source3/modules/vfs_aio_fork.c

   1 /*
   2  * Simulate the Posix AIO using mmap/fork
   3  *
   4  * Copyright (C) Volker Lendecke 2008
   5  * Copyright (C) Jeremy Allison 2010
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with this program; if not, write to the Free Software
  19  * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  20  */
  21
  22 #include "includes.h"
  23 #include "system/filesys.h"
  24 #include "system/shmem.h"
  25 #include "smbd/smbd.h"
  26 #include "smbd/globals.h"
  27 #include "lib/async_req/async_sock.h"
  28 #include "lib/util/tevent_unix.h"
  29 #include "lib/sys_rw.h"
  30
  31 #if !defined(HAVE_STRUCT_MSGHDR_MSG_CONTROL) && !defined(HAVE_STRUCT_MSGHDR_MSG_ACCRIGHTS)
  32 # error Can not pass file descriptors
  33 #endif
  34
  35 #undef recvmsg
  36
  37 #ifndef MAP_FILE
  38 #define MAP_FILE 0
  39 #endif
  40
  41 struct aio_fork_config {
  42         bool erratic_testing_mode;
  43 };
  44
  45 struct mmap_area {
  46         size_t size;
  47         volatile void *ptr;
  48 };
  49
  50 static int mmap_area_destructor(struct mmap_area *area)
  51 {
  52         munmap((void *)area->ptr, area->size);
  53         return 0;
  54 }
  55
  56 static struct mmap_area *mmap_area_init(TALLOC_CTX *mem_ctx, size_t size)
  57 {
  58         struct mmap_area *result;
  59         int fd;
  60
  61         result = talloc(mem_ctx, struct mmap_area);
  62         if (result == NULL) {
  63                 DEBUG(0, ("talloc failed\n"));
  64                 goto fail;
  65         }
  66
  67         fd = open("/dev/zero", O_RDWR);
  68         if (fd == -1) {
  69                 DEBUG(3, ("open(\"/dev/zero\") failed: %s\n",
  70                           strerror(errno)));
  71                 goto fail;
  72         }
  73
  74         result->ptr = mmap(NULL, size, PROT_READ|PROT_WRITE,
  75                            MAP_SHARED|MAP_FILE, fd, 0);
  76         close(fd);
  77         if (result->ptr == MAP_FAILED) {
  78                 DEBUG(1, ("mmap failed: %s\n", strerror(errno)));
  79                 goto fail;
  80         }
  81
  82         result->size = size;
  83         talloc_set_destructor(result, mmap_area_destructor);
  84
  85         return result;
  86
  87 fail:
  88         TALLOC_FREE(result);
  89         return NULL;
  90 }
  91
  92 enum cmd_type {
  93         READ_CMD,
  94         WRITE_CMD,
  95         FSYNC_CMD
  96 };
  97
  98 static const char *cmd_type_str(enum cmd_type cmd)
  99 {
 100         const char *result;
 101
 102         switch (cmd) {
 103         case READ_CMD:
 104                 result = "READ";
 105                 break;
 106         case WRITE_CMD:
 107                 result = "WRITE";
 108                 break;
 109         case FSYNC_CMD:
 110                 result = "FSYNC";
 111                 break;
 112         default:
 113                 result = "<UNKNOWN>";
 114                 break;
 115         }
 116         return result;
 117 }
 118
 119 struct rw_cmd {
 120         size_t n;
 121         off_t offset;
 122         enum cmd_type cmd;
 123         bool erratic_testing_mode;
 124 };
 125
 126 struct rw_ret {
 127         ssize_t size;
 128         int ret_errno;
 129 };
 130
 131 struct aio_child_list;
 132
 133 struct aio_child {
 134         struct aio_child *prev, *next;
 135         struct aio_child_list *list;
 136         pid_t pid;
 137         int sockfd;
 138         struct mmap_area *map;
 139         bool dont_delete;       /* Marked as in use since last cleanup */
 140         bool busy;
 141 };
 142
 143 struct aio_child_list {
 144         struct aio_child *children;
 145         struct tevent_timer *cleanup_event;
 146 };
 147
 148 static void free_aio_children(void **p)
 149 {
 150         TALLOC_FREE(*p);
 151 }
 152
 153 static ssize_t read_fd(int fd, void *ptr, size_t nbytes, int *recvfd)
