File:  [NetBSD Developer Wiki] / wikisrc / ports / xen / howto.mdwn
Revision 1.149: download - view: text, annotated - select for diffs
Thu Jul 26 11:00:10 2018 UTC (2 years, 9 months ago) by maxv
Branches: MAIN
CVS tags: HEAD
Reduce the noise in the introduction, use a table.

    1: [[!meta title="Xen HowTo"]]
    3: Introduction
    4: ============
    6: [![[Xen
    7: screenshot]](](
    9: Xen is a hypervisor for x86 hardware, which supports running multiple guest
   10: operating systems on a single physical machine.  Xen is a Type 1 or
   11: bare-metal hypervisor; one uses the Xen kernel to control the CPU,
   12: memory and console, a dom0 operating system which mediates access to
   13: other hardware (e.g., disks, network, USB), and one or more domU
   14: operating systems which operate in an unprivileged virtualized
   15: environment.  IO requests from the domU systems are forwarded by the
   16: Xen hypervisor to the dom0 to be fulfilled.
   18: Xen supports different styles of guest:
   20: [[!table data="""
   21: Style of guest	|Supported by NetBSD
   22: PV		|Yes
   23: HVM		|Yes
   24: PVHVM		|No
   25: PVH		|No
   26: """]]
   28: In Para-Virtualized (PV) mode, the guest OS does not attempt to access
   29: hardware directly, but instead makes hypercalls to the hypervisor; PV
   30: guests must be specifically coded for Xen. In HVM mode, no guest
   31: modification is required; however, hardware support is required, such
   32: as VT-x on Intel CPUs and SVM on AMD CPUs.
   34: There are further features for IOMMU virtualization, Intel's VT-d and
   35: AMD's AMD-Vi.  TODO: Explain whether Xen on NetBSD makes use of these
   36: features.  TODO: Review by someone who really understands this.
   38: Generally any machine that runs NetBSD/amd64 will work with Xen and PV
   39: guests.  In theory i386 computers (without x86_64/amd64 support) can
   40: be used for Xen <= 4.2, but we have no recent reports of this working
   41: (this is a hint).  For HVM guests, hardware support is needed, but it
   42: is common on recent machines.
   44: At boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   45: The dom0 can start one or more domUs.  (Booting is explained in detail
   46: in the dom0 section.)
   48: NetBSD supports Xen in that it can serve as dom0, be used as a domU,
   49: and that Xen kernels and tools are available in pkgsrc.  This HOWTO
   50: attempts to address both the case of running a NetBSD dom0 on hardware
   51: and running domUs under it (NetBSD and other), and also running NetBSD
   52: as a domU in a VPS.
   54: Xen 3.1 in pkgsrc used to support "PCI passthrough", which means that
   55: specific PCI devices can be made available to a specific domU instead
   56: of the dom0.  This can be useful to let a domU run X11, or access some
   57: network interface or other peripheral.
   59: NetBSD 6 and earlier supported Xen 2; support was removed from NetBSD
   60: 7.  Xen 2 has been removed from pkgsrc.
   62: Prerequisites
   63: -------------
   65: Installing NetBSD/Xen is not extremely difficult, but it is more
   66: complex than a normal installation of NetBSD.
   67: In general, this HOWTO is occasionally overly restrictive about how
   68: things must be done, guiding the reader to stay on the established
   69: path when there are no known good reasons to stray.
   71: This HOWTO presumes a basic familiarity with the Xen system
   72: architecture, with installing NetBSD on i386/amd64 hardware, and with
   73: installing software from pkgsrc.  See also the [Xen
   74: website](
   76: Versions of Xen and NetBSD
   77: ==========================
   79: Most of the installation concepts and instructions are independent
   80: of Xen version and NetBSD version.  This section gives advice on
   81: which version to choose.  Versions not in pkgsrc and older unsupported
   82: versions of NetBSD are intentionally ignored.
   84: The term "amd64" is used to refer to both the NetBSD port and to the
   85: hardware architecture on which it runs.  (Such hardware is made by
   86: both Intel and AMD, and in 2016 a normal PC has this CPU
   87: architecture.)
   89: Xen versions
   90: ------------
   92: In NetBSD, Xen is provided in pkgsrc, via matching pairs of packages
   93: xenkernel and xentools.  We will refer only to the kernel versions,
   94: but note that both packages must be installed together and must have
   95: matching versions.
   97: Versions available in pkgsrc:
   99: [[!table data="""
  100: Xen Version	|Package Name	|Xen CPU Support	|EOL'ed By Upstream
  101: 4.2		|xenkernel42	|32bit, 64bit		|Yes
  102: 4.5		|xenkernel45	|64bit			|Yes
  103: 4.6		|xenkernel46	|64bit			|Partially
  104: 4.8		|xenkernel48	|64bit			|No
  105: 4.11		|xenkernel411	|64bit			|No
  106: """]]
  108: See also the [Xen Security Advisory page](
  110: Note: Xen 4.2 was the last version to support 32bit CPUs.
  112: Xen command program
  113: -------------------
  115: Early Xen used a program called xm to manipulate the system from the
  116: dom0.  Starting in 4.1, a replacement program with similar behavior
  117: called xl is provided, but it does not work well in 4.1.  In 4.2, both
  118: xm and xl work fine.  4.4 is the last version that has xm.
  120: You must make a global choice to use xm or xl, because it affects not
  121: only which command you use, but the command used by rc.d scripts
  122: (specifically xendomains) and which daemons should be run.  The
  123: xentools packages provide xl for 4.2 and up.
  125: In 4.2, you can choose to use xm by simply changing the ctl_command
  126: variable and setting xend=YES in rc.conf.
  128: With xl, virtual devices are configured in parallel, which can cause
  129: problems if they are written assuming serial operation (e.g., updating
  130: firewall rules without explicit locking).  There is now locking for
  131: the provided scripts, which works for normal casses (e.g, file-backed
  132: xbd, where a vnd must be allocated).  But, as of 201612, it has not
  133: been adequately tested for a complex custom setup with a large number
  134: of interfaces.
