File:  [NetBSD Developer Wiki] / wikisrc / ports / xen / howto.mdwn
Revision 1.204: download - view: text, annotated - select for diffs
Mon Mar 15 00:25:54 2021 UTC (3 months, 1 week ago) by gdt
Branches: MAIN
CVS tags: HEAD
adjust pv/hvm wording

    1: [[!meta title="Xen Status and HowTo"]]
    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 document provides status on what Xen things work on NetBSD
   12: (upstream documentation might say something works if it works on some
   13: particular Linux system).
   14: 
   15: This document is also a HOWTO that presumes a basic familiarity with
   16: the Xen system architecture, with installing NetBSD on amd64 hardware,
   17: and with installing software from pkgsrc.  See also the [Xen
   18: website](http://www.xenproject.org/).
   19: 
   20: If this document says that something works, and you find that it does
   21: not, it is best to ask on port-xen and if you are correct to file a
   22: PR.
   23: 
   24: [[!toc]]
   25: 
   26: # Overview
   27: 
   28: The basic concept of Xen is that the hypervisor (xenkernel) runs on
   29: the hardware, and runs a privileged domain ("dom0") that can access
   30: disks/networking/etc.  One then runs additional unprivileged domains
   31: (each a "domU"), presumably to do something useful.
   32: 
   33: This HOWTO addresses how to run a NetBSD dom0 (and hence also build
   34: xen itself).  It also addresses how to run domUs in that environment,
   35: and how to deal with having a domU in a Xen environment run by someone
   36: else and/or not running NetBSD.
   37: 
   38: There are many choices one can make; the HOWTO recommends the standard
   39: approach and limits discussion of alternatives in many cases.
   40: 
   41: ## Guest Styles
   42: 
   43: Xen supports different styles of guests.
   44: 
   45: [[!table data="""
   46: Style of guest	|description				|NetBSD dom0?	|NetBSD domU?
   47: PV		|Paravirtualization (hypercalls)   	|yes		|yes
   48: HVM		|unmodified guest - domU perceives HW	|can run guests	|yes
   49: PVHVM		|Like HVM but also PV drivers		|can run guests	|current only
   50: PVH		|Lightweight HVM (no qemu) and PV	|not yet	|current only
   51: """]]
   52: 
   53: In Para-Virtualized (PV) mode, the guest OS does not attempt to access
   54: hardware directly, but instead makes hypercalls to the hypervisor; PV
   55: guests must be specifically coded for Xen.
   56: See [PV](https://wiki.xen.org/wiki/Paravirtualization_(PV\)).
   57: 
   58: In HVM mode, no guest modification is required; however, hardware
   59: support is required, such as VT-x on Intel CPUs and SVM on AMD CPUs.
   60: The dom0 runs qemu to emulate hardware.  It is therefore non-sensical
   61: to have an HVM dom0.
   62: 
   63: In PVHVM mode, the guest runs as HVM, but additionally can use PV
   64: drivers for efficiency.  Therefore it is non-sensical for to have a
   65: PVHVM dom0.  See [PV on HVM](https://wiki.xen.org/wiki/PV_on_HVM).
   66: 
   67: There have been two PVH modes: original PVH and PVHv2.  Original PVH
   68: was based on PV mode and is no longer relevant at all.  Therefore
   69: PVHv2 is abreviated PVH.  PVHv2 is basically lightweight HVM with PV
   70: drivers.  A critical feature of it is that qemu is not needed; the
   71: hypervisor can do the emulation that is required.  Thus, a dom0 can be
   72: PVHv2.  The source code uses PVH and config files use pvh, but NB that
   73: this refers to PVHv2.  See
   74: [PVH(v2)](https://wiki.xenproject.org/wiki/PVH_(v2\)_Domu).
   75: 
   76: At system boot, the dom0 kernel is loaded as a module with Xen as the kernel.
   77: The dom0 can start one or more domUs.  (Booting is explained in detail
   78: in the dom0 section.)
   79: 
   80: ## CPU Architecture
   81: 
   82: Xen runs on x86_64 hardware (the NetBSD amd64 port).
   83: 
   84: There is a concept of Xen running on ARM, but there are no reports of this working with NetBSD.
   85: 
   86: The dom0 system should be amd64.  (Instructions for i386PAE dom0 have been removed from the HOWTO.)
   87: 
   88: The domU can be i386 PAE or amd64.
   89: i386 PAE at one point was considered as [faster](https://lists.xen.org/archives/html/xen-devel/2012-07/msg00085.html) than amd64.
   90: However, as of 2021 it is normal to use amd64 as the domU architecture, and use of i386 is dwindling.
   91: 
   92: ## Xen Versions
   93: 
   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.
   98: 
   99: Versions available in pkgsrc:
  100: 
  101: [[!table data="""
  102: Xen Version	|Package Name	|Xen CPU Support	|EOL'ed By Upstream
  103: 4.11		|xenkernel411	|x86_64			|No
  104: 4.13		|xenkernel413	|x86_64			|No
  105: """]]
  106: 
  107: See also the [Xen Security Advisory page](http://xenbits.xen.org/xsa/).
