1 \input texinfo @c -*- texinfo -*-
3 @setfilename qemu-doc.info
7 @documentencoding UTF-8
9 @settitle QEMU version @value{VERSION} User Documentation
16 * QEMU: (qemu-doc). The QEMU Emulator User Documentation.
23 @center @titlefont{QEMU version @value{VERSION}}
25 @center @titlefont{User Documentation}
36 * QEMU PC System emulator::
37 * QEMU System emulator for non PC targets::
39 * QEMU User space emulator::
40 * Implementation notes::
41 * Deprecated features::
42 * Supported build platforms::
54 * intro_features:: Features
60 QEMU is a FAST! processor emulator using dynamic translation to
61 achieve good emulation speed.
63 @cindex operating modes
64 QEMU has two operating modes:
67 @cindex system emulation
68 @item Full system emulation. In this mode, QEMU emulates a full system (for
69 example a PC), including one or several processors and various
70 peripherals. It can be used to launch different Operating Systems
71 without rebooting the PC or to debug system code.
73 @cindex user mode emulation
74 @item User mode emulation. In this mode, QEMU can launch
75 processes compiled for one CPU on another CPU. It can be used to
76 launch the Wine Windows API emulator (@url{https://www.winehq.org}) or
77 to ease cross-compilation and cross-debugging.
81 QEMU has the following features:
84 @item QEMU can run without a host kernel driver and yet gives acceptable
85 performance. It uses dynamic translation to native code for reasonable speed,
86 with support for self-modifying code and precise exceptions.
88 @item It is portable to several operating systems (GNU/Linux, *BSD, Mac OS X,
89 Windows) and architectures.
91 @item It performs accurate software emulation of the FPU.
94 QEMU user mode emulation has the following features:
96 @item Generic Linux system call converter, including most ioctls.
98 @item clone() emulation using native CPU clone() to use Linux scheduler for threads.
100 @item Accurate signal handling by remapping host signals to target signals.
103 QEMU full system emulation has the following features:
106 QEMU uses a full software MMU for maximum portability.
109 QEMU can optionally use an in-kernel accelerator, like kvm. The accelerators
110 execute most of the guest code natively, while
111 continuing to emulate the rest of the machine.
114 Various hardware devices can be emulated and in some cases, host
115 devices (e.g. serial and parallel ports, USB, drives) can be used
116 transparently by the guest Operating System. Host device passthrough
117 can be used for talking to external physical peripherals (e.g. a
118 webcam, modem or tape drive).
121 Symmetric multiprocessing (SMP) support. Currently, an in-kernel
122 accelerator is required to use more than one host CPU for emulation.
127 @node QEMU PC System emulator
128 @chapter QEMU PC System emulator
129 @cindex system emulation (PC)
132 * pcsys_introduction:: Introduction
133 * pcsys_quickstart:: Quick Start
134 * sec_invocation:: Invocation
135 * pcsys_keys:: Keys in the graphical frontends
136 * mux_keys:: Keys in the character backend multiplexer
137 * pcsys_monitor:: QEMU Monitor
138 * disk_images:: Disk Images
139 * pcsys_network:: Network emulation
140 * pcsys_other_devs:: Other Devices
141 * direct_linux_boot:: Direct Linux Boot
142 * pcsys_usb:: USB emulation
143 * vnc_security:: VNC security
144 * network_tls:: TLS setup for network services
145 * gdb_usage:: GDB usage
146 * pcsys_os_specific:: Target OS specific information
149 @node pcsys_introduction
150 @section Introduction
152 @c man begin DESCRIPTION
154 The QEMU PC System emulator simulates the
155 following peripherals:
159 i440FX host PCI bridge and PIIX3 PCI to ISA bridge
161 Cirrus CLGD 5446 PCI VGA card or dummy VGA card with Bochs VESA
162 extensions (hardware level, including all non standard modes).
164 PS/2 mouse and keyboard
166 2 PCI IDE interfaces with hard disk and CD-ROM support
170 PCI and ISA network adapters
174 IPMI BMC, either and internal or external one
176 Creative SoundBlaster 16 sound card
178 ENSONIQ AudioPCI ES1370 sound card
180 Intel 82801AA AC97 Audio compatible sound card
182 Intel HD Audio Controller and HDA codec
184 Adlib (OPL2) - Yamaha YM3812 compatible chip
186 Gravis Ultrasound GF1 sound card
188 CS4231A compatible sound card
190 PCI UHCI, OHCI, EHCI or XHCI USB controller and a virtual USB-1.1 hub.
193 SMP is supported with up to 255 CPUs.
195 QEMU uses the PC BIOS from the Seabios project and the Plex86/Bochs LGPL
198 QEMU uses YM3812 emulation by Tatsuyuki Satoh.
200 QEMU uses GUS emulation (GUSEMU32 @url{http://www.deinmeister.de/gusemu/})
201 by Tibor "TS" Schütz.
203 Note that, by default, GUS shares IRQ(7) with parallel ports and so
204 QEMU must be told to not have parallel ports to have working GUS.
207 qemu-system-i386 dos.img -soundhw gus -parallel none
212 qemu-system-i386 dos.img -device gus,irq=5
215 Or some other unclaimed IRQ.
217 CS4231A is the chip used in Windows Sound System and GUSMAX products
221 @node pcsys_quickstart
225 Download and uncompress the linux image (@file{linux.img}) and type:
228 qemu-system-i386 linux.img
231 Linux should boot and give you a prompt.
237 @c man begin SYNOPSIS
238 @command{qemu-system-i386} [@var{options}] [@var{disk_image}]
243 @var{disk_image} is a raw hard disk image for IDE hard disk 0. Some
244 targets do not need a disk image.
246 @include qemu-options.texi
250 @subsection Device URL Syntax
251 @c TODO merge this with section Disk Images
255 In addition to using normal file images for the emulated storage devices,
256 QEMU can also use networked resources such as iSCSI devices. These are
257 specified using a special URL syntax.
261 iSCSI support allows QEMU to access iSCSI resources directly and use as
262 images for the guest storage. Both disk and cdrom images are supported.
264 Syntax for specifying iSCSI LUNs is
265 ``iscsi://<target-ip>[:<port>]/<target-iqn>/<lun>''
267 By default qemu will use the iSCSI initiator-name
268 'iqn.2008-11.org.linux-kvm[:<name>]' but this can also be set from the command
269 line or a configuration file.
271 Since version Qemu 2.4 it is possible to specify a iSCSI request timeout to detect
272 stalled requests and force a reestablishment of the session. The timeout
273 is specified in seconds. The default is 0 which means no timeout. Libiscsi
274 1.15.0 or greater is required for this feature.
276 Example (without authentication):
278 qemu-system-i386 -iscsi initiator-name=iqn.2001-04.com.example:my-initiator \
279 -cdrom iscsi://192.0.2.1/iqn.2001-04.com.example/2 \
280 -drive file=iscsi://192.0.2.1/iqn.2001-04.com.example/1
283 Example (CHAP username/password via URL):
285 qemu-system-i386 -drive file=iscsi://user%password@@192.0.2.1/iqn.2001-04.com.example/1
288 Example (CHAP username/password via environment variables):
290 LIBISCSI_CHAP_USERNAME="user" \
291 LIBISCSI_CHAP_PASSWORD="password" \
292 qemu-system-i386 -drive file=iscsi://192.0.2.1/iqn.2001-04.com.example/1
296 QEMU supports NBD (Network Block Devices) both using TCP protocol as well
297 as Unix Domain Sockets.
299 Syntax for specifying a NBD device using TCP
300 ``nbd:<server-ip>:<port>[:exportname=<export>]''
302 Syntax for specifying a NBD device using Unix Domain Sockets
303 ``nbd:unix:<domain-socket>[:exportname=<export>]''
307 qemu-system-i386 --drive file=nbd:192.0.2.1:30000
310 Example for Unix Domain Sockets
312 qemu-system-i386 --drive file=nbd:unix:/tmp/nbd-socket
316 QEMU supports SSH (Secure Shell) access to remote disks.
320 qemu-system-i386 -drive file=ssh://user@@host/path/to/disk.img
321 qemu-system-i386 -drive file.driver=ssh,file.user=user,file.host=host,file.port=22,file.path=/path/to/disk.img
324 Currently authentication must be done using ssh-agent. Other
325 authentication methods may be supported in future.
328 Sheepdog is a distributed storage system for QEMU.
329 QEMU supports using either local sheepdog devices or remote networked
332 Syntax for specifying a sheepdog device
334 sheepdog[+tcp|+unix]://[host:port]/vdiname[?socket=path][#snapid|#tag]
339 qemu-system-i386 --drive file=sheepdog://192.0.2.1:30000/MyVirtualMachine
342 See also @url{https://sheepdog.github.io/sheepdog/}.
345 GlusterFS is a user space distributed file system.
346 QEMU supports the use of GlusterFS volumes for hosting VM disk images using
347 TCP, Unix Domain Sockets and RDMA transport protocols.
349 Syntax for specifying a VM disk image on GlusterFS volume is
353 gluster[+type]://[host[:port]]/volume/path[?socket=...][,debug=N][,logfile=...]
356 'json:@{"driver":"qcow2","file":@{"driver":"gluster","volume":"testvol","path":"a.img","debug":N,"logfile":"...",
357 @ "server":[@{"type":"tcp","host":"...","port":"..."@},
358 @ @{"type":"unix","socket":"..."@}]@}@}'
365 qemu-system-x86_64 --drive file=gluster://192.0.2.1/testvol/a.img,
366 @ file.debug=9,file.logfile=/var/log/qemu-gluster.log
369 qemu-system-x86_64 'json:@{"driver":"qcow2",
370 @ "file":@{"driver":"gluster",
371 @ "volume":"testvol","path":"a.img",
372 @ "debug":9,"logfile":"/var/log/qemu-gluster.log",
373 @ "server":[@{"type":"tcp","host":"1.2.3.4","port":24007@},
374 @ @{"type":"unix","socket":"/var/run/glusterd.socket"@}]@}@}'
375 qemu-system-x86_64 -drive driver=qcow2,file.driver=gluster,file.volume=testvol,file.path=/path/a.img,
376 @ file.debug=9,file.logfile=/var/log/qemu-gluster.log,
377 @ file.server.0.type=tcp,file.server.0.host=1.2.3.4,file.server.0.port=24007,
378 @ file.server.1.type=unix,file.server.1.socket=/var/run/glusterd.socket
381 See also @url{http://www.gluster.org}.
383 @item HTTP/HTTPS/FTP/FTPS
384 QEMU supports read-only access to files accessed over http(s) and ftp(s).
386 Syntax using a single filename:
388 <protocol>://[<username>[:<password>]@@]<host>/<path>
394 'http', 'https', 'ftp', or 'ftps'.
397 Optional username for authentication to the remote server.
400 Optional password for authentication to the remote server.
403 Address of the remote server.
406 Path on the remote server, including any query string.
409 The following options are also supported:
412 The full URL when passing options to the driver explicitly.
415 The amount of data to read ahead with each range request to the remote server.
416 This value may optionally have the suffix 'T', 'G', 'M', 'K', 'k' or 'b'. If it
417 does not have a suffix, it will be assumed to be in bytes. The value must be a
418 multiple of 512 bytes. It defaults to 256k.
421 Whether to verify the remote server's certificate when connecting over SSL. It
422 can have the value 'on' or 'off'. It defaults to 'on'.
425 Send this cookie (it can also be a list of cookies separated by ';') with
426 each outgoing request. Only supported when using protocols such as HTTP
427 which support cookies, otherwise ignored.
430 Set the timeout in seconds of the CURL connection. This timeout is the time
431 that CURL waits for a response from the remote server to get the size of the
432 image to be downloaded. If not set, the default timeout of 5 seconds is used.
