1 .. include:: <isonum.txt>
7 :Copyright: |copy| 2016, NVIDIA CORPORATION. All rights reserved.
8 :Author: Neo Jia <cjia@nvidia.com>
9 :Author: Kirti Wankhede <kwankhede@nvidia.com>
11 This program is free software; you can redistribute it and/or modify
12 it under the terms of the GNU General Public License version 2 as
13 published by the Free Software Foundation.
16 Virtual Function I/O (VFIO) Mediated devices[1]
17 ===============================================
19 The number of use cases for virtualizing DMA devices that do not have built-in
20 SR_IOV capability is increasing. Previously, to virtualize such devices,
21 developers had to create their own management interfaces and APIs, and then
22 integrate them with user space software. To simplify integration with user space
23 software, we have identified common requirements and a unified management
24 interface for such devices.
26 The VFIO driver framework provides unified APIs for direct device access. It is
27 an IOMMU/device-agnostic framework for exposing direct device access to user
28 space in a secure, IOMMU-protected environment. This framework is used for
29 multiple devices, such as GPUs, network adapters, and compute accelerators. With
30 direct device access, virtual machines or user space applications have direct
31 access to the physical device. This framework is reused for mediated devices.
33 The mediated core driver provides a common interface for mediated device
34 management that can be used by drivers of different devices. This module
35 provides a generic interface to perform these operations:
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
41 The mediated core driver also provides an interface to register a bus driver.
42 For example, the mediated VFIO mdev driver is designed for mediated devices and
43 supports VFIO APIs. The mediated bus driver adds a mediated device to and
44 removes it from a VFIO group.
46 The following high-level block diagram shows the main components and interfaces
47 in the VFIO mediated driver framework. The diagram shows NVIDIA, Intel, and IBM
48 devices as examples, as these devices are the first devices to use this module::
52 | +-----------+ | mdev_register_driver() +--------------+
53 | | | +<------------------------+ |
55 | | bus | +------------------------>+ vfio_mdev.ko |<-> VFIO user
56 | | driver | | probe()/remove() | | APIs
57 | | | | +--------------+
63 | +-----------+ | mdev_register_device() +--------------+
64 | | | +<------------------------+ |
65 | | | | | nvidia.ko |<-> physical
66 | | | +------------------------>+ | device
67 | | | | callbacks +--------------+
69 | | device | | mdev_register_device() +--------------+
70 | | interface | |<------------------------+ |
71 | | | | | i915.ko |<-> physical
72 | | | +------------------------>+ | device
73 | | | | callbacks +--------------+
75 | | | | mdev_register_device() +--------------+
76 | | | +<------------------------+ |
77 | | | | | ccw_device.ko|<-> physical
78 | | | +------------------------>+ | device
79 | | | | callbacks +--------------+
84 Registration Interfaces
85 =======================
87 The mediated core driver provides the following types of registration
90 * Registration interface for a mediated bus driver
91 * Physical device driver interface
93 Registration Interface for a Mediated Bus Driver
94 ------------------------------------------------
96 The registration interface for a mediated bus driver provides the following
97 structure to represent a mediated device's driver::
100 * struct mdev_driver [2] - Mediated device's driver
102 * @probe: called when new device created
103 * @remove: called when device removed
104 * @driver: device driver structure
108 int (*probe) (struct device *dev);
109 void (*remove) (struct device *dev);
110 struct device_driver driver;
113 A mediated bus driver for mdev should use this structure in the function calls
114 to register and unregister itself with the core driver:
118 extern int mdev_register_driver(struct mdev_driver *drv,
119 struct module *owner);
123 extern void mdev_unregister_driver(struct mdev_driver *drv);
125 The mediated bus driver is responsible for adding mediated devices to the VFIO
126 group when devices are bound to the driver and removing mediated devices from
127 the VFIO when devices are unbound from the driver.
130 Physical Device Driver Interface
131 --------------------------------
133 The physical device driver interface provides the mdev_parent_ops[3] structure
134 to define the APIs to manage work in the mediated core driver that is related
135 to the physical device.
137 The structures in the mdev_parent_ops structure are as follows:
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
143 The functions in the mdev_parent_ops structure are as follows:
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
148 (Note that mdev-core provides no implicit serialization of create/remove
149 callbacks per mdev parent device, per mdev type, or any other categorization.
150 Vendor drivers are expected to be fully asynchronous in this respect or
151 provide their own internal resource protection.)
153 The callbacks in the mdev_parent_ops structure are as follows:
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
162 A driver should use the mdev_parent_ops structure in the function call to
163 register itself with the mdev core driver::
165 extern int mdev_register_device(struct device *dev,
166 const struct mdev_parent_ops *ops);
168 However, the mdev_parent_ops structure is not required in the function call
169 that a driver should use to unregister itself with the mdev core driver::
171 extern void mdev_unregister_device(struct device *dev);
174 Mediated Device Management Interface Through sysfs
175 ==================================================
177 The management interface through sysfs enables user space software, such as
178 libvirt, to query and configure mediated devices in a hardware-agnostic fashion.