  108: 
  109: Older Xen had a python-based management tool called xm; this has been
  110: replaced by xl.
  111: 
  112: ## NetBSD versions
  113: 
  114: Xen has been supported in NetBSD for a long time, at least since 2005.
  115: Initially Xen was PV only.
  116: 
  117: NetBSD Xen has always supported PV, in both dom0 and domU; for a long
  118: time this was the only way.  NetBSD >=8 as a dom0 supports HVM mode in
  119: domUs (HVM as a dom0 does not make sense).
  120: 
  121: Support for PVHVM and PVH is available only in NetBSD-current; this is
  122: currently somewhat experimental, although PVHVM appears reasonably
  123: solid.
  124: 
  125: NetBSD up to and including NetBSD 9 as a dom0 cannot safely run SMP.
  126: Even if one added "options MULTIPROCESSOR" and configured multiple
  127: vcpus, the kernel is likely to crash because of drivers without
  128: adequate locking.
  129: 
  130: NetBSD-current supports SMP in dom0, and XEN3_DOM0 includes "options
  131: MULTIPROCESSOR".
  132: 
  133: NetBSD (since NetBSD 6), when run as a domU, can run SMP, using
  134: multiple CPUs if provided.  The XEN3_DOMU kernel is built
  135: with "options MULITPROCESSOR".
  136: 
  137: Note that while Xen 4.13 is current, the kernel support is still
  138: called XEN3, because the hypercall interface has not changed
  139: significantly.
  140: 
  141: # Creating a NetBSD dom0
  142: 
  143: In order to install a NetBSD as a dom0, one first installs a normal
  144: NetBSD system, and then pivot the install to a dom0 install by
  145: changing the kernel and boot configuration.
  146: 
  147: In 2018-05, trouble booting a dom0 was reported with 256M of RAM: with
  148: 512M it worked reliably.  This does not make sense, but if you see
  149: "not ELF" after Xen boots, try increasing dom0 RAM.
  150: 
  151: ## Installation of NetBSD
  152: 
  153: [Install NetBSD/amd64](/guide/inst/) just as you would if you were not
  154: using Xen.  Therefore, use the most recent release, or a build from
  155: the most recent stable branch.  Alternatively, use -current, being
  156: mindful of all the usual caveats of lower stability of current, and
  157: likely a bit more so.
  158: 
  159: ## Installation of Xen
  160: 
  161: ### Building Xen
  162: 
  163: Use the most recent version of Xen in pkgsrc, unless the DESCR says that it is not suitable.
  164: Therefore, choose 4.13.
  165: In the dom0, install xenkernel413 and xentools413 from pkgsrc.
  166: 
  167: Once this is done, copy the Xen kernel from where pkgsrc puts it to
  168: where the boot process will be able to find it:
  169: 
  170: [[!template id=programlisting text="""
  171: # cp -p /usr/pkg/xen413-kernel/xen.gz /
  172: """]]
  173: 
  174: Then, place a NetBSD XEN3_DOM0 kernel in the `/` directory. Such
  175: kernel can either be taken from a local release build.sh run, compiled
  176: manually, or downloaded from the NetBSD FTP, for example at:
  177: 
  178: [[!template id=programlisting text="""
  179: ftp.netbsd.org/pub/NetBSD/NetBSD-9.1/amd64/binary/kernel/netbsd-XEN3_DOM0.gz
  180: """]]
  181: 
  182: ### Configuring booting
  183: 
  184: Read boot.cfg(8) carefully.  Add lines to /boot.cfg to boot Xen,
  185: adjusting for your root filesystem:
  186: 
  187: [[!template id=filecontent name="/boot.cfg" text="""
  188: menu=Xen:load /netbsd-XEN3_DOM0.gz root=wd0a console=pc;multiboot /xen.gz dom0_mem=512M
  189: menu=Xen single user:load /netbsd-XEN3_DOM0.gz root=wd0a console=pc -s;multiboot /xen.gz dom0_mem=512M
  190: """]]
  191: 
  192: This specifies that the dom0 should have 512MB of ram, leaving the rest
  193: to be allocated for domUs.
  194: 
  195: NB: This says add, not replace, so that you will be able to more
  196: easily boot a NetBSD kernel without Xen.  Once Xen boots ok, you may
  197: want to set it as default.  It is highly likely that you will have
  198: trouble at some point, and keeping an up-to-date GENERIC for use in
  199: fixing problems is the standard prudent approach.
  200: 
  201: \todo Explain why rndseed is not set with Xen as part of the dom0
  202: subconfiguration.
  203: 
  204: Note that you are likely to have to set root= because the boot device
  205: from /boot is not passed via Xen to the dom0 kernel.  With one disk,
  206: it will work, but e.g. plugging in USB disk to a machine with root on
  207: wd0a causes boot to fail.
