This VDUSE driver enables implementing software-emulated vDPA devices in userspace. The vDPA device is created by ioctl(VDUSE_CREATE_DEV) on /dev/vduse/control. Then a char device interface (/dev/vduse/$NAME) is exported to userspace for device emulation. In order to make the device emulation more secure, the device's control path is handled in kernel. A message mechnism is introduced to forward some dataplane related control messages to userspace. And in the data path, the DMA buffer will be mapped into userspace address space through different ways depending on the vDPA bus to which the vDPA device is attached. In virtio-vdpa case, the MMU-based software IOTLB is used to achieve that. And in vhost-vdpa case, the DMA buffer is reside in a userspace memory region which can be shared to the VDUSE userspace processs via transferring the shmfd. For more details on VDUSE design and usage, please see the follow-on Documentation commit. NB(mst): when merging this with b542e383d8c0 ("eventfd: Make signal recursion protection a task bit") replace eventfd_signal_count with eventfd_signal_allowed, and drop the previous ("eventfd: Export eventfd_wake_count to modules").
Showing the changes to include/uapi/linux (1 of 6 files). Show all
Unified · Split
| @@ -0,0 +1,306 @@ | ||
| 1 | +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ | |
| 2 | +#ifndef _UAPI_VDUSE_H_ | |
| 3 | +#define _UAPI_VDUSE_H_ | |
| 4 | + | |
| 5 | +#include <linux/types.h> | |
| 6 | + | |
| 7 | +#define VDUSE_BASE 0x81 | |
| 8 | + | |
| 9 | +/* The ioctls for control device (/dev/vduse/control) */ | |
| 10 | + | |
| 11 | +#define VDUSE_API_VERSION 0 | |
| 12 | + | |
| 13 | +/* | |
| 14 | + * Get the version of VDUSE API that kernel supported (VDUSE_API_VERSION). | |
| 15 | + * This is used for future extension. | |
| 16 | + */ | |
| 17 | +#define VDUSE_GET_API_VERSION _IOR(VDUSE_BASE, 0x00, __u64) | |
| 18 | + | |
| 19 | +/* Set the version of VDUSE API that userspace supported. */ | |
| 20 | +#define VDUSE_SET_API_VERSION _IOW(VDUSE_BASE, 0x01, __u64) | |
| 21 | + | |
| 22 | +/** | |
| 23 | + * struct vduse_dev_config - basic configuration of a VDUSE device | |
| 24 | + * @name: VDUSE device name, needs to be NUL terminated | |
| 25 | + * @vendor_id: virtio vendor id | |
| 26 | + * @device_id: virtio device id | |
| 27 | + * @features: virtio features | |
| 28 | + * @vq_num: the number of virtqueues | |
| 29 | + * @vq_align: the allocation alignment of virtqueue's metadata | |
| 30 | + * @reserved: for future use, needs to be initialized to zero | |
| 31 | + * @config_size: the size of the configuration space | |
| 32 | + * @config: the buffer of the configuration space | |
| 33 | + * | |
| 34 | + * Structure used by VDUSE_CREATE_DEV ioctl to create VDUSE device. | |
| 35 | + */ | |
| 36 | +struct vduse_dev_config { | |
| 37 | +#define VDUSE_NAME_MAX 256 | |
| 38 | + char name[VDUSE_NAME_MAX]; | |
| 39 | + __u32 vendor_id; | |
| 40 | + __u32 device_id; | |
| 41 | + __u64 features; | |
| 42 | + __u32 vq_num; | |
| 43 | + __u32 vq_align; | |
| 44 | + __u32 reserved[13]; | |
| 45 | + __u32 config_size; | |
| 46 | + __u8 config[]; | |
| 47 | +}; | |
| 48 | + | |
| 49 | +/* Create a VDUSE device which is represented by a char device (/dev/vduse/$NAME) */ | |
| 50 | +#define VDUSE_CREATE_DEV _IOW(VDUSE_BASE, 0x02, struct vduse_dev_config) | |
