File:  [NetBSD Developer Wiki] / wikisrc / ports / xen / howto.mdwn
Revision 1.202: download - view: text, annotated - select for diffs
Tue Mar 9 13:57:53 2021 UTC (2 months ago) by gson
Branches: MAIN
CVS tags: HEAD
There is no boot.cfg keyword "losad".  Presumably "load" was intended.

    1: [[!meta title="Xen HowTo and Status"]]
    2: 
    3: Xen is a Type 1 hypervisor which supports running multiple guest operating
    4: systems on a single physical machine. One uses the Xen kernel to control the
    5: CPU, memory and console, a dom0 operating system which mediates access to
    6: other hardware (e.g., disks, network, USB), and one or more domU operating
    7: systems which operate in an unprivileged virtualized environment. IO requests
    8: from the domU systems are forwarded by the Xen hypervisor to the dom0 to be
    9: fulfilled.
   10: 
   11: This HOWTO presumes a basic familiarity with the Xen system
   12: architecture, with installing NetBSD on amd64 hardware, and with
   13: installing software from pkgsrc.  See also the [Xen
   14: website](http://www.xenproject.org/).
   15: 
   16: This document also provides status on what Xen things work on NetBSD
   17: (upstream documentation might say something works if it works on some
   18: particular Linux system).
   19: 
   20: [[!toc]]
   21: 
   22: # Overview
   23: 
   24: The basic concept of Xen is that the hypervisor (xenkernel) runs on
   25: the hardware, and runs a privileged domain ("dom0") that can access
   26: disks/networking/etc.  One then runs additonal unprivileged domains
   27: (each a "domU"), presumably to do something useful.
   28: 
   29: This HOWTO addresses how to run a NetBSD dom0 (and hence also build
   30: xen itself).  It also addresses how to run domUs in that environment,
   31: and how to deal with having a domU in a Xen environment run by someone
   32: else and/or not running NetBSD.
   33: 
   34: There are many choices one can make; the HOWTO recommends the standard
   35: approach and limits discussion of alternatives in many cases.
   36: 
   37: ## Guest Styles
   38: 
   39: Xen supports different styles of guests.
   40: 
   41: [[!table data="""
   42: Style of guest	|Supported by NetBSD
   43: PV		|Yes (dom0, domU)
   44: HVM		|Yes (domU)
   45: PVHVM		|current-only (domU)
   46: PVH		|current-only (domU, dom0 not yet)
   47: """]]
   48: 
   49: In Para-Virtualized (PV) mode, the guest OS does not attempt to access
   50: hardware directly, but instead makes hypercalls to the hypervisor; PV
   51: guests must be specifically coded for Xen.
   52: See [PV](https://wiki.xen.org/wiki/Paravirtualization_(PV\)).
   53: 
   54: In HVM mode, no guest modification is required; however, hardware
   55: support is required, such as VT-x on Intel CPUs and SVM on AMD CPUs.
   56: The dom0 runs qemu to emulate hardware.  It is therefore non-sensical
   57: to have an HVM dom0.
   58: 
   59: In PVHVM mode, the guest runs as HVM, but additionally can use PV
   60: drivers for efficiency.  Therefore it is non-sensical for to have a
   61: PVHVM dom0.  See [PV on HVM](https://wiki.xen.org/wiki/PV_on_HVM).
   62: 
   63: There have been two PVH modes: original PVH and PVHv2.  Original PVH
   64: was based on PV mode and is no longer relevant at all.  PVHv2 is
   65: basically lightweight HVM with PV drivers.  A critical feature of it
   66: is that qemu is not needed; the hypervisor can do the emulation that
   67: is required.  Thus, a dom0 can be PVHv2.
   68: The source code uses PVH and config files use pvh; this refers to PVHv2.
   69: See [PVH(v2)](https://wiki.xenproject.org/wiki/PVH_(v2\)_Domu).
   70: 
   71: At system boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   72: The dom0 can start one or more domUs.  (Booting is explained in detail
   73: in the dom0 section.)
   74: 
   75: ## CPU Architecture
   76: 
   77: Xen runs on x86_64 hardware (the NetBSD amd64 port).
   78: 
   79: There is a concept of Xen running on ARM, but there are no reports of this working with NetBSD.
   80: 
   81: The dom0 system should be amd64.  (Instructions for i386PAE dom0 have been removed from the HOWTO.)
   82: 
   83: The domU can be i386PAE or amd64.
   84: i386PAE at one point was considered as [faster](https://lists.xen.org/archives/html/xen-devel/2012-07/msg00085.html) than amd64.
   85: 
   86: ## Xen Versions
   87: 
   88: In NetBSD, Xen is provided in pkgsrc, via matching pairs of packages
   89: xenkernel and xentools.  We will refer only to the kernel versions,
   90: but note that both packages must be installed together and must have
   91: matching versions.