  208: 
  209: Beware that userconf statements must be attached to the dom0 load, and
  210: may not be at top-level, because then they would try to configure the
  211: hypervisor, if there is a way to pass them via multiboot .  It appears
  212: that adding `userconf=pckbc` to `/boot.cfg` causes Xen to crash very
  213: early with a heap overflow.
  214: 
  215: ### Console selection
  216: 
  217: See boot_console(8).  Understand that you should start from a place of
  218: having console setup correct for booting GENERIC before trying to
  219: configure Xen.
  220: 
  221: Generally for GENERIC, one sets the console in bootxx_ffsv1 or
  222: equivalent, and this is passed on to /boot (where one typically does
  223: not set the console).  This configuration of bootxx_ffsv1 should also
  224: be in place for Xen systems, to allow seeing messages from /boot and
  225: use of a keyboard to select a line from the menu.  And, one should
  226: have a working boot path to GENERIC for rescue situations.
  227: 
  228: With GENERIC, the boot options are passed on to /netbsd, but there is
  229: currently no mechanism to pass these via multiboot to the hypervisor.
  230: Thus, in addition to configuring the console in the boot blocks, one
  231: must also configure it for Xen.
  232: 
  233: By default, the hypervisor (Xen itself) will use some sort of vga
  234: device as the console, much like GENERIC uses by default.  The vga
  235: console is relinquished at the conclusion of hypervisor boot, before
  236: the dom0 is started.  Xen when using a vga console does not process
  237: console input.
  238: 
  239: The hypervisor can be configured to use a serial port console, e.g.
  240: [[!template id=filecontent name="/boot.cfg" text="""
  241: menu=Xen:load /netbsd-XEN3_DOM0.gz console=com0;multiboot /xen.gz dom0_mem=512M console=com1 com1=9600,8n1
  242: """]]
  243: This example uses the first serial port (Xen counts from 1; this is
  244: what NetBSD would call com0), and sets speed and parity.  (The dom0 is
  245: then configured to use the same serial port in this example.)
  246: 
  247: With the hypervisor configured for a serial console, it can get input,
  248: and there is a notion of passing this input to the dom0.  \todo
  249: Explain why, if Xen has a serial console, the dom0 console is
  250: typically also configured to open that same serial port, instead of
  251: getting the passthrough input via the xen console.
  252: 
  253: One also configures the console for the dom0.  While one might expect
  254: console=pc to be default, following behavior of GENERIC, a hasty read
  255: of the code suggests there is no default and booting without a
  256: selected console might lead to a panic.  Also, there is merit in
  257: explicit configuration.  Therefore the standard approach is to place
  258: console=pc as part of the load statement for the dom0 kernel, or
  259: alternatively console=com0.
  260: 
  261: The NetBSD dom0 kernel will attach xencons(4) (the man page does not
  262: exist), but this is not used as a console.  It is used to obtain the
  263: messages from the hypervisor's console; run `xl dmesg` to see them.
  264: 
  265: ### Tuning
  266: 
  267: In an attempt to add performance, one can also add `dom0_max_vcpus=1 dom0_vcpus_pin`,
  268: to force only one vcpu to be provided (since NetBSD dom0 can't use
  269: more) and to pin that vcpu to a physical CPU. Xen has
  270: [many boot options](http://xenbits.xenproject.org/docs/4.13-testing/misc/xen-command-line.html),
  271: and other than dom0 memory and max_vcpus, they are generally not
  272: necessary.
  273: \todo Revisit this advice with current.
  274: \todo Explain if anyone has ever actually measured that this helps.
  275: 
  276: ### rc.conf
  277: 
  278: Ensure that the boot scripts installed in
  279: `/usr/pkg/share/examples/rc.d` are in `/etc/rc.d`, either because you
  280: have `PKG_RCD_SCRIPTS=yes`, or manually.  (This is not special to Xen,
  281: but a normal part of pkgsrc usage.)
  282: 
  283: Set `xencommons=YES` in rc.conf:
  284: 
  285: [[!template id=filecontent name="/etc/rc.conf" text="""
  286: xencommons=YES
  287: """]]
  288: 
  289: \todo Recommend for/against xen-watchdog.
  290: 
  291: ### Testing
  292: 
  293: Now, reboot so that you are running a DOM0 kernel under Xen, rather
  294: than GENERIC without Xen.
  295: 
  296: Once the reboot is done, use `xl` to inspect Xen's boot messages,
  297: available resources, and running domains.  For example:
  298: 
  299: [[!template id=programlisting text="""
  300: # xl dmesg
  301: ... xen's boot info ...
  302: # xl info
  303: ... available memory, etc ...
  304: # xl list
  305: Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  306: Domain-0           0       64    0  r----     58.1
  307: """]]
  308: 
  309: Xen logs will be in /var/log/xen.
