File:  [NetBSD Developer Wiki] / wikisrc / ports / xen / howto.mdwn
Revision 1.148: download - view: text, annotated - select for diffs
Thu Jul 26 10:49:59 2018 UTC (2 years, 9 months ago) by maxv
Branches: MAIN
CVS tags: HEAD
Remove the "Stability" section, it is useless.

    1: [[!meta title="Xen HowTo"]]
    3: Introduction
    4: ============
    6: [![[Xen
    7: screenshot]](](
    9: Xen is a hypervisor (or virtual machine monitor) for x86 hardware
   10: (i686-class or higher), which supports running multiple guest
   11: operating systems on a single physical machine.  Xen is a Type 1 or
   12: bare-metal hypervisor; one uses the Xen kernel to control the CPU,
   13: memory and console, a dom0 operating system which mediates access to
   14: other hardware (e.g., disks, network, USB), and one or more domU
   15: operating systems which operate in an unprivileged virtualized
   16: environment.  IO requests from the domU systems are forwarded by the
   17: hypervisor (Xen) to the dom0 to be fulfilled.
   19: Xen supports two styles of guests.  The original is Para-Virtualized
   20: (PV) which means that the guest OS does not attempt to access hardware
   21: directly, but instead makes hypercalls to the hypervisor.  This is
   22: analogous to a user-space program making system calls.  (The dom0
   23: operating system uses PV calls for some functions, such as updating
   24: memory mapping page tables, but has direct hardware access for disk
   25: and network.)   PV guests must be specifically coded for Xen.
   27: The more recent style is HVM, which means that the guest does not have
   28: code for Xen and need not be aware that it is running under Xen.
   29: Attempts to access hardware registers are trapped and emulated.  This
   30: style is less efficient but can run unmodified guests.
   32: Generally any machine that runs NetBSD/amd64 will work with Xen and PV
   33: guests.  In theory i386 computers (without x86_64/amd64 support) can
   34: be used for Xen <= 4.2, but we have no recent reports of this working
   35: (this is a hint).  For HVM guests, hardware support is needed, but it
   36: is common on recent machines.  For Intel CPUs, one needs the VT-x
   37: extension, shown in "cpuctl identify 0" as VMX.  For AMD CPUs, one
   38: needs the AMD-V extensions, shown in "cpuctl identify 0" as SVM.
   39: There are further features for IOMMU virtualization, Intel's VT-d and
   40: AMD's AMD-Vi.  TODO: Explain whether Xen on NetBSD makes use of these
   41: features.  TODO: Review by someone who really understands this.
   43: Note that a FreeBSD dom0 requires VT-x and VT-d (or equivalent); this
   44: is because the FreeBSD dom0 does not run in PV mode.
   46: At boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   47: The dom0 can start one or more domUs.  (Booting is explained in detail
   48: in the dom0 section.)
   50: NetBSD supports Xen in that it can serve as dom0, be used as a domU,
   51: and that Xen kernels and tools are available in pkgsrc.  This HOWTO
   52: attempts to address both the case of running a NetBSD dom0 on hardware
   53: and running domUs under it (NetBSD and other), and also running NetBSD
   54: as a domU in a VPS.
   56: Xen 3.1 in pkgsrc used to support "PCI passthrough", which means that
   57: specific PCI devices can be made available to a specific domU instead
   58: of the dom0.  This can be useful to let a domU run X11, or access some
   59: network interface or other peripheral.
   61: NetBSD 6 and earlier supported Xen 2; support was removed from NetBSD
   62: 7.  Xen 2 has been removed from pkgsrc.
   64: Prerequisites
   65: -------------
   67: Installing NetBSD/Xen is not extremely difficult, but it is more
   68: complex than a normal installation of NetBSD.
   69: In general, this HOWTO is occasionally overly restrictive about how
   70: things must be done, guiding the reader to stay on the established
   71: path when there are no known good reasons to stray.
   73: This HOWTO presumes a basic familiarity with the Xen system
   74: architecture, with installing NetBSD on i386/amd64 hardware, and with
   75: installing software from pkgsrc.  See also the [Xen
   76: website](
   78: Versions of Xen and NetBSD
   79: ==========================
   81: Most of the installation concepts and instructions are independent
   82: of Xen version and NetBSD version.  This section gives advice on
   83: which version to choose.  Versions not in pkgsrc and older unsupported
   84: versions of NetBSD are intentionally ignored.
   86: The term "amd64" is used to refer to both the NetBSD port and to the
   87: hardware architecture on which it runs.  (Such hardware is made by
   88: both Intel and AMD, and in 2016 a normal PC has this CPU
   89: architecture.)
   91: Xen versions
   92: ------------
   94: In NetBSD, Xen is provided in pkgsrc, via matching pairs of packages
   95: xenkernel and xentools.  We will refer only to the kernel versions,
   96: but note that both packages must be installed together and must have
   97: matching versions.
   99: Versions available in pkgsrc:
  101: [[!table data="""
  102: Xen Version	|Package Name	|Xen CPU Support	|EOL'ed By Upstream
  103: 4.2		|xenkernel42	|32bit, 64bit		|Yes
  104: 4.5		|xenkernel45	|64bit			|Yes
  105: 4.6		|xenkernel46	|64bit			|Partially
  106: 4.8		|xenkernel48	|64bit			|No
  107: 4.11		|xenkernel411	|64bit			|No
  108: """]]
  110: See also the [Xen Security Advisory page](
  112: Note: Xen 4.2 was the last version to support 32bit CPUs.