  136: NetBSD versions
  137: ---------------
  139: The netbsd-7, netbsd-8, and -current branches are all reasonable
  140: choices, with more or less the same considerations for non-Xen use.
  141: NetBSD 8 is recommended as the stable version of the most recent
  142: release for production use.
  144: For developing Xen, netbsd-current may be appropriate.
  146: As of NetBSD 6, a NetBSD domU will support multiple vcpus.  There is
  147: no SMP support for NetBSD as dom0.  (The dom0 itself doesn't really
  148: need SMP for dom0 functions; the lack of support is really a problem
  149: when using a dom0 as a normal computer.)
  151: Note: NetBSD support is called XEN3. However, it does support Xen 4,
  152: because the hypercall interface has remained identical.
  154: Architecture
  155: ------------
  157: Xen itself can run on i386 (Xen < 4.2) or amd64 hardware (all Xen
  158: versions).  (Practically, almost any computer where one would want to
  159: run Xen today supports amd64.)
  161: Xen, the dom0 system, and each domU system can be either i386 or
  162: amd64.  When building a xenkernel package, one obtains an i386 Xen
  163: kernel on an i386 host, and an amd64 Xen kernel on an amd64 host.  If
  164: the Xen kernel is i386, then the dom0 kernel and all domU kernels must
  165: be i386.  With an amd64 Xen kernel, an amd64 dom0 kernel is known to
  166: work, and an i386 dom0 kernel should in theory work.  An amd64
  167: Xen/dom0 is known to support both i386 and amd64 domUs.
  169: i386 dom0 and domU kernels must be PAE (except for an i386 Xen 3.1
  170: kernel, where one can use non-PAE for dom0 and all domUs); PAE kernels
  171: are included in the NetBSD default build.  (Note that emacs (at least)
  172: fails if run on i386 with PAE when built without, and vice versa,
  173: presumably due to bugs in the undump code.)
  175: Because of the above, the standard approach is to use an amd64 Xen
  176: kernel and NetBSD/amd64 for the dom0.  For domUs, NetBSD/i386 (with
  177: the PAE kernel) and NetBSD/amd64 are in widespread use, and there is
  178: little to no Xen-specific reason to prefer one over the other.
  180: Note that to use an i386 dom0 with Xen 4.5 or higher, one must build
  181: (or obtain from pre-built packages) an amd64 Xen kernel and install
  182: that on the system.  (One must also use a PAE i386 kernel, but this is
  183: also required with an i386 Xen kernel.).  Almost no one in the
  184: NetBSD/Xen community does this, and the standard, well-tested,
  185: approach is to use an amd64 dom0.
  187: A [posting on
  188: xen-devel](
  189: explained that PV system call overhead was higher on amd64, and thus
  190: there is some notion that i386 guests are faster.  It goes on to
  191: caution that the total situation is complex and not entirely
  192: understood. On top of that caution, the post is about Linux, not
  193: NetBSD.  TODO: Include link to benchmarks, if someone posts them.
  195: Recommendation
  196: --------------
  198: Therefore, this HOWTO recommends running xenkernel46, xl, the NetBSD 7
  199: stable branch, and therefore to use an amd64 kernel as the dom0.
  200: Either the i386PAE or amd64 version of NetBSD may be used as domUs.
  202: A tentative replacement recommendation is xenkernel48, xl, and NetBSD
  203: 8.
  205: Because bugs are fixed quite often, and because of Xen security
  206: advisories, it is good to stay up to date with NetBSD (tracking a
  207: stable branch), with the Xen kernel (tracking a Xen version via
  208: pkgsrc), and with the Xen tools.  Specifically, NetBSD (-7 and
  209: -current) got an important fix affecting dom0/domU timesharing in
  210: November, 2015, and xentools46 got a fix to enable Ubuntu guests to
  211: boot in December, 2016.
  213: NetBSD as a dom0
  214: ================
  216: NetBSD can be used as a dom0 and works very well.  The following
  217: sections address installation, updating NetBSD, and updating Xen.
  218: Note that it doesn't make sense to talk about installing a dom0 OS
  219: without also installing Xen itself.  We first address installing
  220: NetBSD, which is not yet a dom0, and then adding Xen, pivoting the
  221: NetBSD install to a dom0 install by just changing the kernel and boot
  222: configuration.
  224: For experimenting with Xen, a machine with as little as 1G of RAM and
  225: 100G of disk can work.  For running many domUs in productions, far
  226: more will be needed; e.g. 4-8G and 1T of disk is reasonable for a
  227: half-dozen domUs of 512M and 32G each.  Basically, the RAM and disk
  228: have to be bigger than the sum of the RAM/disk needs of the dom0 and
  229: all the domUs.
  231: In 2018-05, trouble booting a dom0 was reported with 256M of RAM: with
  232: 512M it worked reliably.  This does not make sense, but if you see
  233: "not ELF" after Xen boots, try increasing dom0 RAM.
  235: Styles of dom0 operation
  236: ------------------------
  238: There are two basic ways to use Xen.  The traditional method is for
  239: the dom0 to do absolutely nothing other than providing support to some
  240: number of domUs.  Such a system was probably installed for the sole
  241: purpose of hosting domUs, and sits in a server room on a UPS.
  243: The other way is to put Xen under a normal-usage computer, so that the
  244: dom0 is what the computer would have been without Xen, perhaps a
  245: desktop or laptop.  Then, one can run domUs at will.  Purists will
  246: deride this as less secure than the previous approach, and for a
  247: computer whose purpose is to run domUs, they are right.  But Xen and a
  248: dom0 (without domUs) is not meaningfully less secure than the same
  249: things running without Xen.  One can boot Xen or boot regular NetBSD
  250: alternately with little problems, simply refraining from starting the
  251: Xen daemons when not running Xen.