 154 {
 155         struct msghdr msg;
 156         struct iovec iov[1];
 157         ssize_t n;
 158 #ifndef HAVE_STRUCT_MSGHDR_MSG_CONTROL
 159         int newfd;
 160
 161         ZERO_STRUCT(msg);
 162         msg.msg_accrights = (caddr_t) &newfd;
 163         msg.msg_accrightslen = sizeof(int);
 164 #else
 165
 166         union {
 167           struct cmsghdr        cm;
 168           char                          control[CMSG_SPACE(sizeof(int))];
 169         } control_un;
 170         struct cmsghdr  *cmptr;
 171
 172         ZERO_STRUCT(msg);
 173         ZERO_STRUCT(control_un);
 174
 175         msg.msg_control = control_un.control;
 176         msg.msg_controllen = sizeof(control_un.control);
 177 #endif
 178
 179         msg.msg_name = NULL;
 180         msg.msg_namelen = 0;
 181
 182         iov[0].iov_base = (void *)ptr;
 183         iov[0].iov_len = nbytes;
 184         msg.msg_iov = iov;
 185         msg.msg_iovlen = 1;
 186
 187         if ( (n = recvmsg(fd, &msg, 0)) <= 0) {
 188                 return(n);
 189         }
 190
 191 #ifdef  HAVE_STRUCT_MSGHDR_MSG_CONTROL
 192         if ((cmptr = CMSG_FIRSTHDR(&msg)) != NULL
 193             && cmptr->cmsg_len == CMSG_LEN(sizeof(int))) {
 194                 if (cmptr->cmsg_level != SOL_SOCKET) {
 195                         DEBUG(10, ("control level != SOL_SOCKET"));
 196                         errno = EINVAL;
 197                         return -1;
 198                 }
 199                 if (cmptr->cmsg_type != SCM_RIGHTS) {
 200                         DEBUG(10, ("control type != SCM_RIGHTS"));
 201                         errno = EINVAL;
 202                         return -1;
 203                 }
 204                 memcpy(recvfd, CMSG_DATA(cmptr), sizeof(*recvfd));
 205         } else {
 206                 *recvfd = -1;           /* descriptor was not passed */
 207         }
 208 #else
 209         if (msg.msg_accrightslen == sizeof(int)) {
 210                 *recvfd = newfd;
 211         }
 212         else {
 213                 *recvfd = -1;           /* descriptor was not passed */
 214         }
 215 #endif
 216
 217         return(n);
 218 }
 219
 220 static ssize_t write_fd(int fd, void *ptr, size_t nbytes, int sendfd)
 221 {
 222         struct msghdr   msg;
 223         struct iovec    iov[1];
 224
 225 #ifdef  HAVE_STRUCT_MSGHDR_MSG_CONTROL
 226         union {
 227                 struct cmsghdr  cm;
 228                 char control[CMSG_SPACE(sizeof(int))];
 229         } control_un;
 230         struct cmsghdr  *cmptr;
 231
 232         ZERO_STRUCT(msg);
 233         ZERO_STRUCT(control_un);
 234
 235         msg.msg_control = control_un.control;
 236         msg.msg_controllen = sizeof(control_un.control);
 237
 238         cmptr = CMSG_FIRSTHDR(&msg);
 239         cmptr->cmsg_len = CMSG_LEN(sizeof(int));
 240         cmptr->cmsg_level = SOL_SOCKET;
 241         cmptr->cmsg_type = SCM_RIGHTS;
 242         memcpy(CMSG_DATA(cmptr), &sendfd, sizeof(sendfd));
 243 #else
 244         ZERO_STRUCT(msg);
 245         msg.msg_accrights = (caddr_t) &sendfd;
 246         msg.msg_accrightslen = sizeof(int);
 247 #endif
 248
 249         msg.msg_name = NULL;
 250         msg.msg_namelen = 0;
 251
 252         ZERO_STRUCT(iov);
 253         iov[0].iov_base = (void *)ptr;
 254         iov[0].iov_len = nbytes;
 255         msg.msg_iov = iov;
 256         msg.msg_iovlen = 1;
 257
 258         return (sendmsg(fd, &msg, 0));
 259 }
 260
 261 static void aio_child_cleanup(struct tevent_context *event_ctx,
 262                               struct tevent_timer *te,
 263                               struct timeval now,
 264                               void *private_data)
 265 {
 266         struct aio_child_list *list = talloc_get_type_abort(
 267                 private_data, struct aio_child_list);