  310: 
  311: ### Issues with xencommons
  312: 
  313: `xencommons` starts `xenstored`, which stores data on behalf of dom0 and
  314: domUs.  It does not currently work to stop and start xenstored.
  315: Certainly all domUs should be shutdown first, following the sort order
  316: of the rc.d scripts.  However, the dom0 sets up state with xenstored,
  317: and is not notified when xenstored exits, leading to not recreating
  318: the state when the new xenstored starts.  Until there's a mechanism to
  319: make this work, one should not expect to be able to restart xenstored
  320: (and thus xencommons).  There is currently no reason to expect that
  321: this will get fixed any time soon.
  322: \todo Confirm if this is still true in 2020.
  323: 
  324: ## Xen-specific NetBSD issues
  325: 
  326: There are (at least) two additional things different about NetBSD as a
  327: dom0 kernel compared to hardware.
  328: 
  329: One is that through NetBSD 9 the module ABI is different because some
  330: of the #defines change, so there are separate sets of modules in
  331: /stand.  In NetBSD-current, there is only one set of modules.
  332: 
  333: The other difference is that XEN3_DOM0 does not have exactly the same
  334: options as GENERIC.  While it is debatable whether or not this is a
  335: bug, users should be aware of this and can simply add missing config
  336: items if desired.
  337: 
  338: Finally, there have been occasional reports of trouble with X11
  339: servers in NetBSD as a dom0.
  340: 
  341: ## Updating Xen in a dom0
  342: 
  343: Basically, update the xenkernel and xentools packages and copy the new
  344: Xen kernel into place, and reboot.  This procedure should be usable to
  345: update to a new Xen release, but the reader is reminded that having a
  346: non-Xen boot methods was recommended earlier.
  347: 
  348: ## Updating NetBSD in a dom0
  349: 
  350: This is just like updating NetBSD on bare hardware, assuming the new
  351: version supports the version of Xen you are running.  Generally, one
  352: replaces the kernel and reboots, and then overlays userland binaries
  353: and adjusts `/etc`.
  354: 
  355: Note that one should update both the non-Xen kernel typically used for
  356: rescue purposes, as well as the DOM0 kernel used with Xen.
  357: 
  358: ## anita (for testing NetBSD)
  359: 
  360: With a NetBSD dom0, even without any domUs, one should be able to run
  361: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
  362: root, because anita must create a domU):
  363: 
  364: [[!template id=programlisting text="""
  365: anita --vmm=xl test file:///usr/obj/i386/
  366: """]]
  367: 
  368: # Unprivileged domains (domU)
  369: 
  370: This section describes general concepts about domUs.  It does not
  371: address specific domU operating systems or how to install them.  The
  372: config files for domUs are typically in `/usr/pkg/etc/xen`, and are
  373: typically named so that the file name, domU name and the domU's host
  374: name match.
  375: 
  376: The domU is provided with CPU and memory by Xen, configured by the
  377: dom0.  The domU is provided with disk and network by the dom0,
  378: mediated by Xen, and configured in the dom0.
  379: 
  380: Entropy in domUs can be an issue; physical disks and network are on
  381: the dom0.  NetBSD's /dev/random system works, but is often challenged.
  382: 
  383: ## Config files
  384: 
  385: See /usr/pkg/share/examples/xen/xlexample* for a very small number of
  386: examples for running GNU/Linux.
  387: 
  388: The following is an example minimal domain configuration file. The domU
  389: serves as a network file server.
  390: 
  391: [[!template id=filecontent name="/usr/pkg/etc/xen/foo" text="""
  392: name = "domU-id"
  393: kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
  394: memory = 1024
  395: vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
  396: disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
  397:          'file:/n0/xen/foo-wd1,0x1,w' ]
  398: """]]
  399: 
  400: The domain will have name given in the `name` setting.  The kernel has the
  401: host/domU name in it, so that on the dom0 one can update the various
  402: domUs independently.  The `vif` line causes an interface to be provided,
  403: with a specific mac address (do not reuse MAC addresses!), in bridge
  404: mode.  Two disks are provided, and they are both writable; the bits
  405: are stored in files and Xen attaches them to a vnd(4) device in the
  406: dom0 on domain creation.  The system treats xbd0 as the boot device
  407: without needing explicit configuration.
  408: 
  409: There is not type line; that implicitly defines a pv domU.  Otherwise,
  410: one sets type to the lower-case version of the domU type in the table
  411: above, e.g. `type = "hvm"`.
  412: 
  413: By convention, domain config files are kept in `/usr/pkg/etc/xen`.  Note
  414: that "xl create" takes the name of a config file, while other commands
  415: take the name of a domain.
  416: 
  417: Examples of commands:
  418: 
  419: [[!template id=programlisting text="""
  420: xl create /usr/pkg/etc/xen/foo
  421: xl console domU-id
  422: xl create -c /usr/pkg/etc/xen/foo
  423: xl shutdown domU-id
  424: xl list
  425: """]]
  426: 
  427: Typing `^]` will exit the console session.  Shutting down a domain is
  428: equivalent to pushing the power button; a NetBSD domU will receive a
  429: power-press event and do a clean shutdown.  Shutting down the dom0
  430: will trigger controlled shutdowns of all configured domUs.