  253: Note that NetBSD as dom0 does not support multiple CPUs.  This will
  254: limit the performance of the Xen/dom0 workstation approach.  In theory
  255: the only issue is that the "backend drivers" are not yet MPSAFE:
  258: Installation of NetBSD
  259: ----------------------
  261: First,
  262: [install NetBSD/amd64](/guide/inst/)
  263: just as you would if you were not using Xen.
  264: However, the partitioning approach is very important.
  266: If you want to use RAIDframe for the dom0, there are no special issues
  267: for Xen.  Typically one provides RAID storage for the dom0, and the
  268: domU systems are unaware of RAID.  The 2nd-stage loader bootxx_* skips
  269: over a RAID1 header to find /boot from a file system within a RAID
  270: partition; this is no different when booting Xen.
  272: There are 4 styles of providing backing storage for the virtual disks
  273: used by domUs: raw partitions, LVM, file-backed vnd(4), and SAN.
  275: With raw partitions, one has a disklabel (or gpt) partition sized for
  276: each virtual disk to be used by the domU.  (If you are able to predict
  277: how domU usage will evolve, please add an explanation to the HOWTO.
  278: Seriously, needs tend to change over time.)
  280: One can use [lvm(8)](/guide/lvm/) to create logical devices to use
  281: for domU disks.  This is almost as efficient as raw disk partitions
  282: and more flexible.  Hence raw disk partitions should typically not
  283: be used.
  285: One can use files in the dom0 file system, typically created by dd'ing
  286: /dev/zero to create a specific size.  This is somewhat less efficient,
  287: but very convenient, as one can cp the files for backup, or move them
  288: between dom0 hosts.
  290: Finally, in theory one can place the files backing the domU disks in a
  291: SAN.  (This is an invitation for someone who has done this to add a
  292: HOWTO page.)
  294: Installation of Xen
  295: -------------------
  297: In the dom0, install sysutils/xenkernel42 and sysutils/xentools42 from
  298: pkgsrc (or another matching pair).  See [the pkgsrc
  299: documentation]( for help with
  300: pkgsrc.  Ensure that your packages are recent; the HOWTO does not
  301: contemplate old builds.
  304: For Xen 3.1, support for HVM guests is in sysutils/xentool3-hvm.  More
  305: recent versions have HVM support integrated in the main xentools
  306: package.  It is entirely reasonable to run only PV guests.
  308: Next you need to install the selected Xen kernel itself, which is
  309: installed by pkgsrc as "/usr/pkg/xen*-kernel/xen.gz".  Copy it to /.
  310: For debugging, one may copy xen-debug.gz; this is conceptually similar
  311: to DIAGNOSTIC and DEBUG in NetBSD.  xen-debug.gz is basically only
  312: useful with a serial console.  Then, place a NetBSD XEN3_DOM0 kernel
  313: in /, copied from releasedir/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
  314: of a NetBSD build.  If using i386, use
  315: releasedir/i386/binary/kernel/netbsd-XEN3PAE_DOM0.gz.  (If using Xen
  316: 3.1 and i386, you may use XEN3_DOM0 with the non-PAE Xen.  But you
  317: should not use Xen 3.1.)  Both xen and the NetBSD kernel may be (and
  318: typically are) left compressed.
  320: In a dom0, kernfs is mandatory for xend to communicate with the
  321: kernel, so ensure that /kern is in fstab.  (A standard NetBSD install
  322: should already mount /kern.)
  324: Because you already installed NetBSD, you have a working boot setup
  325: with an MBR bootblock, either bootxx_ffsv1 or bootxx_ffsv2 at the
  326: beginning of your root file system, have /boot, and likely also
  327: /boot.cfg.  (If not, fix before continuing!)
  329: Add a line to to /boot.cfg to boot Xen.  See boot.cfg(5) for an
  330: example.  The basic line is
  332:         menu=Xen:load /netbsd-XEN3_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
  334: which specifies that the dom0 should have 512M, leaving the rest to be
  335: allocated for domUs.  To use a serial console, use
  337:         menu=Xen:load /netbsd-XEN3_DOM0.gz;multiboot /xen.gz dom0_mem=512M console=com1 com1=9600,8n1
  339: which will use the first serial port for Xen (which counts starting
  340: from 1, unlike NetBSD which counts starting from 0), forcing
  341: speed/parity.  Because the NetBSD command line lacks a
  342: "console=pc" argument, it will use the default "xencons" console device,
  343: which directs the console I/O through Xen to the same console device Xen
  344: itself uses (in this case, the serial port).
  346: In an attempt to add performance, one can also add
  348:         dom0_max_vcpus=1 dom0_vcpus_pin
  350: to force only one vcpu to be provided (since NetBSD dom0 can't use
  351: more) and to pin that vcpu to a physical CPU.  TODO: benchmark this.
  353: Xen has [many boot
  354: options](,
  355: and other than dom0 memory and max_vcpus, they are generally not
  356: necessary.
  358: As with non-Xen systems, you should have a line to boot /netbsd (a
  359: kernel that works without Xen).  Consider a line to boot /netbsd.ok (a
  360: fallback version of the non-Xen kernel, updated manually when you are
  361: sure /netbsd is ok).  Consider also a line to boot fallback versions
  362: of Xen and the dom0 kernel, but note that non-Xen NetBSD can be used
  363: to resolve Xen booting issues.
  365: Probably you want a default=N line to choose Xen in the absence of
  366: intervention.
  368: Now, reboot so that you are running a DOM0 kernel under Xen, rather
  369: than GENERIC without Xen.
  371: Using grub (historic)
  372: ---------------------
  374: Before NetBSD's native bootloader could support Xen, the use of
  375: grub was recommended.  If necessary, see the
  376: [old grub information](/ports/xen/howto-grub).