 268         struct aio_child *child, *next;
 269
 270         TALLOC_FREE(list->cleanup_event);
 271
 272         for (child = list->children; child != NULL; child = next) {
 273                 next = child->next;
 274
 275                 if (child->busy) {
 276                         DEBUG(10, ("child %d currently active\n",
 277                                    (int)child->pid));
 278                         continue;
 279                 }
 280
 281                 if (child->dont_delete) {
 282                         DEBUG(10, ("Child %d was active since last cleanup\n",
 283                                    (int)child->pid));
 284                         child->dont_delete = false;
 285                         continue;
 286                 }
 287
 288                 DEBUG(10, ("Child %d idle for more than 30 seconds, "
 289                            "deleting\n", (int)child->pid));
 290
 291                 TALLOC_FREE(child);
 292                 child = next;
 293         }
 294
 295         if (list->children != NULL) {
 296                 /*
 297                  * Re-schedule the next cleanup round
 298                  */
 299                 list->cleanup_event = tevent_add_timer(server_event_context(), list,
 300                                                       timeval_add(&now, 30, 0),
 301                                                       aio_child_cleanup, list);
 302
 303         }
 304 }
 305
 306 static struct aio_child_list *init_aio_children(struct vfs_handle_struct *handle)
 307 {
 308         struct aio_child_list *data = NULL;
 309
 310         if (SMB_VFS_HANDLE_TEST_DATA(handle)) {
 311                 SMB_VFS_HANDLE_GET_DATA(handle, data, struct aio_child_list,
 312                                         return NULL);
 313         }
 314
 315         if (data == NULL) {
 316                 data = talloc_zero(NULL, struct aio_child_list);
 317                 if (data == NULL) {
 318                         return NULL;
 319                 }
 320         }
 321
 322         /*
 323          * Regardless of whether the child_list had been around or not, make
 324          * sure that we have a cleanup timed event. This timed event will
 325          * delete itself when it finds that no children are around anymore.
 326          */
 327
 328         if (data->cleanup_event == NULL) {
 329                 data->cleanup_event = tevent_add_timer(server_event_context(), data,
 330                                                       timeval_current_ofs(30, 0),
 331                                                       aio_child_cleanup, data);
 332                 if (data->cleanup_event == NULL) {
 333                         TALLOC_FREE(data);
 334                         return NULL;
 335                 }
 336         }
 337
 338         if (!SMB_VFS_HANDLE_TEST_DATA(handle)) {
 339                 SMB_VFS_HANDLE_SET_DATA(handle, data, free_aio_children,
 340                                         struct aio_child_list, return False);
 341         }
 342
 343         return data;
 344 }
 345
 346 static void aio_child_loop(int sockfd, struct mmap_area *map)
 347 {
 348         while (true) {
 349                 int fd = -1;
 350                 ssize_t ret;
 351                 struct rw_cmd cmd_struct;
 352                 struct rw_ret ret_struct;
 353
 354                 ret = read_fd(sockfd, &cmd_struct, sizeof(cmd_struct), &fd);
 355                 if (ret != sizeof(cmd_struct)) {
 356                         DEBUG(10, ("read_fd returned %d: %s\n", (int)ret,
 357                                    strerror(errno)));
 358                         exit(1);
 359                 }
 360
 361                 DEBUG(10, ("aio_child_loop: %s %d bytes at %d from fd %d\n",
 362                            cmd_type_str(cmd_struct.cmd),
