4 Copyright (c) 2014 Virtual Open Systems Sarl.
6 This work is licensed under the terms of the GNU GPL, version 2 or later.
7 See the COPYING file in the top-level directory.
10 This protocol is aiming to complement the ioctl interface used to control the
11 vhost implementation in the Linux kernel. It implements the control plane needed
12 to establish virtqueue sharing with a user space process on the same host. It
13 uses communication over a Unix domain socket to share file descriptors in the
14 ancillary data of the message.
16 The protocol defines 2 sides of the communication, master and slave. Master is
17 the application that shares its virtqueues, in our case QEMU. Slave is the
18 consumer of the virtqueues.
20 In the current implementation QEMU is the Master, and the Slave is intended to
21 be a software Ethernet switch running in user space, such as Snabbswitch.
23 Master and slave can be either a client (i.e. connecting) or server (listening)
24 in the socket communication.
29 Note that all numbers are in the machine native byte order. A vhost-user message
30 consists of 3 header fields and a payload:
32 ------------------------------------
33 | request | flags | size | payload |
34 ------------------------------------
36 * Request: 32-bit type of the request
37 * Flags: 32-bit bit field:
38 - Lower 2 bits are the version (currently 0x01)
39 - Bit 2 is the reply flag - needs to be sent on each reply from the slave
40 * Size - 32-bit size of the payload
43 Depending on the request type, payload can be:
45 * A single 64-bit integer
50 u64: a 64-bit unsigned integer
52 * A vring state description
60 * A vring address description
61 --------------------------------------------------------------
62 | index | flags | size | descriptor | used | available | log |
63 --------------------------------------------------------------
65 Index: a 32-bit vring index
66 Flags: a 32-bit vring flags
67 Descriptor: a 64-bit user address of the vring descriptor table
68 Used: a 64-bit user address of the vring used ring
69 Available: a 64-bit user address of the vring available ring
70 Log: a 64-bit guest address for logging
72 * Memory regions description
73 ---------------------------------------------------
74 | num regions | padding | region0 | ... | region7 |
75 ---------------------------------------------------
77 Num regions: a 32-bit number of regions
81 -----------------------------------------------------
82 | guest address | size | user address | mmap offset |
83 -----------------------------------------------------
85 Guest address: a 64-bit guest address of the region
87 User address: a 64-bit user address
88 mmap offset: 64-bit offset where region starts in the mapped memory
91 ---------------------------
92 | log size | log offset |
93 ---------------------------
94 log size: size of area used for logging
95 log offset: offset from start of supplied file descriptor
96 where logging starts (i.e. where guest address 0 would be logged)
98 In QEMU the vhost-user message is implemented with the following struct:
100 typedef struct VhostUserMsg {
101 VhostUserRequest request;
106 struct vhost_vring_state state;
107 struct vhost_vring_addr addr;
108 VhostUserMemory memory;
111 } QEMU_PACKED VhostUserMsg;
116 The protocol for vhost-user is based on the existing implementation of vhost
117 for the Linux Kernel. Most messages that can be sent via the Unix domain socket
118 implementing vhost-user have an equivalent ioctl to the kernel implementation.
120 The communication consists of master sending message requests and slave sending
121 message replies. Most of the requests don't require replies. Here is a list of
125 * VHOST_GET_PROTOCOL_FEATURES
126 * VHOST_GET_VRING_BASE
127 * VHOST_SET_LOG_BASE (if VHOST_USER_PROTOCOL_F_LOG_SHMFD)
129 There are several messages that the master sends with file descriptors passed
130 in the ancillary data:
132 * VHOST_SET_MEM_TABLE
133 * VHOST_SET_LOG_BASE (if VHOST_USER_PROTOCOL_F_LOG_SHMFD)
135 * VHOST_SET_VRING_KICK
136 * VHOST_SET_VRING_CALL
137 * VHOST_SET_VRING_ERR
139 If Master is unable to send the full message or receives a wrong reply it will
140 close the connection. An optional reconnection mechanism can be implemented.
142 Any protocol extensions are gated by protocol feature bits,
143 which allows full backwards compatibility on both master
145 As older slaves don't support negotiating protocol features,
146 a feature bit was dedicated for this purpose:
147 #define VHOST_USER_F_PROTOCOL_FEATURES 30
149 Starting and stopping rings
150 ----------------------
151 Client must only process each ring when it is started.
153 Client must only pass data between the ring and the
154 backend, when the ring is enabled.
156 If ring is started but disabled, client must process the
157 ring without talking to the backend.
159 For example, for a networking device, in the disabled state
160 client must not supply any new RX packets, but must process
161 and discard any TX packets.
163 If VHOST_USER_F_PROTOCOL_FEATURES has not been negotiated, the ring is initialized
166 If VHOST_USER_F_PROTOCOL_FEATURES has been negotiated, the ring is initialized
167 in a disabled state. Client must not pass data to/from the backend until ring is enabled by
168 VHOST_USER_SET_VRING_ENABLE with parameter 1, or after it has been disabled by
169 VHOST_USER_SET_VRING_ENABLE with parameter 0.
171 Each ring is initialized in a stopped state, client must not process it until
172 ring is started, or after it has been stopped.