  431: 
  432: ## CPU and memory
  433: 
  434: A domain is provided with some number of vcpus, up to the number
  435: of CPUs seen by the hypervisor. For a domU, it is controlled
  436: from the config file by the "vcpus = N" directive.
  437: 
  438: A domain is provided with memory; this is controlled in the config
  439: file by "memory = N" (in megabytes).  In the straightforward case, the
  440: sum of the the memory allocated to the dom0 and all domUs must be less
  441: than the available memory.
  442: 
  443: ## Balloon driver
  444: 
  445: Xen provides a `balloon` driver, which can be used to let domains use
  446: more memory temporarily.
  447: 
  448: \todo Explain how to set up a aystem to use the balloon scheme in a
  449: useful manner.
  450: 
  451: ## Virtual disks
  452: 
  453: In domU config files, the disks are defined as a sequence of 3-tuples:
  454: 
  455:  * The first element is "method:/path/to/disk". Common methods are
  456:    "file:" for a file-backed vnd, and "phy:" for something that is already
  457:    a device, such as an LVM logical volume.
  458: 
  459:  * The second element is an artifact of how virtual disks are passed to
  460:    Linux, and a source of confusion with NetBSD Xen usage.  Linux domUs
  461:    are given a device name to associate with the disk, and values like
  462:    "hda1" or "sda1" are common.  In a NetBSD domU, the first disk appears
  463:    as xbd0, the second as xbd1, and so on.  However, xl demands a
  464:    second argument.  The name given is converted to a major/minor by
  465:    calling stat(2) on the name in /dev and this is passed to the domU.
  466:    In the general case, the dom0 and domU can be different operating
  467:    systems, and it is an unwarranted assumption that they have consistent
  468:    numbering in /dev, or even that the dom0 OS has a /dev.  With NetBSD
  469:    as both dom0 and domU, using values of 0x0 for the first disk and 0x1
  470:    for the second works fine and avoids this issue.  For a GNU/Linux
  471:    guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
  472:    /dev/hda1.
  473: 
  474:  * The third element is "w" for writable disks, and "r" for read-only
  475:    disks.
  476: 
  477: Example:
  478: [[!template id=filecontent name="/usr/pkg/etc/xen/foo" text="""
  479: disk = [ 'file:/n0/xen/foo-wd0,0x0,w' ]
  480: """]]
  481: 
  482: Note that NetBSD by default creates only vnd[0123].  If you need more
  483: than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
  484: dom0.
  485: 
  486: ## Virtual Networking
  487: 
  488: Xen provides virtual Ethernets, each of which connects the dom0 and a
  489: domU.  For each virtual network, there is an interface "xvifN.M" in
  490: the dom0, and a matching interface xennetM (NetBSD name) in domU index N.
  491: The interfaces behave as if there is an Ethernet with two
  492: adapters connected.  From this primitive, one can construct various
  493: configurations.  We focus on two common and useful cases for which
  494: there are existing scripts: bridging and NAT.
  495: 
  496: With bridging (in the example above), the domU perceives itself to be
  497: on the same network as the dom0.  For server virtualization, this is
  498: usually best.  Bridging is accomplished by creating a bridge(4) device
  499: and adding the dom0's physical interface and the various xvifN.0
  500: interfaces to the bridge.  One specifies "bridge=bridge0" in the domU
  501: config file.  The bridge must be set up already in the dom0; an
  502: example /etc/ifconfig.bridge0 is:
  503: 
  504: [[!template id=filecontent name="/etc/ifconfig.bridge0" text="""
  505: create
  506: up
  507: !brconfig bridge0 add wm0
  508: """]]
  509: 
  510: With NAT, the domU perceives itself to be behind a NAT running on the
  511: dom0.  This is often appropriate when running Xen on a workstation.
  512: TODO: NAT appears to be configured by "vif = [ '' ]".
  513: 
  514: The MAC address specified is the one used for the interface in the new
  515: domain.  The interface in dom0 will use this address XOR'd with
  516: 00:00:00:01:00:00.  Random MAC addresses are assigned if not given.
  517: 
  518: ## Starting domains automatically
  519: 
  520: To start domains `domU-netbsd` and `domU-linux` at boot and shut them
  521: down cleanly on dom0 shutdown, add the following in rc.conf:
  522: 
  523: [[!template id=filecontent name="/etc/rc.conf" text="""
  524: xendomains="domU-netbsd domU-linux"
  525: """]]
  526: 
  527: # domU setup for specific systems
  528: 
  529: Creating domUs is almost entirely independent of operating system.  We
  530: have already presented the basics of config files in the previous system.
  531: 
  532: Of course, this section presumes that you have a working dom0.
