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