 363                            (int)cmd_struct.n, (int)cmd_struct.offset, fd));
 364
 365                 if (cmd_struct.erratic_testing_mode) {
 366                         /*
 367                          * For developer testing, we want erratic behaviour for
 368                          * async I/O times
 369                          */
 370                         uint8_t randval;
 371                         unsigned msecs;
 372                         /*
 373                          * use generate_random_buffer, we just forked from a
 374                          * common parent state
 375                          */
 376                         generate_random_buffer(&randval, sizeof(randval));
 377                         msecs = randval + 20;
 378                         DEBUG(10, ("delaying for %u msecs\n", msecs));
 379                         smb_msleep(msecs);
 380                 }
 381
 382                 ZERO_STRUCT(ret_struct);
 383
 384                 switch (cmd_struct.cmd) {
 385                 case READ_CMD:
 386                         ret_struct.size = sys_pread(
 387                                 fd, (void *)map->ptr, cmd_struct.n,
 388                                 cmd_struct.offset);
 389 #if 0
 390 /* This breaks "make test" when run with aio_fork module. */
 391 #ifdef DEVELOPER
 392                         ret_struct.size = MAX(1, ret_struct.size * 0.9);
 393 #endif
 394 #endif
 395                         break;
 396                 case WRITE_CMD:
 397                         ret_struct.size = sys_pwrite(
 398                                 fd, (void *)map->ptr, cmd_struct.n,
 399                                 cmd_struct.offset);
 400                         break;
 401                 case FSYNC_CMD:
 402                         ret_struct.size = fsync(fd);
 403                         break;
 404                 default:
 405                         ret_struct.size = -1;
 406                         errno = EINVAL;
 407                 }
 408
 409                 DEBUG(10, ("aio_child_loop: syscall returned %d\n",
 410                            (int)ret_struct.size));
 411
 412                 if (ret_struct.size == -1) {
 413                         ret_struct.ret_errno = errno;
 414                 }
 415
 416                 /*
 417                  * Close the fd before telling our parent we're done. The
 418                  * parent might close and re-open the file very quickly, and
 419                  * with system-level share modes (GPFS) we would get an
 420                  * unjustified SHARING_VIOLATION.
 421                  */
 422                 close(fd);
 423
 424                 ret = write_data(sockfd, (char *)&ret_struct,
 425                                  sizeof(ret_struct));
 426                 if (ret != sizeof(ret_struct)) {
 427                         DEBUG(10, ("could not write ret_struct: %s\n",
 428                                    strerror(errno)));
 429                         exit(2);
 430                 }
 431         }
 432 }
 433
 434 static int aio_child_destructor(struct aio_child *child)
 435 {
 436         char c=0;
 437
 438         SMB_ASSERT(!child->busy);
 439
 440         DEBUG(10, ("aio_child_destructor: removing child %d on fd %d\n",
 441                         child->pid, child->sockfd));
 442
 443         /*
 444          * closing the sockfd makes the child not return from recvmsg() on RHEL
 445          * 5.5 so instead force the child to exit by writing bad data to it
 446          */
 447         write(child->sockfd, &c, sizeof(c));
 448         close(child->sockfd);
 449         DLIST_REMOVE(child->list->children, child);
 450         return 0;
 451 }
 452
 453 /*
 454  * We have to close all fd's in open files, we might incorrectly hold a system
 455  * level share mode on a file.