435 Note that when passing options to qemu explicitly, @option{driver} is the value
438 Example: boot from a remote Fedora 20 live ISO image
440 qemu-system-x86_64 --drive media=cdrom,file=http://dl.fedoraproject.org/pub/fedora/linux/releases/20/Live/x86_64/Fedora-Live-Desktop-x86_64-20-1.iso,readonly
442 qemu-system-x86_64 --drive media=cdrom,file.driver=http,file.url=http://dl.fedoraproject.org/pub/fedora/linux/releases/20/Live/x86_64/Fedora-Live-Desktop-x86_64-20-1.iso,readonly
445 Example: boot from a remote Fedora 20 cloud image using a local overlay for
446 writes, copy-on-read, and a readahead of 64k
448 qemu-img create -f qcow2 -o backing_file='json:@{"file.driver":"http",, "file.url":"https://dl.fedoraproject.org/pub/fedora/linux/releases/20/Images/x86_64/Fedora-x86_64-20-20131211.1-sda.qcow2",, "file.readahead":"64k"@}' /tmp/Fedora-x86_64-20-20131211.1-sda.qcow2
450 qemu-system-x86_64 -drive file=/tmp/Fedora-x86_64-20-20131211.1-sda.qcow2,copy-on-read=on
453 Example: boot from an image stored on a VMware vSphere server with a self-signed
454 certificate using a local overlay for writes, a readahead of 64k and a timeout
457 qemu-img create -f qcow2 -o backing_file='json:@{"file.driver":"https",, "file.url":"https://user:password@@vsphere.example.com/folder/test/test-flat.vmdk?dcPath=Datacenter&dsName=datastore1",, "file.sslverify":"off",, "file.readahead":"64k",, "file.timeout":10@}' /tmp/test.qcow2
459 qemu-system-x86_64 -drive file=/tmp/test.qcow2
467 @section Keys in the graphical frontends
471 During the graphical emulation, you can use special key combinations to change
472 modes. The default key mappings are shown below, but if you use @code{-alt-grab}
473 then the modifier is Ctrl-Alt-Shift (instead of Ctrl-Alt) and if you use
474 @code{-ctrl-grab} then the modifier is the right Ctrl key (instead of Ctrl-Alt):
491 Restore the screen's un-scaled dimensions
495 Switch to virtual console 'n'. Standard console mappings are:
498 Target system display
507 Toggle mouse and keyboard grab.
513 @kindex Ctrl-PageDown
514 In the virtual consoles, you can use @key{Ctrl-Up}, @key{Ctrl-Down},
515 @key{Ctrl-PageUp} and @key{Ctrl-PageDown} to move in the back log.
520 @section Keys in the character backend multiplexer
524 During emulation, if you are using a character backend multiplexer
525 (which is the default if you are using @option{-nographic}) then
526 several commands are available via an escape sequence. These
527 key sequences all start with an escape character, which is @key{Ctrl-a}
528 by default, but can be changed with @option{-echr}. The list below assumes
529 you're using the default.
540 Save disk data back to file (if -snapshot)
543 Toggle console timestamps
546 Send break (magic sysrq in Linux)
549 Rotate between the frontends connected to the multiplexer (usually
550 this switches between the monitor and the console)
552 @kindex Ctrl-a Ctrl-a
553 Send the escape character to the frontend
560 The HTML documentation of QEMU for more precise information and Linux
561 user mode emulator invocation.
571 @section QEMU Monitor
574 The QEMU monitor is used to give complex commands to the QEMU
575 emulator. You can use it to:
580 Remove or insert removable media images
581 (such as CD-ROM or floppies).
584 Freeze/unfreeze the Virtual Machine (VM) and save or restore its state
587 @item Inspect the VM state without an external debugger.
593 The following commands are available:
595 @include qemu-monitor.texi
597 @include qemu-monitor-info.texi
599 @subsection Integer expressions
601 The monitor understands integers expressions for every integer
602 argument. You can use register names to get the value of specifics
603 CPU registers by prefixing them with @emph{$}.
608 QEMU supports many disk image formats, including growable disk images
609 (their size increase as non empty sectors are written), compressed and
610 encrypted disk images.
613 * disk_images_quickstart:: Quick start for disk image creation
614 * disk_images_snapshot_mode:: Snapshot mode
615 * vm_snapshots:: VM snapshots
616 * qemu_img_invocation:: qemu-img Invocation
617 * qemu_nbd_invocation:: qemu-nbd Invocation
618 * disk_images_formats:: Disk image file formats
619 * host_drives:: Using host drives
620 * disk_images_fat_images:: Virtual FAT disk images
621 * disk_images_nbd:: NBD access
622 * disk_images_sheepdog:: Sheepdog disk images
623 * disk_images_iscsi:: iSCSI LUNs
624 * disk_images_gluster:: GlusterFS disk images
625 * disk_images_ssh:: Secure Shell (ssh) disk images
626 * disk_images_nvme:: NVMe userspace driver
627 * disk_image_locking:: Disk image file locking
630 @node disk_images_quickstart
631 @subsection Quick start for disk image creation
633 You can create a disk image with the command:
635 qemu-img create myimage.img mysize
637 where @var{myimage.img} is the disk image filename and @var{mysize} is its
638 size in kilobytes. You can add an @code{M} suffix to give the size in
639 megabytes and a @code{G} suffix for gigabytes.
641 See @ref{qemu_img_invocation} for more information.
643 @node disk_images_snapshot_mode
644 @subsection Snapshot mode
646 If you use the option @option{-snapshot}, all disk images are
647 considered as read only. When sectors in written, they are written in
648 a temporary file created in @file{/tmp}. You can however force the
649 write back to the raw disk images by using the @code{commit} monitor
650 command (or @key{C-a s} in the serial console).
653 @subsection VM snapshots
655 VM snapshots are snapshots of the complete virtual machine including
656 CPU state, RAM, device state and the content of all the writable
657 disks. In order to use VM snapshots, you must have at least one non
658 removable and writable block device using the @code{qcow2} disk image
659 format. Normally this device is the first virtual hard drive.
661 Use the monitor command @code{savevm} to create a new VM snapshot or
662 replace an existing one. A human readable name can be assigned to each
663 snapshot in addition to its numerical ID.
665 Use @code{loadvm} to restore a VM snapshot and @code{delvm} to remove
666 a VM snapshot. @code{info snapshots} lists the available snapshots
667 with their associated information:
670 (qemu) info snapshots
671 Snapshot devices: hda
672 Snapshot list (from hda):
673 ID TAG VM SIZE DATE VM CLOCK
674 1 start 41M 2006-08-06 12:38:02 00:00:14.954
675 2 40M 2006-08-06 12:43:29 00:00:18.633
676 3 msys 40M 2006-08-06 12:44:04 00:00:23.514
679 A VM snapshot is made of a VM state info (its size is shown in
680 @code{info snapshots}) and a snapshot of every writable disk image.
681 The VM state info is stored in the first @code{qcow2} non removable
682 and writable block device. The disk image snapshots are stored in
683 every disk image. The size of a snapshot in a disk image is difficult
684 to evaluate and is not shown by @code{info snapshots} because the
685 associated disk sectors are shared among all the snapshots to save
686 disk space (otherwise each snapshot would need a full copy of all the
689 When using the (unrelated) @code{-snapshot} option
690 (@ref{disk_images_snapshot_mode}), you can always make VM snapshots,
691 but they are deleted as soon as you exit QEMU.
693 VM snapshots currently have the following known limitations:
696 They cannot cope with removable devices if they are removed or
697 inserted after a snapshot is done.
699 A few device drivers still have incomplete snapshot support so their
700 state is not saved or restored properly (in particular USB).
703 @node qemu_img_invocation
704 @subsection @code{qemu-img} Invocation
706 @include qemu-img.texi
708 @node qemu_nbd_invocation
709 @subsection @code{qemu-nbd} Invocation
711 @include qemu-nbd.texi
713 @include docs/qemu-block-drivers.texi
716 @section Network emulation
718 QEMU can simulate several network cards (e.g. PCI or ISA cards on the PC
719 target) and can connect them to a network backend on the host or an emulated
720 hub. The various host network backends can either be used to connect the NIC of
721 the guest to a real network (e.g. by using a TAP devices or the non-privileged
722 user mode network stack), or to other guest instances running in another QEMU
723 process (e.g. by using the socket host network backend).
725 @subsection Using TAP network interfaces
727 This is the standard way to connect QEMU to a real network. QEMU adds
728 a virtual network device on your host (called @code{tapN}), and you
729 can then configure it as if it was a real ethernet card.
731 @subsubsection Linux host
733 As an example, you can download the @file{linux-test-xxx.tar.gz}
734 archive and copy the script @file{qemu-ifup} in @file{/etc} and
735 configure properly @code{sudo} so that the command @code{ifconfig}
736 contained in @file{qemu-ifup} can be executed as root. You must verify
737 that your host kernel supports the TAP network interfaces: the
738 device @file{/dev/net/tun} must be present.
740 See @ref{sec_invocation} to have examples of command lines using the
741 TAP network interfaces.
743 @subsubsection Windows host
745 There is a virtual ethernet driver for Windows 2000/XP systems, called
746 TAP-Win32. But it is not included in standard QEMU for Windows,
747 so you will need to get it separately. It is part of OpenVPN package,
748 so download OpenVPN from : @url{https://openvpn.net/}.
750 @subsection Using the user mode network stack
752 By using the option @option{-net user} (default configuration if no
753 @option{-net} option is specified), QEMU uses a completely user mode
754 network stack (you don't need root privilege to use the virtual
755 network). The virtual network configuration is the following:
759 guest (10.0.2.15) <------> Firewall/DHCP server <-----> Internet
762 ----> DNS server (10.0.2.3)
764 ----> SMB server (10.0.2.4)
767 The QEMU VM behaves as if it was behind a firewall which blocks all
768 incoming connections. You can use a DHCP client to automatically
769 configure the network in the QEMU VM. The DHCP server assign addresses
770 to the hosts starting from 10.0.2.15.
772 In order to check that the user mode network is working, you can ping
773 the address 10.0.2.2 and verify that you got an address in the range
774 10.0.2.x from the QEMU virtual DHCP server.
776 Note that ICMP traffic in general does not work with user mode networking.
777 @code{ping}, aka. ICMP echo, to the local router (10.0.2.2) shall work,
778 however. If you're using QEMU on Linux >= 3.0, it can use unprivileged ICMP
779 ping sockets to allow @code{ping} to the Internet. The host admin has to set
780 the ping_group_range in order to grant access to those sockets. To allow ping
781 for GID 100 (usually users group):
784 echo 100 100 > /proc/sys/net/ipv4/ping_group_range
787 When using the built-in TFTP server, the router is also the TFTP
790 When using the @option{'-netdev user,hostfwd=...'} option, TCP or UDP
791 connections can be redirected from the host to the guest. It allows for
792 example to redirect X11, telnet or SSH connections.
796 QEMU can simulate several hubs. A hub can be thought of as a virtual connection
797 between several network devices. These devices can be for example QEMU virtual
798 ethernet cards or virtual Host ethernet devices (TAP devices). You can connect
799 guest NICs or host network backends to such a hub using the @option{-netdev
800 hubport} or @option{-nic hubport} options. The legacy @option{-net} option
801 also connects the given device to the emulated hub with ID 0 (i.e. the default
802 hub) unless you specify a netdev with @option{-net nic,netdev=xxx} here.
804 @subsection Connecting emulated networks between QEMU instances
806 Using the @option{-netdev socket} (or @option{-nic socket} or
807 @option{-net socket}) option, it is possible to create emulated
808 networks that span several QEMU instances.
809 See the description of the @option{-netdev socket} option in the
810 @ref{sec_invocation,,Invocation chapter} to have a basic example.
