ceph/src/spdk/dpdk/drivers/net/virtio/virtio_user/vhost_user.c

   1 /* SPDX-License-Identifier: BSD-3-Clause
   2  * Copyright(c) 2010-2016 Intel Corporation
   3  */
   4
   5 #include <sys/socket.h>
   6 #include <sys/types.h>
   7 #include <sys/stat.h>
   8 #include <unistd.h>
   9 #include <fcntl.h>
  10 #include <sys/un.h>
  11 #include <string.h>
  12 #include <errno.h>
  13
  14 #include <rte_string_fns.h>
  15 #include <rte_fbarray.h>
  16
  17 #include "vhost.h"
  18 #include "virtio_user_dev.h"
  19
  20 /* The version of the protocol we support */
  21 #define VHOST_USER_VERSION    0x1
  22
  23 #define VHOST_MEMORY_MAX_NREGIONS 8
  24 struct vhost_memory {
  25         uint32_t nregions;
  26         uint32_t padding;
  27         struct vhost_memory_region regions[VHOST_MEMORY_MAX_NREGIONS];
  28 };
  29
  30 struct vhost_user_msg {
  31         enum vhost_user_request request;
  32
  33 #define VHOST_USER_VERSION_MASK     0x3
  34 #define VHOST_USER_REPLY_MASK       (0x1 << 2)
  35         uint32_t flags;
  36         uint32_t size; /* the following payload size */
  37         union {
  38 #define VHOST_USER_VRING_IDX_MASK   0xff
  39 #define VHOST_USER_VRING_NOFD_MASK  (0x1 << 8)
  40                 uint64_t u64;
  41                 struct vhost_vring_state state;
  42                 struct vhost_vring_addr addr;
  43                 struct vhost_memory memory;
  44         } payload;
  45         int fds[VHOST_MEMORY_MAX_NREGIONS];
  46 } __rte_packed;
  47
  48 #define VHOST_USER_HDR_SIZE offsetof(struct vhost_user_msg, payload.u64)
  49 #define VHOST_USER_PAYLOAD_SIZE \
  50         (sizeof(struct vhost_user_msg) - VHOST_USER_HDR_SIZE)
  51
  52 static int
  53 vhost_user_write(int fd, void *buf, int len, int *fds, int fd_num)
  54 {
  55         int r;
  56         struct msghdr msgh;
  57         struct iovec iov;
  58         size_t fd_size = fd_num * sizeof(int);
  59         char control[CMSG_SPACE(fd_size)];
  60         struct cmsghdr *cmsg;
  61
  62         memset(&msgh, 0, sizeof(msgh));
  63         memset(control, 0, sizeof(control));
  64
  65         iov.iov_base = (uint8_t *)buf;
  66         iov.iov_len = len;
  67
  68         msgh.msg_iov = &iov;
  69         msgh.msg_iovlen = 1;
  70         msgh.msg_control = control;
  71         msgh.msg_controllen = sizeof(control);
  72
  73         cmsg = CMSG_FIRSTHDR(&msgh);
  74         cmsg->cmsg_len = CMSG_LEN(fd_size);
  75         cmsg->cmsg_level = SOL_SOCKET;
  76         cmsg->cmsg_type = SCM_RIGHTS;
  77         memcpy(CMSG_DATA(cmsg), fds, fd_size);
  78
  79         do {
  80                 r = sendmsg(fd, &msgh, 0);
  81         } while (r < 0 && errno == EINTR);
  82
  83         return r;
  84 }
  85
  86 static int
  87 vhost_user_read(int fd, struct vhost_user_msg *msg)
  88 {
  89         uint32_t valid_flags = VHOST_USER_REPLY_MASK | VHOST_USER_VERSION;
  90         int ret, sz_hdr = VHOST_USER_HDR_SIZE, sz_payload;
  91
  92         ret = recv(fd, (void *)msg, sz_hdr, 0);
  93         if (ret < sz_hdr) {
  94                 PMD_DRV_LOG(ERR, "Failed to recv msg hdr: %d instead of %d.",
  95                             ret, sz_hdr);
  96                 goto fail;
  97         }
  98
  99         /* validate msg flags */
 100         if (msg->flags != (valid_flags)) {
 101                 PMD_DRV_LOG(ERR, "Failed to recv msg: flags %x instead of %x.",
 102                             msg->flags, valid_flags);
 103                 goto fail;
 104         }
 105
 106         sz_payload = msg->size;
 107
 108         if ((size_t)sz_payload > sizeof(msg->payload))
 109                 goto fail;
 110