  114: Xen command program
  115: -------------------
  117: Early Xen used a program called xm to manipulate the system from the
  118: dom0.  Starting in 4.1, a replacement program with similar behavior
  119: called xl is provided, but it does not work well in 4.1.  In 4.2, both
  120: xm and xl work fine.  4.4 is the last version that has xm.
  122: You must make a global choice to use xm or xl, because it affects not
  123: only which command you use, but the command used by rc.d scripts
  124: (specifically xendomains) and which daemons should be run.  The
  125: xentools packages provide xl for 4.2 and up.
  127: In 4.2, you can choose to use xm by simply changing the ctl_command
  128: variable and setting xend=YES in rc.conf.
  130: With xl, virtual devices are configured in parallel, which can cause
  131: problems if they are written assuming serial operation (e.g., updating
  132: firewall rules without explicit locking).  There is now locking for
  133: the provided scripts, which works for normal casses (e.g, file-backed
  134: xbd, where a vnd must be allocated).  But, as of 201612, it has not
  135: been adequately tested for a complex custom setup with a large number
  136: of interfaces.
  138: NetBSD versions
  139: ---------------
  141: The netbsd-7, netbsd-8, and -current branches are all reasonable
  142: choices, with more or less the same considerations for non-Xen use.
  143: NetBSD 8 is recommended as the stable version of the most recent
  144: release for production use.
  146: For developing Xen, netbsd-current may be appropriate.
  148: As of NetBSD 6, a NetBSD domU will support multiple vcpus.  There is
  149: no SMP support for NetBSD as dom0.  (The dom0 itself doesn't really
  150: need SMP for dom0 functions; the lack of support is really a problem
  151: when using a dom0 as a normal computer.)
  153: Note: NetBSD support is called XEN3. However, it does support Xen 4,
  154: because the hypercall interface has remained identical.
  156: Architecture
  157: ------------
  159: Xen itself can run on i386 (Xen < 4.2) or amd64 hardware (all Xen
  160: versions).  (Practically, almost any computer where one would want to
  161: run Xen today supports amd64.)
  163: Xen, the dom0 system, and each domU system can be either i386 or
  164: amd64.  When building a xenkernel package, one obtains an i386 Xen
  165: kernel on an i386 host, and an amd64 Xen kernel on an amd64 host.  If
  166: the Xen kernel is i386, then the dom0 kernel and all domU kernels must
  167: be i386.  With an amd64 Xen kernel, an amd64 dom0 kernel is known to
  168: work, and an i386 dom0 kernel should in theory work.  An amd64
  169: Xen/dom0 is known to support both i386 and amd64 domUs.
  171: i386 dom0 and domU kernels must be PAE (except for an i386 Xen 3.1
  172: kernel, where one can use non-PAE for dom0 and all domUs); PAE kernels
  173: are included in the NetBSD default build.  (Note that emacs (at least)
  174: fails if run on i386 with PAE when built without, and vice versa,
  175: presumably due to bugs in the undump code.)
  177: Because of the above, the standard approach is to use an amd64 Xen
  178: kernel and NetBSD/amd64 for the dom0.  For domUs, NetBSD/i386 (with
  179: the PAE kernel) and NetBSD/amd64 are in widespread use, and there is
  180: little to no Xen-specific reason to prefer one over the other.
  182: Note that to use an i386 dom0 with Xen 4.5 or higher, one must build
  183: (or obtain from pre-built packages) an amd64 Xen kernel and install
  184: that on the system.  (One must also use a PAE i386 kernel, but this is
  185: also required with an i386 Xen kernel.).  Almost no one in the
  186: NetBSD/Xen community does this, and the standard, well-tested,
  187: approach is to use an amd64 dom0.
  189: A [posting on
  190: xen-devel](
  191: explained that PV system call overhead was higher on amd64, and thus
  192: there is some notion that i386 guests are faster.  It goes on to
  193: caution that the total situation is complex and not entirely
  194: understood. On top of that caution, the post is about Linux, not
  195: NetBSD.  TODO: Include link to benchmarks, if someone posts them.
  197: Recommendation
  198: --------------
  200: Therefore, this HOWTO recommends running xenkernel46, xl, the NetBSD 7
  201: stable branch, and therefore to use an amd64 kernel as the dom0.
  202: Either the i386PAE or amd64 version of NetBSD may be used as domUs.
  204: A tentative replacement recommendation is xenkernel48, xl, and NetBSD
  205: 8.
  207: Because bugs are fixed quite often, and because of Xen security
  208: advisories, it is good to stay up to date with NetBSD (tracking a
  209: stable branch), with the Xen kernel (tracking a Xen version via
  210: pkgsrc), and with the Xen tools.  Specifically, NetBSD (-7 and
  211: -current) got an important fix affecting dom0/domU timesharing in
  212: November, 2015, and xentools46 got a fix to enable Ubuntu guests to
  213: boot in December, 2016.
  215: NetBSD as a dom0
  216: ================
  218: NetBSD can be used as a dom0 and works very well.  The following
  219: sections address installation, updating NetBSD, and updating Xen.
  220: Note that it doesn't make sense to talk about installing a dom0 OS
  221: without also installing Xen itself.  We first address installing
  222: NetBSD, which is not yet a dom0, and then adding Xen, pivoting the
  223: NetBSD install to a dom0 install by just changing the kernel and boot
  224: configuration.
  226: For experimenting with Xen, a machine with as little as 1G of RAM and
  227: 100G of disk can work.  For running many domUs in productions, far
  228: more will be needed; e.g. 4-8G and 1T of disk is reasonable for a
  229: half-dozen domUs of 512M and 32G each.  Basically, the RAM and disk
  230: have to be bigger than the sum of the RAM/disk needs of the dom0 and
  231: all the domUs.