   92: 
   93: Versions available in pkgsrc:
   94: 
   95: [[!table data="""
   96: Xen Version	|Package Name	|Xen CPU Support	|EOL'ed By Upstream
   97: 4.11		|xenkernel411	|x86_64			|No
   98: 4.13		|xenkernel413	|x86_64			|No
   99: """]]
  100: 
  101: See also the [Xen Security Advisory page](http://xenbits.xen.org/xsa/).
  102: 
  103: Older Xen had a python-based management tool called xm, now replaced
  104: by xl.
  105: 
  106: ## NetBSD versions
  107: 
  108: Xen has been supported in NetBSD for a long time, at least since 2005.
  109: Initially Xen was PV only.
  110: 
  111: NetBSD 8 and up support PV and HVM modes.
  112: 
  113: Support for PVHVM and PVH is available only in NetBSD-current.
  114: 
  115: NetBSD up to and including NetBSD 9 as a dom0 does not run SMP,
  116: because some drivers are not yet safe for this.  NetBSD-current
  117: supports SMP in dom0.
  118: 
  119: NetBSD, when run as a domU, can and does typically run SMP.
  120: 
  121: Note that while Xen 4.13 is current, the kernel support is still
  122: called XEN3, because the hypercall interface has not changed
  123: significantly.
  124: 
  125: # Creating a NetBSD dom0
  126: 
  127: In order to install a NetBSD as a dom0, one first installs a normal
  128: NetBSD system, and then pivot the install to a dom0 install by
  129: changing the kernel and boot configuration.
  130: 
  131: In 2018-05, trouble booting a dom0 was reported with 256M of RAM: with
  132: 512M it worked reliably.  This does not make sense, but if you see
  133: "not ELF" after Xen boots, try increasing dom0 RAM.
  134: 
  135: ## Installation of NetBSD
  136: 
  137: [Install NetBSD/amd64](/guide/inst/) just as you would if you were not
  138: using Xen.  Therefore, use the most recent release, or a build from
  139: the most recent stable branch.  Alternatively, use -current, being
  140: mindful of all the usual caveats of lower stability of current, and
  141: likely a bit more so.
  142: 
  143: ## Installation of Xen
  144: 
  145: ### Building Xen
  146: 
  147: Use the most recent version of Xen in pkgsrc, unless the DESCR says that it is not suitable.
  148: Therefore, choose 4.13.
  149: In the dom0, install xenkernel413 and xentools413 from pkgsrc.
  150: 
  151: Once this is done, copy the Xen kernel from where pkgsrc puts it to
  152: where the boot process will be able to find it:
  153: 
  154: [[!template id=programlisting text="""
  155: # cp -p /usr/pkg/xen413-kernel/xen.gz /
  156: """]]
  157: 
  158: Then, place a NetBSD XEN3_DOM0 kernel in the `/` directory. Such
  159: kernel can either be taken from a local release build.sh run, compiled
  160: manually, or downloaded from the NetBSD FTP, for example at:
  161: 
  162: [[!template id=programlisting text="""
  163: ftp.netbsd.org/pub/NetBSD/NetBSD-9.1/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
  164: """]]
  165: 
  166: ### Configuring booting
  167: 
  168: Read boot.cfg(8) carefully.  Add lines to /boot.cfg to boot Xen,
  169: adjusting for your root filesystem:
  170: 
  171: [[!template id=filecontent name="/boot.cfg" text="""
  172: menu=Xen:load /netbsd-XEN3_DOM0.gz root=wd0a console=pc;multiboot /xen.gz dom0_mem=512M
  173: menu=Xen single user:load /netbsd-XEN3_DOM0.gz root=wd0a console=pc -s;multiboot /xen.gz dom0_mem=512M
  174: """]]
  175: 
  176: This specifies that the dom0 should have 512MB of ram, leaving the rest
  177: to be allocated for domUs.
  178: 
  179: NB: This says add, not replace, so that you will be able to more
  180: easily boot a NetBSD kernel without Xen.  Once Xen boots ok, you may
  181: want to set it as default.  It is highly likely that you will have
  182: trouble at some point, and keeping an up-to-date GENERIC for use in
  183: fixing problems is the standard prudent approach.
  184: 
  185: \todo Explain why rndseed is not set with Xen as part of the dom0
  186: subconfiguration.
  187: 
  188: Note that you are likely to have to set root= because the boot device
  189: from /boot is not passed via Xen to the dom0 kernel.  With one disk,
  190: it will work, but e.g. plugging in USB disk to a machine with root on
  191: wd0a causes boot to fail.