812 @node pcsys_other_devs
813 @section Other Devices
815 @subsection Inter-VM Shared Memory device
817 On Linux hosts, a shared memory device is available. The basic syntax
821 qemu-system-x86_64 -device ivshmem-plain,memdev=@var{hostmem}
824 where @var{hostmem} names a host memory backend. For a POSIX shared
825 memory backend, use something like
828 -object memory-backend-file,size=1M,share,mem-path=/dev/shm/ivshmem,id=@var{hostmem}
831 If desired, interrupts can be sent between guest VMs accessing the same shared
832 memory region. Interrupt support requires using a shared memory server and
833 using a chardev socket to connect to it. The code for the shared memory server
834 is qemu.git/contrib/ivshmem-server. An example syntax when using the shared
838 # First start the ivshmem server once and for all
839 ivshmem-server -p @var{pidfile} -S @var{path} -m @var{shm-name} -l @var{shm-size} -n @var{vectors}
841 # Then start your qemu instances with matching arguments
842 qemu-system-x86_64 -device ivshmem-doorbell,vectors=@var{vectors},chardev=@var{id}
843 -chardev socket,path=@var{path},id=@var{id}
846 When using the server, the guest will be assigned a VM ID (>=0) that allows guests
847 using the same server to communicate via interrupts. Guests can read their
848 VM ID from a device register (see ivshmem-spec.txt).
850 @subsubsection Migration with ivshmem
852 With device property @option{master=on}, the guest will copy the shared
853 memory on migration to the destination host. With @option{master=off},
854 the guest will not be able to migrate with the device attached. In the
855 latter case, the device should be detached and then reattached after
856 migration using the PCI hotplug support.
858 At most one of the devices sharing the same memory can be master. The
859 master must complete migration before you plug back the other devices.
861 @subsubsection ivshmem and hugepages
863 Instead of specifying the <shm size> using POSIX shm, you may specify
864 a memory backend that has hugepage support:
867 qemu-system-x86_64 -object memory-backend-file,size=1G,mem-path=/dev/hugepages/my-shmem-file,share,id=mb1
868 -device ivshmem-plain,memdev=mb1
871 ivshmem-server also supports hugepages mount points with the
872 @option{-m} memory path argument.
874 @node direct_linux_boot
875 @section Direct Linux Boot
877 This section explains how to launch a Linux kernel inside QEMU without
878 having to make a full bootable image. It is very useful for fast Linux
883 qemu-system-i386 -kernel arch/i386/boot/bzImage -hda root-2.4.20.img -append "root=/dev/hda"
886 Use @option{-kernel} to provide the Linux kernel image and
887 @option{-append} to give the kernel command line arguments. The
888 @option{-initrd} option can be used to provide an INITRD image.
890 When using the direct Linux boot, a disk image for the first hard disk
891 @file{hda} is required because its boot sector is used to launch the
894 If you do not need graphical output, you can disable it and redirect
895 the virtual serial port and the QEMU monitor to the console with the
896 @option{-nographic} option. The typical command line is:
898 qemu-system-i386 -kernel arch/i386/boot/bzImage -hda root-2.4.20.img \
899 -append "root=/dev/hda console=ttyS0" -nographic
902 Use @key{Ctrl-a c} to switch between the serial console and the
903 monitor (@pxref{pcsys_keys}).
906 @section USB emulation
908 QEMU can emulate a PCI UHCI, OHCI, EHCI or XHCI USB controller. You can
909 plug virtual USB devices or real host USB devices (only works with certain
910 host operating systems). QEMU will automatically create and connect virtual
911 USB hubs as necessary to connect multiple USB devices.
918 @subsection Connecting USB devices
920 USB devices can be connected with the @option{-device usb-...} command line
921 option or the @code{device_add} monitor command. Available devices are:
925 Virtual Mouse. This will override the PS/2 mouse emulation when activated.
927 Pointer device that uses absolute coordinates (like a touchscreen).
928 This means QEMU is able to report the mouse position without having
929 to grab the mouse. Also overrides the PS/2 mouse emulation when activated.
930 @item usb-storage,drive=@var{drive_id}
931 Mass storage device backed by @var{drive_id} (@pxref{disk_images})
933 USB attached SCSI device, see
934 @url{https://git.qemu.org/?p=qemu.git;a=blob_plain;f=docs/usb-storage.txt,usb-storage.txt}
937 Bulk-only transport storage device, see
938 @url{https://git.qemu.org/?p=qemu.git;a=blob_plain;f=docs/usb-storage.txt,usb-storage.txt}
939 for details here, too
940 @item usb-mtp,x-root=@var{dir}
941 Media transfer protocol device, using @var{dir} as root of the file tree
942 that is presented to the guest.
943 @item usb-host,hostbus=@var{bus},hostaddr=@var{addr}
944 Pass through the host device identified by @var{bus} and @var{addr}
945 @item usb-host,vendorid=@var{vendor},productid=@var{product}
946 Pass through the host device identified by @var{vendor} and @var{product} ID
947 @item usb-wacom-tablet
948 Virtual Wacom PenPartner tablet. This device is similar to the @code{tablet}
949 above but it can be used with the tslib library because in addition to touch
950 coordinates it reports touch pressure.
952 Standard USB keyboard. Will override the PS/2 keyboard (if present).
953 @item usb-serial,chardev=@var{id}
954 Serial converter. This emulates an FTDI FT232BM chip connected to host character
956 @item usb-braille,chardev=@var{id}
957 Braille device. This will use BrlAPI to display the braille output on a real
958 or fake device referenced by @var{id}.
959 @item usb-net[,netdev=@var{id}]
960 Network adapter that supports CDC ethernet and RNDIS protocols. @var{id}
961 specifies a netdev defined with @code{-netdev @dots{},id=@var{id}}.
962 For instance, user-mode networking can be used with
964 qemu-system-i386 [...] -netdev user,id=net0 -device usb-net,netdev=net0
967 Smartcard reader device
971 Bluetooth dongle for the transport layer of HCI. It is connected to HCI
972 scatternet 0 by default (corresponds to @code{-bt hci,vlan=0}).
973 Note that the syntax for the @code{-device usb-bt-dongle} option is not as
974 useful yet as it was with the legacy @code{-usbdevice} option. So to
975 configure an USB bluetooth device, you might need to use
976 "@code{-usbdevice bt}[:@var{hci-type}]" instead. This configures a
977 bluetooth dongle whose type is specified in the same format as with
978 the @option{-bt hci} option, @pxref{bt-hcis,,allowed HCI types}. If
979 no type is given, the HCI logic corresponds to @code{-bt hci,vlan=0}.
980 This USB device implements the USB Transport Layer of HCI. Example
983 @command{qemu-system-i386} [...@var{OPTIONS}...] @option{-usbdevice} bt:hci,vlan=3 @option{-bt} device:keyboard,vlan=3
987 @node host_usb_devices
988 @subsection Using host USB devices on a Linux host
990 WARNING: this is an experimental feature. QEMU will slow down when
991 using it. USB devices requiring real time streaming (i.e. USB Video
992 Cameras) are not supported yet.
995 @item If you use an early Linux 2.4 kernel, verify that no Linux driver
996 is actually using the USB device. A simple way to do that is simply to
997 disable the corresponding kernel module by renaming it from @file{mydriver.o}
998 to @file{mydriver.o.disabled}.
1000 @item Verify that @file{/proc/bus/usb} is working (most Linux distributions should enable it by default). You should see something like that:
1006 @item Since only root can access to the USB devices directly, you can either launch QEMU as root or change the permissions of the USB devices you want to use. For testing, the following suffices:
1008 chown -R myuid /proc/bus/usb
1011 @item Launch QEMU and do in the monitor:
1014 Device 1.2, speed 480 Mb/s
1015 Class 00: USB device 1234:5678, USB DISK
1017 You should see the list of the devices you can use (Never try to use
1018 hubs, it won't work).
1020 @item Add the device in QEMU by using:
1022 device_add usb-host,vendorid=0x1234,productid=0x5678
1025 Normally the guest OS should report that a new USB device is plugged.
1026 You can use the option @option{-device usb-host,...} to do the same.
1028 @item Now you can try to use the host USB device in QEMU.
1032 When relaunching QEMU, you may have to unplug and plug again the USB
1033 device to make it work again (this is a bug).
1036 @section VNC security
1038 The VNC server capability provides access to the graphical console
1039 of the guest VM across the network. This has a number of security
1040 considerations depending on the deployment scenarios.
1044 * vnc_sec_password::
1045 * vnc_sec_certificate::
1046 * vnc_sec_certificate_verify::
1047 * vnc_sec_certificate_pw::
1049 * vnc_sec_certificate_sasl::
1053 @subsection Without passwords
1055 The simplest VNC server setup does not include any form of authentication.
1056 For this setup it is recommended to restrict it to listen on a UNIX domain
1057 socket only. For example
1060 qemu-system-i386 [...OPTIONS...] -vnc unix:/home/joebloggs/.qemu-myvm-vnc
1063 This ensures that only users on local box with read/write access to that
1064 path can access the VNC server. To securely access the VNC server from a
1065 remote machine, a combination of netcat+ssh can be used to provide a secure
1068 @node vnc_sec_password
1069 @subsection With passwords
1071 The VNC protocol has limited support for password based authentication. Since
1072 the protocol limits passwords to 8 characters it should not be considered
1073 to provide high security. The password can be fairly easily brute-forced by
1074 a client making repeat connections. For this reason, a VNC server using password
1075 authentication should be restricted to only listen on the loopback interface
1076 or UNIX domain sockets. Password authentication is not supported when operating
1077 in FIPS 140-2 compliance mode as it requires the use of the DES cipher. Password
1078 authentication is requested with the @code{password} option, and then once QEMU
1079 is running the password is set with the monitor. Until the monitor is used to
1080 set the password all clients will be rejected.
1083 qemu-system-i386 [...OPTIONS...] -vnc :1,password -monitor stdio
1084 (qemu) change vnc password
1089 @node vnc_sec_certificate
1090 @subsection With x509 certificates
1092 The QEMU VNC server also implements the VeNCrypt extension allowing use of
1093 TLS for encryption of the session, and x509 certificates for authentication.
1094 The use of x509 certificates is strongly recommended, because TLS on its
1095 own is susceptible to man-in-the-middle attacks. Basic x509 certificate
1096 support provides a secure session, but no authentication. This allows any
1097 client to connect, and provides an encrypted session.
1100 qemu-system-i386 [...OPTIONS...] -vnc :1,tls,x509=/etc/pki/qemu -monitor stdio
1103 In the above example @code{/etc/pki/qemu} should contain at least three files,
1104 @code{ca-cert.pem}, @code{server-cert.pem} and @code{server-key.pem}. Unprivileged
1105 users will want to use a private directory, for example @code{$HOME/.pki/qemu}.
1106 NB the @code{server-key.pem} file should be protected with file mode 0600 to
1107 only be readable by the user owning it.
1109 @node vnc_sec_certificate_verify
1110 @subsection With x509 certificates and client verification
1112 Certificates can also provide a means to authenticate the client connecting.
1113 The server will request that the client provide a certificate, which it will
1114 then validate against the CA certificate. This is a good choice if deploying
1115 in an environment with a private internal certificate authority.
1118 qemu-system-i386 [...OPTIONS...] -vnc :1,tls,x509verify=/etc/pki/qemu -monitor stdio
1122 @node vnc_sec_certificate_pw
1123 @subsection With x509 certificates, client verification and passwords
1125 Finally, the previous method can be combined with VNC password authentication
1126 to provide two layers of authentication for clients.
1129 qemu-system-i386 [...OPTIONS...] -vnc :1,password,tls,x509verify=/etc/pki/qemu -monitor stdio
1130 (qemu) change vnc password
1137 @subsection With SASL authentication
1139 The SASL authentication method is a VNC extension, that provides an
1140 easily extendable, pluggable authentication method. This allows for
1141 integration with a wide range of authentication mechanisms, such as
1142 PAM, GSSAPI/Kerberos, LDAP, SQL databases, one-time keys and more.
1143 The strength of the authentication depends on the exact mechanism
1144 configured. If the chosen mechanism also provides a SSF layer, then
1145 it will encrypt the datastream as well.
