xref: /dpdk/doc/guides/prog_guide/vhost_lib.rst (revision 2a1e2da1bb34c7cc0145fdd8940047ae325277bc)
1..  SPDX-License-Identifier: BSD-3-Clause
2    Copyright(c) 2010-2016 Intel Corporation.
3
4Vhost Library
5=============
6
7The vhost library implements a user space virtio net server allowing the user
8to manipulate the virtio ring directly. In another words, it allows the user
9to fetch/put packets from/to the VM virtio net device. To achieve this, a
10vhost library should be able to:
11
12* Access the guest memory:
13
14  For QEMU, this is done by using the ``-object memory-backend-file,share=on,...``
15  option. Which means QEMU will create a file to serve as the guest RAM.
16  The ``share=on`` option allows another process to map that file, which
17  means it can access the guest RAM.
18
19* Know all the necessary information about the vring:
20
21  Information such as where the available ring is stored. Vhost defines some
22  messages (passed through a Unix domain socket file) to tell the backend all
23  the information it needs to know how to manipulate the vring.
24
25
26Vhost API Overview
27------------------
28
29The following is an overview of some key Vhost API functions:
30
31* ``rte_vhost_driver_register(path, flags)``
32
33  This function registers a vhost driver into the system. ``path`` specifies
34  the Unix domain socket file path.
35
36  Currently supported flags are:
37
38  - ``RTE_VHOST_USER_CLIENT``
39
40    DPDK vhost-user will act as the client when this flag is given. See below
41    for an explanation.
42
43  - ``RTE_VHOST_USER_NO_RECONNECT``
44
45    When DPDK vhost-user acts as the client it will keep trying to reconnect
46    to the server (QEMU) until it succeeds. This is useful in two cases:
47
48    * When QEMU is not started yet.
49    * When QEMU restarts (for example due to a guest OS reboot).
50
51    This reconnect option is enabled by default. However, it can be turned off
52    by setting this flag.
53
54  - ``RTE_VHOST_USER_DEQUEUE_ZERO_COPY``
55
56    Dequeue zero copy will be enabled when this flag is set. It is disabled by
57    default.
58
59    There are some truths (including limitations) you might want to know while
60    setting this flag:
61
62    * zero copy is not good for small packets (typically for packet size below
63      512).
64
65    * zero copy is really good for VM2VM case. For iperf between two VMs, the
66      boost could be above 70% (when TSO is enableld).
67
68    * for VM2NIC case, the ``nb_tx_desc`` has to be small enough: <= 64 if virtio
69      indirect feature is not enabled and <= 128 if it is enabled.
70
71      This is because when dequeue zero copy is enabled, guest Tx used vring will
72      be updated only when corresponding mbuf is freed. Thus, the nb_tx_desc
73      has to be small enough so that the PMD driver will run out of available
74      Tx descriptors and free mbufs timely. Otherwise, guest Tx vring would be
75      starved.
76
77    * Guest memory should be backended with huge pages to achieve better
78      performance. Using 1G page size is the best.
79
80      When dequeue zero copy is enabled, the guest phys address and host phys
81      address mapping has to be established. Using non-huge pages means far
82      more page segments. To make it simple, DPDK vhost does a linear search
83      of those segments, thus the fewer the segments, the quicker we will get
84      the mapping. NOTE: we may speed it by using tree searching in future.
85
86    * zero copy can not work when using vfio-pci with iommu mode currently, this
87      is because we don't setup iommu dma mapping for guest memory. If you have
88      to use vfio-pci driver, please insert vfio-pci kernel module in noiommu
89      mode.
90
91  - ``RTE_VHOST_USER_IOMMU_SUPPORT``
92
93    IOMMU support will be enabled when this flag is set. It is disabled by
94    default.
95
96    Enabling this flag makes possible to use guest vIOMMU to protect vhost
97    from accessing memory the virtio device isn't allowed to, when the feature
98    is negotiated and an IOMMU device is declared.
99
100    However, this feature enables vhost-user's reply-ack protocol feature,
101    which implementation is buggy in Qemu v2.7.0-v2.9.0 when doing multiqueue.
102    Enabling this flag with these Qemu version results in Qemu being blocked
103    when multiple queue pairs are declared.
104
105* ``rte_vhost_driver_set_features(path, features)``
106
107  This function sets the feature bits the vhost-user driver supports. The
108  vhost-user driver could be vhost-user net, yet it could be something else,
109  say, vhost-user SCSI.
110
111* ``rte_vhost_driver_callback_register(path, vhost_device_ops)``
112
113  This function registers a set of callbacks, to let DPDK applications take
114  the appropriate action when some events happen. The following events are
115  currently supported:
116
117  * ``new_device(int vid)``
118
119    This callback is invoked when a virtio device becomes ready. ``vid``
120    is the vhost device ID.
121
122  * ``destroy_device(int vid)``
123
124    This callback is invoked when a virtio device is paused or shut down.
125
126  * ``vring_state_changed(int vid, uint16_t queue_id, int enable)``
127
128    This callback is invoked when a specific queue's state is changed, for
129    example to enabled or disabled.
130
131  * ``features_changed(int vid, uint64_t features)``
132
133    This callback is invoked when the features is changed. For example,
134    ``VHOST_F_LOG_ALL`` will be set/cleared at the start/end of live
135    migration, respectively.