  192: 
  193: Beware that userconf statements must be attached to the dom0 load, and
  194: may not be at top-level, because then they would try to configure the
  195: hypervisor, if there is a way to pass them via multiboot .  It appears
  196: that adding `userconf=pckbc` to `/boot.cfg` causes Xen to crash very
  197: early with a heap overflow.
  198: 
  199: ### Console selection
  200: 
  201: See boot_console(8).  Understand that you should start from a place of
  202: having console setup correct for booting GENERIC before trying to
  203: configure Xen.
  204: 
  205: Generally for GENERIC, one sets the console in bootxx_ffsv1 or
  206: equivalent, and this is passed on to /boot (where one typically does
  207: not set the console).  This configuration of bootxx_ffsv1 should also
  208: be in place for Xen systems, to allow seeing messages from /boot and
  209: use of a keyboard to select a line from the menu.  And, one should
  210: have a working boot path to GENERIC for rescue situations.
  211: 
  212: With GENERIC, the boot options are passed on to /netbsd, but there is
  213: currently no mechanism to pass these via multiboot to the hypervisor.
  214: Thus, in addition to configuring the console in the boot blocks, one
  215: must also configure it for Xen.
  216: 
  217: By default, the hypervisor (Xen itself) will use some sort of vga
  218: device as the console, much like GENERIC uses by default.  The vga
  219: console is relinquished at the conclusion of hypervisor boot, before
  220: the dom0 is started.  Xen when using a vga console does not process
  221: console input.
  222: 
  223: The hypervisor can be configured to use a serial port console, e.g.
  224: [[!template id=filecontent name="/boot.cfg" text="""
  225: menu=Xen:load /netbsd-XEN3_DOM0.gz console=com0;multiboot /xen.gz dom0_mem=512M console=com1 com1=9600,8n1
  226: """]]
  227: This example uses the first serial port (Xen counts from 1; this is
  228: what NetBSD would call com0), and sets speed and parity.  (The dom0 is
  229: then configured to use the same serial port in this example.)
  230: 
  231: With the hypervisor configured for a serial console, it can get input,
  232: and there is a notion of passing this input to the dom0.  \todo
  233: Explain why, if Xen has a serial console, the dom0 console is
  234: typically also configured to open that same serial port, instead of
  235: getting the passthrough input via the xen console.
  236: 
  237: One also configures the console for the dom0.  While one might expect
  238: console=pc to be default, following behavior of GENERIC, a hasty read
  239: of the code suggests there is no default and booting without a
  240: selected console might lead to a panic.  Also, there is merit in
  241: explicit configuration.  Therefore the standard approach is to place
  242: console=pc as part of the load statement for the dom0 kernel, or
  243: alternatively console=com0.
  244: 
  245: The NetBSD dom0 kernel will attach xencons(4) (the man page does not
  246: exist), but this is not used as a console.  It is used to obtain the
  247: messages from the hypervisor's console; run `xl dmesg` to see them.
  248: 
  249: ### Tuning
  250: 
  251: In an attempt to add performance, one can also add `dom0_max_vcpus=1 dom0_vcpus_pin`,
  252: to force only one vcpu to be provided (since NetBSD dom0 can't use
  253: more) and to pin that vcpu to a physical CPU. Xen has
  254: [many boot options](http://xenbits.xenproject.org/docs/4.13-testing/misc/xen-command-line.html),
  255: and other than dom0 memory and max_vcpus, they are generally not
  256: necessary.
  257: \todo Revisit this advice with current.
  258: \todo Explain if anyone has ever actually measured that this helps.
  259: 
  260: ### rc.conf
  261: 
  262: Ensure that the boot scripts installed in
  263: `/usr/pkg/share/examples/rc.d` are in `/etc/rc.d`, either because you
  264: have `PKG_RCD_SCRIPTS=yes`, or manually.  (This is not special to Xen,
  265: but a normal part of pkgsrc usage.)
  266: 
  267: Set `xencommons=YES` in rc.conf:
  268: 
  269: [[!template id=filecontent name="/etc/rc.conf" text="""
  270: xencommons=YES
  271: """]]
  272: 
  273: \todo Recommend for/against xen-watchdog.
  274: 
  275: ### Testing
  276: 
  277: Now, reboot so that you are running a DOM0 kernel under Xen, rather
  278: than GENERIC without Xen.
  279: 
  280: Once the reboot is done, use `xl` to inspect Xen's boot messages,
  281: available resources, and running domains.  For example:
  282: 
  283: [[!template id=programlisting text="""
  284: # xl dmesg
  285: ... xen's boot info ...
  286: # xl info
  287: ... available memory, etc ...
  288: # xl list
  289: Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  290: Domain-0           0       64    0  r----     58.1
  291: """]]
  292: 
  293: Xen logs will be in /var/log/xen.