1147 Refer to the later docs on how to choose the exact SASL mechanism
1148 used for authentication, but assuming use of one supporting SSF,
1149 then QEMU can be launched with:
1152 qemu-system-i386 [...OPTIONS...] -vnc :1,sasl -monitor stdio
1155 @node vnc_sec_certificate_sasl
1156 @subsection With x509 certificates and SASL authentication
1158 If the desired SASL authentication mechanism does not supported
1159 SSF layers, then it is strongly advised to run it in combination
1160 with TLS and x509 certificates. This provides securely encrypted
1161 data stream, avoiding risk of compromising of the security
1162 credentials. This can be enabled, by combining the 'sasl' option
1163 with the aforementioned TLS + x509 options:
1166 qemu-system-i386 [...OPTIONS...] -vnc :1,tls,x509,sasl -monitor stdio
1169 @node vnc_setup_sasl
1171 @subsection Configuring SASL mechanisms
1173 The following documentation assumes use of the Cyrus SASL implementation on a
1174 Linux host, but the principles should apply to any other SASL implementation
1175 or host. When SASL is enabled, the mechanism configuration will be loaded from
1176 system default SASL service config /etc/sasl2/qemu.conf. If running QEMU as an
1177 unprivileged user, an environment variable SASL_CONF_PATH can be used to make
1178 it search alternate locations for the service config file.
1180 If the TLS option is enabled for VNC, then it will provide session encryption,
1181 otherwise the SASL mechanism will have to provide encryption. In the latter
1182 case the list of possible plugins that can be used is drastically reduced. In
1183 fact only the GSSAPI SASL mechanism provides an acceptable level of security
1184 by modern standards. Previous versions of QEMU referred to the DIGEST-MD5
1185 mechanism, however, it has multiple serious flaws described in detail in
1186 RFC 6331 and thus should never be used any more. The SCRAM-SHA-1 mechanism
1187 provides a simple username/password auth facility similar to DIGEST-MD5, but
1188 does not support session encryption, so can only be used in combination with
1191 When not using TLS the recommended configuration is
1195 keytab: /etc/qemu/krb5.tab
1198 This says to use the 'GSSAPI' mechanism with the Kerberos v5 protocol, with
1199 the server principal stored in /etc/qemu/krb5.tab. For this to work the
1200 administrator of your KDC must generate a Kerberos principal for the server,
1201 with a name of 'qemu/somehost.example.com@@EXAMPLE.COM' replacing
1202 'somehost.example.com' with the fully qualified host name of the machine
1203 running QEMU, and 'EXAMPLE.COM' with the Kerberos Realm.
1205 When using TLS, if username+password authentication is desired, then a
1206 reasonable configuration is
1209 mech_list: scram-sha-1
1210 sasldb_path: /etc/qemu/passwd.db
1213 The @code{saslpasswd2} program can be used to populate the @code{passwd.db}
1216 Other SASL configurations will be left as an exercise for the reader. Note that
1217 all mechanisms, except GSSAPI, should be combined with use of TLS to ensure a
1218 secure data channel.
1222 @section TLS setup for network services
1224 Almost all network services in QEMU have the ability to use TLS for
1225 session data encryption, along with x509 certificates for simple
1226 client authentication. What follows is a description of how to
1227 generate certificates suitable for usage with QEMU, and applies to
1228 the VNC server, character devices with the TCP backend, NBD server
1229 and client, and migration server and client.
1231 At a high level, QEMU requires certificates and private keys to be
1232 provided in PEM format. Aside from the core fields, the certificates
1233 should include various extension data sets, including v3 basic
1234 constraints data, key purpose, key usage and subject alt name.
1236 The GnuTLS package includes a command called @code{certtool} which can
1237 be used to easily generate certificates and keys in the required format
1238 with expected data present. Alternatively a certificate management
1239 service may be used.
1241 At a minimum it is necessary to setup a certificate authority, and
1242 issue certificates to each server. If using x509 certificates for
1243 authentication, then each client will also need to be issued a
1246 Assuming that the QEMU network services will only ever be exposed to
1247 clients on a private intranet, there is no need to use a commercial
1248 certificate authority to create certificates. A self-signed CA is
1249 sufficient, and in fact likely to be more secure since it removes
1250 the ability of malicious 3rd parties to trick the CA into mis-issuing
1251 certs for impersonating your services. The only likely exception
1252 where a commercial CA might be desirable is if enabling the VNC
1253 websockets server and exposing it directly to remote browser clients.
1254 In such a case it might be useful to use a commercial CA to avoid
1255 needing to install custom CA certs in the web browsers.
1257 The recommendation is for the server to keep its certificates in either
1258 @code{/etc/pki/qemu} or for unprivileged users in @code{$HOME/.pki/qemu}.
1262 * tls_generate_server::
1263 * tls_generate_client::
1266 @node tls_generate_ca
1267 @subsection Setup the Certificate Authority
1269 This step only needs to be performed once per organization / organizational
1270 unit. First the CA needs a private key. This key must be kept VERY secret
1271 and secure. If this key is compromised the entire trust chain of the certificates
1272 issued with it is lost.
1275 # certtool --generate-privkey > ca-key.pem
1278 To generate a self-signed certificate requires one core piece of information,
1279 the name of the organization. A template file @code{ca.info} should be
1280 populated with the desired data to avoid having to deal with interactive
1281 prompts from certtool:
1283 # cat > ca.info <<EOF
1284 cn = Name of your organization
1288 # certtool --generate-self-signed \
1289 --load-privkey ca-key.pem
1290 --template ca.info \
1291 --outfile ca-cert.pem
1294 The @code{ca} keyword in the template sets the v3 basic constraints extension
1295 to indicate this certificate is for a CA, while @code{cert_signing_key} sets
1296 the key usage extension to indicate this will be used for signing other keys.
1297 The generated @code{ca-cert.pem} file should be copied to all servers and
1298 clients wishing to utilize TLS support in the VNC server. The @code{ca-key.pem}
1299 must not be disclosed/copied anywhere except the host responsible for issuing
1302 @node tls_generate_server
1303 @subsection Issuing server certificates
1305 Each server (or host) needs to be issued with a key and certificate. When connecting
1306 the certificate is sent to the client which validates it against the CA certificate.
1307 The core pieces of information for a server certificate are the hostnames and/or IP
1308 addresses that will be used by clients when connecting. The hostname / IP address
1309 that the client specifies when connecting will be validated against the hostname(s)
1310 and IP address(es) recorded in the server certificate, and if no match is found
1311 the client will close the connection.
1313 Thus it is recommended that the server certificate include both the fully qualified
1314 and unqualified hostnames. If the server will have permanently assigned IP address(es),
1315 and clients are likely to use them when connecting, they may also be included in the
1316 certificate. Both IPv4 and IPv6 addresses are supported. Historically certificates
1317 only included 1 hostname in the @code{CN} field, however, usage of this field for
1318 validation is now deprecated. Instead modern TLS clients will validate against the
1319 Subject Alt Name extension data, which allows for multiple entries. In the future
1320 usage of the @code{CN} field may be discontinued entirely, so providing SAN
1321 extension data is strongly recommended.
1323 On the host holding the CA, create template files containing the information
1324 for each server, and use it to issue server certificates.
1327 # cat > server-hostNNN.info <<EOF
1328 organization = Name of your organization
1329 cn = hostNNN.foo.example.com
1331 dns_name = hostNNN.foo.example.com
1332 ip_address = 10.0.1.87
1333 ip_address = 192.8.0.92
1334 ip_address = 2620:0:cafe::87
1335 ip_address = 2001:24::92
1340 # certtool --generate-privkey > server-hostNNN-key.pem
1341 # certtool --generate-certificate \
1342 --load-ca-certificate ca-cert.pem \
1343 --load-ca-privkey ca-key.pem \
1344 --load-privkey server-hostNNN-key.pem \
1345 --template server-hostNNN.info \
1346 --outfile server-hostNNN-cert.pem
1349 The @code{dns_name} and @code{ip_address} fields in the template are setting
1350 the subject alt name extension data. The @code{tls_www_server} keyword is the
1351 key purpose extension to indicate this certificate is intended for usage in
1352 a web server. Although QEMU network services are not in fact HTTP servers
1353 (except for VNC websockets), setting this key purpose is still recommended.
1354 The @code{encryption_key} and @code{signing_key} keyword is the key usage
1355 extension to indicate this certificate is intended for usage in the data
1358 The @code{server-hostNNN-key.pem} and @code{server-hostNNN-cert.pem} files
1359 should now be securely copied to the server for which they were generated,
1360 and renamed to @code{server-key.pem} and @code{server-cert.pem} when added
1361 to the @code{/etc/pki/qemu} directory on the target host. The @code{server-key.pem}
1362 file is security sensitive and should be kept protected with file mode 0600
1363 to prevent disclosure.
1365 @node tls_generate_client
1366 @subsection Issuing client certificates
1368 The QEMU x509 TLS credential setup defaults to enabling client verification
1369 using certificates, providing a simple authentication mechanism. If this
1370 default is used, each client also needs to be issued a certificate. The client
1371 certificate contains enough metadata to uniquely identify the client with the
1372 scope of the certificate authority. The client certificate would typically
1373 include fields for organization, state, city, building, etc.
1375 Once again on the host holding the CA, create template files containing the
1376 information for each client, and use it to issue client certificates.
1380 # cat > client-hostNNN.info <<EOF
1383 locality = City Of London
1384 organization = Name of your organization
1385 cn = hostNNN.foo.example.com
1390 # certtool --generate-privkey > client-hostNNN-key.pem
1391 # certtool --generate-certificate \
1392 --load-ca-certificate ca-cert.pem \
1393 --load-ca-privkey ca-key.pem \
1394 --load-privkey client-hostNNN-key.pem \
1395 --template client-hostNNN.info \
1396 --outfile client-hostNNN-cert.pem
1399 The subject alt name extension data is not required for clients, so the
1400 the @code{dns_name} and @code{ip_address} fields are not included.
1401 The @code{tls_www_client} keyword is the key purpose extension to indicate
1402 this certificate is intended for usage in a web client. Although QEMU
1403 network clients are not in fact HTTP clients, setting this key purpose is
1404 still recommended. The @code{encryption_key} and @code{signing_key} keyword
1405 is the key usage extension to indicate this certificate is intended for
1406 usage in the data session.
1408 The @code{client-hostNNN-key.pem} and @code{client-hostNNN-cert.pem} files
1409 should now be securely copied to the client for which they were generated,
1410 and renamed to @code{client-key.pem} and @code{client-cert.pem} when added
1411 to the @code{/etc/pki/qemu} directory on the target host. The @code{client-key.pem}
1412 file is security sensitive and should be kept protected with file mode 0600
1413 to prevent disclosure.
1415 If a single host is going to be using TLS in both a client and server
1416 role, it is possible to create a single certificate to cover both roles.
1417 This would be quite common for the migration and NBD services, where a
1418 QEMU process will be started by accepting a TLS protected incoming migration,
1419 and later itself be migrated out to another host. To generate a single
1420 certificate, simply include the template data from both the client and server
1421 instructions in one.
1424 # cat > both-hostNNN.info <<EOF
1427 locality = City Of London
1428 organization = Name of your organization
1429 cn = hostNNN.foo.example.com
1431 dns_name = hostNNN.foo.example.com
1432 ip_address = 10.0.1.87
1433 ip_address = 192.8.0.92
1434 ip_address = 2620:0:cafe::87
1435 ip_address = 2001:24::92
1441 # certtool --generate-privkey > both-hostNNN-key.pem
1442 # certtool --generate-certificate \
1443 --load-ca-certificate ca-cert.pem \
1444 --load-ca-privkey ca-key.pem \
1445 --load-privkey both-hostNNN-key.pem \
1446 --template both-hostNNN.info \
1447 --outfile both-hostNNN-cert.pem
1450 When copying the PEM files to the target host, save them twice,
1451 once as @code{server-cert.pem} and @code{server-key.pem}, and
1452 again as @code{client-cert.pem} and @code{client-key.pem}.