  233: In 2018-05, trouble booting a dom0 was reported with 256M of RAM: with
  234: 512M it worked reliably.  This does not make sense, but if you see
  235: "not ELF" after Xen boots, try increasing dom0 RAM.
  237: Styles of dom0 operation
  238: ------------------------
  240: There are two basic ways to use Xen.  The traditional method is for
  241: the dom0 to do absolutely nothing other than providing support to some
  242: number of domUs.  Such a system was probably installed for the sole
  243: purpose of hosting domUs, and sits in a server room on a UPS.
  245: The other way is to put Xen under a normal-usage computer, so that the
  246: dom0 is what the computer would have been without Xen, perhaps a
  247: desktop or laptop.  Then, one can run domUs at will.  Purists will
  248: deride this as less secure than the previous approach, and for a
  249: computer whose purpose is to run domUs, they are right.  But Xen and a
  250: dom0 (without domUs) is not meaningfully less secure than the same
  251: things running without Xen.  One can boot Xen or boot regular NetBSD
  252: alternately with little problems, simply refraining from starting the
  253: Xen daemons when not running Xen.
  255: Note that NetBSD as dom0 does not support multiple CPUs.  This will
  256: limit the performance of the Xen/dom0 workstation approach.  In theory
  257: the only issue is that the "backend drivers" are not yet MPSAFE:
  260: Installation of NetBSD
  261: ----------------------
  263: First,
  264: [install NetBSD/amd64](/guide/inst/)
  265: just as you would if you were not using Xen.
  266: However, the partitioning approach is very important.
  268: If you want to use RAIDframe for the dom0, there are no special issues
  269: for Xen.  Typically one provides RAID storage for the dom0, and the
  270: domU systems are unaware of RAID.  The 2nd-stage loader bootxx_* skips
  271: over a RAID1 header to find /boot from a file system within a RAID
  272: partition; this is no different when booting Xen.
  274: There are 4 styles of providing backing storage for the virtual disks
  275: used by domUs: raw partitions, LVM, file-backed vnd(4), and SAN.
  277: With raw partitions, one has a disklabel (or gpt) partition sized for
  278: each virtual disk to be used by the domU.  (If you are able to predict
  279: how domU usage will evolve, please add an explanation to the HOWTO.
  280: Seriously, needs tend to change over time.)
  282: One can use [lvm(8)](/guide/lvm/) to create logical devices to use
  283: for domU disks.  This is almost as efficient as raw disk partitions
  284: and more flexible.  Hence raw disk partitions should typically not
  285: be used.
  287: One can use files in the dom0 file system, typically created by dd'ing
  288: /dev/zero to create a specific size.  This is somewhat less efficient,
  289: but very convenient, as one can cp the files for backup, or move them
  290: between dom0 hosts.
  292: Finally, in theory one can place the files backing the domU disks in a
  293: SAN.  (This is an invitation for someone who has done this to add a
  294: HOWTO page.)
  296: Installation of Xen
  297: -------------------
  299: In the dom0, install sysutils/xenkernel42 and sysutils/xentools42 from
  300: pkgsrc (or another matching pair).  See [the pkgsrc
  301: documentation]( for help with
  302: pkgsrc.  Ensure that your packages are recent; the HOWTO does not
  303: contemplate old builds.
  306: For Xen 3.1, support for HVM guests is in sysutils/xentool3-hvm.  More
  307: recent versions have HVM support integrated in the main xentools
  308: package.  It is entirely reasonable to run only PV guests.
  310: Next you need to install the selected Xen kernel itself, which is
  311: installed by pkgsrc as "/usr/pkg/xen*-kernel/xen.gz".  Copy it to /.
  312: For debugging, one may copy xen-debug.gz; this is conceptually similar
  313: to DIAGNOSTIC and DEBUG in NetBSD.  xen-debug.gz is basically only
  314: useful with a serial console.  Then, place a NetBSD XEN3_DOM0 kernel
  315: in /, copied from releasedir/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
  316: of a NetBSD build.  If using i386, use
  317: releasedir/i386/binary/kernel/netbsd-XEN3PAE_DOM0.gz.  (If using Xen
  318: 3.1 and i386, you may use XEN3_DOM0 with the non-PAE Xen.  But you
  319: should not use Xen 3.1.)  Both xen and the NetBSD kernel may be (and
  320: typically are) left compressed.
  322: In a dom0, kernfs is mandatory for xend to communicate with the
  323: kernel, so ensure that /kern is in fstab.  (A standard NetBSD install
  324: should already mount /kern.)
  326: Because you already installed NetBSD, you have a working boot setup
  327: with an MBR bootblock, either bootxx_ffsv1 or bootxx_ffsv2 at the
  328: beginning of your root file system, have /boot, and likely also
  329: /boot.cfg.  (If not, fix before continuing!)
  331: Add a line to to /boot.cfg to boot Xen.  See boot.cfg(5) for an
  332: example.  The basic line is
  334:         menu=Xen:load /netbsd-XEN3_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
  336: which specifies that the dom0 should have 512M, leaving the rest to be
  337: allocated for domUs.  To use a serial console, use
  339:         menu=Xen:load /netbsd-XEN3_DOM0.gz;multiboot /xen.gz dom0_mem=512M console=com1 com1=9600,8n1
  341: which will use the first serial port for Xen (which counts starting
  342: from 1, unlike NetBSD which counts starting from 0), forcing
  343: speed/parity.  Because the NetBSD command line lacks a
  344: "console=pc" argument, it will use the default "xencons" console device,
  345: which directs the console I/O through Xen to the same console device Xen
  346: itself uses (in this case, the serial port).