  294: 
  295: ### Issues with xencommons
  296: 
  297: `xencommons` starts `xenstored`, which stores data on behalf of dom0 and
  298: domUs.  It does not currently work to stop and start xenstored.
  299: Certainly all domUs should be shutdown first, following the sort order
  300: of the rc.d scripts.  However, the dom0 sets up state with xenstored,
  301: and is not notified when xenstored exits, leading to not recreating
  302: the state when the new xenstored starts.  Until there's a mechanism to
  303: make this work, one should not expect to be able to restart xenstored
  304: (and thus xencommons).  There is currently no reason to expect that
  305: this will get fixed any time soon.
  306: \todo Confirm if this is still true in 2020.
  307: 
  308: ## Xen-specific NetBSD issues
  309: 
  310: There are (at least) two additional things different about NetBSD as a
  311: dom0 kernel compared to hardware.
  312: 
  313: One is that through NetBSD 9 the module ABI is different because some
  314: of the #defines change, so there are separate sets of modules in
  315: /stand.  In NetBSD-current, there is only one set of modules.
  316: 
  317: The other difference is that XEN3_DOM0 does not have exactly the same
  318: options as GENERIC.  While it is debatable whether or not this is a
  319: bug, users should be aware of this and can simply add missing config
  320: items if desired.
  321: 
  322: Finally, there have been occasional reports of trouble with X11
  323: servers in NetBSD as a dom0.
  324: 
  325: ## Updating Xen in a dom0
  326: 
  327: Basically, update the xenkernel and xentools packages and copy the new
  328: Xen kernel into place, and reboot.  This procedure should be usable to
  329: update to a new Xen release, but the reader is reminded that having a
  330: non-Xen boot methods was recommended earlier.
  331: 
  332: ## Updating NetBSD in a dom0
  333: 
  334: This is just like updating NetBSD on bare hardware, assuming the new
  335: version supports the version of Xen you are running.  Generally, one
  336: replaces the kernel and reboots, and then overlays userland binaries
  337: and adjusts `/etc`.
  338: 
  339: Note that one should update both the non-Xen kernel typically used for
  340: rescue purposes, as well as the DOM0 kernel used with Xen.
  341: 
  342: ## anita (for testing NetBSD)
  343: 
  344: With a NetBSD dom0, even without any domUs, one should be able to run
  345: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
  346: root, because anita must create a domU):
  347: 
  348: [[!template id=programlisting text="""
  349: anita --vmm=xl test file:///usr/obj/i386/
  350: """]]
  351: 
  352: # Unprivileged domains (domU)
  353: 
  354: This section describes general concepts about domUs.  It does not
  355: address specific domU operating systems or how to install them.  The
  356: config files for domUs are typically in `/usr/pkg/etc/xen`, and are
  357: typically named so that the file name, domU name and the domU's host
  358: name match.
  359: 
  360: The domU is provided with CPU and memory by Xen, configured by the
  361: dom0.  The domU is provided with disk and network by the dom0,
  362: mediated by Xen, and configured in the dom0.
  363: 
  364: Entropy in domUs can be an issue; physical disks and network are on
  365: the dom0.  NetBSD's /dev/random system works, but is often challenged.
  366: 
  367: ## Config files
  368: 
  369: See /usr/pkg/share/examples/xen/xlexample* for a very small number of
  370: examples for running GNU/Linux.
  371: 
  372: The following is an example minimal domain configuration file. The domU
  373: serves as a network file server.
  374: 
  375: [[!template id=filecontent name="/usr/pkg/etc/xen/foo" text="""
  376: name = "domU-id"
  377: kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
  378: memory = 1024
  379: vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
  380: disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
  381:          'file:/n0/xen/foo-wd1,0x1,w' ]
  382: """]]
  383: 
  384: The domain will have name given in the `name` setting.  The kernel has the
  385: host/domU name in it, so that on the dom0 one can update the various
  386: domUs independently.  The `vif` line causes an interface to be provided,
  387: with a specific mac address (do not reuse MAC addresses!), in bridge
  388: mode.  Two disks are provided, and they are both writable; the bits
  389: are stored in files and Xen attaches them to a vnd(4) device in the
  390: dom0 on domain creation.  The system treats xbd0 as the boot device
  391: without needing explicit configuration.
  392: 
  393: There is not type line; that implicitly defines a pv domU.  Otherwise,
  394: one sets type to the lower-case version of the domU type in the table
  395: above, e.g. `type = "hvm"`.
  396: 
  397: By convention, domain config files are kept in `/usr/pkg/etc/xen`.  Note
  398: that "xl create" takes the name of a config file, while other commands
  399: take the name of a domain.