 111         if (sz_payload) {
 112                 ret = recv(fd, (void *)((char *)msg + sz_hdr), sz_payload, 0);
 113                 if (ret < sz_payload) {
 114                         PMD_DRV_LOG(ERR,
 115                                 "Failed to recv msg payload: %d instead of %d.",
 116                                 ret, msg->size);
 117                         goto fail;
 118                 }
 119         }
 120
 121         return 0;
 122
 123 fail:
 124         return -1;
 125 }
 126
 127 struct walk_arg {
 128         struct vhost_memory *vm;
 129         int *fds;
 130         int region_nr;
 131 };
 132
 133 static int
 134 update_memory_region(const struct rte_memseg_list *msl __rte_unused,
 135                 const struct rte_memseg *ms, void *arg)
 136 {
 137         struct walk_arg *wa = arg;
 138         struct vhost_memory_region *mr;
 139         uint64_t start_addr, end_addr;
 140         size_t offset;
 141         int i, fd;
 142
 143         fd = rte_memseg_get_fd_thread_unsafe(ms);
 144         if (fd < 0) {
 145                 PMD_DRV_LOG(ERR, "Failed to get fd, ms=%p rte_errno=%d",
 146                         ms, rte_errno);
 147                 return -1;
 148         }
 149
 150         if (rte_memseg_get_fd_offset_thread_unsafe(ms, &offset) < 0) {
 151                 PMD_DRV_LOG(ERR, "Failed to get offset, ms=%p rte_errno=%d",
 152                         ms, rte_errno);
 153                 return -1;
 154         }
 155
 156         start_addr = (uint64_t)(uintptr_t)ms->addr;
 157         end_addr = start_addr + ms->len;
 158
 159         for (i = 0; i < wa->region_nr; i++) {
 160                 if (wa->fds[i] != fd)
 161                         continue;
 162
 163                 mr = &wa->vm->regions[i];
 164
 165                 if (mr->userspace_addr + mr->memory_size < end_addr)
 166                         mr->memory_size = end_addr - mr->userspace_addr;
 167
 168                 if (mr->userspace_addr > start_addr) {
 169                         mr->userspace_addr = start_addr;
 170                         mr->guest_phys_addr = start_addr;
 171                 }
 172
 173                 if (mr->mmap_offset > offset)
 174                         mr->mmap_offset = offset;
 175
 176                 PMD_DRV_LOG(DEBUG, "index=%d fd=%d offset=0x%" PRIx64
 177                         " addr=0x%" PRIx64 " len=%" PRIu64, i, fd,
 178                         mr->mmap_offset, mr->userspace_addr,
 179                         mr->memory_size);
 180
 181                 return 0;
 182         }
 183
 184         if (i >= VHOST_MEMORY_MAX_NREGIONS) {
 185                 PMD_DRV_LOG(ERR, "Too many memory regions");
 186                 return -1;
 187         }
 188
 189         mr = &wa->vm->regions[i];
 190         wa->fds[i] = fd;
 191
 192         mr->guest_phys_addr = start_addr;
 193         mr->userspace_addr = start_addr;
 194         mr->memory_size = ms->len;
 195         mr->mmap_offset = offset;
 196
 197         PMD_DRV_LOG(DEBUG, "index=%d fd=%d offset=0x%" PRIx64
 198                 " addr=0x%" PRIx64 " len=%" PRIu64, i, fd,
 199                 mr->mmap_offset, mr->userspace_addr,
 200                 mr->memory_size);
 201
 202         wa->region_nr++;
 203
 204         return 0;
 205 }
 206
 207 static int
 208 prepare_vhost_memory_user(struct vhost_user_msg *msg, int fds[])
 209 {
 210         struct walk_arg wa;
 211
 212         wa.region_nr = 0;
 213         wa.vm = &msg->payload.memory;
 214         wa.fds = fds;
 215
 216         /*
 217          * The memory lock has already been taken by memory subsystem
 218          * or virtio_user_start_device().