  533: 
  534: ## Creating a NetBSD PV domU
  535: 
  536: See the earlier config file, and adjust memory.  Decide on how much
  537: storage you will provide, and prepare it (file or LVM).
  538: 
  539: While the kernel will be obtained from the dom0 file system, the same
  540: file should be present in the domU as /netbsd so that tools like
  541: savecore(8) can work.   (This is helpful but not necessary.)
  542: 
  543: The kernel must be specifically built for Xen, to use PV interfacesas
  544: a domU.  NetBSD release builds provide the following kernels:
  545: 
  546:         i386 XEN3PAE_DOMU
  547:         amd64 XEN3_DOMU
  548: 
  549: This will boot NetBSD, but this is not that useful if the disk is
  550: empty.  One approach is to unpack sets onto the disk outside of Xen
  551: (by mounting it, just as you would prepare a physical disk for a
  552: system you can't run the installer on).
  553: 
  554: A second approach is to run an INSTALL kernel, which has a miniroot
  555: and can load sets from the network.  To do this, copy the INSTALL
  556: kernel to / and change the kernel line in the config file to:
  557: 
  558:         kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
  559: 
  560: Then, start the domain as "xl create -c configfile".
  561: 
  562: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
  563: line should be used in the config file.
  564: 
  565:     disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
  566: 
  567: After booting the domain, the option to install via CDROM may be
  568: selected.  The CDROM device should be changed to `xbd1d`.
  569: 
  570: Once done installing, "halt -p" the new domain (don't reboot or halt:
  571: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
  572: config file), switch the config file back to the XEN3_DOMU kernel,
  573: and start the new domain again. Now it should be able to use "root on
  574: xbd0a" and you should have a functional NetBSD domU.
  575: 
  576: TODO: check if this is still accurate.
  577: When the new domain is booting you'll see some warnings about *wscons*
  578: and the pseudo-terminals. These can be fixed by editing the files
  579: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
  580: `/etc/ttys`, except *console*, like this:
  581: 
  582:     console "/usr/libexec/getty Pc"         vt100   on secure
  583:     ttyE0   "/usr/libexec/getty Pc"         vt220   off secure
  584:     ttyE1   "/usr/libexec/getty Pc"         vt220   off secure
  585:     ttyE2   "/usr/libexec/getty Pc"         vt220   off secure
  586:     ttyE3   "/usr/libexec/getty Pc"         vt220   off secure
  587: 
  588: Finally, all screens must be commented out from `/etc/wscons.conf`.
  589: 
  590: One should also run `powerd` in a domU, but this should not need
  591: configuring.  With powerd, the domain will run a controlled shutdown
  592: if `xl shutdown -R` or `xl shutdown -H` is used on the dom0, via
  593: receiving a synthetic `power button pressed` signal.  In 9 and
  594: current, `powerd` is run by default under Xen kernels (or if ACPI is
  595: present), and it can be added to rc.conf if not.
  596: 
  597: It is not strictly necessary to have a kernel (as /netbsd) in the domU
  598: file system.  However, various programs (e.g. netstat) will use that
  599: kernel to look up symbols to read from kernel virtual memory.  If
  600: /netbsd is not the running kernel, those lookups will fail.  (This is
  601: not really a Xen-specific issue, but because the domU kernel is
  602: obtained from the dom0, it is far more likely to be out of sync or
  603: missing with Xen.)
  604: 
  605: Note that NetBSD by default creates only xbd[0123].  If you need more
  606: virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
  607: 
  608: ## Creating a Linux PV domU
  609: 
  610: Creating unprivileged Linux domains isn't much different from
  611: unprivileged NetBSD domains, but there are some details to know.
  612: 
  613: First, the second parameter passed to the disk declaration (the '0x1' in
  614: the example below)
  615: 
  616:     disk = [ 'phy:/dev/wd0e,0x1,w' ]
  617: 
  618: does matter to Linux. It wants a Linux device number here (e.g. 0x300
  619: for hda).  Linux builds device numbers as: (major \<\< 8 + minor).
  620: So, hda1 which has major 3 and minor 1 on a Linux system will have
  621: device number 0x301.  Alternatively, devices names can be used (hda,
  622: hdb, ...)  as xentools has a table to map these names to devices
  623: numbers.  To export a partition to a Linux guest we can use:
  624: 
  625:         disk = [ 'phy:/dev/wd0e,0x300,w' ]
  626:         root = "/dev/hda1 ro"
  627: 
  628: and it will appear as /dev/hda on the Linux system, and be used as root
  629: partition.
  630: 
  631: To install the Linux system on the partition to be exported to the
  632: guest domain, the following method can be used: install
  633: sysutils/e2fsprogs from pkgsrc.  Use mke2fs to format the partition
  634: that will be the root partition of your Linux domain, and mount it.