  400: 
  401: Examples of commands:
  402: 
  403: [[!template id=programlisting text="""
  404: xl create /usr/pkg/etc/xen/foo
  405: xl console domU-id
  406: xl create -c /usr/pkg/etc/xen/foo
  407: xl shutdown domU-id
  408: xl list
  409: """]]
  410: 
  411: Typing `^]` will exit the console session.  Shutting down a domain is
  412: equivalent to pushing the power button; a NetBSD domU will receive a
  413: power-press event and do a clean shutdown.  Shutting down the dom0
  414: will trigger controlled shutdowns of all configured domUs.
  415: 
  416: ## CPU and memory
  417: 
  418: A domain is provided with some number of vcpus, up to the number
  419: of CPUs seen by the hypervisor. For a domU, it is controlled
  420: from the config file by the "vcpus = N" directive.
  421: 
  422: A domain is provided with memory; this is controlled in the config
  423: file by "memory = N" (in megabytes).  In the straightforward case, the
  424: sum of the the memory allocated to the dom0 and all domUs must be less
  425: than the available memory.
  426: 
  427: ## Balloon driver
  428: 
  429: Xen provides a `balloon` driver, which can be used to let domains use
  430: more memory temporarily.
  431: 
  432: \todo Explain how to set up a aystem to use the balloon scheme in a
  433: useful manner.
  434: 
  435: ## Virtual disks
  436: 
  437: In domU config files, the disks are defined as a sequence of 3-tuples:
  438: 
  439:  * The first element is "method:/path/to/disk". Common methods are
  440:    "file:" for a file-backed vnd, and "phy:" for something that is already
  441:    a device, such as an LVM logical volume.
  442: 
  443:  * The second element is an artifact of how virtual disks are passed to
  444:    Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
  445:    are given a device name to associate with the disk, and values like
  446:    "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
  447:    as xbd0, the second as xbd1, and so on.  However, xl demands a
  448:    second argument.  The name given is converted to a major/minor by
  449:    calling stat(2) on the name in /dev and this is passed to the domU.
  450:    In the general case, the dom0 and domU can be different operating
  451:    systems, and it is an unwarranted assumption that they have consistent
  452:    numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
  453:    as both dom0 and domU, using values of 0x0 for the first disk and 0x1
  454:    for the second works fine and avoids this issue.  For a GNU/Linux
  455:    guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
  456:    /dev/hda1.
  457: 
  458:  * The third element is "w" for writable disks, and "r" for read-only
  459:    disks.
  460: 
  461: Example:
  462: [[!template id=filecontent name="/usr/pkg/etc/xen/foo" text="""
  463: disk = [ 'file:/n0/xen/foo-wd0,0x0,w' ]
  464: """]]
  465: 
  466: Note that NetBSD by default creates only vnd[0123].  If you need more
  467: than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
  468: dom0.
  469: 
  470: ## Virtual Networking
  471: 
  472: Xen provides virtual Ethernets, each of which connects the dom0 and a
  473: domU.  For each virtual network, there is an interface "xvifN.M" in
  474: the dom0, and a matching interface xennetM (NetBSD name) in domU index N.
  475: The interfaces behave as if there is an Ethernet with two
  476: adapters connected.  From this primitive, one can construct various
  477: configurations.  We focus on two common and useful cases for which
  478: there are existing scripts: bridging and NAT.
  479: 
  480: With bridging (in the example above), the domU perceives itself to be
  481: on the same network as the dom0.  For server virtualization, this is
  482: usually best.  Bridging is accomplished by creating a bridge(4) device
  483: and adding the dom0's physical interface and the various xvifN.0
  484: interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
  485: config file.  The bridge must be set up already in the dom0; an
  486: example /etc/ifconfig.bridge0 is:
  487: 
  488: [[!template id=filecontent name="/etc/ifconfig.bridge0" text="""
  489: create
  490: up
  491: !brconfig bridge0 add wm0
  492: """]]
  493: 
  494: With NAT, the domU perceives itself to be behind a NAT running on the
  495: dom0.  This is often appropriate when running Xen on a workstation.
  496: TODO: NAT appears to be configured by "vif = [ '' ]".
  497: 
  498: The MAC address specified is the one used for the interface in the new
  499: domain.  The interface in dom0 will use this address XOR'd with
  500: 00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
  501: 
  502: ## Starting domains automatically
  503: 
  504: To start domains `domU-netbsd` and `domU-linux` at boot and shut them
  505: down cleanly on dom0 shutdown, add the following in rc.conf:
  506: 
  507: [[!template id=filecontent name="/etc/rc.conf" text="""
  508: xendomains="domU-netbsd domU-linux"
  509: """]]
  510: 
  511: # domU setup for specific systems
  512: 
  513: Creating domUs is almost entirely independent of operating system.  We
  514: have already presented the basics of config files in the previous system.