  378: The [HowTo on Installing into
  379: RAID-1](
  380: explains how to set up booting a dom0 with Xen using grub with
  381: NetBSD's RAIDframe.  (This is obsolete with the use of NetBSD's native
  382: boot.  Now, just create a system with RAID-1, and alter /boot.cfg as
  383: described above.)
  385: Configuring Xen
  386: ---------------
  388: Xen logs will be in /var/log/xen.
  390: Now, you have a system that will boot Xen and the dom0 kernel, but not
  391: do anything else special.  Make sure that you have rebooted into Xen.
  392: There will be no domUs, and none can be started because you still have
  393: to configure the dom0 daemons.
  395: The daemons which should be run vary with Xen version and with whether
  396: one is using xm or xl.  The Xen 3.1, 3.3 and 4.1 packages use xm.  Xen
  397: 4.2 and up packages use xl.  To use xm with 4.2, edit xendomains to
  398: use xm instead.
  400: For 3.1 and 3.3, you should enable xend and xenbackendd:
  402:         xend=YES
  403:         xenbackendd=YES
  405: For 4.1 and up, you should enable xencommons.  Not enabling xencommons
  406: will result in a hang; it is necessary to hit ^C on the console to let
  407: the machine finish booting.  If you are using xm (default in 4.1, or
  408: if you changed xendomains in 4.2), you should also enable xend:
  410:         xend=YES # only if using xm, and only installed <= 4.2
  411:         xencommons=YES
  413: TODO: Recommend for/against xen-watchdog.
  415: After you have configured the daemons and either started them (in the
  416: order given) or rebooted, use xm or xl to inspect Xen's boot messages,
  417: available resources, and running domains.  An example with xl follows:
  419:         # xl dmesg
  420: 	[xen's boot info]
  421:         # xl info
  422: 	[available memory, etc.]
  423:         # xl list
  424:         Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  425:         Domain-0           0       64    0  r----     58.1
  427: ### Issues with xencommons
  429: xencommons starts xenstored, which stores data on behalf of dom0 and
  430: domUs.  It does not currently work to stop and start xenstored.
  431: Certainly all domUs should be shutdown first, following the sort order
  432: of the rc.d scripts.  However, the dom0 sets up state with xenstored,
  433: and is not notified when xenstored exits, leading to not recreating
  434: the state when the new xenstored starts.  Until there's a mechanism to
  435: make this work, one should not expect to be able to restart xenstored
  436: (and thus xencommons).  There is currently no reason to expect that
  437: this will get fixed any time soon.
  439: ### No-longer needed advice about devices
  441: The installation of NetBSD should already have created devices for xen
  442: (xencons, xenevt, xsd_kva), but if they are not present, create them:
  444:         cd /dev && sh MAKEDEV xen
  446: anita (for testing NetBSD)
  447: --------------------------
  449: With the setup so far (assuming 4.2/xl), one should be able to run
  450: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
  451: root, because anita must create a domU):
  453:         anita --vmm=xl test file:///usr/obj/i386/
  455: Alternatively, one can use --vmm=xm to use xm-based domU creation
  456: instead (and must, on Xen <= 4.1).   TODO: confirm that anita xl really works.
  458: Xen-specific NetBSD issues
  459: --------------------------
  461: There are (at least) two additional things different about NetBSD as a
  462: dom0 kernel compared to hardware.
  464: One is that the module ABI is different because some of the #defines
  465: change, so one must build modules for Xen.  As of netbsd-7, the build
  466: system does this automatically.  TODO: check this.  (Before building
  467: Xen modules was added, it was awkward to use modules to the point
  468: where it was considered that it did not work.)
  470: The other difference is that XEN3_DOM0 does not have exactly the same
  471: options as GENERIC.  While it is debatable whether or not this is a
  472: bug, users should be aware of this and can simply add missing config
  473: items if desired.
  475: Updating NetBSD in a dom0
  476: -------------------------
  478: This is just like updating NetBSD on bare hardware, assuming the new
  479: version supports the version of Xen you are running.  Generally, one
  480: replaces the kernel and reboots, and then overlays userland binaries
  481: and adjusts /etc.
  483: Note that one must update both the non-Xen kernel typically used for
  484: rescue purposes and the DOM0 kernel used with Xen.
  486: Converting from grub to /boot
  487: -----------------------------
  489: These instructions were [TODO: will be] used to convert a system from
  490: grub to /boot.  The system was originally installed in February of
  491: 2006 with a RAID1 setup and grub to boot Xen 2, and has been updated
  492: over time.  Before these commands, it was running NetBSD 6 i386, Xen
  493: 4.1 and grub, much like the message linked earlier in the grub
  494: section.
  496:         # Install MBR bootblocks on both disks. 
  497:         fdisk -i /dev/rwd0d
  498:         fdisk -i /dev/rwd1d
  499:         # Install NetBSD primary boot loader (/ is FFSv1) into RAID1 components.
  500:         installboot -v /dev/rwd0d /usr/mdec/bootxx_ffsv1
  501:         installboot -v /dev/rwd1d /usr/mdec/bootxx_ffsv1
  502:         # Install secondary boot loader
  503:         cp -p /usr/mdec/boot /
  504:         # Create boot.cfg following earlier guidance:
  505:         menu=Xen:load /netbsd-XEN3PAE_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
  506:         menu=Xen.ok:load /netbsd-XEN3PAE_DOM0.ok.gz console=pc;multiboot /xen.ok.gz dom0_mem=512M
  507:         menu=GENERIC:boot
  508:         menu=GENERIC single-user:boot -s
  509:         menu=GENERIC.ok:boot netbsd.ok
  510:         menu=GENERIC.ok single-user:boot netbsd.ok -s
  511:         menu=Drop to boot prompt:prompt
  512:         default=1
  513:         timeout=30
  515: TODO: actually do this and fix it if necessary.
