xref: /qemu/docs/about/deprecated.rst (revision bcbd8c0edf712a7d55d9bf020be4b0f21ce3c35c)
1.. _Deprecated features:
2
3Deprecated features
4===================
5
6In general features are intended to be supported indefinitely once
7introduced into QEMU. In the event that a feature needs to be removed,
8it will be listed in this section. The feature will remain functional for the
9release in which it was deprecated and one further release. After these two
10releases, the feature is liable to be removed. Deprecated features may also
11generate warnings on the console when QEMU starts up, or if activated via a
12monitor command, however, this is not a mandatory requirement.
13
14As a special exception to this general timeframe, rather than have an
15indefinite lifetime, versioned machine types are only intended to be
16supported for a period of 6 years, equivalent to 18 QEMU releases. All
17versioned machine types will be automatically marked deprecated after an
18initial 3 years (9 QEMU releases) has passed, and will then be deleted after
19a further 3 year period has passed. It is recommended that a deprecated
20machine type is only used for incoming migrations and restore of saved state,
21for pre-existing VM deployments. They should be scheduled for updating to a
22newer machine type during an appropriate service window. Newly deployed VMs
23should exclusively use a non-deprecated machine type, with use of the most
24recent version highly recommended. Non-versioned machine types follow the
25general feature deprecation policy.
26
27What follows is a list of all features currently marked as
28deprecated.
29
30System emulator command line arguments
31--------------------------------------
32
33Short-form boolean options (since 6.0)
34''''''''''''''''''''''''''''''''''''''
35
36Boolean options such as ``share=on``/``share=off`` could be written
37in short form as ``share`` and ``noshare``.  This is now deprecated
38and will cause a warning.
39
40``delay`` option for socket character devices (since 6.0)
41'''''''''''''''''''''''''''''''''''''''''''''''''''''''''
42
43The replacement for the ``nodelay`` short-form boolean option is ``nodelay=on``
44rather than ``delay=off``.
45
46Plugin argument passing through ``arg=<string>`` (since 6.1)
47''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
48
49Passing TCG plugins arguments through ``arg=`` is redundant is makes the
50command-line less readable, especially when the argument itself consist of a
51name and a value, e.g. ``-plugin plugin_name,arg="arg_name=arg_value"``.
52Therefore, the usage of ``arg`` is redundant. Single-word arguments are treated
53as short-form boolean values, and passed to plugins as ``arg_name=on``.
54However, short-form booleans are deprecated and full explicit ``arg_name=on``
55form is preferred.
56
57``-smp`` (Unsupported "parameter=1" SMP configurations) (since 9.0)
58'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
59
60Specified CPU topology parameters must be supported by the machine.
61
62In the SMP configuration, users should provide the CPU topology parameters that
63are supported by the target machine.
64
65However, historically it was allowed for users to specify the unsupported
66topology parameter as "1", which is meaningless. So support for this kind of
67configurations (e.g. -smp drawers=1,books=1,clusters=1 for x86 PC machine) is
68marked deprecated since 9.0, users have to ensure that all the topology members
69described with -smp are supported by the target machine.
70
71``-old-param`` option for booting Arm kernels via param_struct (since 10.0)
72'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
73
74The ``-old-param`` command line option is specific to Arm targets:
75it is used when directly booting a guest kernel to pass it the
76command line and other information via the old ``param_struct`` ABI,
77rather than the newer ATAGS or DTB mechanisms. This option was only
78ever needed to support ancient kernels on some old board types
79like the ``akita`` or ``terrier``; it has been deprecated in the
80kernel since 2001. None of the board types QEMU supports need
81``param_struct`` support, so this option has been deprecated and will
82be removed in a future QEMU version.
83
84User-mode emulator command line arguments
85-----------------------------------------
86
87``-p`` (since 9.0)
88''''''''''''''''''
89
90The ``-p`` option pretends to control the host page size.  However,
91it is not possible to change the host page size, and using the
92option only causes failures.
93
94QEMU Machine Protocol (QMP) commands
95------------------------------------
96
97``blockdev-open-tray``, ``blockdev-close-tray`` argument ``device`` (since 2.8)
98'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
99
100Use argument ``id`` instead.
101
102``eject`` argument ``device`` (since 2.8)
103'''''''''''''''''''''''''''''''''''''''''
104
105Use argument ``id`` instead.
106
107``blockdev-change-medium`` argument ``device`` (since 2.8)
108''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
109
110Use argument ``id`` instead.