  635: Then copy the files from a working Linux system, make adjustments in
  636: `/etc` (fstab, network config).  It should also be possible to extract
  637: binary packages such as .rpm or .deb directly to the mounted partition
  638: using the appropriate tool, possibly running under NetBSD's Linux
  639: emulation.  Once the file system has been populated, umount it.  If
  640: desirable, the file system can be converted to ext3 using tune2fs -j.
  641: It should now be possible to boot the Linux guest domain, using one of
  642: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
  643: 
  644: To get the Linux console right, you need to add:
  645: 
  646:     extra = "xencons=tty1"
  647: 
  648: to your configuration since not all Linux distributions auto-attach a
  649: tty to the xen console.
  650: 
  651: ## Creating a NetBSD HVM domU
  652: 
  653: Use type='hvm', probably.  Use a GENERIC kernel within the disk image.
  654: 
  655: ## Creating a NetBSD PVH domU
  656: 
  657: This only works with a current kernel in the domU.
  658: 
  659: Use type='pvh'.  Probably, use a GENERIC kernel within the disk image,
  660: which in current has PV support.
  661: 
  662: \todo Verify.
  663: 
  664: \todo Verify if one can have current PVH domU on a 9 dom0.
  665: 
  666: ## Creating a Solaris domU
  667: 
  668: See possibly outdated
  669: [Solaris domU instructions](/ports/xen/howto-solaris/).
  670: 
  671: ## PCI passthrough: Using PCI devices in guest domains
  672: 
  673: NB: PCI passthrough only works on some Xen versions and as of 2020 it
  674: is not clear that it works on any version in pkgsrc.  \todo Reports
  675: confirming or denying this notion should be sent to port-xen@.
  676: 
  677: The dom0 can give other domains access to selected PCI
  678: devices. This can allow, for example, a non-privileged domain to have
  679: access to a physical network interface or disk controller.  However,
  680: keep in mind that giving a domain access to a PCI device most likely
  681: will give the domain read/write access to the whole physical memory,
  682: as PCs don't have an IOMMU to restrict memory access to DMA-capable
  683: device.  Also, it's not possible to export ISA devices to non-dom0
  684: domains, which means that the primary VGA adapter can't be exported.
  685: A guest domain trying to access the VGA registers will panic.
  686: 
  687: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
  688: not been ported to later versions at this time.
  689: 
  690: For a PCI device to be exported to a domU, is has to be attached to
  691: the "pciback" driver in dom0.  Devices passed to the dom0 via the
  692: pciback.hide boot parameter will attach to "pciback" instead of the
  693: usual driver.  The list of devices is specified as "(bus:dev.func)",
  694: where bus and dev are 2-digit hexadecimal numbers, and func a
  695: single-digit number:
  696: 
  697:         pciback.hide=(00:0a.0)(00:06.0)
  698: 
  699: pciback devices should show up in the dom0's boot messages, and the
  700: devices should be listed in the `/kern/xen/pci` directory.
  701: 
  702: PCI devices to be exported to a domU are listed in the "pci" array of
  703: the domU's config file, with the format "0000:bus:dev.func".
  704: 
  705:         pci = [ '0000:00:06.0', '0000:00:0a.0' ]
  706: 
  707: In the domU an "xpci" device will show up, to which one or more pci
  708: buses will attach.  Then the PCI drivers will attach to PCI buses as
  709: usual.  Note that the default NetBSD DOMU kernels do not have "xpci"
  710: or any PCI drivers built in by default; you have to build your own
  711: kernel to use PCI devices in a domU.  Here's a kernel config example;
  712: note that only the "xpci" lines are unusual.
  713: 
  714:         include         "arch/i386/conf/XEN3_DOMU"
  715: 
  716:         # Add support for PCI buses to the XEN3_DOMU kernel
  717:         xpci* at xenbus ?
  718:         pci* at xpci ?
  719: 
  720:         # PCI USB controllers
  721:         uhci*   at pci? dev ? function ?        # Universal Host Controller (Intel)
  722: 
  723:         # USB bus support
  724:         usb*    at uhci?
  725: 
  726:         # USB Hubs
  727:         uhub*   at usb?
  728:         uhub*   at uhub? port ? configuration ? interface ?
  729: 
  730:         # USB Mass Storage
  731:         umass*  at uhub? port ? configuration ? interface ?
  732:         wd*     at umass?
  733:         # SCSI controllers
  734:         ahc*    at pci? dev ? function ?        # Adaptec [23]94x, aic78x0 SCSI
  735: 
  736:         # SCSI bus support (for both ahc and umass)
  737:         scsibus* at scsi?
  738: 
  739:         # SCSI devices
  740:         sd*     at scsibus? target ? lun ?      # SCSI disk drives
  741:         cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
  742: 
  743: 
  744: # Miscellaneous Information
  745: 
  746: ## Nesting under Linux KVM
  747: 
  748: It is possible to run a Xen and a NetBSD dom0 under Linux KVM.  One
  749: can enable virtio in the dom0 for greater speed.