 456  */
 457
 458 static struct files_struct *close_fsp_fd(struct files_struct *fsp,
 459                                          void *private_data)
 460 {
 461         if ((fsp->fh != NULL) && (fsp->fh->fd != -1)) {
 462                 close(fsp->fh->fd);
 463                 fsp->fh->fd = -1;
 464         }
 465         return NULL;
 466 }
 467
 468 static int create_aio_child(struct smbd_server_connection *sconn,
 469                             struct aio_child_list *children,
 470                             size_t map_size,
 471                             struct aio_child **presult)
 472 {
 473         struct aio_child *result;
 474         int fdpair[2];
 475         int ret;
 476
 477         fdpair[0] = fdpair[1] = -1;
 478
 479         result = talloc_zero(children, struct aio_child);
 480         if (result == NULL) {
 481                 return ENOMEM;
 482         }
 483
 484         if (socketpair(AF_UNIX, SOCK_STREAM, 0, fdpair) == -1) {
 485                 ret = errno;
 486                 DEBUG(10, ("socketpair() failed: %s\n", strerror(errno)));
 487                 goto fail;
 488         }
 489
 490         DEBUG(10, ("fdpair = %d/%d\n", fdpair[0], fdpair[1]));
 491
 492         result->map = mmap_area_init(result, map_size);
 493         if (result->map == NULL) {
 494                 ret = errno;
 495                 DEBUG(0, ("Could not create mmap area\n"));
 496                 goto fail;
 497         }
 498
 499         result->pid = fork();
 500         if (result->pid == -1) {
 501                 ret = errno;
 502                 DEBUG(0, ("fork failed: %s\n", strerror(errno)));
 503                 goto fail;
 504         }
 505
 506         if (result->pid == 0) {
 507                 close(fdpair[0]);
 508                 result->sockfd = fdpair[1];
 509                 files_forall(sconn, close_fsp_fd, NULL);
 510                 aio_child_loop(result->sockfd, result->map);
 511         }
 512
 513         DEBUG(10, ("Child %d created with sockfd %d\n",
 514                         result->pid, fdpair[0]));
 515
 516         result->sockfd = fdpair[0];
 517         close(fdpair[1]);
 518
 519         result->list = children;
 520         DLIST_ADD(children->children, result);
 521
 522         talloc_set_destructor(result, aio_child_destructor);
 523
 524         *presult = result;
 525
 526         return 0;
 527
 528  fail:
 529         if (fdpair[0] != -1) close(fdpair[0]);
 530         if (fdpair[1] != -1) close(fdpair[1]);
 531         TALLOC_FREE(result);
 532
 533         return ret;
 534 }
 535
 536 static int get_idle_child(struct vfs_handle_struct *handle,
 537                           struct aio_child **pchild)
 538 {
 539         struct aio_child_list *children;
 540         struct aio_child *child;
 541
 542         children = init_aio_children(handle);
 543         if (children == NULL) {
 544                 return ENOMEM;
 545         }
 546
 547         for (child = children->children; child != NULL; child = child->next) {
 548                 if (!child->busy) {
 549                         break;
 550                 }
 551         }
 552
 553         if (child == NULL) {
 554                 int ret;
 555
 556                 DEBUG(10, ("no idle child found, creating new one\n"));
 557
 558                 ret = create_aio_child(handle->conn->sconn, children,
 559                                           128*1024, &child);
 560                 if (ret != 0) {
 561                         DEBUG(10, ("create_aio_child failed: %s\n",
 562                                    strerror(errno)));
 563                         return ret;
 564                 }
 565         }
 566
 567         child->dont_delete = true;
 568         child->busy = true;
 569
 570         *pchild = child;
 571         return 0;
 572 }
 573
 574 struct aio_fork_pread_state {
 575         struct aio_child *child;
 576         ssize_t ret;
 577         int err;
 578 };
 579
 580 static void aio_fork_pread_done(struct tevent_req *subreq);
 581
 582 static struct tevent_req *aio_fork_pread_send(struct vfs_handle_struct *handle,
 583                                               TALLOC_CTX *mem_ctx,
 584                                               struct tevent_context *ev,
 585                                               struct files_struct *fsp,
 586                                               void *data,
 587                                               size_t n, off_t offset)
 588 {
 589         struct tevent_req *req, *subreq;
 590         struct aio_fork_pread_state *state;
 591         struct rw_cmd cmd;
 592         ssize_t written;
 593         int err;
 594         struct aio_fork_config *config;
 595
 596         SMB_VFS_HANDLE_GET_DATA(handle, config,
 597                                 struct aio_fork_config,
 598                                 return NULL);
 599
 600         req = tevent_req_create(mem_ctx, &state, struct aio_fork_pread_state);
 601         if (req == NULL) {
 602                 return NULL;
 603         }
 604
 605         if (n > 128*1024) {
 606                 /* TODO: support variable buffers */
 607                 tevent_req_error(req, EINVAL);
 608                 return tevent_req_post(req, ev);
 609         }
 610
 611         err = get_idle_child(handle, &state->child);
 612         if (err != 0) {
 613                 tevent_req_error(req, err);
 614                 return tevent_req_post(req, ev);
 615         }
 616
 617         ZERO_STRUCT(cmd);
 618         cmd.n = n;
 619         cmd.offset = offset;
 620         cmd.cmd = READ_CMD;
 621         cmd.erratic_testing_mode = config->erratic_testing_mode;
 622
 623         DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
 624                    (int)state->child->pid));
 625
 626         /*
 627          * Not making this async. We're writing into an empty unix
 628          * domain socket. This should never block.