| 51 | + | |
| 52 | +/* | |
| 53 | + * Destroy a VDUSE device. Make sure there are no more references | |
| 54 | + * to the char device (/dev/vduse/$NAME). | |
| 55 | + */ | |
| 56 | +#define VDUSE_DESTROY_DEV _IOW(VDUSE_BASE, 0x03, char[VDUSE_NAME_MAX]) | |
| 57 | + | |
| 58 | +/* The ioctls for VDUSE device (/dev/vduse/$NAME) */ | |
| 59 | + | |
| 60 | +/** | |
| 61 | + * struct vduse_iotlb_entry - entry of IOTLB to describe one IOVA region [start, last] | |
| 62 | + * @offset: the mmap offset on returned file descriptor | |
| 63 | + * @start: start of the IOVA region | |
| 64 | + * @last: last of the IOVA region | |
| 65 | + * @perm: access permission of the IOVA region | |
| 66 | + * | |
| 67 | + * Structure used by VDUSE_IOTLB_GET_FD ioctl to find an overlapped IOVA region. | |
| 68 | + */ | |
| 69 | +struct vduse_iotlb_entry { | |
| 70 | + __u64 offset; | |
| 71 | + __u64 start; | |
| 72 | + __u64 last; | |
| 73 | +#define VDUSE_ACCESS_RO 0x1 | |
| 74 | +#define VDUSE_ACCESS_WO 0x2 | |
| 75 | +#define VDUSE_ACCESS_RW 0x3 | |
| 76 | + __u8 perm; | |
| 77 | +}; | |
| 78 | + | |
| 79 | +/* | |
| 80 | + * Find the first IOVA region that overlaps with the range [start, last] | |
| 81 | + * and return the corresponding file descriptor. Return -EINVAL means the | |
| 82 | + * IOVA region doesn't exist. Caller should set start and last fields. | |
| 83 | + */ | |
| 84 | +#define VDUSE_IOTLB_GET_FD _IOWR(VDUSE_BASE, 0x10, struct vduse_iotlb_entry) | |
| 85 | + | |
| 86 | +/* | |
| 87 | + * Get the negotiated virtio features. It's a subset of the features in | |
| 88 | + * struct vduse_dev_config which can be accepted by virtio driver. It's | |
| 89 | + * only valid after FEATURES_OK status bit is set. | |
| 90 | + */ | |
| 91 | +#define VDUSE_DEV_GET_FEATURES _IOR(VDUSE_BASE, 0x11, __u64) | |
| 92 | + | |
| 93 | +/** | |
| 94 | + * struct vduse_config_data - data used to update configuration space | |
| 95 | + * @offset: the offset from the beginning of configuration space | |
| 96 | + * @length: the length to write to configuration space | |
| 97 | + * @buffer: the buffer used to write from | |
| 98 | + * | |
| 99 | + * Structure used by VDUSE_DEV_SET_CONFIG ioctl to update device | |
| 100 | + * configuration space. | |
| 101 | + */ | |
| 102 | +struct vduse_config_data { | |
| 103 | + __u32 offset; | |
| 104 | + __u32 length; | |
| 105 | + __u8 buffer[]; | |
| 106 | +}; | |
| 107 | + | |
| 108 | +/* Set device configuration space */ | |
| 109 | +#define VDUSE_DEV_SET_CONFIG _IOW(VDUSE_BASE, 0x12, struct vduse_config_data) | |
| 110 | + | |
| 111 | +/* | |
| 112 | + * Inject a config interrupt. It's usually used to notify virtio driver | |
| 113 | + * that device configuration space has changed. | |
| 114 | + */ | |
| 115 | +#define VDUSE_DEV_INJECT_CONFIG_IRQ _IO(VDUSE_BASE, 0x13) | |
| 116 | + | |
| 117 | +/** | |
| 118 | + * struct vduse_vq_config - basic configuration of a virtqueue | |
| 119 | + * @index: virtqueue index | |
| 120 | + * @max_size: the max size of virtqueue | |
| 121 | + * @reserved: for future use, needs to be initialized to zero | |
| 122 | + * | |
| 123 | + * Structure used by VDUSE_VQ_SETUP ioctl to setup a virtqueue. | |
| 124 | + */ | |
| 125 | +struct vduse_vq_config { | |
| 126 | + __u32 index; | |
| 127 | + __u16 max_size; | |
| 128 | + __u16 reserved[13]; | |