1454 @node tls_creds_setup
1455 @subsection TLS x509 credential configuration
1457 QEMU has a standard mechanism for loading x509 credentials that will be
1458 used for network services and clients. It requires specifying the
1459 @code{tls-creds-x509} class name to the @code{--object} command line
1460 argument for the system emulators. Each set of credentials loaded should
1461 be given a unique string identifier via the @code{id} parameter. A single
1462 set of TLS credentials can be used for multiple network backends, so VNC,
1463 migration, NBD, character devices can all share the same credentials. Note,
1464 however, that credentials for use in a client endpoint must be loaded
1465 separately from those used in a server endpoint.
1467 When specifying the object, the @code{dir} parameters specifies which
1468 directory contains the credential files. This directory is expected to
1469 contain files with the names mentioned previously, @code{ca-cert.pem},
1470 @code{server-key.pem}, @code{server-cert.pem}, @code{client-key.pem}
1471 and @code{client-cert.pem} as appropriate. It is also possible to
1472 include a set of pre-generated Diffie-Hellman (DH) parameters in a file
1473 @code{dh-params.pem}, which can be created using the
1474 @code{certtool --generate-dh-params} command. If omitted, QEMU will
1475 dynamically generate DH parameters when loading the credentials.
1477 The @code{endpoint} parameter indicates whether the credentials will
1478 be used for a network client or server, and determines which PEM
1481 The @code{verify} parameter determines whether x509 certificate
1482 validation should be performed. This defaults to enabled, meaning
1483 clients will always validate the server hostname against the
1484 certificate subject alt name fields and/or CN field. It also
1485 means that servers will request that clients provide a certificate
1486 and validate them. Verification should never be turned off for
1487 client endpoints, however, it may be turned off for server endpoints
1488 if an alternative mechanism is used to authenticate clients. For
1489 example, the VNC server can use SASL to authenticate clients
1492 To load server credentials with client certificate validation
1496 $QEMU -object tls-creds-x509,id=tls0,dir=/etc/pki/qemu,endpoint=server
1499 while to load client credentials use
1502 $QEMU -object tls-creds-x509,id=tls0,dir=/etc/pki/qemu,endpoint=client
1505 Network services which support TLS will all have a @code{tls-creds}
1506 parameter which expects the ID of the TLS credentials object. For
1510 $QEMU -vnc 0.0.0.0:0,tls-creds=tls0
1516 QEMU has a primitive support to work with gdb, so that you can do
1517 'Ctrl-C' while the virtual machine is running and inspect its state.
1519 In order to use gdb, launch QEMU with the '-s' option. It will wait for a
1522 qemu-system-i386 -s -kernel arch/i386/boot/bzImage -hda root-2.4.20.img \
1523 -append "root=/dev/hda"
1524 Connected to host network interface: tun0
1525 Waiting gdb connection on port 1234
1528 Then launch gdb on the 'vmlinux' executable:
1533 In gdb, connect to QEMU:
1535 (gdb) target remote localhost:1234
1538 Then you can use gdb normally. For example, type 'c' to launch the kernel:
1543 Here are some useful tips in order to use gdb on system code:
1547 Use @code{info reg} to display all the CPU registers.
1549 Use @code{x/10i $eip} to display the code at the PC position.
1551 Use @code{set architecture i8086} to dump 16 bit code. Then use
1552 @code{x/10i $cs*16+$eip} to dump the code at the PC position.
1555 Advanced debugging options:
1557 The default single stepping behavior is step with the IRQs and timer service routines off. It is set this way because when gdb executes a single step it expects to advance beyond the current instruction. With the IRQs and timer service routines on, a single step might jump into the one of the interrupt or exception vectors instead of executing the current instruction. This means you may hit the same breakpoint a number of times before executing the instruction gdb wants to have executed. Because there are rare circumstances where you want to single step into an interrupt vector the behavior can be controlled from GDB. There are three commands you can query and set the single step behavior:
1559 @item maintenance packet qqemu.sstepbits
1561 This will display the MASK bits used to control the single stepping IE:
1563 (gdb) maintenance packet qqemu.sstepbits
1564 sending: "qqemu.sstepbits"
1565 received: "ENABLE=1,NOIRQ=2,NOTIMER=4"
1567 @item maintenance packet qqemu.sstep
1569 This will display the current value of the mask used when single stepping IE:
1571 (gdb) maintenance packet qqemu.sstep
1572 sending: "qqemu.sstep"
1575 @item maintenance packet Qqemu.sstep=HEX_VALUE
1577 This will change the single step mask, so if wanted to enable IRQs on the single step, but not timers, you would use:
1579 (gdb) maintenance packet Qqemu.sstep=0x5
1580 sending: "qemu.sstep=0x5"
1585 @node pcsys_os_specific
1586 @section Target OS specific information
1590 To have access to SVGA graphic modes under X11, use the @code{vesa} or
1591 the @code{cirrus} X11 driver. For optimal performances, use 16 bit
1592 color depth in the guest and the host OS.
1594 When using a 2.6 guest Linux kernel, you should add the option
1595 @code{clock=pit} on the kernel command line because the 2.6 Linux
1596 kernels make very strict real time clock checks by default that QEMU
1597 cannot simulate exactly.
1599 When using a 2.6 guest Linux kernel, verify that the 4G/4G patch is
1600 not activated because QEMU is slower with this patch. The QEMU
1601 Accelerator Module is also much slower in this case. Earlier Fedora
1602 Core 3 Linux kernel (< 2.6.9-1.724_FC3) were known to incorporate this
1603 patch by default. Newer kernels don't have it.
1607 If you have a slow host, using Windows 95 is better as it gives the
1608 best speed. Windows 2000 is also a good choice.
1610 @subsubsection SVGA graphic modes support
1612 QEMU emulates a Cirrus Logic GD5446 Video
1613 card. All Windows versions starting from Windows 95 should recognize
1614 and use this graphic card. For optimal performances, use 16 bit color
1615 depth in the guest and the host OS.
1617 If you are using Windows XP as guest OS and if you want to use high
1618 resolution modes which the Cirrus Logic BIOS does not support (i.e. >=
1619 1280x1024x16), then you should use the VESA VBE virtual graphic card
1620 (option @option{-std-vga}).
1622 @subsubsection CPU usage reduction
1624 Windows 9x does not correctly use the CPU HLT
1625 instruction. The result is that it takes host CPU cycles even when
1626 idle. You can install the utility from
1627 @url{https://web.archive.org/web/20060212132151/http://www.user.cityline.ru/~maxamn/amnhltm.zip}
1628 to solve this problem. Note that no such tool is needed for NT, 2000 or XP.
1630 @subsubsection Windows 2000 disk full problem
1632 Windows 2000 has a bug which gives a disk full problem during its
1633 installation. When installing it, use the @option{-win2k-hack} QEMU
1634 option to enable a specific workaround. After Windows 2000 is
1635 installed, you no longer need this option (this option slows down the
1638 @subsubsection Windows 2000 shutdown
1640 Windows 2000 cannot automatically shutdown in QEMU although Windows 98
1641 can. It comes from the fact that Windows 2000 does not automatically
1642 use the APM driver provided by the BIOS.
1644 In order to correct that, do the following (thanks to Struan
1645 Bartlett): go to the Control Panel => Add/Remove Hardware & Next =>
1646 Add/Troubleshoot a device => Add a new device & Next => No, select the
1647 hardware from a list & Next => NT Apm/Legacy Support & Next => Next
1648 (again) a few times. Now the driver is installed and Windows 2000 now
1649 correctly instructs QEMU to shutdown at the appropriate moment.
1651 @subsubsection Share a directory between Unix and Windows
1653 See @ref{sec_invocation} about the help of the option
1654 @option{'-netdev user,smb=...'}.
1656 @subsubsection Windows XP security problem
1658 Some releases of Windows XP install correctly but give a security
1661 A problem is preventing Windows from accurately checking the
1662 license for this computer. Error code: 0x800703e6.
1665 The workaround is to install a service pack for XP after a boot in safe
1666 mode. Then reboot, and the problem should go away. Since there is no
1667 network while in safe mode, its recommended to download the full
1668 installation of SP1 or SP2 and transfer that via an ISO or using the
1669 vvfat block device ("-hdb fat:directory_which_holds_the_SP").
1671 @subsection MS-DOS and FreeDOS
1673 @subsubsection CPU usage reduction
1675 DOS does not correctly use the CPU HLT instruction. The result is that
1676 it takes host CPU cycles even when idle. You can install the utility from
1677 @url{https://web.archive.org/web/20051222085335/http://www.vmware.com/software/dosidle210.zip}
1678 to solve this problem.
1680 @node QEMU System emulator for non PC targets
1681 @chapter QEMU System emulator for non PC targets
1683 QEMU is a generic emulator and it emulates many non PC
1684 machines. Most of the options are similar to the PC emulator. The
1685 differences are mentioned in the following sections.
1688 * PowerPC System emulator::
1689 * Sparc32 System emulator::
1690 * Sparc64 System emulator::
1691 * MIPS System emulator::
1692 * ARM System emulator::
1693 * ColdFire System emulator::
1694 * Cris System emulator::
1695 * Microblaze System emulator::
1696 * SH4 System emulator::
1697 * Xtensa System emulator::
1700 @node PowerPC System emulator
1701 @section PowerPC System emulator
1702 @cindex system emulation (PowerPC)
1704 Use the executable @file{qemu-system-ppc} to simulate a complete PREP
1705 or PowerMac PowerPC system.
1707 QEMU emulates the following PowerMac peripherals:
1711 UniNorth or Grackle PCI Bridge
1713 PCI VGA compatible card with VESA Bochs Extensions
1715 2 PMAC IDE interfaces with hard disk and CD-ROM support
1721 VIA-CUDA with ADB keyboard and mouse.
1724 QEMU emulates the following PREP peripherals:
1730 PCI VGA compatible card with VESA Bochs Extensions
1732 2 IDE interfaces with hard disk and CD-ROM support
1736 NE2000 network adapters
1740 PREP Non Volatile RAM
1742 PC compatible keyboard and mouse.
1745 QEMU uses the Open Hack'Ware Open Firmware Compatible BIOS available at
1746 @url{http://perso.magic.fr/l_indien/OpenHackWare/index.htm}.
1748 Since version 0.9.1, QEMU uses OpenBIOS @url{https://www.openbios.org/}
1749 for the g3beige and mac99 PowerMac machines. OpenBIOS is a free (GPL
1750 v2) portable firmware implementation. The goal is to implement a 100%
1751 IEEE 1275-1994 (referred to as Open Firmware) compliant firmware.
1753 @c man begin OPTIONS
1755 The following options are specific to the PowerPC emulation:
1759 @item -g @var{W}x@var{H}[x@var{DEPTH}]
1761 Set the initial VGA graphic mode. The default is 800x600x32.
1763 @item -prom-env @var{string}
1765 Set OpenBIOS variables in NVRAM, for example:
1768 qemu-system-ppc -prom-env 'auto-boot?=false' \
1769 -prom-env 'boot-device=hd:2,\yaboot' \
1770 -prom-env 'boot-args=conf=hd:2,\yaboot.conf'
1773 These variables are not used by Open Hack'Ware.
1780 More information is available at
1781 @url{http://perso.magic.fr/l_indien/qemu-ppc/}.
1783 @node Sparc32 System emulator
1784 @section Sparc32 System emulator
1785 @cindex system emulation (Sparc32)
1787 Use the executable @file{qemu-system-sparc} to simulate the following
1788 Sun4m architecture machines:
1803 SPARCstation Voyager
1810 The emulation is somewhat complete. SMP up to 16 CPUs is supported,
1811 but Linux limits the number of usable CPUs to 4.
