1.. include:: <isonum.txt>
2
3=====================
4VFIO Mediated devices
5=====================
6
7:Copyright: |copy| 2016, NVIDIA CORPORATION. All rights reserved.
8:Author: Neo Jia <cjia@nvidia.com>
9:Author: Kirti Wankhede <kwankhede@nvidia.com>
10
11This program is free software; you can redistribute it and/or modify
12it under the terms of the GNU General Public License version 2 as
13published by the Free Software Foundation.
14
15
16Virtual Function I/O (VFIO) Mediated devices[1]
17===============================================
18
19The number of use cases for virtualizing DMA devices that do not have built-in
20SR_IOV capability is increasing. Previously, to virtualize such devices,
21developers had to create their own management interfaces and APIs, and then
22integrate them with user space software. To simplify integration with user space
23software, we have identified common requirements and a unified management
24interface for such devices.
25
26The VFIO driver framework provides unified APIs for direct device access. It is
27an IOMMU/device-agnostic framework for exposing direct device access to user
28space in a secure, IOMMU-protected environment. This framework is used for
29multiple devices, such as GPUs, network adapters, and compute accelerators. With
30direct device access, virtual machines or user space applications have direct
31access to the physical device. This framework is reused for mediated devices.
32
33The mediated core driver provides a common interface for mediated device
34management that can be used by drivers of different devices. This module
35provides a generic interface to perform these operations:
36
37* Create and destroy a mediated device
38* Add a mediated device to and remove it from a mediated bus driver
39* Add a mediated device to and remove it from an IOMMU group
40
41The mediated core driver also provides an interface to register a bus driver.
42For example, the mediated VFIO mdev driver is designed for mediated devices and
43supports VFIO APIs. The mediated bus driver adds a mediated device to and
44removes it from a VFIO group.
45
46The following high-level block diagram shows the main components and interfaces
47in the VFIO mediated driver framework. The diagram shows NVIDIA, Intel, and IBM
48devices as examples, as these devices are the first devices to use this module::
49
50     +---------------+
51     |               |
52     | +-----------+ |  mdev_register_driver() +--------------+
53     | |           | +<------------------------+              |
54     | |  mdev     | |                         |              |
55     | |  bus      | +------------------------>+ vfio_mdev.ko |<-> VFIO user
56     | |  driver   | |     probe()/remove()    |              |    APIs
57     | |           | |                         +--------------+
58     | +-----------+ |
59     |               |
60     |  MDEV CORE    |
61     |   MODULE      |
62     |   mdev.ko     |
63     | +-----------+ |  mdev_register_device() +--------------+
64     | |           | +<------------------------+              |
65     | |           | |                         |  nvidia.ko   |<-> physical
66     | |           | +------------------------>+              |    device
67     | |           | |        callbacks        +--------------+
68     | | Physical  | |
69     | |  device   | |  mdev_register_device() +--------------+
70     | | interface | |<------------------------+              |
71     | |           | |                         |  i915.ko     |<-> physical
72     | |           | +------------------------>+              |    device
73     | |           | |        callbacks        +--------------+
74     | |           | |
75     | |           | |  mdev_register_device() +--------------+
76     | |           | +<------------------------+              |
77     | |           | |                         | ccw_device.ko|<-> physical
78     | |           | +------------------------>+              |    device
79     | |           | |        callbacks        +--------------+
80     | +-----------+ |
81     +---------------+
82
83
84Registration Interfaces
85=======================
86
87The mediated core driver provides the following types of registration
88interfaces:
89
90* Registration interface for a mediated bus driver
91* Physical device driver interface
92
93Registration Interface for a Mediated Bus Driver
94------------------------------------------------
95
96The registration interface for a mediated bus driver provides the following
97structure to represent a mediated device's driver::
98
99     /*
100      * struct mdev_driver [2] - Mediated device's driver
101      * @name: driver name
102      * @probe: called when new device created
103      * @remove: called when device removed
104      * @driver: device driver structure
105      */
106     struct mdev_driver {
107	     const char *name;
108	     int  (*probe)  (struct device *dev);
109	     void (*remove) (struct device *dev);
110	     struct device_driver    driver;
111     };
112
113A mediated bus driver for mdev should use this structure in the function calls
114to register and unregister itself with the core driver:
115
116* Register::
117
118    extern int  mdev_register_driver(struct mdev_driver *drv,
119				   struct module *owner);
120
121* Unregister::
122
123    extern void mdev_unregister_driver(struct mdev_driver *drv);
124
125The mediated bus driver is responsible for adding mediated devices to the VFIO
126group when devices are bound to the driver and removing mediated devices from
127the VFIO when devices are unbound from the driver.
128
129
130Physical Device Driver Interface
131--------------------------------
132
133The physical device driver interface provides the mdev_parent_ops[3] structure
134to define the APIs to manage work in the mediated core driver that is related
135to the physical device.