| 129 | +}; | |
| 130 | + | |
| 131 | +/* | |
| 132 | + * Setup the specified virtqueue. Make sure all virtqueues have been | |
| 133 | + * configured before the device is attached to vDPA bus. | |
| 134 | + */ | |
| 135 | +#define VDUSE_VQ_SETUP _IOW(VDUSE_BASE, 0x14, struct vduse_vq_config) | |
| 136 | + | |
| 137 | +/** | |
| 138 | + * struct vduse_vq_state_split - split virtqueue state | |
| 139 | + * @avail_index: available index | |
| 140 | + */ | |
| 141 | +struct vduse_vq_state_split { | |
| 142 | + __u16 avail_index; | |
| 143 | +}; | |
| 144 | + | |
| 145 | +/** | |
| 146 | + * struct vduse_vq_state_packed - packed virtqueue state | |
| 147 | + * @last_avail_counter: last driver ring wrap counter observed by device | |
| 148 | + * @last_avail_idx: device available index | |
| 149 | + * @last_used_counter: device ring wrap counter | |
| 150 | + * @last_used_idx: used index | |
| 151 | + */ | |
| 152 | +struct vduse_vq_state_packed { | |
| 153 | + __u16 last_avail_counter; | |
| 154 | + __u16 last_avail_idx; | |
| 155 | + __u16 last_used_counter; | |
| 156 | + __u16 last_used_idx; | |
| 157 | +}; | |
| 158 | + | |
| 159 | +/** | |
| 160 | + * struct vduse_vq_info - information of a virtqueue | |
| 161 | + * @index: virtqueue index | |
| 162 | + * @num: the size of virtqueue | |
| 163 | + * @desc_addr: address of desc area | |
| 164 | + * @driver_addr: address of driver area | |
| 165 | + * @device_addr: address of device area | |
| 166 | + * @split: split virtqueue state | |
| 167 | + * @packed: packed virtqueue state | |
| 168 | + * @ready: ready status of virtqueue | |
| 169 | + * | |
| 170 | + * Structure used by VDUSE_VQ_GET_INFO ioctl to get virtqueue's information. | |
| 171 | + */ | |
| 172 | +struct vduse_vq_info { | |
| 173 | + __u32 index; | |
| 174 | + __u32 num; | |
| 175 | + __u64 desc_addr; | |
| 176 | + __u64 driver_addr; | |
| 177 | + __u64 device_addr; | |
| 178 | + union { | |
| 179 | + struct vduse_vq_state_split split; | |
| 180 | + struct vduse_vq_state_packed packed; | |
| 181 | + }; | |
| 182 | + __u8 ready; | |
| 183 | +}; | |
| 184 | + | |
| 185 | +/* Get the specified virtqueue's information. Caller should set index field. */ | |
| 186 | +#define VDUSE_VQ_GET_INFO _IOWR(VDUSE_BASE, 0x15, struct vduse_vq_info) | |
| 187 | + | |
| 188 | +/** | |
| 189 | + * struct vduse_vq_eventfd - eventfd configuration for a virtqueue | |
| 190 | + * @index: virtqueue index | |
| 191 | + * @fd: eventfd, -1 means de-assigning the eventfd | |
| 192 | + * | |
| 193 | + * Structure used by VDUSE_VQ_SETUP_KICKFD ioctl to setup kick eventfd. | |
| 194 | + */ | |
| 195 | +struct vduse_vq_eventfd { | |
| 196 | + __u32 index; | |
| 197 | +#define VDUSE_EVENTFD_DEASSIGN -1 | |
| 198 | + int fd; | |
| 199 | +}; | |
| 200 | + | |
| 201 | +/* | |
| 202 | + * Setup kick eventfd for specified virtqueue. The kick eventfd is used | |
| 203 | + * by VDUSE kernel module to notify userspace to consume the avail vring. | |
| 204 | + */ | |
| 205 | +#define VDUSE_VQ_SETUP_KICKFD _IOW(VDUSE_BASE, 0x16, struct vduse_vq_eventfd) | |
| 206 | + | |
| 207 | +/* | |
| 208 | + * Inject an interrupt for specific virtqueue. It's used to notify virtio driver | |
| 209 | + * to consume the used vring. | |
| 210 | + */ | |
| 211 | +#define VDUSE_VQ_INJECT_IRQ _IOW(VDUSE_BASE, 0x17, __u32) | |
| 212 | + | |