1813 QEMU emulates the following sun4m peripherals:
1819 TCX or cgthree Frame buffer
1821 Lance (Am7990) Ethernet
1823 Non Volatile RAM M48T02/M48T08
1825 Slave I/O: timers, interrupt controllers, Zilog serial ports, keyboard
1826 and power/reset logic
1828 ESP SCSI controller with hard disk and CD-ROM support
1830 Floppy drive (not on SS-600MP)
1832 CS4231 sound device (only on SS-5, not working yet)
1835 The number of peripherals is fixed in the architecture. Maximum
1836 memory size depends on the machine type, for SS-5 it is 256MB and for
1839 Since version 0.8.2, QEMU uses OpenBIOS
1840 @url{https://www.openbios.org/}. OpenBIOS is a free (GPL v2) portable
1841 firmware implementation. The goal is to implement a 100% IEEE
1842 1275-1994 (referred to as Open Firmware) compliant firmware.
1844 A sample Linux 2.6 series kernel and ram disk image are available on
1845 the QEMU web site. There are still issues with NetBSD and OpenBSD, but
1846 most kernel versions work. Please note that currently older Solaris kernels
1847 don't work probably due to interface issues between OpenBIOS and
1850 @c man begin OPTIONS
1852 The following options are specific to the Sparc32 emulation:
1856 @item -g @var{W}x@var{H}x[x@var{DEPTH}]
1858 Set the initial graphics mode. For TCX, the default is 1024x768x8 with the
1859 option of 1024x768x24. For cgthree, the default is 1024x768x8 with the option
1860 of 1152x900x8 for people who wish to use OBP.
1862 @item -prom-env @var{string}
1864 Set OpenBIOS variables in NVRAM, for example:
1867 qemu-system-sparc -prom-env 'auto-boot?=false' \
1868 -prom-env 'boot-device=sd(0,2,0):d' -prom-env 'boot-args=linux single'
1871 @item -M [SS-4|SS-5|SS-10|SS-20|SS-600MP|LX|Voyager|SPARCClassic] [|SPARCbook]
1873 Set the emulated machine type. Default is SS-5.
1879 @node Sparc64 System emulator
1880 @section Sparc64 System emulator
1881 @cindex system emulation (Sparc64)
1883 Use the executable @file{qemu-system-sparc64} to simulate a Sun4u
1884 (UltraSPARC PC-like machine), Sun4v (T1 PC-like machine), or generic
1885 Niagara (T1) machine. The Sun4u emulator is mostly complete, being
1886 able to run Linux, NetBSD and OpenBSD in headless (-nographic) mode. The
1887 Sun4v emulator is still a work in progress.
1889 The Niagara T1 emulator makes use of firmware and OS binaries supplied in the S10image/ directory
1890 of the OpenSPARC T1 project @url{http://download.oracle.com/technetwork/systems/opensparc/OpenSPARCT1_Arch.1.5.tar.bz2}
1891 and is able to boot the disk.s10hw2 Solaris image.
1893 qemu-system-sparc64 -M niagara -L /path-to/S10image/ \
1895 -drive if=pflash,readonly=on,file=/S10image/disk.s10hw2
1899 QEMU emulates the following peripherals:
1903 UltraSparc IIi APB PCI Bridge
1905 PCI VGA compatible card with VESA Bochs Extensions
1907 PS/2 mouse and keyboard
1909 Non Volatile RAM M48T59
1911 PC-compatible serial ports
1913 2 PCI IDE interfaces with hard disk and CD-ROM support
1918 @c man begin OPTIONS
1920 The following options are specific to the Sparc64 emulation:
1924 @item -prom-env @var{string}
1926 Set OpenBIOS variables in NVRAM, for example:
1929 qemu-system-sparc64 -prom-env 'auto-boot?=false'
1932 @item -M [sun4u|sun4v|niagara]
1934 Set the emulated machine type. The default is sun4u.
1940 @node MIPS System emulator
1941 @section MIPS System emulator
1942 @cindex system emulation (MIPS)
1944 Four executables cover simulation of 32 and 64-bit MIPS systems in
1945 both endian options, @file{qemu-system-mips}, @file{qemu-system-mipsel}
1946 @file{qemu-system-mips64} and @file{qemu-system-mips64el}.
1947 Five different machine types are emulated:
1951 A generic ISA PC-like machine "mips"
1953 The MIPS Malta prototype board "malta"
1955 An ACER Pica "pica61". This machine needs the 64-bit emulator.
1957 MIPS emulator pseudo board "mipssim"
1959 A MIPS Magnum R4000 machine "magnum". This machine needs the 64-bit emulator.
1962 The generic emulation is supported by Debian 'Etch' and is able to
1963 install Debian into a virtual disk image. The following devices are
1968 A range of MIPS CPUs, default is the 24Kf
1970 PC style serial port
1977 The Malta emulation supports the following devices:
1981 Core board with MIPS 24Kf CPU and Galileo system controller
1983 PIIX4 PCI/USB/SMbus controller
1985 The Multi-I/O chip's serial device
1987 PCI network cards (PCnet32 and others)
1989 Malta FPGA serial device
1991 Cirrus (default) or any other PCI VGA graphics card
1994 The ACER Pica emulation supports:
2000 PC-style IRQ and DMA controllers
2007 The mipssim pseudo board emulation provides an environment similar
2008 to what the proprietary MIPS emulator uses for running Linux.
2013 A range of MIPS CPUs, default is the 24Kf
2015 PC style serial port
2017 MIPSnet network emulation
2020 The MIPS Magnum R4000 emulation supports:
2026 PC-style IRQ controller
2036 @node ARM System emulator
2037 @section ARM System emulator
2038 @cindex system emulation (ARM)
2040 Use the executable @file{qemu-system-arm} to simulate a ARM
2041 machine. The ARM Integrator/CP board is emulated with the following
2046 ARM926E, ARM1026E, ARM946E, ARM1136 or Cortex-A8 CPU
2050 SMC 91c111 Ethernet adapter
2052 PL110 LCD controller
2054 PL050 KMI with PS/2 keyboard and mouse.
2056 PL181 MultiMedia Card Interface with SD card.
2059 The ARM Versatile baseboard is emulated with the following devices:
2063 ARM926E, ARM1136 or Cortex-A8 CPU
2065 PL190 Vectored Interrupt Controller
2069 SMC 91c111 Ethernet adapter
2071 PL110 LCD controller
2073 PL050 KMI with PS/2 keyboard and mouse.
2075 PCI host bridge. Note the emulated PCI bridge only provides access to
2076 PCI memory space. It does not provide access to PCI IO space.
2077 This means some devices (eg. ne2k_pci NIC) are not usable, and others
2078 (eg. rtl8139 NIC) are only usable when the guest drivers use the memory
2079 mapped control registers.
2081 PCI OHCI USB controller.
2083 LSI53C895A PCI SCSI Host Bus Adapter with hard disk and CD-ROM devices.
2085 PL181 MultiMedia Card Interface with SD card.
2088 Several variants of the ARM RealView baseboard are emulated,
2089 including the EB, PB-A8 and PBX-A9. Due to interactions with the
2090 bootloader, only certain Linux kernel configurations work out
2091 of the box on these boards.
2093 Kernels for the PB-A8 board should have CONFIG_REALVIEW_HIGH_PHYS_OFFSET
2094 enabled in the kernel, and expect 512M RAM. Kernels for The PBX-A9 board
2095 should have CONFIG_SPARSEMEM enabled, CONFIG_REALVIEW_HIGH_PHYS_OFFSET
2096 disabled and expect 1024M RAM.
2098 The following devices are emulated:
2102 ARM926E, ARM1136, ARM11MPCore, Cortex-A8 or Cortex-A9 MPCore CPU
2104 ARM AMBA Generic/Distributed Interrupt Controller
2108 SMC 91c111 or SMSC LAN9118 Ethernet adapter
2110 PL110 LCD controller
2112 PL050 KMI with PS/2 keyboard and mouse
2116 PCI OHCI USB controller
2118 LSI53C895A PCI SCSI Host Bus Adapter with hard disk and CD-ROM devices
2120 PL181 MultiMedia Card Interface with SD card.
2123 The XScale-based clamshell PDA models ("Spitz", "Akita", "Borzoi"
2124 and "Terrier") emulation includes the following peripherals:
2128 Intel PXA270 System-on-chip (ARM V5TE core)
2132 IBM/Hitachi DSCM microdrive in a PXA PCMCIA slot - not in "Akita"
2134 On-chip OHCI USB controller
2136 On-chip LCD controller
2138 On-chip Real Time Clock
2140 TI ADS7846 touchscreen controller on SSP bus
2142 Maxim MAX1111 analog-digital converter on I@math{^2}C bus
2144 GPIO-connected keyboard controller and LEDs
2146 Secure Digital card connected to PXA MMC/SD host
2150 WM8750 audio CODEC on I@math{^2}C and I@math{^2}S busses
2153 The Palm Tungsten|E PDA (codename "Cheetah") emulation includes the
2158 Texas Instruments OMAP310 System-on-chip (ARM 925T core)
2160 ROM and RAM memories (ROM firmware image can be loaded with -option-rom)
2162 On-chip LCD controller
2164 On-chip Real Time Clock
2166 TI TSC2102i touchscreen controller / analog-digital converter / Audio
2167 CODEC, connected through MicroWire and I@math{^2}S busses
2169 GPIO-connected matrix keypad
2171 Secure Digital card connected to OMAP MMC/SD host
2176 Nokia N800 and N810 internet tablets (known also as RX-34 and RX-44 / 48)
2177 emulation supports the following elements:
2181 Texas Instruments OMAP2420 System-on-chip (ARM 1136 core)
2183 RAM and non-volatile OneNAND Flash memories
2185 Display connected to EPSON remote framebuffer chip and OMAP on-chip
2186 display controller and a LS041y3 MIPI DBI-C controller
2188 TI TSC2301 (in N800) and TI TSC2005 (in N810) touchscreen controllers
2189 driven through SPI bus
2191 National Semiconductor LM8323-controlled qwerty keyboard driven
2192 through I@math{^2}C bus
2194 Secure Digital card connected to OMAP MMC/SD host
2196 Three OMAP on-chip UARTs and on-chip STI debugging console
2198 A Bluetooth(R) transceiver and HCI connected to an UART
2200 Mentor Graphics "Inventra" dual-role USB controller embedded in a TI
2201 TUSB6010 chip - only USB host mode is supported
2203 TI TMP105 temperature sensor driven through I@math{^2}C bus
2205 TI TWL92230C power management companion with an RTC on I@math{^2}C bus
2207 Nokia RETU and TAHVO multi-purpose chips with an RTC, connected
2211 The Luminary Micro Stellaris LM3S811EVB emulation includes the following
2218 64k Flash and 8k SRAM.
2220 Timers, UARTs, ADC and I@math{^2}C interface.
2222 OSRAM Pictiva 96x16 OLED with SSD0303 controller on I@math{^2}C bus.
2225 The Luminary Micro Stellaris LM3S6965EVB emulation includes the following
2232 256k Flash and 64k SRAM.
2234 Timers, UARTs, ADC, I@math{^2}C and SSI interfaces.
2236 OSRAM Pictiva 128x64 OLED with SSD0323 controller connected via SSI.
2239 The Freecom MusicPal internet radio emulation includes the following
2244 Marvell MV88W8618 ARM core.
2246 32 MB RAM, 256 KB SRAM, 8 MB flash.
2250 MV88W8xx8 Ethernet controller
2252 MV88W8618 audio controller, WM8750 CODEC and mixer
2254 128×64 display with brightness control
2256 2 buttons, 2 navigation wheels with button function
2259 The Siemens SX1 models v1 and v2 (default) basic emulation.
2260 The emulation includes the following elements:
2264 Texas Instruments OMAP310 System-on-chip (ARM 925T core)
2266 ROM and RAM memories (ROM firmware image can be loaded with -pflash)
2268 1 Flash of 16MB and 1 Flash of 8MB
2272 On-chip LCD controller
2274 On-chip Real Time Clock
2276 Secure Digital card connected to OMAP MMC/SD host
2281 A Linux 2.6 test image is available on the QEMU web site. More
2282 information is available in the QEMU mailing-list archive.