179 This management interface provides flexibility to the underlying physical
180 device's driver to support features such as:
182 * Mediated device hot plug
183 * Multiple mediated devices in a single virtual machine
184 * Multiple mediated devices from different physical devices
186 Links in the mdev_bus Class Directory
187 -------------------------------------
188 The /sys/class/mdev_bus/ directory contains links to devices that are registered
189 with the mdev core driver.
191 Directories and files under the sysfs for Each Physical Device
192 --------------------------------------------------------------
196 |- [parent physical device]
197 |--- Vendor-specific-attributes [optional]
198 |--- [mdev_supported_types]
202 | | |--- available_instances
209 | | |--- available_instances
216 | |--- available_instances
221 * [mdev_supported_types]
223 The list of currently supported mediated device types and their details.
225 [<type-id>], device_api, and available_instances are mandatory attributes
226 that should be provided by vendor driver.
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
234 sprintf(buf, "%s-%s", dev_driver_string(parent->dev), group->name);
236 (or using mdev_parent_dev(mdev) to arrive at the parent device outside
237 of the core mdev code)
241 This attribute should show which device API is being created, for example,
242 "vfio-pci" for a PCI device.
244 * available_instances
246 This attribute should show the number of devices of type <type-id> that can be
251 This directory contains links to the devices of type <type-id> that have been
256 This attribute should show human readable name. This is optional attribute.
260 This attribute should show brief features/description of the type. This is
263 Directories and Files Under the sysfs for Each mdev Device
264 ----------------------------------------------------------
268 |- [parent phy device]
271 |--- mdev_type {link to its type}
272 |--- vendor-specific-attributes [optional]
274 * remove (write only)
276 Writing '1' to the 'remove' file destroys the mdev device. The vendor driver can
277 fail the remove() callback if that device is active and the vendor driver
278 doesn't support hot unplug.
282 # echo 1 > /sys/bus/mdev/devices/$mdev_UUID/remove
284 Mediated device Hot plug
285 ------------------------
287 Mediated devices can be created and assigned at runtime. The procedure to hot
288 plug a mediated device is the same as the procedure to hot plug a PCI device.
290 Translation APIs for Mediated Devices
291 =====================================
293 The following APIs are provided for translating user pfn to host pfn in a VFIO
296 extern int vfio_pin_pages(struct device *dev, unsigned long *user_pfn,
297 int npage, int prot, unsigned long *phys_pfn);
299 extern int vfio_unpin_pages(struct device *dev, unsigned long *user_pfn,
302 These functions call back into the back-end IOMMU module by using the pin_pages
303 and unpin_pages callbacks of the struct vfio_iommu_driver_ops[4]. Currently
304 these callbacks are supported in the TYPE1 IOMMU module. To enable them for
305 other IOMMU backend modules, such as PPC64 sPAPR module, they need to provide
306 these two callback functions.
308 Using the Sample Code
309 =====================
311 mtty.c in samples/vfio-mdev/ directory is a sample driver program to
312 demonstrate how to use the mediated device framework.
314 The sample driver creates an mdev device that simulates a serial port over a PCI
317 1. Build and load the mtty.ko module.
319 This step creates a dummy device, /sys/devices/virtual/mtty/mtty/
321 Files in this device directory in sysfs are similar to the following::
323 # tree /sys/devices/virtual/mtty/mtty/
324 /sys/devices/virtual/mtty/mtty/
325 |-- mdev_supported_types
327 | | |-- available_instances
333 | |-- available_instances
339 | `-- sample_mtty_dev
341 | |-- autosuspend_delay_ms
343 | |-- runtime_active_time
345 | `-- runtime_suspended_time
346 |-- subsystem -> ../../../../class/mtty
349 2. Create a mediated device by using the dummy device that you created in the
352 # echo "83b8f4f2-509f-382f-3c1e-e6bfe0fa1001" > \
353 /sys/devices/virtual/mtty/mtty/mdev_supported_types/mtty-2/create
355 3. Add parameters to qemu-kvm::
358 sysfsdev=/sys/bus/mdev/devices/83b8f4f2-509f-382f-3c1e-e6bfe0fa1001
362 In the Linux guest VM, with no hardware on the host, the device appears
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
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
382 In the Linux guest VM, dmesg output for the device is as follows:
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
389 5. In the Linux guest VM, check the serial ports::
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
396 6. Using minicom or any terminal emulation program, open port /dev/ttyS1 or
397 /dev/ttyS2 with hardware flow control disabled.
399 7. Type data on the minicom terminal or send data to the terminal emulation
400 program and read the data.
402 Data is loop backed from hosts mtty driver.
404 8. Destroy the mediated device that you created::
406 # echo 1 > /sys/bus/mdev/devices/83b8f4f2-509f-382f-3c1e-e6bfe0fa1001/remove
411 1. See Documentation/vfio.txt for more information on VFIO.
412 2. struct mdev_driver in include/linux/mdev.h
413 3. struct mdev_parent_ops in include/linux/mdev.h
414 4. struct vfio_iommu_driver_ops in include/linux/vfio.h