111
112``block_set_io_throttle`` argument ``device`` (since 2.8)
113'''''''''''''''''''''''''''''''''''''''''''''''''''''''''
114
115Use argument ``id`` instead.
116
117``blockdev-add`` empty string argument ``backing`` (since 2.10)
118'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
119
120Use argument value ``null`` instead.
121
122``block-commit`` arguments ``base`` and ``top`` (since 3.1)
123'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
124
125Use arguments ``base-node`` and ``top-node`` instead.
126
127``nbd-server-add`` and ``nbd-server-remove`` (since 5.2)
128''''''''''''''''''''''''''''''''''''''''''''''''''''''''
129
130Use the more generic commands ``block-export-add`` and ``block-export-del``
131instead.  As part of this deprecation, where ``nbd-server-add`` used a
132single ``bitmap``, the new ``block-export-add`` uses a list of ``bitmaps``.
133
134``query-qmp-schema`` return value member ``values`` (since 6.2)
135'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
136
137Member ``values`` in return value elements with meta-type ``enum`` is
138deprecated.  Use ``members`` instead.
139
140``drive-backup`` (since 6.2)
141''''''''''''''''''''''''''''
142
143Use ``blockdev-backup`` in combination with ``blockdev-add`` instead.
144This change primarily separates the creation/opening process of the backup
145target with explicit, separate steps. ``blockdev-backup`` uses mostly the
146same arguments as ``drive-backup``, except the ``format`` and ``mode``
147options are removed in favor of using explicit ``blockdev-create`` and
148``blockdev-add`` calls. See :doc:`/interop/live-block-operations` for
149details.
150
151``query-migrationthreads`` (since 9.2)
152''''''''''''''''''''''''''''''''''''''
153
154To be removed with no replacement, as it reports only a limited set of
155threads (for example, it only reports source side of multifd threads,
156without reporting any destination threads, or non-multifd source threads).
157For debugging purpose, please use ``-name $VM,debug-threads=on`` instead.
158
159Incorrectly typed ``device_add`` arguments (since 6.2)
160''''''''''''''''''''''''''''''''''''''''''''''''''''''
161
162Due to shortcomings in the internal implementation of ``device_add``, QEMU
163incorrectly accepts certain invalid arguments: Any object or list arguments are
164silently ignored. Other argument types are not checked, but an implicit
165conversion happens, so that e.g. string values can be assigned to integer
166device properties or vice versa.
167
168This is a bug in QEMU that will be fixed in the future so that previously
169accepted incorrect commands will return an error. Users should make sure that
170all arguments passed to ``device_add`` are consistent with the documented
171property types.
172
173Host Architectures
174------------------
175
176Big endian MIPS since 7.2; 32-bit little endian MIPS since 9.2
177''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
178
179As Debian 10 ("Buster") moved into LTS the big endian 32 bit version of
180MIPS moved out of support making it hard to maintain our
181cross-compilation CI tests of the architecture. As we no longer have
182CI coverage support may bitrot away before the deprecation process
183completes.
184
185Likewise, the little endian variant of 32 bit MIPS is not supported by
186Debian 13 ("Trixie") and newer.
187
18864 bit little endian MIPS is still a supported host architecture.
189
190System emulation on 32-bit x86 hosts (since 8.0)
191''''''''''''''''''''''''''''''''''''''''''''''''
192
193Support for 32-bit x86 host deployments is increasingly uncommon in mainstream
194OS distributions given the widespread availability of 64-bit x86 hardware.
195The QEMU project no longer considers 32-bit x86 support for system emulation to
196be an effective use of its limited resources, and thus intends to discontinue
197it. Since all recent x86 hardware from the past >10 years is capable of the
19864-bit x86 extensions, a corresponding 64-bit OS should be used instead.
199
200TCG Plugin support not enabled by default on 32-bit hosts (since 9.2)
201'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
202
203While it is still possible to enable TCG plugin support for 32-bit
204hosts there are a number of potential pitfalls when instrumenting
20564-bit guests. The plugin APIs typically pass most addresses as
206uint64_t but practices like encoding that address in a host pointer
207for passing as user-data will lose data. As most software analysis
208benefits from having plenty of host memory it seems reasonable to
209encourage users to use 64 bit builds of QEMU for analysis work
210whatever targets they are instrumenting.