2284 @c man begin OPTIONS
2286 The following options are specific to the ARM emulation:
2291 Enable semihosting syscall emulation.
2293 On ARM this implements the "Angel" interface.
2295 Note that this allows guest direct access to the host filesystem,
2296 so should only be used with trusted guest OS.
2302 @node ColdFire System emulator
2303 @section ColdFire System emulator
2304 @cindex system emulation (ColdFire)
2305 @cindex system emulation (M68K)
2307 Use the executable @file{qemu-system-m68k} to simulate a ColdFire machine.
2308 The emulator is able to boot a uClinux kernel.
2310 The M5208EVB emulation includes the following devices:
2314 MCF5208 ColdFire V2 Microprocessor (ISA A+ with EMAC).
2316 Three Two on-chip UARTs.
2318 Fast Ethernet Controller (FEC)
2321 The AN5206 emulation includes the following devices:
2325 MCF5206 ColdFire V2 Microprocessor.
2330 @c man begin OPTIONS
2332 The following options are specific to the ColdFire emulation:
2337 Enable semihosting syscall emulation.
2339 On M68K this implements the "ColdFire GDB" interface used by libgloss.
2341 Note that this allows guest direct access to the host filesystem,
2342 so should only be used with trusted guest OS.
2348 @node Cris System emulator
2349 @section Cris System emulator
2350 @cindex system emulation (Cris)
2354 @node Microblaze System emulator
2355 @section Microblaze System emulator
2356 @cindex system emulation (Microblaze)
2360 @node SH4 System emulator
2361 @section SH4 System emulator
2362 @cindex system emulation (SH4)
2366 @node Xtensa System emulator
2367 @section Xtensa System emulator
2368 @cindex system emulation (Xtensa)
2370 Two executables cover simulation of both Xtensa endian options,
2371 @file{qemu-system-xtensa} and @file{qemu-system-xtensaeb}.
2372 Two different machine types are emulated:
2376 Xtensa emulator pseudo board "sim"
2378 Avnet LX60/LX110/LX200 board
2381 The sim pseudo board emulation provides an environment similar
2382 to one provided by the proprietary Tensilica ISS.
2387 A range of Xtensa CPUs, default is the DC232B
2389 Console and filesystem access via semihosting calls
2392 The Avnet LX60/LX110/LX200 emulation supports:
2396 A range of Xtensa CPUs, default is the DC232B
2400 OpenCores 10/100 Mbps Ethernet MAC
2403 @c man begin OPTIONS
2405 The following options are specific to the Xtensa emulation:
2410 Enable semihosting syscall emulation.
2412 Xtensa semihosting provides basic file IO calls, such as open/read/write/seek/select.
2413 Tensilica baremetal libc for ISS and linux platform "sim" use this interface.
2415 Note that this allows guest direct access to the host filesystem,
2416 so should only be used with trusted guest OS.
2422 @node QEMU Guest Agent
2423 @chapter QEMU Guest Agent invocation
2425 @include qemu-ga.texi
2427 @node QEMU User space emulator
2428 @chapter QEMU User space emulator
2431 * Supported Operating Systems ::
2433 * Linux User space emulator::
2434 * BSD User space emulator ::
2437 @node Supported Operating Systems
2438 @section Supported Operating Systems
2440 The following OS are supported in user space emulation:
2444 Linux (referred as qemu-linux-user)
2446 BSD (referred as qemu-bsd-user)
2452 QEMU user space emulation has the following notable features:
2455 @item System call translation:
2456 QEMU includes a generic system call translator. This means that
2457 the parameters of the system calls can be converted to fix
2458 endianness and 32/64-bit mismatches between hosts and targets.
2459 IOCTLs can be converted too.
2461 @item POSIX signal handling:
2462 QEMU can redirect to the running program all signals coming from
2463 the host (such as @code{SIGALRM}), as well as synthesize signals from
2464 virtual CPU exceptions (for example @code{SIGFPE} when the program
2465 executes a division by zero).
2467 QEMU relies on the host kernel to emulate most signal system
2468 calls, for example to emulate the signal mask. On Linux, QEMU
2469 supports both normal and real-time signals.
2472 On Linux, QEMU can emulate the @code{clone} syscall and create a real
2473 host thread (with a separate virtual CPU) for each emulated thread.
2474 Note that not all targets currently emulate atomic operations correctly.
2475 x86 and ARM use a global lock in order to preserve their semantics.
2478 QEMU was conceived so that ultimately it can emulate itself. Although
2479 it is not very useful, it is an important test to show the power of the
2482 @node Linux User space emulator
2483 @section Linux User space emulator
2488 * Command line options::
2493 @subsection Quick Start
2495 In order to launch a Linux process, QEMU needs the process executable
2496 itself and all the target (x86) dynamic libraries used by it.
2500 @item On x86, you can just try to launch any process by using the native
2504 qemu-i386 -L / /bin/ls
2507 @code{-L /} tells that the x86 dynamic linker must be searched with a
2510 @item Since QEMU is also a linux process, you can launch QEMU with
2511 QEMU (NOTE: you can only do that if you compiled QEMU from the sources):
2514 qemu-i386 -L / qemu-i386 -L / /bin/ls
2517 @item On non x86 CPUs, you need first to download at least an x86 glibc
2518 (@file{qemu-runtime-i386-XXX-.tar.gz} on the QEMU web page). Ensure that
2519 @code{LD_LIBRARY_PATH} is not set:
2522 unset LD_LIBRARY_PATH
2525 Then you can launch the precompiled @file{ls} x86 executable:
2528 qemu-i386 tests/i386/ls
2530 You can look at @file{scripts/qemu-binfmt-conf.sh} so that
2531 QEMU is automatically launched by the Linux kernel when you try to
2532 launch x86 executables. It requires the @code{binfmt_misc} module in the
2535 @item The x86 version of QEMU is also included. You can try weird things such as:
2537 qemu-i386 /usr/local/qemu-i386/bin/qemu-i386 \
2538 /usr/local/qemu-i386/bin/ls-i386
2544 @subsection Wine launch
2548 @item Ensure that you have a working QEMU with the x86 glibc
2549 distribution (see previous section). In order to verify it, you must be
2553 qemu-i386 /usr/local/qemu-i386/bin/ls-i386
2556 @item Download the binary x86 Wine install
2557 (@file{qemu-XXX-i386-wine.tar.gz} on the QEMU web page).
2559 @item Configure Wine on your account. Look at the provided script
2560 @file{/usr/local/qemu-i386/@/bin/wine-conf.sh}. Your previous
2561 @code{$@{HOME@}/.wine} directory is saved to @code{$@{HOME@}/.wine.org}.
2563 @item Then you can try the example @file{putty.exe}:
2566 qemu-i386 /usr/local/qemu-i386/wine/bin/wine \
2567 /usr/local/qemu-i386/wine/c/Program\ Files/putty.exe
2572 @node Command line options
2573 @subsection Command line options
2576 @command{qemu-i386} [@option{-h]} [@option{-d]} [@option{-L} @var{path}] [@option{-s} @var{size}] [@option{-cpu} @var{model}] [@option{-g} @var{port}] [@option{-B} @var{offset}] [@option{-R} @var{size}] @var{program} [@var{arguments}...]
2583 Set the x86 elf interpreter prefix (default=/usr/local/qemu-i386)
2585 Set the x86 stack size in bytes (default=524288)
2587 Select CPU model (-cpu help for list and additional feature selection)
2588 @item -E @var{var}=@var{value}
2589 Set environment @var{var} to @var{value}.
2591 Remove @var{var} from the environment.
2593 Offset guest address by the specified number of bytes. This is useful when
2594 the address region required by guest applications is reserved on the host.
2595 This option is currently only supported on some hosts.
2597 Pre-allocate a guest virtual address space of the given size (in bytes).
2598 "G", "M", and "k" suffixes may be used when specifying the size.
2605 Activate logging of the specified items (use '-d help' for a list of log items)
2607 Act as if the host page size was 'pagesize' bytes
2609 Wait gdb connection to port
2611 Run the emulation in single step mode.
2614 Environment variables:
2618 Print system calls and arguments similar to the 'strace' program
2619 (NOTE: the actual 'strace' program will not work because the user
2620 space emulator hasn't implemented ptrace). At the moment this is
2621 incomplete. All system calls that don't have a specific argument
2622 format are printed with information for six arguments. Many
2623 flag-style arguments don't have decoders and will show up as numbers.
2626 @node Other binaries
2627 @subsection Other binaries
2629 @cindex user mode (Alpha)
2630 @command{qemu-alpha} TODO.
2632 @cindex user mode (ARM)
2633 @command{qemu-armeb} TODO.
2635 @cindex user mode (ARM)
2636 @command{qemu-arm} is also capable of running ARM "Angel" semihosted ELF
2637 binaries (as implemented by the arm-elf and arm-eabi Newlib/GDB
2638 configurations), and arm-uclinux bFLT format binaries.
2640 @cindex user mode (ColdFire)
2641 @cindex user mode (M68K)
2642 @command{qemu-m68k} is capable of running semihosted binaries using the BDM
2643 (m5xxx-ram-hosted.ld) or m68k-sim (sim.ld) syscall interfaces, and
2644 coldfire uClinux bFLT format binaries.
2646 The binary format is detected automatically.
2648 @cindex user mode (Cris)
2649 @command{qemu-cris} TODO.
2651 @cindex user mode (i386)
2652 @command{qemu-i386} TODO.
2653 @command{qemu-x86_64} TODO.
2655 @cindex user mode (Microblaze)
2656 @command{qemu-microblaze} TODO.
2658 @cindex user mode (MIPS)
2659 @command{qemu-mips} TODO.
2660 @command{qemu-mipsel} TODO.
2662 @cindex user mode (NiosII)
2663 @command{qemu-nios2} TODO.
2665 @cindex user mode (PowerPC)
2666 @command{qemu-ppc64abi32} TODO.
2667 @command{qemu-ppc64} TODO.
2668 @command{qemu-ppc} TODO.
2670 @cindex user mode (SH4)
2671 @command{qemu-sh4eb} TODO.
2672 @command{qemu-sh4} TODO.
2674 @cindex user mode (SPARC)
2675 @command{qemu-sparc} can execute Sparc32 binaries (Sparc32 CPU, 32 bit ABI).
2677 @command{qemu-sparc32plus} can execute Sparc32 and SPARC32PLUS binaries
2678 (Sparc64 CPU, 32 bit ABI).
2680 @command{qemu-sparc64} can execute some Sparc64 (Sparc64 CPU, 64 bit ABI) and
2681 SPARC32PLUS binaries (Sparc64 CPU, 32 bit ABI).
2683 @node BSD User space emulator
2684 @section BSD User space emulator
2689 * BSD Command line options::
2693 @subsection BSD Status
2697 target Sparc64 on Sparc64: Some trivial programs work.
2700 @node BSD Quick Start
2701 @subsection Quick Start
2703 In order to launch a BSD process, QEMU needs the process executable
2704 itself and all the target dynamic libraries used by it.
2708 @item On Sparc64, you can just try to launch any process by using the native
2712 qemu-sparc64 /bin/ls
2717 @node BSD Command line options
2718 @subsection Command line options
2721 @command{qemu-sparc64} [@option{-h]} [@option{-d]} [@option{-L} @var{path}] [@option{-s} @var{size}] [@option{-bsd} @var{type}] @var{program} [@var{arguments}...]
2728 Set the library root path (default=/)
2730 Set the stack size in bytes (default=524288)
2731 @item -ignore-environment
2732 Start with an empty environment. Without this option,
2733 the initial environment is a copy of the caller's environment.