174 Client must start ring upon receiving a kick (that is, detecting that file
175 descriptor is readable) on the descriptor specified by
176 VHOST_USER_SET_VRING_KICK, and stop ring upon receiving
177 VHOST_USER_GET_VRING_BASE.
179 While processing the rings (whether they are enabled or not), client must
180 support changing some configuration aspects on the fly.
182 Multiple queue support
183 ----------------------
185 Multiple queue is treated as a protocol extension, hence the slave has to
186 implement protocol features first. The multiple queues feature is supported
187 only when the protocol feature VHOST_USER_PROTOCOL_F_MQ (bit 0) is set.
189 The max number of queues the slave supports can be queried with message
190 VHOST_USER_GET_PROTOCOL_FEATURES. Master should stop when the number of
191 requested queues is bigger than that.
193 As all queues share one connection, the master uses a unique index for each
194 queue in the sent message to identify a specified queue. One queue pair
195 is enabled initially. More queues are enabled dynamically, by sending
196 message VHOST_USER_SET_VRING_ENABLE.
201 During live migration, the master may need to track the modifications
202 the slave makes to the memory mapped regions. The client should mark
203 the dirty pages in a log. Once it complies to this logging, it may
204 declare the VHOST_F_LOG_ALL vhost feature.
206 To start/stop logging of data/used ring writes, server may send messages
207 VHOST_USER_SET_FEATURES with VHOST_F_LOG_ALL and VHOST_USER_SET_VRING_ADDR with
208 VHOST_VRING_F_LOG in ring's flags set to 1/0, respectively.
210 All the modifications to memory pointed by vring "descriptor" should
211 be marked. Modifications to "used" vring should be marked if
212 VHOST_VRING_F_LOG is part of ring's flags.
214 Dirty pages are of size:
215 #define VHOST_LOG_PAGE 0x1000
217 The log memory fd is provided in the ancillary data of
218 VHOST_USER_SET_LOG_BASE message when the slave has
219 VHOST_USER_PROTOCOL_F_LOG_SHMFD protocol feature.
221 The size of the log is supplied as part of VhostUserMsg
222 which should be large enough to cover all known guest
223 addresses. Log starts at the supplied offset in the
224 supplied file descriptor.
225 The log covers from address 0 to the maximum of guest
226 regions. In pseudo-code, to mark page at "addr" as dirty:
228 page = addr / VHOST_LOG_PAGE
229 log[page / 8] |= 1 << page % 8
231 Where addr is the guest physical address.
233 Use atomic operations, as the log may be concurrently manipulated.
235 Note that when logging modifications to the used ring (when VHOST_VRING_F_LOG
236 is set for this ring), log_guest_addr should be used to calculate the log
237 offset: the write to first byte of the used ring is logged at this offset from
238 log start. Also note that this value might be outside the legal guest physical
239 address range (i.e. does not have to be covered by the VhostUserMemory table),
240 but the bit offset of the last byte of the ring must fall within
241 the size supplied by VhostUserLog.
243 VHOST_USER_SET_LOG_FD is an optional message with an eventfd in
244 ancillary data, it may be used to inform the master that the log has
247 Once the source has finished migration, rings will be stopped by
248 the source. No further update must be done before rings are
254 #define VHOST_USER_PROTOCOL_F_MQ 0
255 #define VHOST_USER_PROTOCOL_F_LOG_SHMFD 1
256 #define VHOST_USER_PROTOCOL_F_RARP 2
261 * VHOST_USER_GET_FEATURES
264 Equivalent ioctl: VHOST_GET_FEATURES
268 Get from the underlying vhost implementation the features bitmask.
269 Feature bit VHOST_USER_F_PROTOCOL_FEATURES signals slave support for
270 VHOST_USER_GET_PROTOCOL_FEATURES and VHOST_USER_SET_PROTOCOL_FEATURES.
272 * VHOST_USER_SET_FEATURES
275 Ioctl: VHOST_SET_FEATURES
278 Enable features in the underlying vhost implementation using a bitmask.
279 Feature bit VHOST_USER_F_PROTOCOL_FEATURES signals slave support for
280 VHOST_USER_GET_PROTOCOL_FEATURES and VHOST_USER_SET_PROTOCOL_FEATURES.
282 * VHOST_USER_GET_PROTOCOL_FEATURES
285 Equivalent ioctl: VHOST_GET_FEATURES
289 Get the protocol feature bitmask from the underlying vhost implementation.
290 Only legal if feature bit VHOST_USER_F_PROTOCOL_FEATURES is present in
291 VHOST_USER_GET_FEATURES.
292 Note: slave that reported VHOST_USER_F_PROTOCOL_FEATURES must support
293 this message even before VHOST_USER_SET_FEATURES was called.
295 * VHOST_USER_SET_PROTOCOL_FEATURES
298 Ioctl: VHOST_SET_FEATURES
301 Enable protocol features in the underlying vhost implementation.
302 Only legal if feature bit VHOST_USER_F_PROTOCOL_FEATURES is present in
303 VHOST_USER_GET_FEATURES.
304 Note: slave that reported VHOST_USER_F_PROTOCOL_FEATURES must support
305 this message even before VHOST_USER_SET_FEATURES was called.