211
212TCG Plugin support not enabled by default with TCI (since 9.2)
213''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
214
215While the TCG interpreter can interpret the TCG ops used by plugins it
216is going to be so much slower it wouldn't make sense for any serious
217instrumentation. Due to implementation differences there will also be
218anomalies in things like memory instrumentation.
219
22032-bit host operating systems (since 10.0)
221''''''''''''''''''''''''''''''''''''''''''
222
223Keeping 32-bit host support alive is a substantial burden for the
224QEMU project.  Thus QEMU will in future drop the support for all
22532-bit host systems.
226
227linux-user mode CPUs
228--------------------
229
230iwMMXt emulation and the ``pxa`` CPUs (since 10.0)
231''''''''''''''''''''''''''''''''''''''''''''''''''
232
233The ``pxa`` CPU family (``pxa250``, ``pxa255``, ``pxa260``,
234``pxa261``, ``pxa262``, ``pxa270-a0``, ``pxa270-a1``, ``pxa270``,
235``pxa270-b0``, ``pxa270-b1``, ``pxa270-c0``, ``pxa270-c5``) are no
236longer used in system emulation, because all the machine types which
237used these CPUs were removed in the QEMU 9.2 release. These CPUs can
238now only be used in linux-user mode, and to do that you would have to
239explicitly select one of these CPUs with the ``-cpu`` command line
240option or the ``QEMU_CPU`` environment variable.
241
242We don't believe that anybody is using the iwMMXt emulation, and we do
243not have any tests to validate it or any real hardware or similar
244known-good implementation to test against. GCC is in the process of
245dropping their support for iwMMXt codegen. These CPU types are
246therefore deprecated in QEMU, and will be removed in a future release.
247
248System emulator CPUs
249--------------------
250
251``power5+`` and ``power7+`` CPU names (since 9.0)
252'''''''''''''''''''''''''''''''''''''''''''''''''
253
254The character "+" in device (and thus also CPU) names is not allowed
255in the QEMU object model anymore. ``power5+``, ``power5+_v2.1``,
256``power7+`` and ``power7+_v2.1`` are currently still supported via
257an alias, but for consistency these will get removed in a future
258release, too. Use ``power5p_v2.1`` and ``power7p_v2.1`` instead.
259
260``Sun-UltraSparc-IIIi+`` and ``Sun-UltraSparc-IV+`` CPU names (since 9.1)
261'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
262
263The character "+" in device (and thus also CPU) names is not allowed
264in the QEMU object model anymore. ``Sun-UltraSparc-IIIi+`` and
265``Sun-UltraSparc-IV+`` are currently still supported via a workaround,
266but for consistency these will get removed in a future release, too.
267Use ``Sun-UltraSparc-IIIi-plus`` and ``Sun-UltraSparc-IV-plus`` instead.
268
269PPC 405 CPUs (since 10.0)
270'''''''''''''''''''''''''
271
272The PPC 405 CPU has no known users and the ``ref405ep`` machine was
273removed in QEMU 10.0. Since the IBM POWER [8-11] processors uses an
274embedded 405 for power management (OCC) and other internal tasks, it
275is theoretically possible to use QEMU to model them. Let's keep the
276CPU implementation for a while before removing all support.
277
278System emulator machines
279------------------------
280
281Arm ``virt`` machine ``dtb-kaslr-seed`` property (since 7.1)
282''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
283
284The ``dtb-kaslr-seed`` property on the ``virt`` board has been
285deprecated; use the new name ``dtb-randomness`` instead. The new name
286better reflects the way this property affects all random data within
287the device tree blob, not just the ``kaslr-seed`` node.
288
289Big-Endian variants of MicroBlaze ``petalogix-ml605`` and ``xlnx-zynqmp-pmu`` machines (since 9.2)
290''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
291
292Both ``petalogix-ml605`` and ``xlnx-zynqmp-pmu`` were added for little endian
293CPUs. Big endian support is not tested.
294
295Backend options
296---------------
297
298Using non-persistent backing file with pmem=on (since 6.1)
299''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
300
301This option is used when ``memory-backend-file`` is consumed by emulated NVDIMM
302device. However enabling ``memory-backend-file.pmem`` option, when backing file
303is (a) not DAX capable or (b) not on a filesystem that support direct mapping
304of persistent memory, is not safe and may lead to data loss or corruption in case
305of host crash.