2734 @item -E @var{var}=@var{value}
2735 Set environment @var{var} to @var{value}.
2737 Remove @var{var} from the environment.
2739 Set the type of the emulated BSD Operating system. Valid values are
2740 FreeBSD, NetBSD and OpenBSD (default).
2747 Activate logging of the specified items (use '-d help' for a list of log items)
2749 Act as if the host page size was 'pagesize' bytes
2751 Run the emulation in single step mode.
2755 @include qemu-tech.texi
2757 @node Deprecated features
2758 @appendix Deprecated features
2760 In general features are intended to be supported indefinitely once
2761 introduced into QEMU. In the event that a feature needs to be removed,
2762 it will be listed in this appendix. The feature will remain functional
2763 for 2 releases prior to actual removal. Deprecated features may also
2764 generate warnings on the console when QEMU starts up, or if activated
2765 via a monitor command, however, this is not a mandatory requirement.
2767 Prior to the 2.10.0 release there was no official policy on how
2768 long features would be deprecated prior to their removal, nor
2769 any documented list of which features were deprecated. Thus
2770 any features deprecated prior to 2.10.0 will be treated as if
2771 they were first deprecated in the 2.10.0 release.
2773 What follows is a list of all features currently marked as
2776 @section Build options
2780 Previously QEMU has supported building against both GTK 2.x
2781 and 3.x series APIs. Support for the GTK 2.x builds will be
2782 discontinued, so maintainers should switch to using GTK 3.x,
2783 which is the default.
2787 Previously QEMU has supported building against both SDL 1.2
2788 and 2.0 series APIs. Support for the SDL 1.2 builds will be
2789 discontinued, so maintainers should switch to using SDL 2.0,
2790 which is the default.
2792 @section System emulator command line arguments
2794 @subsection -no-kvm (since 1.3.0)
2796 The ``-no-kvm'' argument is now a synonym for setting
2797 ``-machine accel=tcg''.
2799 @subsection -vnc tls (since 2.5.0)
2801 The ``-vnc tls'' argument is now a synonym for setting
2802 ``-object tls-creds-anon,id=tls0'' combined with
2803 ``-vnc tls-creds=tls0'
2805 @subsection -vnc x509 (since 2.5.0)
2807 The ``-vnc x509=/path/to/certs'' argument is now a
2809 ``-object tls-creds-x509,dir=/path/to/certs,id=tls0,verify-peer=no''
2810 combined with ``-vnc tls-creds=tls0'
2812 @subsection -vnc x509verify (since 2.5.0)
2814 The ``-vnc x509verify=/path/to/certs'' argument is now a
2816 ``-object tls-creds-x509,dir=/path/to/certs,id=tls0,verify-peer=yes''
2817 combined with ``-vnc tls-creds=tls0'
2819 @subsection -tftp (since 2.6.0)
2821 The ``-tftp /some/dir'' argument is replaced by either
2822 ``-netdev user,id=x,tftp=/some/dir '' (for pluggable NICs, accompanied
2823 with ``-device ...,netdev=x''), or ``-nic user,tftp=/some/dir''
2824 (for embedded NICs). The new syntax allows different settings to be
2827 @subsection -bootp (since 2.6.0)
2829 The ``-bootp /some/file'' argument is replaced by either
2830 ``-netdev user,id=x,bootp=/some/file '' (for pluggable NICs, accompanied
2831 with ``-device ...,netdev=x''), or ``-nic user,bootp=/some/file''
2832 (for embedded NICs). The new syntax allows different settings to be
2835 @subsection -redir (since 2.6.0)
2837 The ``-redir [tcp|udp]:hostport:[guestaddr]:guestport'' argument is
2839 ``-netdev user,id=x,hostfwd=[tcp|udp]:[hostaddr]:hostport-[guestaddr]:guestport''
2840 (for pluggable NICs, accompanied with ``-device ...,netdev=x'') or
2841 ``-nic user,hostfwd=[tcp|udp]:[hostaddr]:hostport-[guestaddr]:guestport''
2842 (for embedded NICs). The new syntax allows different settings to be
2845 @subsection -smb (since 2.6.0)
2847 The ``-smb /some/dir'' argument is replaced by either
2848 ``-netdev user,id=x,smb=/some/dir '' (for pluggable NICs, accompanied
2849 with ``-device ...,netdev=x''), or ``-nic user,smb=/some/dir''
2850 (for embedded NICs). The new syntax allows different settings to be
2853 @subsection -drive cyls=...,heads=...,secs=...,trans=... (since 2.10.0)
2855 The drive geometry arguments are replaced by the the geometry arguments
2856 that can be specified with the ``-device'' parameter.
2858 @subsection -drive serial=... (since 2.10.0)
2860 The drive serial argument is replaced by the the serial argument
2861 that can be specified with the ``-device'' parameter.
2863 @subsection -drive addr=... (since 2.10.0)
2865 The drive addr argument is replaced by the the addr argument
2866 that can be specified with the ``-device'' parameter.
2868 @subsection -usbdevice (since 2.10.0)
2870 The ``-usbdevice DEV'' argument is now a synonym for setting
2871 the ``-device usb-DEV'' argument instead. The deprecated syntax
2872 would automatically enable USB support on the machine type.
2873 If using the new syntax, USB support must be explicitly
2874 enabled via the ``-machine usb=on'' argument.
2876 @subsection -nodefconfig (since 2.11.0)
2878 The ``-nodefconfig`` argument is a synonym for ``-no-user-config``.
2880 @subsection -balloon (since 2.12.0)
2882 The @option{--balloon virtio} argument has been superseded by
2883 @option{--device virtio-balloon}.
2885 @subsection -machine s390-squash-mcss=on|off (since 2.12.0)
2887 The ``s390-squash-mcss=on`` property has been obsoleted by allowing the
2888 cssid to be chosen freely. Instead of squashing subchannels into the
2889 default channel subsystem image for guests that do not support multiple
2890 channel subsystems, all devices can be put into the default channel
2893 @subsection -fsdev handle (since 2.12.0)
2895 The ``handle'' fsdev backend does not support symlinks and causes the 9p
2896 filesystem in the guest to fail a fair amount of tests from the PJD POSIX
2897 filesystem test suite. Also it requires the CAP_DAC_READ_SEARCH capability,
2898 which is not the recommended way to run QEMU. This backend should not be
2899 used and it will be removed with no replacement.
2901 @subsection -no-frame (since 2.12.0)
2903 The @code{--no-frame} argument works with SDL 1.2 only. The other user
2904 interfaces never implemented this in the first place. So this will be
2905 removed together with SDL 1.2 support.
2907 @subsection -rtc-td-hack (since 2.12.0)
2909 The @code{-rtc-td-hack} option has been replaced by
2910 @code{-rtc driftfix=slew}.
2912 @subsection -localtime (since 2.12.0)
2914 The @code{-localtime} option has been replaced by @code{-rtc base=localtime}.
2916 @subsection -startdate (since 2.12.0)
2918 The @code{-startdate} option has been replaced by @code{-rtc base=@var{date}}.
2920 @subsection -virtioconsole (since 3.0.0)
2922 Option @option{-virtioconsole} has been replaced by
2923 @option{-device virtconsole}.
2925 @subsection -clock (since 3.0.0)
2927 The @code{-clock} option is ignored since QEMU version 1.7.0. There is no
2928 replacement since it is not needed anymore.
2930 @section QEMU Machine Protocol (QMP) commands
2932 @subsection block-dirty-bitmap-add "autoload" parameter (since 2.12.0)
2934 "autoload" parameter is now ignored. All bitmaps are automatically loaded
2937 @subsection query-cpus (since 2.12.0)
2939 The ``query-cpus'' command is replaced by the ``query-cpus-fast'' command.
2941 @subsection query-cpus-fast "arch" output member (since 3.0.0)
2943 The ``arch'' output member of the ``query-cpus-fast'' command is
2944 replaced by the ``target'' output member.
2946 @section System emulator devices
2948 @subsection ivshmem (since 2.6.0)
2950 The ``ivshmem'' device type is replaced by either the ``ivshmem-plain''
2951 or ``ivshmem-doorbell`` device types.
2953 @subsection Page size support < 4k for embedded PowerPC CPUs (since 2.12.0)
2955 qemu-system-ppcemb will be removed. qemu-system-ppc (or qemu-system-ppc64)
2956 should be used instead. That means that embedded 4xx PowerPC CPUs will not
2957 support page sizes < 4096 any longer.
2959 @section System emulator machines
2961 @section Device options
2963 @subsection Block device options
2965 @subsubsection "backing": "" (since 2.12.0)
2967 In order to prevent QEMU from automatically opening an image's backing
2968 chain, use ``"backing": null'' instead.
2970 @subsection vio-spapr-device device options
2972 @subsubsection "irq": "" (since 3.0.0)
2974 The ``irq'' property is obsoleted.
2976 @node Supported build platforms
2977 @appendix Supported build platforms
2979 QEMU aims to support building and executing on multiple host OS platforms.
2980 This appendix outlines which platforms are the major build targets. These
2981 platforms are used as the basis for deciding upon the minimum required
2982 versions of 3rd party software QEMU depends on. The supported platforms
2983 are the targets for automated testing performed by the project when patches
2984 are submitted for review, and tested before and after merge.
2986 If a platform is not listed here, it does not imply that QEMU won't work.
2987 If an unlisted platform has comparable software versions to a listed platform,
2988 there is every expectation that it will work. Bug reports are welcome for
2989 problems encountered on unlisted platforms unless they are clearly older
2990 vintage than what is described here.
2992 Note that when considering software versions shipped in distros as support
2993 targets, QEMU considers only the version number, and assumes the features in
2994 that distro match the upstream release with the same version. In other words,
2995 if a distro backports extra features to the software in their distro, QEMU
2996 upstream code will not add explicit support for those backports, unless the
2997 feature is auto-detectable in a manner that works for the upstream releases
3000 The Repology site @url{https://repology.org} is a useful resource to identify
3001 currently shipped versions of software in various operating systems, though
3002 it does not cover all distros listed below.
3006 For distributions with frequent, short-lifetime releases, the project will
3007 aim to support all versions that are not end of life by their respective
3008 vendors. For the purposes of identifying supported software versions, the
3009 project will look at Fedora, Ubuntu, and openSUSE distros. Other short-
3010 lifetime distros will be assumed to ship similar software versions.
3012 For distributions with long-lifetime releases, the project will aim to support
3013 the most recent major version at all times. Support for the previous major
3014 version will be dropped 2 years after the new major version is released. For
3015 the purposes of identifying supported software versions, the project will look
3016 at RHEL, Debian, Ubuntu LTS, and SLES distros. Other long-lifetime distros will
3017 be assumed to ship similar software versions.
3021 The project supports building with current versions of the MinGW toolchain,
3026 The project supports building with the two most recent versions of macOS, with
3027 the current homebrew package set available.
3031 The project aims to support the all the versions which are not end of life.
3035 The project aims to support the most recent major version at all times. Support
3036 for the previous major version will be dropped 2 years after the new major
3037 version is released.
3041 The project aims to support the all the versions which are not end of life.
3046 QEMU is a trademark of Fabrice Bellard.
3048 QEMU is released under the
3049 @url{https://www.gnu.org/licenses/gpl-2.0.txt,GNU General Public License},
3050 version 2. Parts of QEMU have specific licenses, see file
3051 @url{https://git.qemu.org/?p=qemu.git;a=blob_plain;f=LICENSE,LICENSE}.
3065 @section Concept Index
3066 This is the main index. Should we combine all keywords in one index? TODO
3069 @node Function Index
3070 @section Function Index
3071 This index could be used for command line options and monitor functions.
3074 @node Keystroke Index
3075 @section Keystroke Index
3077 This is a list of all keystrokes which have a special function
3078 in system emulation.
3083 @section Program Index
3086 @node Data Type Index
3087 @section Data Type Index
3089 This index could be used for qdev device names and options.
3093 @node Variable Index
3094 @section Variable Index