307 * VHOST_USER_SET_OWNER
310 Equivalent ioctl: VHOST_SET_OWNER
313 Issued when a new connection is established. It sets the current Master
314 as an owner of the session. This can be used on the Slave as a
315 "session start" flag.
317 * VHOST_USER_RESET_OWNER
322 This is no longer used. Used to be sent to request disabling
323 all rings, but some clients interpreted it to also discard
324 connection state (this interpretation would lead to bugs).
325 It is recommended that clients either ignore this message,
326 or use it to disable all rings.
328 * VHOST_USER_SET_MEM_TABLE
331 Equivalent ioctl: VHOST_SET_MEM_TABLE
332 Master payload: memory regions description
334 Sets the memory map regions on the slave so it can translate the vring
335 addresses. In the ancillary data there is an array of file descriptors
336 for each memory mapped region. The size and ordering of the fds matches
337 the number and ordering of memory regions.
339 * VHOST_USER_SET_LOG_BASE
342 Equivalent ioctl: VHOST_SET_LOG_BASE
346 Sets logging shared memory space.
347 When slave has VHOST_USER_PROTOCOL_F_LOG_SHMFD protocol
348 feature, the log memory fd is provided in the ancillary data of
349 VHOST_USER_SET_LOG_BASE message, the size and offset of shared
350 memory area provided in the message.
353 * VHOST_USER_SET_LOG_FD
356 Equivalent ioctl: VHOST_SET_LOG_FD
359 Sets the logging file descriptor, which is passed as ancillary data.
361 * VHOST_USER_SET_VRING_NUM
364 Equivalent ioctl: VHOST_SET_VRING_NUM
365 Master payload: vring state description
367 Sets the number of vrings for this owner.
369 * VHOST_USER_SET_VRING_ADDR
372 Equivalent ioctl: VHOST_SET_VRING_ADDR
373 Master payload: vring address description
376 Sets the addresses of the different aspects of the vring.
378 * VHOST_USER_SET_VRING_BASE
381 Equivalent ioctl: VHOST_SET_VRING_BASE
382 Master payload: vring state description
384 Sets the base offset in the available vring.
386 * VHOST_USER_GET_VRING_BASE
389 Equivalent ioctl: VHOST_USER_GET_VRING_BASE
390 Master payload: vring state description
391 Slave payload: vring state description
393 Get the available vring base offset.
395 * VHOST_USER_SET_VRING_KICK
398 Equivalent ioctl: VHOST_SET_VRING_KICK
401 Set the event file descriptor for adding buffers to the vring. It
402 is passed in the ancillary data.
403 Bits (0-7) of the payload contain the vring index. Bit 8 is the
404 invalid FD flag. This flag is set when there is no file descriptor
405 in the ancillary data. This signals that polling should be used
406 instead of waiting for a kick.
408 * VHOST_USER_SET_VRING_CALL
411 Equivalent ioctl: VHOST_SET_VRING_CALL
414 Set the event file descriptor to signal when buffers are used. It
415 is passed in the ancillary data.
416 Bits (0-7) of the payload contain the vring index. Bit 8 is the
417 invalid FD flag. This flag is set when there is no file descriptor
418 in the ancillary data. This signals that polling will be used
419 instead of waiting for the call.
421 * VHOST_USER_SET_VRING_ERR
424 Equivalent ioctl: VHOST_SET_VRING_ERR
427 Set the event file descriptor to signal when error occurs. It
428 is passed in the ancillary data.
429 Bits (0-7) of the payload contain the vring index. Bit 8 is the
430 invalid FD flag. This flag is set when there is no file descriptor
431 in the ancillary data.
433 * VHOST_USER_GET_QUEUE_NUM
436 Equivalent ioctl: N/A
440 Query how many queues the backend supports. This request should be
441 sent only when VHOST_USER_PROTOCOL_F_MQ is set in quried protocol
442 features by VHOST_USER_GET_PROTOCOL_FEATURES.
444 * VHOST_USER_SET_VRING_ENABLE
447 Equivalent ioctl: N/A
448 Master payload: vring state description
450 Signal slave to enable or disable corresponding vring.
451 This request should be sent only when VHOST_USER_F_PROTOCOL_FEATURES
454 * VHOST_USER_SEND_RARP
457 Equivalent ioctl: N/A
460 Ask vhost user backend to broadcast a fake RARP to notify the migration
461 is terminated for guest that does not support GUEST_ANNOUNCE.
462 Only legal if feature bit VHOST_USER_F_PROTOCOL_FEATURES is present in
463 VHOST_USER_GET_FEATURES and protocol feature bit VHOST_USER_PROTOCOL_F_RARP
464 is present in VHOST_USER_GET_PROTOCOL_FEATURES.
465 The first 6 bytes of the payload contain the mac address of the guest to
466 allow the vhost user backend to construct and broadcast the fake RARP.