  517: Upgrading Xen versions
  518: ---------------------
  520: Minor version upgrades are trivial.  Just rebuild/replace the
  521: xenkernel version and copy the new xen.gz to / (where /boot.cfg
  522: references it), and reboot.
  524: Major version upgrades are conceptually not difficult, but can run
  525: into all the issues found when installing Xen.  Assuming migration
  526: from 4.1 to 4.2, remove the xenkernel41 and xentools41 packages and
  527: install the xenkernel42 and xentools42 packages.  Copy the 4.2 xen.gz
  528: to /.
  530: Ensure that the contents of /etc/rc.d/xen* are correct.  Specifically,
  531: they must match the package you just installed and not be left over
  532: from some previous installation.
  534: Enable the correct set of daemons; see the configuring section above.
  535: (Upgrading from 3.x to 4.x without doing this will result in a hang.)
  537: Ensure that the domU config files are valid for the new version.
  538: Specifically, for 4.x remove autorestart=True, and ensure that disks
  539: are specified with numbers as the second argument, as the examples
  540: above show, and not NetBSD device names.
  542: Hardware known to work
  543: ----------------------
  545: Arguably, this section is misplaced, and there should be a page of
  546: hardware that runs NetBSD/amd64 well, with the mostly-well-founded
  547: assumption that NetBSD/xen runs fine on any modern hardware that
  548: NetBSD/amd64 runs well on.  Until then, we give motherboard/CPU (and
  549: sometimes RAM) pairs/triples to aid those choosing a motherboard.
  550: Note that Xen systems usually do not run X, so a listing here does not
  551: imply that X works at all.
  553:         Supermicro X9SRL-F, Xeon E5-1650 v2, 96 GiB ECC
  554:         Supermicro ??, Atom C2758 (8 core), 32 GiB ECC
  555:         ASUS M5A78L-M/USB3 AM3+ microATX, AMD Piledriver X8 4000MHz, 16 GiB ECC
  557: Older hardware:
  559:         Intel D915GEV, Pentium4 CPU 3.40GHz, 4GB 533MHz Synchronous DDR2
  560:         INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E6850  @ 3.00GHz"
  561:         INTEL DG33FB, "Intel(R) Core(TM)2 Duo CPU     E8400  @ 3.00GHz"
  563: Running Xen under qemu
  564: ----------------------
  566: The astute reader will note that this section is somewhat twisted.
  567: However, it can be useful to run Xen under qemu either because the
  568: version of NetBSD as a dom0 does not run on the hardware in use, or to
  569: generate automated test cases involving Xen.
  571: In 2015-01, the following combination was reported to mostly work:
  573:         host OS: NetBSD/amd64 6.1.4
  574:         qemu: 2.2.0 from pkgsrc
  575:         Xen kernel: xenkernel42-4.2.5nb1 from pkgsrc
  576:         dom0 kernel: NetBSD/amd64 6.1.5
  577:         Xen tools: xentools42-4.2.5 from pkgsrc
  579: See [PR 47720]( for a problem with dom0
  580: shutdown.
  582: Unprivileged domains (domU)
  583: ===========================
  585: This section describes general concepts about domUs.  It does not
  586: address specific domU operating systems or how to install them.  The
  587: config files for domUs are typically in /usr/pkg/etc/xen, and are
  588: typically named so that the file name, domU name and the domU's host
  589: name match.
  591: The domU is provided with CPU and memory by Xen, configured by the
  592: dom0.  The domU is provided with disk and network by the dom0,
  593: mediated by Xen, and configured in the dom0.
  595: Entropy in domUs can be an issue; physical disks and network are on
  596: the dom0.  NetBSD's /dev/random system works, but is often challenged.
  598: Config files
  599: ------------
  601: There is no good order to present config files and the concepts
  602: surrounding what is being configured.  We first show an example config
  603: file, and then in the various sections give details.
  605: See (at least in xentools41) /usr/pkg/share/examples/xen/xmexample*,
  606: for a large number of well-commented examples, mostly for running
  607: GNU/Linux.
  609: The following is an example minimal domain configuration file
  610: "/usr/pkg/etc/xen/foo".  It is (with only a name change) an actual
  611: known working config file on Xen 4.1 (NetBSD 5 amd64 dom0 and NetBSD 5
  612: i386 domU).  The domU serves as a network file server.
  614:         # -*- mode: python; -*-
  616:         kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
  617:         memory = 1024
  618:         vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
  619:         disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
  620:                  'file:/n0/xen/foo-wd1,0x1,w' ]
  622: The domain will have the same name as the file.  The kernel has the
  623: host/domU name in it, so that on the dom0 one can update the various
  624: domUs independently.  The vif line causes an interface to be provided,
  625: with a specific mac address (do not reuse MAC addresses!), in bridge
  626: mode.  Two disks are provided, and they are both writable; the bits
  627: are stored in files and Xen attaches them to a vnd(4) device in the
  628: dom0 on domain creation.  The system treats xbd0 as the boot device
  629: without needing explicit configuration.
  631: By default xm looks for domain config files in /usr/pkg/etc/xen.  Note
  632: that "xm create" takes the name of a config file, while other commands
  633: take the name of a domain.  To create the domain, connect to the
  634: console, create the domain while attaching the console, shutdown the
  635: domain, and see if it has finished stopping, do (or xl with Xen >=
  636: 4.2):
  638:         xm create foo
  639:         xm console foo
  640:         xm create -c foo
  641:         xm shutdown foo
  642:         xm list
  644: Typing ^] will exit the console session.  Shutting down a domain is
  645: equivalent to pushing the power button; a NetBSD domU will receive a
  646: power-press event and do a clean shutdown.  Shutting down the dom0
  647: will trigger controlled shutdowns of all configured domUs.