  750: 
  751: ## Other nesting
  752: 
  753: In theory, any full emulation should be able to run Xen and a NetBSD
  754: dom0.  The HOWTO does not currently have information about Xen XVM
  755: mode, nvmm, qemu, Virtualbox, etc.
  756: 
  757: ## NetBSD 5 as domU
  758: 
  759: [NetBSD 5 is known to panic.](http://mail-index.netbsd.org/port-xen/2018/04/17/msg009181.html)
  760: (However, NetBSD 5 systems should be updated to a supported version.)
  761: 
  762: # NetBSD as a domU in a VPS
  763: 
  764: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
  765: hardware.  This section explains how to deal with Xen in a domU as a
  766: virtual private server where you do not control or have access to the
  767: dom0.  This is not intended to be an exhaustive list of VPS providers;
  768: only a few are mentioned that specifically support NetBSD.
  769: 
  770: VPS operators provide varying degrees of access and mechanisms for
  771: configuration.  The big issue is usually how one controls which kernel
  772: is booted, because the kernel is nominally in the dom0 file system (to
  773: which VPS users do not normally have access).  A second issue is how
  774: to install NetBSD.
  775: A VPS user may want to compile a kernel for security updates, to run
  776: npf, run IPsec, or any other reason why someone would want to change
  777: their kernel.
  778: 
  779: One approach is to have an administrative interface to upload a kernel,
  780: or to select from a prepopulated list.  Other approaches are pygrub
  781: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
  782: kernel from the domU file system.  This is closer to a regular physical
  783: computer, where someone who controls a machine can replace the kernel.
  784: 
  785: A second issue is multiple CPUs.  With NetBSD 6, domUs support
  786: multiple vcpus, and it is typical for VPS providers to enable multiple
  787: CPUs for NetBSD domUs.
  788: 
  789: ## Complexities due to Xen changes
  790: 
  791: Xen has many security advisories and people running Xen systems make
  792: different choices.
  793: 
  794: ### stub domains
  795: 
  796: Some (Linux only?) dom0 systems use something called "stub domains" to
  797: isolate qemu from the dom0 system, as a security and reliabilty
  798: mechanism when running HVM domUs.  Somehow, NetBSD's GENERIC kernel
  799: ends up using PIO for disks rather than DMA.  Of course, all of this
  800: is emulated, but emulated PIO is unusably slow.  This problem is not
  801: currently understood.
  802: 
  803: ### Grant tables
  804: 
  805: There are multiple versions of using grant tables, and some security
  806: advisories have suggested disabling some versions.  Some versions of
  807: NetBSD apparently only use specific versions and this can lead to
  808: "NetBSD current doesn't run on hosting provider X" situations.
  809: 
  810: \todo Explain better.
  811: 
  812: ## Boot methods
  813: 
  814: ### pvgrub
  815: 
  816: pvgrub is a version of grub that uses PV operations instead of BIOS
  817: calls.  It is booted from the dom0 as the domU kernel, and then reads
  818: /grub/menu.lst and loads a kernel from the domU file system.
  819: 
  820: [Panix](http://www.panix.com/) lets users use pvgrub.  Panix reports
  821: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
  822: (and hence with defaults from "newfs -O 2").  See [Panix's pvgrub
  823: page](http://www.panix.com/v-colo/grub.html), which describes only
  824: Linux but should be updated to cover NetBSD :-).
  825: 
  826: [prgmr.com](http://prgmr.com/) also lets users with pvgrub to boot
  827: their own kernel.  See then [prgmr.com NetBSD
  828: HOWTO](http://wiki.prgmr.com/mediawiki/index.php/NetBSD_as_a_DomU)
  829: (which is in need of updating).
  830: 
  831: It appears that [grub's FFS
  832: code](http://xenbits.xensource.com/hg/xen-unstable.hg/file/bca284f67702/tools/libfsimage/ufs/fsys_ufs.c)
  833: does not support all aspects of modern FFS, but there are also reports
  834: that FFSv2 works fine.  At prgmr, typically one has an ext2 or FAT
  835: partition for the kernel with the intent that grub can understand it,
  836: which leads to /netbsd not being the actual kernel.  One must remember
  837: to update the special boot partition.
  838: 
  839: ### pygrub
  840: 
  841: pygrub runs in the dom0 and looks into the domU file system.  This
  842: implies that the domU must have a kernel in a file system in a format
  843: known to pygrub.
  844: 
  845: pygrub doesn't seem to work to load Linux images under NetBSD dom0,
  846: and is inherently less secure than pvgrub due to running inside dom0. For both these
  847: reasons, pygrub should not be used, and is only still present so that
  848: historical DomU images using it still work.
  849: 
  850: As of 2014, pygrub seems to be of mostly historical
  851: interest. New DomUs should use pvgrub.
  852: 
  853: ## Specific Providers
  854: 
  855: ### Amazon
  856: 
  857: See the [Amazon EC2 page](/amazon_ec2/).

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