136
137The structures in the mdev_parent_ops structure are as follows:
138
139* dev_attr_groups: attributes of the parent device
140* mdev_attr_groups: attributes of the mediated device
141* supported_config: attributes to define supported configurations
142
143The functions in the mdev_parent_ops structure are as follows:
144
145* create: allocate basic resources in a driver for a mediated device
146* remove: free resources in a driver when a mediated device is destroyed
147
148(Note that mdev-core provides no implicit serialization of create/remove
149callbacks per mdev parent device, per mdev type, or any other categorization.
150Vendor drivers are expected to be fully asynchronous in this respect or
151provide their own internal resource protection.)
152
153The callbacks in the mdev_parent_ops structure are as follows:
154
155* open: open callback of mediated device
156* close: close callback of mediated device
157* ioctl: ioctl callback of mediated device
158* read : read emulation callback
159* write: write emulation callback
160* mmap: mmap emulation callback
161
162A driver should use the mdev_parent_ops structure in the function call to
163register itself with the mdev core driver::
164
165	extern int  mdev_register_device(struct device *dev,
166	                                 const struct mdev_parent_ops *ops);
167
168However, the mdev_parent_ops structure is not required in the function call
169that a driver should use to unregister itself with the mdev core driver::
170
171	extern void mdev_unregister_device(struct device *dev);
172
173
174Mediated Device Management Interface Through sysfs
175==================================================
176
177The management interface through sysfs enables user space software, such as
178libvirt, to query and configure mediated devices in a hardware-agnostic fashion.
179This management interface provides flexibility to the underlying physical
180device's driver to support features such as:
181
182* Mediated device hot plug
183* Multiple mediated devices in a single virtual machine
184* Multiple mediated devices from different physical devices
185
186Links in the mdev_bus Class Directory
187-------------------------------------
188The /sys/class/mdev_bus/ directory contains links to devices that are registered
189with the mdev core driver.
190
191Directories and files under the sysfs for Each Physical Device
192--------------------------------------------------------------
193
194::
195
196  |- [parent physical device]
197  |--- Vendor-specific-attributes [optional]
198  |--- [mdev_supported_types]
199  |     |--- [<type-id>]
200  |     |   |--- create
201  |     |   |--- name
202  |     |   |--- available_instances
203  |     |   |--- device_api
204  |     |   |--- description
205  |     |   |--- [devices]
206  |     |--- [<type-id>]
207  |     |   |--- create
208  |     |   |--- name
209  |     |   |--- available_instances
210  |     |   |--- device_api
211  |     |   |--- description
212  |     |   |--- [devices]
213  |     |--- [<type-id>]
214  |          |--- create
215  |          |--- name
216  |          |--- available_instances
217  |          |--- device_api
218  |          |--- description
219  |          |--- [devices]
220
221* [mdev_supported_types]
222
223  The list of currently supported mediated device types and their details.
224
225  [<type-id>], device_api, and available_instances are mandatory attributes
226  that should be provided by vendor driver.
227
228* [<type-id>]
229
230  The [<type-id>] name is created by adding the device driver string as a prefix
231  to the string provided by the vendor driver. This format of this name is as
232  follows::
233
234	sprintf(buf, "%s-%s", dev_driver_string(parent->dev), group->name);
235
236  (or using mdev_parent_dev(mdev) to arrive at the parent device outside
237  of the core mdev code)
238
239* device_api
240
241  This attribute should show which device API is being created, for example,
242  "vfio-pci" for a PCI device.
243
244* available_instances
245
246  This attribute should show the number of devices of type <type-id> that can be
247  created.
248
249* [device]
250
251  This directory contains links to the devices of type <type-id> that have been
252  created.
253
254* name
255
256  This attribute should show human readable name. This is optional attribute.
257
258* description
259
260  This attribute should show brief features/description of the type. This is
261  optional attribute.
262
263Directories and Files Under the sysfs for Each mdev Device
264----------------------------------------------------------
265
266::
267
268  |- [parent phy device]
269  |--- [$MDEV_UUID]
270         |--- remove
271         |--- mdev_type {link to its type}
272         |--- vendor-specific-attributes [optional]
273
274* remove (write only)
275
276Writing '1' to the 'remove' file destroys the mdev device. The vendor driver can
277fail the remove() callback if that device is active and the vendor driver
278doesn't support hot unplug.
279
280Example::
281
282	# echo 1 > /sys/bus/mdev/devices/$mdev_UUID/remove
283
284Mediated device Hot plug
285------------------------
286
287Mediated devices can be created and assigned at runtime. The procedure to hot
288plug a mediated device is the same as the procedure to hot plug a PCI device.