  649: domU kernels
  650: ------------
  652: On a physical computer, the BIOS reads sector 0, and a chain of boot
  653: loaders finds and loads a kernel.  Normally this comes from the root
  654: file system.  With Xen domUs, the process is totally different.  The
  655: normal path is for the domU kernel to be a file in the dom0's
  656: file system.  At the request of the dom0, Xen loads that kernel into a
  657: new domU instance and starts execution.  While domU kernels can be
  658: anyplace, reasonable places to store domU kernels on the dom0 are in /
  659: (so they are near the dom0 kernel), in /usr/pkg/etc/xen (near the
  660: config files), or in /u0/xen (where the vdisks are).
  662: Note that loading the domU kernel from the dom0 implies that boot
  663: blocks, /boot, /boot.cfg, and so on are all ignored in the domU.
  664: See the VPS section near the end for discussion of alternate ways to
  665: obtain domU kernels.
  667: CPU and memory
  668: --------------
  670: A domain is provided with some number of vcpus, less than the number
  671: of CPUs seen by the hypervisor.  (For a dom0, this is controlled by
  672: the boot argument "dom0_max_vcpus=1".)  For a domU, it is controlled
  673: from the config file by the "vcpus = N" directive.
  675: A domain is provided with memory; this is controlled in the config
  676: file by "memory = N" (in megabytes).  In the straightforward case, the
  677: sum of the the memory allocated to the dom0 and all domUs must be less
  678: than the available memory.
  680: Xen also provides a "balloon" driver, which can be used to let domains
  681: use more memory temporarily.  TODO: Explain better, and explain how
  682: well it works with NetBSD.
  684: Virtual disks
  685: -------------
  687: With the file/vnd style, typically one creates a directory,
  688: e.g. /u0/xen, on a disk large enough to hold virtual disks for all
  689: domUs.  Then, for each domU disk, one writes zeros to a file that then
  690: serves to hold the virtual disk's bits; a suggested name is foo-xbd0
  691: for the first virtual disk for the domU called foo.  Writing zeros to
  692: the file serves two purposes.  One is that preallocating the contents
  693: improves performance.  The other is that vnd on sparse files has
  694: failed to work.  TODO: give working/notworking NetBSD versions for
  695: sparse vnd and gnats reference.  Note that the use of file/vnd for Xen
  696: is not really different than creating a file-backed virtual disk for
  697: some other purpose, except that xentools handles the vnconfig
  698: commands.  To create an empty 4G virtual disk, simply do
  700:         dd if=/dev/zero of=foo-xbd0 bs=1m count=4096
  702: Do not use qemu-img-xen, because this will create sparse file.  There
  703: have been recent (2015) reports of sparse vnd(4) devices causing
  704: lockups, but there is apparently no PR.
  706: With the lvm style, one creates logical devices.  They are then used
  707: similarly to vnds.  TODO: Add an example with lvm.
  709: In domU config files, the disks are defined as a sequence of 3-tuples.
  710: The first element is "method:/path/to/disk".  Common methods are
  711: "file:" for file-backed vnd. and "phy:" for something that is already
  712: a (TODO: character or block) device.
  714: The second element is an artifact of how virtual disks are passed to
  715: Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
  716: are given a device name to associate with the disk, and values like
  717: "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
  718: as xbd0, the second as xbd1, and so on.  However, xm/xl demand a
  719: second argument.  The name given is converted to a major/minor by
  720: calling stat(2) on the name in /dev and this is passed to the domU.
  721: In the general case, the dom0 and domU can be different operating
  722: systems, and it is an unwarranted assumption that they have consistent
  723: numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
  724: as both dom0 and domU, using values of 0x0 for the first disk and 0x1
  725: for the second works fine and avoids this issue.  For a GNU/Linux
  726: guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
  727: /dev/hda1.
  729: The third element is "w" for writable disks, and "r" for read-only
  730: disks.
  732: Note that NetBSD by default creates only vnd[0123].  If you need more
  733: than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
  734: dom0.
  736: Note that NetBSD by default creates only xbd[0123].  If you need more
  737: virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
  739: Virtual Networking
  740: ------------------
  742: Xen provides virtual Ethernets, each of which connects the dom0 and a
  743: domU.  For each virtual network, there is an interface "xvifN.M" in
  744: the dom0, and in domU index N, a matching interface xennetM (NetBSD
  745: name).  The interfaces behave as if there is an Ethernet with two
  746: adapters connected.  From this primitive, one can construct various
  747: configurations.  We focus on two common and useful cases for which
  748: there are existing scripts: bridging and NAT.
  750: With bridging (in the example above), the domU perceives itself to be
  751: on the same network as the dom0.  For server virtualization, this is
  752: usually best.  Bridging is accomplished by creating a bridge(4) device
  753: and adding the dom0's physical interface and the various xvifN.0
  754: interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
  755: config file.  The bridge must be set up already in the dom0; an
  756: example /etc/ifconfig.bridge0 is:
  758:         create
  759:         up
  760:         !brconfig bridge0 add wm0
  762: With NAT, the domU perceives itself to be behind a NAT running on the
  763: dom0.  This is often appropriate when running Xen on a workstation.
  764: TODO: NAT appears to be configured by "vif = [ '' ]".
  766: The MAC address specified is the one used for the interface in the new
  767: domain.  The interface in dom0 will use this address XOR'd with
  768: 00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
  770: Sizing domains
  771: --------------
  773: Modern x86 hardware has vast amounts of resources.  However, many
  774: virtual servers can function just fine on far less.  A system with
  775: 512M of RAM and a 4G disk can be a reasonable choice.  Note that it is
  776: far easier to adjust virtual resources than physical ones.  For
  777: memory, it's just a config file edit and a reboot.  For disk, one can
  778: create a new file and vnconfig it (or lvm), and then dump/restore,
  779: just like updating physical disks, but without having to be there and
  780: without those pesky connectors.
