.  This is closer to a regular physical
 1008: computer, where someone who controls a machine can replace the kernel.
 1010: A second issue is multiple CPUs.  With NetBSD 6, domUs support
 1011: multiple vcpus, and it is typical for VPS providers to enable multiple
 1012: CPUs for NetBSD domUs.
 1014: pygrub
 1015: -------
 1017: pygrub runs in the dom0 and looks into the domU filesystem.  This
 1018: implies that the domU must have a kernel in a filesystem in a format
 1019: known to pygrub.  As of 2014, pygrub seems to be of mostly historical
 1020: interest.
 1022: pvgrub
 1023: ------
 1025: pvgrub is a version of grub that uses PV operations instead of BIOS
 1026: calls.  It is booted from the dom0 as the domU kernel, and then reads
 1027: /grub/menu.lst and loads a kernel from the domU filesystem.
 1029: [Panix]( lets users use pvgrub.  Panix reports
 1030: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
 1031: (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
 1032: page](, which describes only
 1033: Linux but should be updated to cover NetBSD :-).
 1035: []( also lets users with pvgrub to boot
 1036: their own kernel.  See then [ NetBSD
 1037: HOWTO](
 1038: (which is in need of updating).
 1040: It appears that [grub's FFS
 1041: code](
 1042: does not support all aspects of modern FFS, but there are also reports
 1043: that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
 1044: partition for the kernel with the intent that grub can understand it,
 1045: which leads to /netbsd not being the actual kernel.  One must remember
 1046: to update the special boot partiion.
 1048: Amazon
 1049: ------
 1051: See the [Amazon EC2 page](../amazon_ec2/).
 1053: Using npf
 1054: ---------
 1056: In standard kernels, npf is a module, and thus cannot be loaded in a
 1057: DOMU kernel.
 1059: TODO: Explain how to compile npf into a custom kernel, answering (but
 1060: note that the problem was caused by not booting the right kernel)
 1061: [this email to
 1062: netbsd-users](
 1064: TODO items for improving NetBSD/xen
 1065: ===================================
 1067: * Make the NetBSD dom0 kernel work with SMP.
 1068: * Test the Xen 4.5 packages adequately to be able to recommend them as
 1069:   the standard approach.
 1070: * Get PCI passthrough working on Xen 4.5
 1071: * Get pvgrub into pkgsrc, either via xentools or separately.
 1072: * grub
 1073:   * Check/add support to pkgsrc grub2 for UFS2 and arbitrary
 1074:     fragsize/blocksize (UFS2 support may be present; the point is to
 1075:     make it so that with any UFS1/UFS2 filesystem setup that works
 1076:     with NetBSD grub will also work).
 1077:     See [pkg/40258](
 1078:   * Push patches upstream.
 1079:   * Get UFS2 patches into pvgrub.
 1080: * Add support for PV ops to a version of /boot, and make it usable as
 1081:   a kernel in Xen, similar to pvgrub.
 1082: * Solve somehow the issue with modules for GENERIC not being loadable
 1083:   in a Xen dom0 or domU kernel.
 1085: Random pointers
 1086: ===============
 1088: TODO: This section contains links from elsewhere not yet integrated
 1089: into the HOWTO.
 1091: *
 1092: *

CVSweb for NetBSD wikisrc <> software: FreeBSD-CVSweb