306Options are:
307
308    - modify VM configuration to set ``pmem=off`` to continue using fake NVDIMM
309      (without persistence guaranties) with backing file on non DAX storage
310    - move backing file to NVDIMM storage and keep ``pmem=on``
311      (to have NVDIMM with persistence guaranties).
312
313Device options
314--------------
315
316Emulated device options
317'''''''''''''''''''''''
318
319``-device nvme-ns,eui64-default=on|off`` (since 7.1)
320^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
321
322In QEMU versions 6.1, 6.2 and 7.0, the ``nvme-ns`` generates an EUI-64
323identifier that is not globally unique. If an EUI-64 identifier is required, the
324user must set it explicitly using the ``nvme-ns`` device parameter ``eui64``.
325
326``-device nvme,use-intel-id=on|off`` (since 7.1)
327^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
328
329The ``nvme`` device originally used a PCI Vendor/Device Identifier combination
330from Intel that was not properly allocated. Since version 5.2, the controller
331has used a properly allocated identifier. Deprecate the ``use-intel-id``
332machine compatibility parameter.
333
334``-device cxl-type3,memdev=xxxx`` (since 8.0)
335^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
336
337The ``cxl-type3`` device initially only used a single memory backend.  With
338the addition of volatile memory support, it is now necessary to distinguish
339between persistent and volatile memory backends.  As such, memdev is deprecated
340in favor of persistent-memdev.
341
342
343RISC-V CPU properties which start with capital 'Z' (since 8.2)
344^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
345
346All RISC-V CPU properties which start with capital 'Z' are being deprecated
347starting in 8.2. The reason is that they were wrongly added with capital 'Z'
348in the past. CPU properties were later added with lower-case names, which
349is the format we want to use from now on.
350
351Users which try to use these deprecated properties will receive a warning
352recommending to switch to their stable counterparts:
353
354- "Zifencei" should be replaced with "zifencei"
355- "Zicsr" should be replaced with "zicsr"
356- "Zihintntl" should be replaced with "zihintntl"
357- "Zihintpause" should be replaced with "zihintpause"
358- "Zawrs" should be replaced with "zawrs"
359- "Zfa" should be replaced with "zfa"
360- "Zfh" should be replaced with "zfh"
361- "Zfhmin" should be replaced with "zfhmin"
362- "Zve32f" should be replaced with "zve32f"
363- "Zve64f" should be replaced with "zve64f"
364- "Zve64d" should be replaced with "zve64d"
365
366``-device sd-card,spec_version=1`` (since 9.1)
367^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
368
369SD physical layer specification v2.00 supersedes the v1.10 one.
370v2.00 is the default since QEMU 3.0.0.
371
372Block device options
373''''''''''''''''''''
374
375``"backing": ""`` (since 2.12)
376^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
377
378In order to prevent QEMU from automatically opening an image's backing
379chain, use ``"backing": null`` instead.
380
381``rbd`` keyvalue pair encoded filenames: ``""`` (since 3.1)
382^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
383
384Options for ``rbd`` should be specified according to its runtime options,
385like other block drivers.  Legacy parsing of keyvalue pair encoded
386filenames is useful to open images with the old format for backing files;
387These image files should be updated to use the current format.
388
389Example of legacy encoding::
390
391  json:{"file.driver":"rbd", "file.filename":"rbd:rbd/name"}
392
393The above, converted to the current supported format::
394
395  json:{"file.driver":"rbd", "file.pool":"rbd", "file.image":"name"}
396
397``iscsi,password=xxx`` (since 8.0)
398^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
399
400Specifying the iSCSI password in plain text on the command line using the
401``password`` option is insecure. The ``password-secret`` option should be
402used instead, to refer to a ``--object secret...`` instance that provides
403a password via a file, or encrypted.
404
405``gluster`` backend (since 9.2)
406^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
407
408According to https://marc.info/?l=fedora-devel-list&m=171934833215726
409the GlusterFS development effectively ended. Unless the development
410gains momentum again, the QEMU project will remove the gluster backend
411in a future release.
412
413
414Character device options
415''''''''''''''''''''''''
416
417Backend ``memory`` (since 9.0)
418^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
419
420``memory`` is a deprecated synonym for ``ringbuf``.
421
422``reconnect`` (since 9.2)
423^^^^^^^^^^^^^^^^^^^^^^^^^
424
425The ``reconnect`` option only allows specifiying second granularity timeouts,
426which is not enough for all types of use cases, use ``reconnect-ms`` instead.