  348: In an attempt to add performance, one can also add
  350:         dom0_max_vcpus=1 dom0_vcpus_pin
  352: to force only one vcpu to be provided (since NetBSD dom0 can't use
  353: more) and to pin that vcpu to a physical CPU.  TODO: benchmark this.
  355: Xen has [many boot
  356: options](,
  357: and other than dom0 memory and max_vcpus, they are generally not
  358: necessary.
  360: As with non-Xen systems, you should have a line to boot /netbsd (a
  361: kernel that works without Xen).  Consider a line to boot /netbsd.ok (a
  362: fallback version of the non-Xen kernel, updated manually when you are
  363: sure /netbsd is ok).  Consider also a line to boot fallback versions
  364: of Xen and the dom0 kernel, but note that non-Xen NetBSD can be used
  365: to resolve Xen booting issues.
  367: Probably you want a default=N line to choose Xen in the absence of
  368: intervention.
  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.  Now, just create a system with RAID-1, and alter /boot.cfg as
  385: described above.)
  387: Configuring Xen
  388: ---------------
  390: Xen logs will be in /var/log/xen.
  392: Now, you have a system that will boot Xen and the dom0 kernel, but not
  393: do anything else special.  Make sure that you have rebooted into Xen.
  394: There will be no domUs, and none can be started because you still have
  395: to configure the dom0 daemons.
  397: The daemons which should be run vary with Xen version and with whether
  398: one is using xm or xl.  The Xen 3.1, 3.3 and 4.1 packages use xm.  Xen
  399: 4.2 and up packages use xl.  To use xm with 4.2, edit xendomains to
  400: use xm instead.
  402: For 3.1 and 3.3, you should enable xend and xenbackendd:
  404:         xend=YES
  405:         xenbackendd=YES
  407: For 4.1 and up, you should enable xencommons.  Not enabling xencommons
  408: will result in a hang; it is necessary to hit ^C on the console to let
  409: the machine finish booting.  If you are using xm (default in 4.1, or
  410: if you changed xendomains in 4.2), you should also enable xend:
  412:         xend=YES # only if using xm, and only installed <= 4.2
  413:         xencommons=YES
  415: TODO: Recommend for/against xen-watchdog.
  417: After you have configured the daemons and either started them (in the
  418: order given) or rebooted, use xm or xl to inspect Xen's boot messages,
  419: available resources, and running domains.  An example with xl follows:
  421:         # xl dmesg
  422: 	[xen's boot info]
  423:         # xl info
  424: 	[available memory, etc.]
  425:         # xl list
  426:         Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  427:         Domain-0           0       64    0  r----     58.1
  429: ### Issues with xencommons
  431: xencommons starts xenstored, which stores data on behalf of dom0 and
  432: domUs.  It does not currently work to stop and start xenstored.
  433: Certainly all domUs should be shutdown first, following the sort order
  434: of the rc.d scripts.  However, the dom0 sets up state with xenstored,
  435: and is not notified when xenstored exits, leading to not recreating
  436: the state when the new xenstored starts.  Until there's a mechanism to
  437: make this work, one should not expect to be able to restart xenstored
  438: (and thus xencommons).  There is currently no reason to expect that
  439: this will get fixed any time soon.
  441: ### No-longer needed advice about devices
  443: The installation of NetBSD should already have created devices for xen
  444: (xencons, xenevt, xsd_kva), but if they are not present, create them:
  446:         cd /dev && sh MAKEDEV xen
  448: anita (for testing NetBSD)
  449: --------------------------
  451: With the setup so far (assuming 4.2/xl), one should be able to run
  452: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
  453: root, because anita must create a domU):
  455:         anita --vmm=xl test file:///usr/obj/i386/
  457: Alternatively, one can use --vmm=xm to use xm-based domU creation
  458: instead (and must, on Xen <= 4.1).   TODO: confirm that anita xl really works.
  460: Xen-specific NetBSD issues
  461: --------------------------
  463: There are (at least) two additional things different about NetBSD as a
  464: dom0 kernel compared to hardware.
  466: One is that the module ABI is different because some of the #defines
  467: change, so one must build modules for Xen.  As of netbsd-7, the build
  468: system does this automatically.  TODO: check this.  (Before building
  469: Xen modules was added, it was awkward to use modules to the point
  470: where it was considered that it did not work.)
  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 boot.cfg following earlier guidance:
  507:         menu=Xen:load /netbsd-XEN3PAE_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
  508:         menu=Xen.ok:load /netbsd-XEN3PAE_DOM0.ok.gz console=pc;multiboot /xen.ok.gz dom0_mem=512M
  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: Minor version upgrades are trivial.  Just rebuild/replace the
  523: xenkernel version and copy the new xen.gz to / (where /boot.cfg
  524: references it), and reboot.
  526: Major version upgrades are conceptually not difficult, but can run
  527: into all the issues found when installing Xen.  Assuming migration
  528: from 4.1 to 4.2, remove the xenkernel41 and xentools41 packages and
  529: install the xenkernel42 and xentools42 packages.  Copy the 4.2 xen.gz
  530: to /.
  532: Ensure that the contents of /etc/rc.d/xen* are correct.  Specifically,
  533: they must match the package you just installed and not be left over
  534: from some previous installation.
  536: Enable the correct set of daemons; see the configuring section above.
  537: (Upgrading from 3.x to 4.x without doing this will result in a hang.)
  539: Ensure that the domU config files are valid for the new version.
  540: Specifically, for 4.x remove autorestart=True, and ensure that disks
  541: are specified with numbers as the second argument, as the examples
  542: above show, and not NetBSD device names.