  782: Starting domains automatically
  783: ------------------------------
  785: To start domains foo at bar at boot and shut them down cleanly on dom0
  786: shutdown, in rc.conf add:
  788:         xendomains="foo bar"
  790: Note that earlier versions of the xentools41 xendomains rc.d script
  791: used xl, when one should use xm with 4.1.
  793: Creating specific unprivileged domains (domU)
  794: =============================================
  796: Creating domUs is almost entirely independent of operating system.  We
  797: have already presented the basics of config files.  Note that you must
  798: have already completed the dom0 setup so that "xl list" (or "xm list")
  799: works.
  801: Creating an unprivileged NetBSD domain (domU)
  802: ---------------------------------------------
  804: See the earlier config file, and adjust memory.  Decide on how much
  805: storage you will provide, and prepare it (file or lvm).
  807: While the kernel will be obtained from the dom0 file system, the same
  808: file should be present in the domU as /netbsd so that tools like
  809: savecore(8) can work.   (This is helpful but not necessary.)
  811: The kernel must be specifically for Xen and for use as a domU.  The
  812: i386 and amd64 provide the following kernels:
  814:         i386 XEN3_DOMU
  815:         i386 XEN3PAE_DOMU
  816:         amd64 XEN3_DOMU
  818: Unless using Xen 3.1 (and you shouldn't) with i386-mode Xen, you must
  819: use the PAE version of the i386 kernel.
  821: This will boot NetBSD, but this is not that useful if the disk is
  822: empty.  One approach is to unpack sets onto the disk outside of xen
  823: (by mounting it, just as you would prepare a physical disk for a
  824: system you can't run the installer on).
  826: A second approach is to run an INSTALL kernel, which has a miniroot
  827: and can load sets from the network.  To do this, copy the INSTALL
  828: kernel to / and change the kernel line in the config file to:
  830:         kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
  832: Then, start the domain as "xl create -c configname".
  834: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
  835: line should be used in the config file.
  837:     disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
  839: After booting the domain, the option to install via CDROM may be
  840: selected.  The CDROM device should be changed to `xbd1d`.
  842: Once done installing, "halt -p" the new domain (don't reboot or halt,
  843: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
  844: config file), switch the config file back to the XEN3_DOMU kernel,
  845: and start the new domain again. Now it should be able to use "root on
  846: xbd0a" and you should have a, functional NetBSD domU.
  848: TODO: check if this is still accurate.
  849: When the new domain is booting you'll see some warnings about *wscons*
  850: and the pseudo-terminals. These can be fixed by editing the files
  851: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
  852: `/etc/ttys`, except *console*, like this:
  854:     console "/usr/libexec/getty Pc"         vt100   on secure
  855:     ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
  856:     ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
  857:     ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
  858:     ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
  860: Finally, all screens must be commented out from `/etc/wscons.conf`.
  862: It is also desirable to add
  864:         powerd=YES
  866: in rc.conf. This way, the domain will be properly shut down if
  867: `xm shutdown -R` or `xm shutdown -H` is used on the dom0.
  869: It is not strictly necessary to have a kernel (as /netbsd) in the domU
  870: file system.  However, various programs (e.g. netstat) will use that
  871: kernel to look up symbols to read from kernel virtual memory.  If
  872: /netbsd is not the running kernel, those lookups will fail.  (This is
  873: not really a Xen-specific issue, but because the domU kernel is
  874: obtained from the dom0, it is far more likely to be out of sync or
  875: missing with Xen.)
  877: Creating an unprivileged Linux domain (domU)
  878: --------------------------------------------
  880: Creating unprivileged Linux domains isn't much different from
  881: unprivileged NetBSD domains, but there are some details to know.
  883: First, the second parameter passed to the disk declaration (the '0x1' in
  884: the example below)
  886:     disk = [ 'phy:/dev/wd0e,0x1,w' ]
  888: does matter to Linux. It wants a Linux device number here (e.g. 0x300
  889: for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
  890: So, hda1 which has major 3 and minor 1 on a Linux system will have
  891: device number 0x301.  Alternatively, devices names can be used (hda,
  892: hdb, ...)  as xentools has a table to map these names to devices
  893: numbers.  To export a partition to a Linux guest we can use:
  895:         disk = [ 'phy:/dev/wd0e,0x300,w' ]
  896:         root = "/dev/hda1 ro"
  898: and it will appear as /dev/hda on the Linux system, and be used as root
  899: partition.
  901: To install the Linux system on the partition to be exported to the
  902: guest domain, the following method can be used: install
  903: sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
  904: that will be the root partition of your Linux domain, and mount it.
  905: Then copy the files from a working Linux system, make adjustments in
  906: `/etc` (fstab, network config).  It should also be possible to extract
  907: binary packages such as .rpm or .deb directly to the mounted partition
  908: using the appropriate tool, possibly running under NetBSD's Linux
  909: emulation.  Once the file system has been populated, umount it.  If
  910: desirable, the file system can be converted to ext3 using tune2fs -j.
  911: It should now be possible to boot the Linux guest domain, using one of
  912: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
  914: To get the Linux console right, you need to add:
  916:     extra = "xencons=tty1"
  918: to your configuration since not all Linux distributions auto-attach a
  919: tty to the xen console.
  921: Creating an unprivileged Solaris domain (domU)
  922: ----------------------------------------------
  924: See possibly outdated
  925: [Solaris domU instructions](/ports/xen/howto-solaris/).
  928: PCI passthrough: Using PCI devices in guest domains
  929: ---------------------------------------------------
  931: The dom0 can give other domains access to selected PCI
  932: devices. This can allow, for example, a non-privileged domain to have
  933: access to a physical network interface or disk controller.  However,
  934: keep in mind that giving a domain access to a PCI device most likely
  935: will give the domain read/write access to the whole physical memory,
  936: as PCs don't have an IOMMU to restrict memory access to DMA-capable
  937: device.  Also, it's not possible to export ISA devices to non-dom0
  938: domains, which means that the primary VGA adapter can't be exported.