 219          */
 220         if (rte_memseg_walk_thread_unsafe(update_memory_region, &wa) < 0)
 221                 return -1;
 222
 223         msg->payload.memory.nregions = wa.region_nr;
 224         msg->payload.memory.padding = 0;
 225
 226         return 0;
 227 }
 228
 229 static struct vhost_user_msg m;
 230
 231 const char * const vhost_msg_strings[] = {
 232         [VHOST_USER_SET_OWNER] = "VHOST_SET_OWNER",
 233         [VHOST_USER_RESET_OWNER] = "VHOST_RESET_OWNER",
 234         [VHOST_USER_SET_FEATURES] = "VHOST_SET_FEATURES",
 235         [VHOST_USER_GET_FEATURES] = "VHOST_GET_FEATURES",
 236         [VHOST_USER_SET_VRING_CALL] = "VHOST_SET_VRING_CALL",
 237         [VHOST_USER_SET_VRING_NUM] = "VHOST_SET_VRING_NUM",
 238         [VHOST_USER_SET_VRING_BASE] = "VHOST_SET_VRING_BASE",
 239         [VHOST_USER_GET_VRING_BASE] = "VHOST_GET_VRING_BASE",
 240         [VHOST_USER_SET_VRING_ADDR] = "VHOST_SET_VRING_ADDR",
 241         [VHOST_USER_SET_VRING_KICK] = "VHOST_SET_VRING_KICK",
 242         [VHOST_USER_SET_MEM_TABLE] = "VHOST_SET_MEM_TABLE",
 243         [VHOST_USER_SET_VRING_ENABLE] = "VHOST_SET_VRING_ENABLE",
 244 };
 245
 246 static int
 247 vhost_user_sock(struct virtio_user_dev *dev,
 248                 enum vhost_user_request req,
 249                 void *arg)
 250 {
 251         struct vhost_user_msg msg;
 252         struct vhost_vring_file *file = 0;
 253         int need_reply = 0;
 254         int fds[VHOST_MEMORY_MAX_NREGIONS];
 255         int fd_num = 0;
 256         int len;
 257         int vhostfd = dev->vhostfd;
 258
 259         RTE_SET_USED(m);
 260
 261         PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);
 262
 263         if (dev->is_server && vhostfd < 0)
 264                 return -1;
 265
 266         msg.request = req;
 267         msg.flags = VHOST_USER_VERSION;
 268         msg.size = 0;
 269
 270         switch (req) {
 271         case VHOST_USER_GET_FEATURES:
 272                 need_reply = 1;
 273                 break;
 274
 275         case VHOST_USER_SET_FEATURES:
 276         case VHOST_USER_SET_LOG_BASE:
 277                 msg.payload.u64 = *((__u64 *)arg);
 278                 msg.size = sizeof(m.payload.u64);
 279                 break;
 280
 281         case VHOST_USER_SET_OWNER:
 282         case VHOST_USER_RESET_OWNER:
 283                 break;
 284
 285         case VHOST_USER_SET_MEM_TABLE:
 286                 if (prepare_vhost_memory_user(&msg, fds) < 0)
 287                         return -1;
 288                 fd_num = msg.payload.memory.nregions;
 289                 msg.size = sizeof(m.payload.memory.nregions);
 290                 msg.size += sizeof(m.payload.memory.padding);
 291                 msg.size += fd_num * sizeof(struct vhost_memory_region);
 292                 break;
 293
 294         case VHOST_USER_SET_LOG_FD:
 295                 fds[fd_num++] = *((int *)arg);
 296                 break;
 297
 298         case VHOST_USER_SET_VRING_NUM:
 299         case VHOST_USER_SET_VRING_BASE:
 300         case VHOST_USER_SET_VRING_ENABLE:
 301                 memcpy(&msg.payload.state, arg, sizeof(msg.payload.state));
 302                 msg.size = sizeof(m.payload.state);
 303                 break;
 304
 305         case VHOST_USER_GET_VRING_BASE:
 306                 memcpy(&msg.payload.state, arg, sizeof(msg.payload.state));
 307                 msg.size = sizeof(m.payload.state);
 308                 need_reply = 1;
 309                 break;
 310
 311         case VHOST_USER_SET_VRING_ADDR:
 312                 memcpy(&msg.payload.addr, arg, sizeof(msg.payload.addr));
 313                 msg.size = sizeof(m.payload.addr);
 314                 break;
 315
 316         case VHOST_USER_SET_VRING_KICK:
 317         case VHOST_USER_SET_VRING_CALL:
 318         case VHOST_USER_SET_VRING_ERR:
 319                 file = arg;
 320                 msg.payload.u64 = file->index & VHOST_USER_VRING_IDX_MASK;
 321                 msg.size = sizeof(m.payload.u64);
 322                 if (file->fd > 0)
 323                         fds[fd_num++] = file->fd;
 324                 else
 325                         msg.payload.u64 |= VHOST_USER_VRING_NOFD_MASK;
 326                 break;
 327
 328         default:
 329                 PMD_DRV_LOG(ERR, "trying to send unhandled msg type");
 330                 return -1;
 331         }
 332
 333         len = VHOST_USER_HDR_SIZE + msg.size;
 334         if (vhost_user_write(vhostfd, &msg, len, fds, fd_num) < 0) {
 335                 PMD_DRV_LOG(ERR, "%s failed: %s",
 336                             vhost_msg_strings[req], strerror(errno));
 337                 return -1;
 338         }
 339
 340         if (need_reply) {
 341                 if (vhost_user_read(vhostfd, &msg) < 0) {
 342                         PMD_DRV_LOG(ERR, "Received msg failed: %s",