  544: Hardware known to work
  545: ----------------------
  547: Arguably, this section is misplaced, and there should be a page of
  548: hardware that runs NetBSD/amd64 well, with the mostly-well-founded
  549: assumption that NetBSD/xen runs fine on any modern hardware that
  550: NetBSD/amd64 runs well on.  Until then, we give motherboard/CPU (and
  551: sometimes RAM) pairs/triples to aid those choosing a motherboard.
  552: Note that Xen systems usually do not run X, so a listing here does not
  553: imply that X works at all.
  555:         Supermicro X9SRL-F, Xeon E5-1650 v2, 96 GiB ECC
  556:         Supermicro ??, Atom C2758 (8 core), 32 GiB ECC
  557:         ASUS M5A78L-M/USB3 AM3+ microATX, AMD Piledriver X8 4000MHz, 16 GiB ECC
  559: Older hardware:
  561:         Intel D915GEV, Pentium4 CPU 3.40GHz, 4GB 533MHz Synchronous DDR2
  562:         INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E6850  @ 3.00GHz"
  563:         INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E8400  @ 3.00GHz"
  565: Running Xen under qemu
  566: ----------------------
  568: The astute reader will note that this section is somewhat twisted.
  569: However, it can be useful to run Xen under qemu either because the
  570: version of NetBSD as a dom0 does not run on the hardware in use, or to
  571: generate automated test cases involving Xen.
  573: In 2015-01, the following combination was reported to mostly work:
  575:         host OS: NetBSD/amd64 6.1.4
  576:         qemu: 2.2.0 from pkgsrc
  577:         Xen kernel: xenkernel42-4.2.5nb1 from pkgsrc
  578:         dom0 kernel: NetBSD/amd64 6.1.5
  579:         Xen tools: xentools42-4.2.5 from pkgsrc
  581: See [PR 47720]( for a problem with dom0
  582: shutdown.
  584: Unprivileged domains (domU)
  585: ===========================
  587: This section describes general concepts about domUs.  It does not
  588: address specific domU operating systems or how to install them.  The
  589: config files for domUs are typically in /usr/pkg/etc/xen, and are
  590: typically named so that the file name, domU name and the domU's host
  591: name match.
  593: The domU is provided with CPU and memory by Xen, configured by the
  594: dom0.  The domU is provided with disk and network by the dom0,
  595: mediated by Xen, and configured in the dom0.
  597: Entropy in domUs can be an issue; physical disks and network are on
  598: the dom0.  NetBSD's /dev/random system works, but is often challenged.
  600: Config files
  601: ------------
  603: There is no good order to present config files and the concepts
  604: surrounding what is being configured.  We first show an example config
  605: file, and then in the various sections give details.
  607: See (at least in xentools41) /usr/pkg/share/examples/xen/xmexample*,
  608: for a large number of well-commented examples, mostly for running
  609: GNU/Linux.
  611: The following is an example minimal domain configuration file
  612: "/usr/pkg/etc/xen/foo".  It is (with only a name change) an actual
  613: known working config file on Xen 4.1 (NetBSD 5 amd64 dom0 and NetBSD 5
  614: i386 domU).  The domU serves as a network file server.
  616:         # -*- mode: python; -*-
  618:         kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
  619:         memory = 1024
  620:         vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
  621:         disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
  622:                  'file:/n0/xen/foo-wd1,0x1,w' ]
  624: The domain will have the same name as the file.  The kernel has the
  625: host/domU name in it, so that on the dom0 one can update the various
  626: domUs independently.  The vif line causes an interface to be provided,
  627: with a specific mac address (do not reuse MAC addresses!), in bridge
  628: mode.  Two disks are provided, and they are both writable; the bits
  629: are stored in files and Xen attaches them to a vnd(4) device in the
  630: dom0 on domain creation.  The system treats xbd0 as the boot device
  631: without needing explicit configuration.
  633: By default xm looks for domain config files in /usr/pkg/etc/xen.  Note
  634: that "xm create" takes the name of a config file, while other commands
  635: take the name of a domain.  To create the domain, connect to the
  636: console, create the domain while attaching the console, shutdown the
  637: domain, and see if it has finished stopping, do (or xl with Xen >=
  638: 4.2):
  640:         xm create foo
  641:         xm console foo
  642:         xm create -c foo
  643:         xm shutdown foo
  644:         xm list
  646: Typing ^] will exit the console session.  Shutting down a domain is
  647: equivalent to pushing the power button; a NetBSD domU will receive a
  648: power-press event and do a clean shutdown.  Shutting down the dom0
  649: will trigger controlled shutdowns of all configured domUs.
  651: domU kernels
  652: ------------
  654: On a physical computer, the BIOS reads sector 0, and a chain of boot
  655: loaders finds and loads a kernel.  Normally this comes from the root
  656: file system.  With Xen domUs, the process is totally different.  The
  657: normal path is for the domU kernel to be a file in the dom0's
  658: file system.  At the request of the dom0, Xen loads that kernel into a
  659: new domU instance and starts execution.  While domU kernels can be
  660: anyplace, reasonable places to store domU kernels on the dom0 are in /
  661: (so they are near the dom0 kernel), in /usr/pkg/etc/xen (near the
  662: config files), or in /u0/xen (where the vdisks are).
  664: Note that loading the domU kernel from the dom0 implies that boot
  665: blocks, /boot, /boot.cfg, and so on are all ignored in the domU.
  666: See the VPS section near the end for discussion of alternate ways to
  667: obtain domU kernels.