 629          */
 630         written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
 631                            fsp->fh->fd);
 632         if (written == -1) {
 633                 err = errno;
 634
 635                 TALLOC_FREE(state->child);
 636
 637                 DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
 638                 tevent_req_error(req, err);
 639                 return tevent_req_post(req, ev);
 640         }
 641
 642         subreq = read_packet_send(state, ev, state->child->sockfd,
 643                                   sizeof(struct rw_ret), NULL, NULL);
 644         if (tevent_req_nomem(subreq, req)) {
 645                 TALLOC_FREE(state->child); /* we sent sth down */
 646                 return tevent_req_post(req, ev);
 647         }
 648         tevent_req_set_callback(subreq, aio_fork_pread_done, req);
 649         return req;
 650 }
 651
 652 static void aio_fork_pread_done(struct tevent_req *subreq)
 653 {
 654         struct tevent_req *req = tevent_req_callback_data(
 655                 subreq, struct tevent_req);
 656         struct aio_fork_pread_state *state = tevent_req_data(
 657                 req, struct aio_fork_pread_state);
 658         ssize_t nread;
 659         uint8_t *buf;
 660         int err;
 661         struct rw_ret *retbuf;
 662
 663         nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
 664         TALLOC_FREE(subreq);
 665         if (nread == -1) {
 666                 TALLOC_FREE(state->child);
 667                 tevent_req_error(req, err);
 668                 return;
 669         }
 670
 671         state->child->busy = false;
 672
 673         retbuf = (struct rw_ret *)buf;
 674         state->ret = retbuf->size;
 675         state->err = retbuf->ret_errno;
 676         tevent_req_done(req);
 677 }
 678
 679 static ssize_t aio_fork_pread_recv(struct tevent_req *req, int *err)
 680 {
 681         struct aio_fork_pread_state *state = tevent_req_data(
 682                 req, struct aio_fork_pread_state);
 683
 684         if (tevent_req_is_unix_error(req, err)) {
 685                 return -1;
 686         }
 687         if (state->ret == -1) {
 688                 *err = state->err;
 689         }
 690         return state->ret;
 691 }
 692
 693 struct aio_fork_pwrite_state {
 694         struct aio_child *child;
 695         ssize_t ret;
 696         int err;
 697 };
 698
 699 static void aio_fork_pwrite_done(struct tevent_req *subreq);
 700
 701 static struct tevent_req *aio_fork_pwrite_send(
 702         struct vfs_handle_struct *handle, TALLOC_CTX *mem_ctx,
 703         struct tevent_context *ev, struct files_struct *fsp,
 704         const void *data, size_t n, off_t offset)
 705 {
 706         struct tevent_req *req, *subreq;
 707         struct aio_fork_pwrite_state *state;
 708         struct rw_cmd cmd;
 709         ssize_t written;
 710         int err;
 711         struct aio_fork_config *config;
 712         SMB_VFS_HANDLE_GET_DATA(handle, config,
 713                                 struct aio_fork_config,
 714                                 return NULL);
 715
 716         req = tevent_req_create(mem_ctx, &state, struct aio_fork_pwrite_state);
 717         if (req == NULL) {
 718                 return NULL;
 719         }
 720
 721         if (n > 128*1024) {
 722                 /* TODO: support variable buffers */
 723                 tevent_req_error(req, EINVAL);
 724                 return tevent_req_post(req, ev);
 725         }
 726
 727         err = get_idle_child(handle, &state->child);
 728         if (err != 0) {
 729                 tevent_req_error(req, err);
 730                 return tevent_req_post(req, ev);
 731         }
 732
 733         ZERO_STRUCT(cmd);
 734         cmd.n = n;
 735         cmd.offset = offset;
 736         cmd.cmd = WRITE_CMD;
 737         cmd.erratic_testing_mode = config->erratic_testing_mode;
 738
 739         DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
 740                    (int)state->child->pid));
 741
 742         /*
 743          * Not making this async. We're writing into an empty unix
 744          * domain socket. This should never block.