289
290Translation APIs for Mediated Devices
291=====================================
292
293The following APIs are provided for translating user pfn to host pfn in a VFIO
294driver::
295
296	extern int vfio_pin_pages(struct device *dev, unsigned long *user_pfn,
297				  int npage, int prot, unsigned long *phys_pfn);
298
299	extern int vfio_unpin_pages(struct device *dev, unsigned long *user_pfn,
300				    int npage);
301
302These functions call back into the back-end IOMMU module by using the pin_pages
303and unpin_pages callbacks of the struct vfio_iommu_driver_ops[4]. Currently
304these callbacks are supported in the TYPE1 IOMMU module. To enable them for
305other IOMMU backend modules, such as PPC64 sPAPR module, they need to provide
306these two callback functions.
307
308Using the Sample Code
309=====================
310
311mtty.c in samples/vfio-mdev/ directory is a sample driver program to
312demonstrate how to use the mediated device framework.
313
314The sample driver creates an mdev device that simulates a serial port over a PCI
315card.
316
3171. Build and load the mtty.ko module.
318
319   This step creates a dummy device, /sys/devices/virtual/mtty/mtty/
320
321   Files in this device directory in sysfs are similar to the following::
322
323     # tree /sys/devices/virtual/mtty/mtty/
324        /sys/devices/virtual/mtty/mtty/
325        |-- mdev_supported_types
326        |   |-- mtty-1
327        |   |   |-- available_instances
328        |   |   |-- create
329        |   |   |-- device_api
330        |   |   |-- devices
331        |   |   `-- name
332        |   `-- mtty-2
333        |       |-- available_instances
334        |       |-- create
335        |       |-- device_api
336        |       |-- devices
337        |       `-- name
338        |-- mtty_dev
339        |   `-- sample_mtty_dev
340        |-- power
341        |   |-- autosuspend_delay_ms
342        |   |-- control
343        |   |-- runtime_active_time
344        |   |-- runtime_status
345        |   `-- runtime_suspended_time
346        |-- subsystem -> ../../../../class/mtty
347        `-- uevent
348
3492. Create a mediated device by using the dummy device that you created in the
350   previous step::
351
352     # echo "83b8f4f2-509f-382f-3c1e-e6bfe0fa1001" >	\
353              /sys/devices/virtual/mtty/mtty/mdev_supported_types/mtty-2/create
354
3553. Add parameters to qemu-kvm::
356
357     -device vfio-pci,\
358      sysfsdev=/sys/bus/mdev/devices/83b8f4f2-509f-382f-3c1e-e6bfe0fa1001
359
3604. Boot the VM.
361
362   In the Linux guest VM, with no hardware on the host, the device appears
363   as  follows::
364
365     # lspci -s 00:05.0 -xxvv
366     00:05.0 Serial controller: Device 4348:3253 (rev 10) (prog-if 02 [16550])
367             Subsystem: Device 4348:3253
368             Physical Slot: 5
369             Control: I/O+ Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr-
370     Stepping- SERR- FastB2B- DisINTx-
371             Status: Cap- 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort-
372     <TAbort- <MAbort- >SERR- <PERR- INTx-
373             Interrupt: pin A routed to IRQ 10
374             Region 0: I/O ports at c150 [size=8]
375             Region 1: I/O ports at c158 [size=8]
376             Kernel driver in use: serial
377     00: 48 43 53 32 01 00 00 02 10 02 00 07 00 00 00 00
378     10: 51 c1 00 00 59 c1 00 00 00 00 00 00 00 00 00 00
379     20: 00 00 00 00 00 00 00 00 00 00 00 00 48 43 53 32
380     30: 00 00 00 00 00 00 00 00 00 00 00 00 0a 01 00 00
381
382     In the Linux guest VM, dmesg output for the device is as follows:
383
384     serial 0000:00:05.0: PCI INT A -> Link[LNKA] -> GSI 10 (level, high) -> IRQ 10
385     0000:00:05.0: ttyS1 at I/O 0xc150 (irq = 10) is a 16550A
386     0000:00:05.0: ttyS2 at I/O 0xc158 (irq = 10) is a 16550A
387
388
3895. In the Linux guest VM, check the serial ports::
390
391     # setserial -g /dev/ttyS*
392     /dev/ttyS0, UART: 16550A, Port: 0x03f8, IRQ: 4
393     /dev/ttyS1, UART: 16550A, Port: 0xc150, IRQ: 10
394     /dev/ttyS2, UART: 16550A, Port: 0xc158, IRQ: 10
395
3966. Using minicom or any terminal emulation program, open port /dev/ttyS1 or
397   /dev/ttyS2 with hardware flow control disabled.
398
3997. Type data on the minicom terminal or send data to the terminal emulation
400   program and read the data.
401
402   Data is loop backed from hosts mtty driver.
403
4048. Destroy the mediated device that you created::
405
406     # echo 1 > /sys/bus/mdev/devices/83b8f4f2-509f-382f-3c1e-e6bfe0fa1001/remove
407
408References
409==========
410
4111. See Documentation/driver-api/vfio.rst for more information on VFIO.
4122. struct mdev_driver in include/linux/mdev.h
4133. struct mdev_parent_ops in include/linux/mdev.h
4144. struct vfio_iommu_driver_ops in include/linux/vfio.h
415