427
428
429Net device options
430''''''''''''''''''
431
432Stream ``reconnect`` (since 9.2)
433^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
434
435The ``reconnect`` option only allows specifiying second granularity timeouts,
436which is not enough for all types of use cases, use ``reconnect-ms`` instead.
437
438VFIO device options
439'''''''''''''''''''
440
441``-device vfio-calxeda-xgmac`` (since 10.0)
442^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
443The vfio-calxeda-xgmac device allows to assign a host Calxeda Highbank
44410Gb XGMAC Ethernet controller device ("calxeda,hb-xgmac" compatibility
445string) to a guest. Calxeda HW has been ewasted now and there is no point
446keeping that device.
447
448``-device vfio-amd-xgbe`` (since 10.0)
449^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
450The vfio-amd-xgbe device allows to assign a host AMD 10GbE controller
451to a guest ("amd,xgbe-seattle-v1a" compatibility string). AMD "Seattle"
452is not supported anymore and there is no point keeping that device.
453
454``-device vfio-platform`` (since 10.0)
455^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
456The vfio-platform device allows to assign a host platform device
457to a guest in a generic manner. Integrating a new device into
458the vfio-platform infrastructure requires some adaptation at
459both kernel and qemu level. No such attempt has been done for years
460and the conclusion is that vfio-platform has not got any traction.
461PCIe passthrough shall be the mainline solution.
462
463CPU device properties
464'''''''''''''''''''''
465
466``pcommit`` on x86 (since 9.1)
467^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
468
469The PCOMMIT instruction was never included in any physical processor.
470It was implemented as a no-op instruction in TCG up to QEMU 9.0, but
471only with ``-cpu max`` (which does not guarantee migration compatibility
472across versions).
473
474``pmu-num=n`` on RISC-V CPUs (since 8.2)
475^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
476
477In order to support more flexible counter configurations this has been replaced
478by a ``pmu-mask`` property. If set of counters is continuous then the mask can
479be calculated with ``((2 ^ n) - 1) << 3``. The least significant three bits
480must be left clear.
481
482
483Backwards compatibility
484-----------------------
485
486Runnability guarantee of CPU models (since 4.1)
487'''''''''''''''''''''''''''''''''''''''''''''''
488
489Previous versions of QEMU never changed existing CPU models in
490ways that introduced additional host software or hardware
491requirements to the VM.  This allowed management software to
492safely change the machine type of an existing VM without
493introducing new requirements ("runnability guarantee").  This
494prevented CPU models from being updated to include CPU
495vulnerability mitigations, leaving guests vulnerable in the
496default configuration.
497
498The CPU model runnability guarantee won't apply anymore to
499existing CPU models.  Management software that needs runnability
500guarantees must resolve the CPU model aliases using the
501``alias-of`` field returned by the ``query-cpu-definitions`` QMP
502command.
503
504While those guarantees are kept, the return value of
505``query-cpu-definitions`` will have existing CPU model aliases
506point to a version that doesn't break runnability guarantees
507(specifically, version 1 of those CPU models).  In future QEMU
508versions, aliases will point to newer CPU model versions
509depending on the machine type, so management software must
510resolve CPU model aliases before starting a virtual machine.
511
512RISC-V "virt" board "riscv,delegate" DT property (since 9.1)
513''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
514
515The "riscv,delegate" DT property was added in QEMU 7.0 as part of
516the AIA APLIC support.  The property changed name during the review
517process in Linux and the correct name ended up being
518"riscv,delegation".  Changing the DT property name will break all
519available firmwares that are using the current (wrong) name.  The
520property is kept as is in 9.1, together with "riscv,delegation", to
521give more time for firmware developers to change their code.
522
523Migration
524---------
525
526``fd:`` URI when used for file migration (since 9.1)
527''''''''''''''''''''''''''''''''''''''''''''''''''''
528
529The ``fd:`` URI can currently provide a file descriptor that
530references either a socket or a plain file. These are two different
531types of migration. In order to reduce ambiguity, the ``fd:`` URI
532usage of providing a file descriptor to a plain file has been
533deprecated in favor of explicitly using the ``file:`` URI with the
534file descriptor being passed as an ``fdset``. Refer to the ``add-fd``
535command documentation for details on the ``fdset`` usage.
536
537``zero-blocks`` capability (since 9.2)
538''''''''''''''''''''''''''''''''''''''
539
540The ``zero-blocks`` capability was part of the block migration which
541doesn't exist anymore since it was removed in QEMU v9.1.
542