 745          */
 746         written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
 747                            fsp->fh->fd);
 748         if (written == -1) {
 749                 err = errno;
 750
 751                 TALLOC_FREE(state->child);
 752
 753                 DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
 754                 tevent_req_error(req, err);
 755                 return tevent_req_post(req, ev);
 756         }
 757
 758         subreq = read_packet_send(state, ev, state->child->sockfd,
 759                                   sizeof(struct rw_ret), NULL, NULL);
 760         if (tevent_req_nomem(subreq, req)) {
 761                 TALLOC_FREE(state->child); /* we sent sth down */
 762                 return tevent_req_post(req, ev);
 763         }
 764         tevent_req_set_callback(subreq, aio_fork_pwrite_done, req);
 765         return req;
 766 }
 767
 768 static void aio_fork_pwrite_done(struct tevent_req *subreq)
 769 {
 770         struct tevent_req *req = tevent_req_callback_data(
 771                 subreq, struct tevent_req);
 772         struct aio_fork_pwrite_state *state = tevent_req_data(
 773                 req, struct aio_fork_pwrite_state);
 774         ssize_t nread;
 775         uint8_t *buf;
 776         int err;
 777         struct rw_ret *retbuf;
 778
 779         nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
 780         TALLOC_FREE(subreq);
 781         if (nread == -1) {
 782                 TALLOC_FREE(state->child);
 783                 tevent_req_error(req, err);
 784                 return;
 785         }
 786
 787         state->child->busy = false;
 788
 789         retbuf = (struct rw_ret *)buf;
 790         state->ret = retbuf->size;
 791         state->err = retbuf->ret_errno;
 792         tevent_req_done(req);
 793 }
 794
 795 static ssize_t aio_fork_pwrite_recv(struct tevent_req *req, int *err)
 796 {
 797         struct aio_fork_pwrite_state *state = tevent_req_data(
 798                 req, struct aio_fork_pwrite_state);
 799
 800         if (tevent_req_is_unix_error(req, err)) {
 801                 return -1;
 802         }
 803         if (state->ret == -1) {
 804                 *err = state->err;
 805         }
 806         return state->ret;
 807 }
 808
 809 struct aio_fork_fsync_state {
 810         struct aio_child *child;
 811         ssize_t ret;
 812         int err;
 813 };
 814
 815 static void aio_fork_fsync_done(struct tevent_req *subreq);
 816
 817 static struct tevent_req *aio_fork_fsync_send(
 818         struct vfs_handle_struct *handle, TALLOC_CTX *mem_ctx,
 819         struct tevent_context *ev, struct files_struct *fsp)
 820 {
 821         struct tevent_req *req, *subreq;
 822         struct aio_fork_fsync_state *state;
 823         struct rw_cmd cmd;
 824         ssize_t written;
 825         int err;
 826         struct aio_fork_config *config;
 827
 828         SMB_VFS_HANDLE_GET_DATA(handle, config,
 829                                 struct aio_fork_config,
 830                                 return NULL);
 831
 832         req = tevent_req_create(mem_ctx, &state, struct aio_fork_fsync_state);
 833         if (req == NULL) {
 834                 return NULL;
 835         }
 836
 837         err = get_idle_child(handle, &state->child);
 838         if (err != 0) {
 839                 tevent_req_error(req, err);
 840                 return tevent_req_post(req, ev);
 841         }
 842
 843         ZERO_STRUCT(cmd);
 844         cmd.cmd = FSYNC_CMD;
 845         cmd.erratic_testing_mode = config->erratic_testing_mode;
 846
 847         DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
 848                    (int)state->child->pid));
 849
 850         /*
 851          * Not making this async. We're writing into an empty unix
 852          * domain socket. This should never block.