  669: CPU and memory
  670: --------------
  672: A domain is provided with some number of vcpus, less than the number
  673: of CPUs seen by the hypervisor.  (For a dom0, this is controlled by
  674: the boot argument "dom0_max_vcpus=1".)  For a domU, it is controlled
  675: from the config file by the "vcpus = N" directive.
  677: A domain is provided with memory; this is controlled in the config
  678: file by "memory = N" (in megabytes).  In the straightforward case, the
  679: sum of the the memory allocated to the dom0 and all domUs must be less
  680: than the available memory.
  682: Xen also provides a "balloon" driver, which can be used to let domains
  683: use more memory temporarily.  TODO: Explain better, and explain how
  684: well it works with NetBSD.
  686: Virtual disks
  687: -------------
  689: With the file/vnd style, typically one creates a directory,
  690: e.g. /u0/xen, on a disk large enough to hold virtual disks for all
  691: domUs.  Then, for each domU disk, one writes zeros to a file that then
  692: serves to hold the virtual disk's bits; a suggested name is foo-xbd0
  693: for the first virtual disk for the domU called foo.  Writing zeros to
  694: the file serves two purposes.  One is that preallocating the contents
  695: improves performance.  The other is that vnd on sparse files has
  696: failed to work.  TODO: give working/notworking NetBSD versions for
  697: sparse vnd and gnats reference.  Note that the use of file/vnd for Xen
  698: is not really different than creating a file-backed virtual disk for
  699: some other purpose, except that xentools handles the vnconfig
  700: commands.  To create an empty 4G virtual disk, simply do
  702:         dd if=/dev/zero of=foo-xbd0 bs=1m count=4096
  704: Do not use qemu-img-xen, because this will create sparse file.  There
  705: have been recent (2015) reports of sparse vnd(4) devices causing
  706: lockups, but there is apparently no PR.
  708: With the lvm style, one creates logical devices.  They are then used
  709: similarly to vnds.  TODO: Add an example with lvm.
  711: In domU config files, the disks are defined as a sequence of 3-tuples.
  712: The first element is "method:/path/to/disk".  Common methods are
  713: "file:" for file-backed vnd. and "phy:" for something that is already
  714: a (TODO: character or block) device.
  716: The second element is an artifact of how virtual disks are passed to
  717: Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
  718: are given a device name to associate with the disk, and values like
  719: "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
  720: as xbd0, the second as xbd1, and so on.  However, xm/xl demand a
  721: second argument.  The name given is converted to a major/minor by
  722: calling stat(2) on the name in /dev and this is passed to the domU.
  723: In the general case, the dom0 and domU can be different operating
  724: systems, and it is an unwarranted assumption that they have consistent
  725: numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
  726: as both dom0 and domU, using values of 0x0 for the first disk and 0x1
  727: for the second works fine and avoids this issue.  For a GNU/Linux
  728: guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
  729: /dev/hda1.
  731: The third element is "w" for writable disks, and "r" for read-only
  732: disks.
  734: Note that NetBSD by default creates only vnd[0123].  If you need more
  735: than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
  736: dom0.
  738: Note that NetBSD by default creates only xbd[0123].  If you need more
  739: virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
  741: Virtual Networking
  742: ------------------
  744: Xen provides virtual Ethernets, each of which connects the dom0 and a
  745: domU.  For each virtual network, there is an interface "xvifN.M" in
  746: the dom0, and in domU index N, a matching interface xennetM (NetBSD
  747: name).  The interfaces behave as if there is an Ethernet with two
  748: adapters connected.  From this primitive, one can construct various
  749: configurations.  We focus on two common and useful cases for which
  750: there are existing scripts: bridging and NAT.
  752: With bridging (in the example above), the domU perceives itself to be
  753: on the same network as the dom0.  For server virtualization, this is
  754: usually best.  Bridging is accomplished by creating a bridge(4) device
  755: and adding the dom0's physical interface and the various xvifN.0
  756: interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
  757: config file.  The bridge must be set up already in the dom0; an
  758: example /etc/ifconfig.bridge0 is:
  760:         create
  761:         up
  762:         !brconfig bridge0 add wm0
  764: With NAT, the domU perceives itself to be behind a NAT running on the
  765: dom0.  This is often appropriate when running Xen on a workstation.
  766: TODO: NAT appears to be configured by "vif = [ '' ]".
  768: The MAC address specified is the one used for the interface in the new
  769: domain.  The interface in dom0 will use this address XOR'd with
  770: 00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
  772: Sizing domains
  773: --------------
  775: Modern x86 hardware has vast amounts of resources.  However, many
  776: virtual servers can function just fine on far less.  A system with
  777: 512M of RAM and a 4G disk can be a reasonable choice.  Note that it is
  778: far easier to adjust virtual resources than physical ones.  For
  779: memory, it's just a config file edit and a reboot.  For disk, one can
  780: create a new file and vnconfig it (or lvm), and then dump/restore,
  781: just like updating physical disks, but without having to be there and
  782: without those pesky connectors.
  784: Starting domains automatically
  785: ------------------------------
  787: To start domains foo at bar at boot and shut them down cleanly on dom0
  788: shutdown, in rc.conf add:
  790:         xendomains="foo bar"
  792: Note that earlier versions of the xentools41 xendomains rc.d script
  793: used xl, when one should use xm with 4.1.
  795: Creating specific unprivileged domains (domU)
  796: =============================================
  798: Creating domUs is almost entirely independent of operating system.  We
  799: have already presented the basics of config files.  Note that you must
  800: have already completed the dom0 setup so that "xl list" (or "xm list")
  801: works.