| 213 | +/* The control messages definition for read(2)/write(2) on /dev/vduse/$NAME */ | |
| 214 | + | |
| 215 | +/** | |
| 216 | + * enum vduse_req_type - request type | |
| 217 | + * @VDUSE_GET_VQ_STATE: get the state for specified virtqueue from userspace | |
| 218 | + * @VDUSE_SET_STATUS: set the device status | |
| 219 | + * @VDUSE_UPDATE_IOTLB: Notify userspace to update the memory mapping for | |
| 220 | + * specified IOVA range via VDUSE_IOTLB_GET_FD ioctl | |
| 221 | + */ | |
| 222 | +enum vduse_req_type { | |
| 223 | + VDUSE_GET_VQ_STATE, | |
| 224 | + VDUSE_SET_STATUS, | |
| 225 | + VDUSE_UPDATE_IOTLB, | |
| 226 | +}; | |
| 227 | + | |
| 228 | +/** | |
| 229 | + * struct vduse_vq_state - virtqueue state | |
| 230 | + * @index: virtqueue index | |
| 231 | + * @split: split virtqueue state | |
| 232 | + * @packed: packed virtqueue state | |
| 233 | + */ | |
| 234 | +struct vduse_vq_state { | |
| 235 | + __u32 index; | |
| 236 | + union { | |
| 237 | + struct vduse_vq_state_split split; | |
| 238 | + struct vduse_vq_state_packed packed; | |
| 239 | + }; | |
| 240 | +}; | |
| 241 | + | |
| 242 | +/** | |
| 243 | + * struct vduse_dev_status - device status | |
| 244 | + * @status: device status | |
| 245 | + */ | |
| 246 | +struct vduse_dev_status { | |
| 247 | + __u8 status; | |
| 248 | +}; | |
| 249 | + | |
| 250 | +/** | |
| 251 | + * struct vduse_iova_range - IOVA range [start, last] | |
| 252 | + * @start: start of the IOVA range | |
| 253 | + * @last: last of the IOVA range | |
| 254 | + */ | |
| 255 | +struct vduse_iova_range { | |
| 256 | + __u64 start; | |
| 257 | + __u64 last; | |
| 258 | +}; | |
| 259 | + | |
| 260 | +/** | |
| 261 | + * struct vduse_dev_request - control request | |
| 262 | + * @type: request type | |
| 263 | + * @request_id: request id | |
| 264 | + * @reserved: for future use | |
| 265 | + * @vq_state: virtqueue state, only index field is available | |
| 266 | + * @s: device status | |
| 267 | + * @iova: IOVA range for updating | |
| 268 | + * @padding: padding | |
| 269 | + * | |
| 270 | + * Structure used by read(2) on /dev/vduse/$NAME. | |
| 271 | + */ | |
| 272 | +struct vduse_dev_request { | |
| 273 | + __u32 type; | |
| 274 | + __u32 request_id; | |
| 275 | + __u32 reserved[4]; | |
| 276 | + union { | |
| 277 | + struct vduse_vq_state vq_state; | |
| 278 | + struct vduse_dev_status s; | |
| 279 | + struct vduse_iova_range iova; | |
| 280 | + __u32 padding[32]; | |
| 281 | + }; | |
| 282 | +}; | |
| 283 | + | |
| 284 | +/** | |
| 285 | + * struct vduse_dev_response - response to control request | |
| 286 | + * @request_id: corresponding request id | |
| 287 | + * @result: the result of request | |
| 288 | + * @reserved: for future use, needs to be initialized to zero | |
| 289 | + * @vq_state: virtqueue state | |
| 290 | + * @padding: padding | |
| 291 | + * | |
| 292 | + * Structure used by write(2) on /dev/vduse/$NAME. | |
| 293 | + */ | |
| 294 | +struct vduse_dev_response { | |
| 295 | + __u32 request_id; | |
| 296 | +#define VDUSE_REQ_RESULT_OK 0x00 | |
| 297 | +#define VDUSE_REQ_RESULT_FAILED 0x01 | |
| 298 | + __u32 result; | |
| 299 | + __u32 reserved[4]; | |
| 300 | + union { | |
| 301 | + struct vduse_vq_state vq_state; | |
| 302 | + __u32 padding[32]; | |
| 303 | + }; | |
| 304 | +}; | |
| 305 | + | |
| 306 | +#endif /* _UAPI_VDUSE_H_ */ | |