 853          */
 854         written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
 855                            fsp->fh->fd);
 856         if (written == -1) {
 857                 err = errno;
 858
 859                 TALLOC_FREE(state->child);
 860
 861                 DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
 862                 tevent_req_error(req, err);
 863                 return tevent_req_post(req, ev);
 864         }
 865
 866         subreq = read_packet_send(state, ev, state->child->sockfd,
 867                                   sizeof(struct rw_ret), NULL, NULL);
 868         if (tevent_req_nomem(subreq, req)) {
 869                 TALLOC_FREE(state->child); /* we sent sth down */
 870                 return tevent_req_post(req, ev);
 871         }
 872         tevent_req_set_callback(subreq, aio_fork_fsync_done, req);
 873         return req;
 874 }
 875
 876 static void aio_fork_fsync_done(struct tevent_req *subreq)
 877 {
 878         struct tevent_req *req = tevent_req_callback_data(
 879                 subreq, struct tevent_req);
 880         struct aio_fork_fsync_state *state = tevent_req_data(
 881                 req, struct aio_fork_fsync_state);
 882         ssize_t nread;
 883         uint8_t *buf;
 884         int err;
 885         struct rw_ret *retbuf;
 886
 887         nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
 888         TALLOC_FREE(subreq);
 889         if (nread == -1) {
 890                 TALLOC_FREE(state->child);
 891                 tevent_req_error(req, err);
 892                 return;
 893         }
 894
 895         state->child->busy = false;
 896
 897         retbuf = (struct rw_ret *)buf;
 898         state->ret = retbuf->size;
 899         state->err = retbuf->ret_errno;
 900         tevent_req_done(req);
 901 }
 902
 903 static int aio_fork_fsync_recv(struct tevent_req *req, int *err)
 904 {
 905         struct aio_fork_fsync_state *state = tevent_req_data(
 906                 req, struct aio_fork_fsync_state);
 907
 908         if (tevent_req_is_unix_error(req, err)) {
 909                 return -1;
 910         }
 911         if (state->ret == -1) {
 912                 *err = state->err;
 913         }
 914         return state->ret;
 915 }
 916
 917 static int aio_fork_connect(vfs_handle_struct *handle, const char *service,
 918                             const char *user)
 919 {
 920         int ret;
 921         struct aio_fork_config *config;
 922         ret = SMB_VFS_NEXT_CONNECT(handle, service, user);
 923
 924         if (ret < 0) {
 925                 return ret;
 926         }
 927
 928         config = talloc_zero(handle->conn, struct aio_fork_config);
 929         if (!config) {
 930                 SMB_VFS_NEXT_DISCONNECT(handle);
 931                 DEBUG(0, ("talloc_zero() failed\n"));
 932                 return -1;
 933         }
 934
 935         config->erratic_testing_mode = lp_parm_bool(SNUM(handle->conn), "vfs_aio_fork",
 936                                                     "erratic_testing_mode", false);
 937
 938         SMB_VFS_HANDLE_SET_DATA(handle, config,
 939                                 NULL, struct aio_fork_config,
 940                                 return -1);
 941
 942         /*********************************************************************
 943          * How many threads to initialize ?
 944          * 100 per process seems insane as a default until you realize that
 945          * (a) Threads terminate after 1 second when idle.
 946          * (b) Throttling is done in SMB2 via the crediting algorithm.
 947          * (c) SMB1 clients are limited to max_mux (50) outstanding
 948          *     requests and Windows clients don't use this anyway.
 949          * Essentially we want this to be unlimited unless smb.conf
 950          * says different.
 951          *********************************************************************/
 952         aio_pending_size = 100;
 953         return 0;
 954 }
 955
 956 static struct vfs_fn_pointers vfs_aio_fork_fns = {
 957         .connect_fn = aio_fork_connect,
 958         .pread_send_fn = aio_fork_pread_send,
 959         .pread_recv_fn = aio_fork_pread_recv,
 960         .pwrite_send_fn = aio_fork_pwrite_send,
 961         .pwrite_recv_fn = aio_fork_pwrite_recv,
 962         .fsync_send_fn = aio_fork_fsync_send,
 963         .fsync_recv_fn = aio_fork_fsync_recv,
 964 };
 965
 966 NTSTATUS vfs_aio_fork_init(void);
 967 NTSTATUS vfs_aio_fork_init(void)
 968 {
 969         return smb_register_vfs(SMB_VFS_INTERFACE_VERSION,
 970                                 "aio_fork", &vfs_aio_fork_fns);
 971 }