  515: 
  516: Of course, this section presumes that you have a working dom0.
  517: 
  518: ## Creating a NetBSD PV domU
  519: 
  520: See the earlier config file, and adjust memory.  Decide on how much
  521: storage you will provide, and prepare it (file or LVM).
  522: 
  523: While the kernel will be obtained from the dom0 file system, the same
  524: file should be present in the domU as /netbsd so that tools like
  525: savecore(8) can work.   (This is helpful but not necessary.)
  526: 
  527: The kernel must be specifically built for Xen, to use PV interfacesas
  528: a domU.  NetBSD release builds provide the following kernels:
  529: 
  530:         i386 XEN3PAE_DOMU
  531:         amd64 XEN3_DOMU
  532: 
  533: This will boot NetBSD, but this is not that useful if the disk is
  534: empty.  One approach is to unpack sets onto the disk outside of Xen
  535: (by mounting it, just as you would prepare a physical disk for a
  536: system you can't run the installer on).
  537: 
  538: A second approach is to run an INSTALL kernel, which has a miniroot
  539: and can load sets from the network.  To do this, copy the INSTALL
  540: kernel to / and change the kernel line in the config file to:
  541: 
  542:         kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
  543: 
  544: Then, start the domain as "xl create -c configfile".
  545: 
  546: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
  547: line should be used in the config file.
  548: 
  549:     disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
  550: 
  551: After booting the domain, the option to install via CDROM may be
  552: selected.  The CDROM device should be changed to `xbd1d`.
  553: 
  554: Once done installing, "halt -p" the new domain (don't reboot or halt:
  555: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
  556: config file), switch the config file back to the XEN3_DOMU kernel,
  557: and start the new domain again. Now it should be able to use "root on
  558: xbd0a" and you should have a functional NetBSD domU.
  559: 
  560: TODO: check if this is still accurate.
  561: When the new domain is booting you'll see some warnings about *wscons*
  562: and the pseudo-terminals. These can be fixed by editing the files
  563: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
  564: `/etc/ttys`, except *console*, like this:
  565: 
  566:     console "/usr/libexec/getty Pc"         vt100   on secure
  567:     ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
  568:     ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
  569:     ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
  570:     ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
  571: 
  572: Finally, all screens must be commented out from `/etc/wscons.conf`.
  573: 
  574: One should also run `powerd` in a domU, but this should not need
  575: configuring.  With powerd, the domain will run a controlled shutdown
  576: if `xl shutdown -R` or `xl shutdown -H` is used on the dom0, via
  577: receiving a synthetic `power button pressed` signal.  In 9 and
  578: current, `powerd` is run by default under Xen kernels (or if ACPI is
  579: present), and it can be added to rc.conf if not.
  580: 
  581: It is not strictly necessary to have a kernel (as /netbsd) in the domU
  582: file system.  However, various programs (e.g. netstat) will use that
  583: kernel to look up symbols to read from kernel virtual memory.  If
  584: /netbsd is not the running kernel, those lookups will fail.  (This is
  585: not really a Xen-specific issue, but because the domU kernel is
  586: obtained from the dom0, it is far more likely to be out of sync or
  587: missing with Xen.)
  588: 
  589: Note that NetBSD by default creates only xbd[0123].  If you need more
  590: virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
  591: 
  592: ## Creating a Linux PV domU
  593: 
  594: Creating unprivileged Linux domains isn't much different from
  595: unprivileged NetBSD domains, but there are some details to know.
  596: 
  597: First, the second parameter passed to the disk declaration (the '0x1' in
  598: the example below)
  599: 
  600:     disk = [ 'phy:/dev/wd0e,0x1,w' ]
  601: 
  602: does matter to Linux. It wants a Linux device number here (e.g. 0x300
  603: for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
  604: So, hda1 which has major 3 and minor 1 on a Linux system will have
  605: device number 0x301.  Alternatively, devices names can be used (hda,
  606: hdb, ...)  as xentools has a table to map these names to devices
  607: numbers.  To export a partition to a Linux guest we can use:
  608: 
  609:         disk = [ 'phy:/dev/wd0e,0x300,w' ]
  610:         root = "/dev/hda1 ro"
  611: 
  612: and it will appear as /dev/hda on the Linux system, and be used as root
  613: partition.
  614: 
  615: To install the Linux system on the partition to be exported to the
  616: guest domain, the following method can be used: install
  617: sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
  618: that will be the root partition of your Linux domain, and mount it.