  939: A guest domain trying to access the VGA registers will panic.
  941: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
  942: not been ported to later versions at this time.
  944: For a PCI device to be exported to a domU, is has to be attached to
  945: the "pciback" driver in dom0.  Devices passed to the dom0 via the
  946: pciback.hide boot parameter will attach to "pciback" instead of the
  947: usual driver.  The list of devices is specified as "(bus:dev.func)",
  948: where bus and dev are 2-digit hexadecimal numbers, and func a
  949: single-digit number:
  951:         pciback.hide=(00:0a.0)(00:06.0)
  953: pciback devices should show up in the dom0's boot messages, and the
  954: devices should be listed in the `/kern/xen/pci` directory.
  956: PCI devices to be exported to a domU are listed in the "pci" array of
  957: the domU's config file, with the format "0000:bus:dev.func".
  959:         pci = [ '0000:00:06.0', '0000:00:0a.0' ]
  961: In the domU an "xpci" device will show up, to which one or more pci
  962: buses will attach.  Then the PCI drivers will attach to PCI buses as
  963: usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
  964: or any PCI drivers built in by default; you have to build your own
  965: kernel to use PCI devices in a domU.  Here's a kernel config example;
  966: note that only the "xpci" lines are unusual.
  968:         include         "arch/i386/conf/XEN3_DOMU"
  970:         # Add support for PCI buses to the XEN3_DOMU kernel
  971:         xpci* at xenbus ?
  972:         pci* at xpci ?
  974:         # PCI USB controllers
  975:         uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
  977:         # USB bus support
  978:         usb*    at uhci?
  980:         # USB Hubs
  981:         uhub*   at usb?
  982:         uhub*   at uhub? port ? configuration ? interface ?
  984:         # USB Mass Storage
  985:         umass*  at uhub? port ? configuration ? interface ?
  986:         wd*     at umass?
  987:         # SCSI controllers
  988:         ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
  990:         # SCSI bus support (for both ahc and umass)
  991:         scsibus* at scsi?
  993:         # SCSI devices
  994:         sd*     at scsibus? target ? lun ?      # SCSI disk drives
  995:         cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
  998: NetBSD as a domU in a VPS
  999: =========================
 1001: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
 1002: hardware.  This section explains how to deal with Xen in a domU as a
 1003: virtual private server where you do not control or have access to the
 1004: dom0.  This is not intended to be an exhaustive list of VPS providers;
 1005: only a few are mentioned that specifically support NetBSD.
 1007: VPS operators provide varying degrees of access and mechanisms for
 1008: configuration.  The big issue is usually how one controls which kernel
 1009: is booted, because the kernel is nominally in the dom0 file system (to
 1010: which VPS users do not normally have access).  A second issue is how
 1011: to install NetBSD.
 1012: A VPS user may want to compile a kernel for security updates, to run
 1013: npf, run IPsec, or any other reason why someone would want to change
 1014: their kernel.
 1016: One approach is to have an administrative interface to upload a kernel,
 1017: or to select from a prepopulated list.  Other approaches are pygrub
 1018: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
 1019: kernel from the domU file system.  This is closer to a regular physical
 1020: computer, where someone who controls a machine can replace the kernel.
 1022: A second issue is multiple CPUs.  With NetBSD 6, domUs support
 1023: multiple vcpus, and it is typical for VPS providers to enable multiple
 1024: CPUs for NetBSD domUs.
 1026: pygrub
 1027: -------
 1029: pygrub runs in the dom0 and looks into the domU file system.  This
 1030: implies that the domU must have a kernel in a file system in a format
 1031: known to pygrub.  As of 2014, pygrub seems to be of mostly historical
 1032: interest.
 1034: pvgrub
 1035: ------
 1037: pvgrub is a version of grub that uses PV operations instead of BIOS
 1038: calls.  It is booted from the dom0 as the domU kernel, and then reads
 1039: /grub/menu.lst and loads a kernel from the domU file system.
 1041: [Panix]( lets users use pvgrub.  Panix reports
 1042: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
 1043: (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
 1044: page](, which describes only
 1045: Linux but should be updated to cover NetBSD :-).
 1047: []( also lets users with pvgrub to boot
 1048: their own kernel.  See then [ NetBSD
 1049: HOWTO](
 1050: (which is in need of updating).
 1052: It appears that [grub's FFS
 1053: code](
 1054: does not support all aspects of modern FFS, but there are also reports
 1055: that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
 1056: partition for the kernel with the intent that grub can understand it,
 1057: which leads to /netbsd not being the actual kernel.  One must remember
 1058: to update the special boot partition.
 1060: Amazon
 1061: ------
 1063: See the [Amazon EC2 page](/amazon_ec2/).
 1065: TODO items for improving NetBSD/xen
 1066: ===================================
 1068: * Make the NetBSD dom0 kernel work with SMP.
 1069: * Test the Xen 4.5 packages adequately to be able to recommend them as
 1070:   the standard approach.
 1071: * Get PCI passthrough working on Xen 4.5
 1072: * Get pvgrub into pkgsrc, either via xentools or separately.
 1073: * grub
 1074:   * Check/add support to pkgsrc grub2 for UFS2 and arbitrary
 1075:     fragsize/blocksize (UFS2 support may be present; the point is to
 1076:     make it so that with any UFS1/UFS2 file system setup that works
 1077:     with NetBSD grub will also work).
 1078:     See [pkg/40258](
 1079:   * Push patches upstream.
 1080:   * Get UFS2 patches into pvgrub.
 1081: * Add support for PV ops to a version of /boot, and make it usable as
 1082:   a kernel in Xen, similar to pvgrub.
 1084: Random pointers
 1085: ===============
 1087: This section contains links from elsewhere not yet integrated into the
 1088: HOWTO, and other guides.
 1090: *
 1091: *
 1092: *

CVSweb for NetBSD wikisrc <> software: FreeBSD-CVSweb