136
137  * ``new_connection(int vid)``
138
139    This callback is invoked on new vhost-user socket connection. If DPDK
140    acts as the server the device should not be deleted before
141    ``destroy_connection`` callback is received.
142
143  * ``destroy_connection(int vid)``
144
145    This callback is invoked when vhost-user socket connection is closed.
146    It indicates that device with id ``vid`` is no longer in use and can be
147    safely deleted.
148
149* ``rte_vhost_driver_disable/enable_features(path, features))``
150
151  This function disables/enables some features. For example, it can be used to
152  disable mergeable buffers and TSO features, which both are enabled by
153  default.
154
155* ``rte_vhost_driver_start(path)``
156
157  This function triggers the vhost-user negotiation. It should be invoked at
158  the end of initializing a vhost-user driver.
159
160* ``rte_vhost_enqueue_burst(vid, queue_id, pkts, count)``
161
162  Transmits (enqueues) ``count`` packets from host to guest.
163
164* ``rte_vhost_dequeue_burst(vid, queue_id, mbuf_pool, pkts, count)``
165
166  Receives (dequeues) ``count`` packets from guest, and stored them at ``pkts``.
167
168* ``rte_vhost_crypto_create(vid, cryptodev_id, sess_mempool, socket_id)``
169
170  As an extension of new_device(), this function adds virtio-crypto workload
171  acceleration capability to the device. All crypto workload is processed by
172  DPDK cryptodev with the device ID of ``cryptodev_id``.
173
174* ``rte_vhost_crypto_free(vid)``
175
176  Frees the memory and vhost-user message handlers created in
177  rte_vhost_crypto_create().
178
179* ``rte_vhost_crypto_fetch_requests(vid, queue_id, ops, nb_ops)``
180
181  Receives (dequeues) ``nb_ops`` virtio-crypto requests from guest, parses
182  them to DPDK Crypto Operations, and fills the ``ops`` with parsing results.
183
184* ``rte_vhost_crypto_finalize_requests(queue_id, ops, nb_ops)``
185
186  After the ``ops`` are dequeued from Cryptodev, finalizes the jobs and
187  notifies the guest(s).
188
189* ``rte_vhost_crypto_set_zero_copy(vid, option)``
190
191  Enable or disable zero copy feature of the vhost crypto backend.
192
193Vhost-user Implementations
194--------------------------
195
196Vhost-user uses Unix domain sockets for passing messages. This means the DPDK
197vhost-user implementation has two options:
198
199* DPDK vhost-user acts as the server.
200
201  DPDK will create a Unix domain socket server file and listen for
202  connections from the frontend.
203
204  Note, this is the default mode, and the only mode before DPDK v16.07.
205
206
207* DPDK vhost-user acts as the client.
208
209  Unlike the server mode, this mode doesn't create the socket file;
210  it just tries to connect to the server (which responses to create the
211  file instead).
212
213  When the DPDK vhost-user application restarts, DPDK vhost-user will try to
214  connect to the server again. This is how the "reconnect" feature works.
215
216  .. Note::
217     * The "reconnect" feature requires **QEMU v2.7** (or above).
218
219     * The vhost supported features must be exactly the same before and
220       after the restart. For example, if TSO is disabled and then enabled,
221       nothing will work and issues undefined might happen.
222
223No matter which mode is used, once a connection is established, DPDK
224vhost-user will start receiving and processing vhost messages from QEMU.
225
226For messages with a file descriptor, the file descriptor can be used directly
227in the vhost process as it is already installed by the Unix domain socket.
228
229The supported vhost messages are:
230
231* ``VHOST_SET_MEM_TABLE``
232* ``VHOST_SET_VRING_KICK``
233* ``VHOST_SET_VRING_CALL``
234* ``VHOST_SET_LOG_FD``
235* ``VHOST_SET_VRING_ERR``
236
237For ``VHOST_SET_MEM_TABLE`` message, QEMU will send information for each
238memory region and its file descriptor in the ancillary data of the message.
239The file descriptor is used to map that region.
240
241``VHOST_SET_VRING_KICK`` is used as the signal to put the vhost device into
242the data plane, and ``VHOST_GET_VRING_BASE`` is used as the signal to remove
243the vhost device from the data plane.
244
245When the socket connection is closed, vhost will destroy the device.
246
247Guest memory requirement
248------------------------
249
250* Memory pre-allocation
251
252  For non-zerocopy, guest memory pre-allocation is not a must. This can help
253  save of memory. If users really want the guest memory to be pre-allocated
254  (e.g., for performance reason), we can add option ``-mem-prealloc`` when
255  starting QEMU. Or, we can lock all memory at vhost side which will force
256  memory to be allocated when mmap at vhost side; option --mlockall in
257  ovs-dpdk is an example in hand.
258
259  For zerocopy, we force the VM memory to be pre-allocated at vhost lib when
260  mapping the guest memory; and also we need to lock the memory to prevent
261  pages being swapped out to disk.
262
263* Memory sharing
264
265  Make sure ``share=on`` QEMU option is given. vhost-user will not work with
266  a QEMU version without shared memory mapping.
267
268Vhost supported vSwitch reference
269---------------------------------
270
271For more vhost details and how to support vhost in vSwitch, please refer to
272the vhost example in the DPDK Sample Applications Guide.
273