  619: Then copy the files from a working Linux system, make adjustments in
  620: `/etc` (fstab, network config).  It should also be possible to extract
  621: binary packages such as .rpm or .deb directly to the mounted partition
  622: using the appropriate tool, possibly running under NetBSD's Linux
  623: emulation.  Once the file system has been populated, umount it.  If
  624: desirable, the file system can be converted to ext3 using tune2fs -j.
  625: It should now be possible to boot the Linux guest domain, using one of
  626: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
  627: 
  628: To get the Linux console right, you need to add:
  629: 
  630:     extra = "xencons=tty1"
  631: 
  632: to your configuration since not all Linux distributions auto-attach a
  633: tty to the xen console.
  634: 
  635: ## Creating a NetBSD HVM domU
  636: 
  637: Use type='hvm', probably.  Use a GENERIC kernel within the disk image.
  638: 
  639: ## Creating a NetBSD PVH domU
  640: 
  641: This only works with a current kernel in the domU.
  642: 
  643: Use type='pvh'.  Probably, use a GENERIC kernel within the disk image,
  644: which in current has PV support.
  645: 
  646: \todo Verify.
  647: 
  648: \todo Verify if one can have current PVH domU on a 9 dom0.
  649: 
  650: ## Creating a Solaris domU
  651: 
  652: See possibly outdated
  653: [Solaris domU instructions](/ports/xen/howto-solaris/).
  654: 
  655: ## PCI passthrough: Using PCI devices in guest domains
  656: 
  657: NB: PCI passthrough only works on some Xen versions and as of 2020 it
  658: is not clear that it works on any version in pkgsrc.  \todo Reports
  659: confirming or denying this notion should be sent to port-xen@.
  660: 
  661: The dom0 can give other domains access to selected PCI
  662: devices. This can allow, for example, a non-privileged domain to have
  663: access to a physical network interface or disk controller.  However,
  664: keep in mind that giving a domain access to a PCI device most likely
  665: will give the domain read/write access to the whole physical memory,
  666: as PCs don't have an IOMMU to restrict memory access to DMA-capable
  667: device.  Also, it's not possible to export ISA devices to non-dom0
  668: domains, which means that the primary VGA adapter can't be exported.
  669: A guest domain trying to access the VGA registers will panic.
  670: 
  671: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
  672: not been ported to later versions at this time.
  673: 
  674: For a PCI device to be exported to a domU, is has to be attached to
  675: the "pciback" driver in dom0.  Devices passed to the dom0 via the
  676: pciback.hide boot parameter will attach to "pciback" instead of the
  677: usual driver.  The list of devices is specified as "(bus:dev.func)",
  678: where bus and dev are 2-digit hexadecimal numbers, and func a
  679: single-digit number:
  680: 
  681:         pciback.hide=(00:0a.0)(00:06.0)
  682: 
  683: pciback devices should show up in the dom0's boot messages, and the
  684: devices should be listed in the `/kern/xen/pci` directory.
  685: 
  686: PCI devices to be exported to a domU are listed in the "pci" array of
  687: the domU's config file, with the format "0000:bus:dev.func".
  688: 
  689:         pci = [ '0000:00:06.0', '0000:00:0a.0' ]
  690: 
  691: In the domU an "xpci" device will show up, to which one or more pci
  692: buses will attach.  Then the PCI drivers will attach to PCI buses as
  693: usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
  694: or any PCI drivers built in by default; you have to build your own
  695: kernel to use PCI devices in a domU.  Here's a kernel config example;
  696: note that only the "xpci" lines are unusual.
  697: 
  698:         include         "arch/i386/conf/XEN3_DOMU"
  699: 
  700:         # Add support for PCI buses to the XEN3_DOMU kernel
  701:         xpci* at xenbus ?
  702:         pci* at xpci ?
  703: 
  704:         # PCI USB controllers
  705:         uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
  706: 
  707:         # USB bus support
  708:         usb*    at uhci?
  709: 
  710:         # USB Hubs
  711:         uhub*   at usb?
  712:         uhub*   at uhub? port ? configuration ? interface ?
  713: 
  714:         # USB Mass Storage
  715:         umass*  at uhub? port ? configuration ? interface ?
  716:         wd*     at umass?
  717:         # SCSI controllers
  718:         ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
  719: 
  720:         # SCSI bus support (for both ahc and umass)
  721:         scsibus* at scsi?
  722: 
  723:         # SCSI devices
  724:         sd*     at scsibus? target ? lun ?      # SCSI disk drives
  725:         cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
  726: 
  727: 
  728: # Miscellaneous Information
  729: 
  730: ## Nesting under Linux KVM
  731: 
  732: It is possible to run a Xen and a NetBSD dom0 under Linux KVM.  One
  733: can enable virtio in the dom0 for greater speed.