  803: Creating an unprivileged NetBSD domain (domU)
  804: ---------------------------------------------
  806: See the earlier config file, and adjust memory.  Decide on how much
  807: storage you will provide, and prepare it (file or lvm).
  809: While the kernel will be obtained from the dom0 file system, the same
  810: file should be present in the domU as /netbsd so that tools like
  811: savecore(8) can work.   (This is helpful but not necessary.)
  813: The kernel must be specifically for Xen and for use as a domU.  The
  814: i386 and amd64 provide the following kernels:
  816:         i386 XEN3_DOMU
  817:         i386 XEN3PAE_DOMU
  818:         amd64 XEN3_DOMU
  820: Unless using Xen 3.1 (and you shouldn't) with i386-mode Xen, you must
  821: use the PAE version of the i386 kernel.
  823: This will boot NetBSD, but this is not that useful if the disk is
  824: empty.  One approach is to unpack sets onto the disk outside of xen
  825: (by mounting it, just as you would prepare a physical disk for a
  826: system you can't run the installer on).
  828: A second approach is to run an INSTALL kernel, which has a miniroot
  829: and can load sets from the network.  To do this, copy the INSTALL
  830: kernel to / and change the kernel line in the config file to:
  832:         kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
  834: Then, start the domain as "xl create -c configname".
  836: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
  837: line should be used in the config file.
  839:     disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
  841: After booting the domain, the option to install via CDROM may be
  842: selected.  The CDROM device should be changed to `xbd1d`.
  844: Once done installing, "halt -p" the new domain (don't reboot or halt,
  845: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
  846: config file), switch the config file back to the XEN3_DOMU kernel,
  847: and start the new domain again. Now it should be able to use "root on
  848: xbd0a" and you should have a, functional NetBSD domU.
  850: TODO: check if this is still accurate.
  851: When the new domain is booting you'll see some warnings about *wscons*
  852: and the pseudo-terminals. These can be fixed by editing the files
  853: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
  854: `/etc/ttys`, except *console*, like this:
  856:     console "/usr/libexec/getty Pc"         vt100   on secure
  857:     ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
  858:     ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
  859:     ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
  860:     ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
  862: Finally, all screens must be commented out from `/etc/wscons.conf`.
  864: It is also desirable to add
  866:         powerd=YES
  868: in rc.conf. This way, the domain will be properly shut down if
  869: `xm shutdown -R` or `xm shutdown -H` is used on the dom0.
  871: It is not strictly necessary to have a kernel (as /netbsd) in the domU
  872: file system.  However, various programs (e.g. netstat) will use that
  873: kernel to look up symbols to read from kernel virtual memory.  If
  874: /netbsd is not the running kernel, those lookups will fail.  (This is
  875: not really a Xen-specific issue, but because the domU kernel is
  876: obtained from the dom0, it is far more likely to be out of sync or
  877: missing with Xen.)
  879: Creating an unprivileged Linux domain (domU)
  880: --------------------------------------------
  882: Creating unprivileged Linux domains isn't much different from
  883: unprivileged NetBSD domains, but there are some details to know.
  885: First, the second parameter passed to the disk declaration (the '0x1' in
  886: the example below)
  888:     disk = [ 'phy:/dev/wd0e,0x1,w' ]
  890: does matter to Linux. It wants a Linux device number here (e.g. 0x300
  891: for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
  892: So, hda1 which has major 3 and minor 1 on a Linux system will have
  893: device number 0x301.  Alternatively, devices names can be used (hda,
  894: hdb, ...)  as xentools has a table to map these names to devices
  895: numbers.  To export a partition to a Linux guest we can use:
  897:         disk = [ 'phy:/dev/wd0e,0x300,w' ]
  898:         root = "/dev/hda1 ro"
  900: and it will appear as /dev/hda on the Linux system, and be used as root
  901: partition.
  903: To install the Linux system on the partition to be exported to the
  904: guest domain, the following method can be used: install
  905: sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
  906: that will be the root partition of your Linux domain, and mount it.
  907: Then copy the files from a working Linux system, make adjustments in
  908: `/etc` (fstab, network config).  It should also be possible to extract
  909: binary packages such as .rpm or .deb directly to the mounted partition
  910: using the appropriate tool, possibly running under NetBSD's Linux
  911: emulation.  Once the file system has been populated, umount it.  If
  912: desirable, the file system can be converted to ext3 using tune2fs -j.
  913: It should now be possible to boot the Linux guest domain, using one of
  914: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
  916: To get the Linux console right, you need to add:
  918:     extra = "xencons=tty1"
  920: to your configuration since not all Linux distributions auto-attach a
  921: tty to the xen console.
  923: Creating an unprivileged Solaris domain (domU)
  924: ----------------------------------------------
  926: See possibly outdated
  927: [Solaris domU instructions](/ports/xen/howto-solaris/).
  930: PCI passthrough: Using PCI devices in guest domains
  931: ---------------------------------------------------
  933: The dom0 can give other domains access to selected PCI
  934: devices. This can allow, for example, a non-privileged domain to have
  935: access to a physical network interface or disk controller.  However,
  936: keep in mind that giving a domain access to a PCI device most likely
  937: will give the domain read/write access to the whole physical memory,
  938: as PCs don't have an IOMMU to restrict memory access to DMA-capable
  939: device.  Also, it's not possible to export ISA devices to non-dom0
  940: domains, which means that the primary VGA adapter can't be exported.
  941: A guest domain trying to access the VGA registers will panic.
  943: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
  944: not been ported to later versions at this time.