 343                                     strerror(errno));
 344                         return -1;
 345                 }
 346
 347                 if (req != msg.request) {
 348                         PMD_DRV_LOG(ERR, "Received unexpected msg type");
 349                         return -1;
 350                 }
 351
 352                 switch (req) {
 353                 case VHOST_USER_GET_FEATURES:
 354                         if (msg.size != sizeof(m.payload.u64)) {
 355                                 PMD_DRV_LOG(ERR, "Received bad msg size");
 356                                 return -1;
 357                         }
 358                         *((__u64 *)arg) = msg.payload.u64;
 359                         break;
 360                 case VHOST_USER_GET_VRING_BASE:
 361                         if (msg.size != sizeof(m.payload.state)) {
 362                                 PMD_DRV_LOG(ERR, "Received bad msg size");
 363                                 return -1;
 364                         }
 365                         memcpy(arg, &msg.payload.state,
 366                                sizeof(struct vhost_vring_state));
 367                         break;
 368                 default:
 369                         PMD_DRV_LOG(ERR, "Received unexpected msg type");
 370                         return -1;
 371                 }
 372         }
 373
 374         return 0;
 375 }
 376
 377 #define MAX_VIRTIO_USER_BACKLOG 1
 378 static int
 379 virtio_user_start_server(struct virtio_user_dev *dev, struct sockaddr_un *un)
 380 {
 381         int ret;
 382         int flag;
 383         int fd = dev->listenfd;
 384
 385         ret = bind(fd, (struct sockaddr *)un, sizeof(*un));
 386         if (ret < 0) {
 387                 PMD_DRV_LOG(ERR, "failed to bind to %s: %s; remove it and try again\n",
 388                             dev->path, strerror(errno));
 389                 return -1;
 390         }
 391         ret = listen(fd, MAX_VIRTIO_USER_BACKLOG);
 392         if (ret < 0)
 393                 return -1;
 394
 395         flag = fcntl(fd, F_GETFL);
 396         if (fcntl(fd, F_SETFL, flag | O_NONBLOCK) < 0) {
 397                 PMD_DRV_LOG(ERR, "fcntl failed, %s", strerror(errno));
 398                 return -1;
 399         }
 400
 401         return 0;
 402 }
 403
 404 /**
 405  * Set up environment to talk with a vhost user backend.
 406  *
 407  * @return
 408  *   - (-1) if fail;
 409  *   - (0) if succeed.
 410  */
 411 static int
 412 vhost_user_setup(struct virtio_user_dev *dev)
 413 {
 414         int fd;
 415         int flag;
 416         struct sockaddr_un un;
 417
 418         fd = socket(AF_UNIX, SOCK_STREAM, 0);
 419         if (fd < 0) {
 420                 PMD_DRV_LOG(ERR, "socket() error, %s", strerror(errno));
 421                 return -1;
 422         }
 423
 424         flag = fcntl(fd, F_GETFD);
 425         if (fcntl(fd, F_SETFD, flag | FD_CLOEXEC) < 0)
 426                 PMD_DRV_LOG(WARNING, "fcntl failed, %s", strerror(errno));
 427
 428         memset(&un, 0, sizeof(un));
 429         un.sun_family = AF_UNIX;
 430         strlcpy(un.sun_path, dev->path, sizeof(un.sun_path));
 431
 432         if (dev->is_server) {
 433                 dev->listenfd = fd;
 434                 if (virtio_user_start_server(dev, &un) < 0) {
 435                         PMD_DRV_LOG(ERR, "virtio-user startup fails in server mode");
 436                         close(fd);
 437                         return -1;
 438                 }
 439                 dev->vhostfd = -1;
 440         } else {
 441                 if (connect(fd, (struct sockaddr *)&un, sizeof(un)) < 0) {
 442                         PMD_DRV_LOG(ERR, "connect error, %s", strerror(errno));
 443                         close(fd);
 444                         return -1;
 445                 }
 446                 dev->vhostfd = fd;
 447         }
 448
 449         return 0;
 450 }
 451
 452 static int
 453 vhost_user_enable_queue_pair(struct virtio_user_dev *dev,
 454                              uint16_t pair_idx,
 455                              int enable)
 456 {
 457         int i;
 458
 459         if (dev->qp_enabled[pair_idx] == enable)
 460                 return 0;
 461
 462         for (i = 0; i < 2; ++i) {
 463                 struct vhost_vring_state state = {
 464                         .index = pair_idx * 2 + i,
 465                         .num   = enable,
 466                 };
 467
 468                 if (vhost_user_sock(dev, VHOST_USER_SET_VRING_ENABLE, &state))
 469                         return -1;
 470         }
 471
 472         dev->qp_enabled[pair_idx] = enable;
 473         return 0;
 474 }
 475
 476 struct virtio_user_backend_ops virtio_ops_user = {
 477         .setup = vhost_user_setup,
 478         .send_request = vhost_user_sock,
 479         .enable_qp = vhost_user_enable_queue_pair
 480 };