  734: 
  735: ## Other nesting
  736: 
  737: In theory, any full emulation should be able to run Xen and a NetBSD
  738: dom0.  The HOWTO does not currently have information about Xen XVM
  739: mode, nvmm, qemu, Virtualbox, etc.
  740: 
  741: ## NetBSD 5 as domU
  742: 
  743: [NetBSD 5 is known to panic.](http://mail-index.netbsd.org/port-xen/2018/04/17/msg009181.html)
  744: (However, NetBSD 5 systems should be updated to a supported version.)
  745: 
  746: # NetBSD as a domU in a VPS
  747: 
  748: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
  749: hardware.  This section explains how to deal with Xen in a domU as a
  750: virtual private server where you do not control or have access to the
  751: dom0.  This is not intended to be an exhaustive list of VPS providers;
  752: only a few are mentioned that specifically support NetBSD.
  753: 
  754: VPS operators provide varying degrees of access and mechanisms for
  755: configuration.  The big issue is usually how one controls which kernel
  756: is booted, because the kernel is nominally in the dom0 file system (to
  757: which VPS users do not normally have access).  A second issue is how
  758: to install NetBSD.
  759: A VPS user may want to compile a kernel for security updates, to run
  760: npf, run IPsec, or any other reason why someone would want to change
  761: their kernel.
  762: 
  763: One approach is to have an administrative interface to upload a kernel,
  764: or to select from a prepopulated list.  Other approaches are pygrub
  765: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
  766: kernel from the domU file system.  This is closer to a regular physical
  767: computer, where someone who controls a machine can replace the kernel.
  768: 
  769: A second issue is multiple CPUs.  With NetBSD 6, domUs support
  770: multiple vcpus, and it is typical for VPS providers to enable multiple
  771: CPUs for NetBSD domUs.
  772: 
  773: ## Complexities due to Xen changes
  774: 
  775: Xen has many security advisories and people running Xen systems make
  776: different choices.
  777: 
  778: ### stub domains
  779: 
  780: Some (Linux only?) dom0 systems use something called "stub domains" to
  781: isolate qemu from the dom0 system, as a security and reliabilty
  782: mechanism when running HVM domUs.  Somehow, NetBSD's GENERIC kernel
  783: ends up using PIO for disks rather than DMA.  Of course, all of this
  784: is emulated, but emulated PIO is unusably slow.  This problem is not
  785: currently understood.
  786: 
  787: ### Grant tables
  788: 
  789: There are multiple versions of using grant tables, and some security
  790: advisories have suggested disabling some versions.  Some versions of
  791: NetBSD apparently only use specific versions and this can lead to
  792: "NetBSD current doesn't run on hosting provider X" situations.
  793: 
  794: \todo Explain better.
  795: 
  796: ## Boot methods
  797: 
  798: ### pvgrub
  799: 
  800: pvgrub is a version of grub that uses PV operations instead of BIOS
  801: calls.  It is booted from the dom0 as the domU kernel, and then reads
  802: /grub/menu.lst and loads a kernel from the domU file system.
  803: 
  804: [Panix](http://www.panix.com/) lets users use pvgrub.  Panix reports
  805: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
  806: (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
  807: page](http://www.panix.com/v-colo/grub.html), which describes only
  808: Linux but should be updated to cover NetBSD :-).
  809: 
  810: [prgmr.com](http://prgmr.com/) also lets users with pvgrub to boot
  811: their own kernel.  See then [prgmr.com NetBSD
  812: HOWTO](http://wiki.prgmr.com/mediawiki/index.php/NetBSD_as_a_DomU)
  813: (which is in need of updating).
  814: 
  815: It appears that [grub's FFS
  816: code](http://xenbits.xensource.com/hg/xen-unstable.hg/file/bca284f67702/tools/libfsimage/ufs/fsys_ufs.c)
  817: does not support all aspects of modern FFS, but there are also reports
  818: that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
  819: partition for the kernel with the intent that grub can understand it,
  820: which leads to /netbsd not being the actual kernel.  One must remember
  821: to update the special boot partition.
  822: 
  823: ### pygrub
  824: 
  825: pygrub runs in the dom0 and looks into the domU file system.  This
  826: implies that the domU must have a kernel in a file system in a format
  827: known to pygrub.
  828: 
  829: pygrub doesn't seem to work to load Linux images under NetBSD dom0,
  830: and is inherently less secure than pvgrub due to running inside dom0. For both these
  831: reasons, pygrub should not be used, and is only still present so that
  832: historical DomU images using it still work.
  833: 
  834: As of 2014, pygrub seems to be of mostly historical
  835: interest. New DomUs should use pvgrub.
  836: 
  837: ## Specific Providers
  838: 
  839: ### Amazon
  840: 
  841: See the [Amazon EC2 page](/amazon_ec2/).

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