  946: For a PCI device to be exported to a domU, is has to be attached to
  947: the "pciback" driver in dom0.  Devices passed to the dom0 via the
  948: pciback.hide boot parameter will attach to "pciback" instead of the
  949: usual driver.  The list of devices is specified as "(bus:dev.func)",
  950: where bus and dev are 2-digit hexadecimal numbers, and func a
  951: single-digit number:
  953:         pciback.hide=(00:0a.0)(00:06.0)
  955: pciback devices should show up in the dom0's boot messages, and the
  956: devices should be listed in the `/kern/xen/pci` directory.
  958: PCI devices to be exported to a domU are listed in the "pci" array of
  959: the domU's config file, with the format "0000:bus:dev.func".
  961:         pci = [ '0000:00:06.0', '0000:00:0a.0' ]
  963: In the domU an "xpci" device will show up, to which one or more pci
  964: buses will attach.  Then the PCI drivers will attach to PCI buses as
  965: usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
  966: or any PCI drivers built in by default; you have to build your own
  967: kernel to use PCI devices in a domU.  Here's a kernel config example;
  968: note that only the "xpci" lines are unusual.
  970:         include         "arch/i386/conf/XEN3_DOMU"
  972:         # Add support for PCI buses to the XEN3_DOMU kernel
  973:         xpci* at xenbus ?
  974:         pci* at xpci ?
  976:         # PCI USB controllers
  977:         uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
  979:         # USB bus support
  980:         usb*    at uhci?
  982:         # USB Hubs
  983:         uhub*   at usb?
  984:         uhub*   at uhub? port ? configuration ? interface ?
  986:         # USB Mass Storage
  987:         umass*  at uhub? port ? configuration ? interface ?
  988:         wd*     at umass?
  989:         # SCSI controllers
  990:         ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
  992:         # SCSI bus support (for both ahc and umass)
  993:         scsibus* at scsi?
  995:         # SCSI devices
  996:         sd*     at scsibus? target ? lun ?      # SCSI disk drives
  997:         cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
 1000: NetBSD as a domU in a VPS
 1001: =========================
 1003: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
 1004: hardware.  This section explains how to deal with Xen in a domU as a
 1005: virtual private server where you do not control or have access to the
 1006: dom0.  This is not intended to be an exhaustive list of VPS providers;
 1007: only a few are mentioned that specifically support NetBSD.
 1009: VPS operators provide varying degrees of access and mechanisms for
 1010: configuration.  The big issue is usually how one controls which kernel
 1011: is booted, because the kernel is nominally in the dom0 file system (to
 1012: which VPS users do not normally have access).  A second issue is how
 1013: to install NetBSD.
 1014: A VPS user may want to compile a kernel for security updates, to run
 1015: npf, run IPsec, or any other reason why someone would want to change
 1016: their kernel.
 1018: One approach is to have an administrative interface to upload a kernel,
 1019: or to select from a prepopulated list.  Other approaches are pygrub
 1020: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
 1021: kernel from the domU file system.  This is closer to a regular physical
 1022: computer, where someone who controls a machine can replace the kernel.
 1024: A second issue is multiple CPUs.  With NetBSD 6, domUs support
 1025: multiple vcpus, and it is typical for VPS providers to enable multiple
 1026: CPUs for NetBSD domUs.
 1028: pygrub
 1029: -------
 1031: pygrub runs in the dom0 and looks into the domU file system.  This
 1032: implies that the domU must have a kernel in a file system in a format
 1033: known to pygrub.  As of 2014, pygrub seems to be of mostly historical
 1034: interest.
 1036: pvgrub
 1037: ------
 1039: pvgrub is a version of grub that uses PV operations instead of BIOS
 1040: calls.  It is booted from the dom0 as the domU kernel, and then reads
 1041: /grub/menu.lst and loads a kernel from the domU file system.
 1043: [Panix]( lets users use pvgrub.  Panix reports
 1044: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
 1045: (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
 1046: page](, which describes only
 1047: Linux but should be updated to cover NetBSD :-).
 1049: []( also lets users with pvgrub to boot
 1050: their own kernel.  See then [ NetBSD
 1051: HOWTO](
 1052: (which is in need of updating).
 1054: It appears that [grub's FFS
 1055: code](
 1056: does not support all aspects of modern FFS, but there are also reports
 1057: that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
 1058: partition for the kernel with the intent that grub can understand it,
 1059: which leads to /netbsd not being the actual kernel.  One must remember
 1060: to update the special boot partition.
 1062: Amazon
 1063: ------
 1065: See the [Amazon EC2 page](/amazon_ec2/).
 1067: TODO items for improving NetBSD/xen
 1068: ===================================
 1070: * Make the NetBSD dom0 kernel work with SMP.
 1071: * Test the Xen 4.5 packages adequately to be able to recommend them as
 1072:   the standard approach.
 1073: * Get PCI passthrough working on Xen 4.5
 1074: * Get pvgrub into pkgsrc, either via xentools or separately.
 1075: * grub
 1076:   * Check/add support to pkgsrc grub2 for UFS2 and arbitrary
 1077:     fragsize/blocksize (UFS2 support may be present; the point is to
 1078:     make it so that with any UFS1/UFS2 file system setup that works
 1079:     with NetBSD grub will also work).
 1080:     See [pkg/40258](
 1081:   * Push patches upstream.
 1082:   * Get UFS2 patches into pvgrub.
 1083: * Add support for PV ops to a version of /boot, and make it usable as
 1084:   a kernel in Xen, similar to pvgrub.
 1086: Random pointers
 1087: ===============
 1089: This section contains links from elsewhere not yet integrated into the
 1090: HOWTO, and other guides.
 1092: *
 1093: *
 1094: *

CVSweb for NetBSD wikisrc <> software: FreeBSD-CVSweb