aio-posix.c

   1 /*
   2  * QEMU aio implementation
   3  *
   4  * Copyright IBM, Corp. 2008
   5  *
   6  * Authors:
   7  *  Anthony Liguori   <aliguori@us.ibm.com>
   8  *
   9  * This work is licensed under the terms of the GNU GPL, version 2.  See
  10  * the COPYING file in the top-level directory.
  11  *
  12  * Contributions after 2012-01-13 are licensed under the terms of the
  13  * GNU GPL, version 2 or (at your option) any later version.
  14  */
  15
  16 #include "qemu/osdep.h"
  17 #include "qemu-common.h"
  18 #include "block/block.h"
  19 #include "qemu/queue.h"
  20 #include "qemu/sockets.h"
  21 #ifdef CONFIG_EPOLL_CREATE1
  22 #include <sys/epoll.h>
  23 #endif
  24
  25 struct AioHandler
  26 {
  27     GPollFD pfd;
  28     IOHandler *io_read;
  29     IOHandler *io_write;
  30     int deleted;
  31     void *opaque;
  32     bool is_external;
  33     QLIST_ENTRY(AioHandler) node;
  34 };
  35
  36 #ifdef CONFIG_EPOLL_CREATE1
  37
  38 /* The fd number threashold to switch to epoll */
  39 #define EPOLL_ENABLE_THRESHOLD 64
  40
  41 static void aio_epoll_disable(AioContext *ctx)
  42 {
  43     ctx->epoll_available = false;
  44     if (!ctx->epoll_enabled) {
  45         return;
  46     }
  47     ctx->epoll_enabled = false;
  48     close(ctx->epollfd);
  49 }
  50
  51 static inline int epoll_events_from_pfd(int pfd_events)
  52 {
  53     return (pfd_events & G_IO_IN ? EPOLLIN : 0) |
  54            (pfd_events & G_IO_OUT ? EPOLLOUT : 0) |
  55            (pfd_events & G_IO_HUP ? EPOLLHUP : 0) |
  56            (pfd_events & G_IO_ERR ? EPOLLERR : 0);
  57 }
  58
  59 static bool aio_epoll_try_enable(AioContext *ctx)
  60 {
  61     AioHandler *node;
  62     struct epoll_event event;
  63
  64     QLIST_FOREACH(node, &ctx->aio_handlers, node) {
  65         int r;
  66         if (node->deleted || !node->pfd.events) {
  67             continue;
  68         }
  69         event.events = epoll_events_from_pfd(node->pfd.events);
  70         event.data.ptr = node;
  71         r = epoll_ctl(ctx->epollfd, EPOLL_CTL_ADD, node->pfd.fd, &event);
  72         if (r) {
  73             return false;
  74         }
  75     }
  76     ctx->epoll_enabled = true;
  77     return true;
  78 }
  79
  80 static void aio_epoll_update(AioContext *ctx, AioHandler *node, bool is_new)
  81 {
  82     struct epoll_event event;
  83     int r;
  84     int ctl;
  85
  86     if (!ctx->epoll_enabled) {
  87         return;
  88     }
  89     if (!node->pfd.events) {
  90         ctl = EPOLL_CTL_DEL;
  91     } else {
  92         event.data.ptr = node;
  93         event.events = epoll_events_from_pfd(node->pfd.events);
  94         ctl = is_new ? EPOLL_CTL_ADD : EPOLL_CTL_MOD;
  95     }
  96
  97     r = epoll_ctl(ctx->epollfd, ctl, node->pfd.fd, &event);
  98     if (r) {
  99         aio_epoll_disable(ctx);
 100     }
 101 }
 102
 103 static int aio_epoll(AioContext *ctx, GPollFD *pfds,
 104                      unsigned npfd, int64_t timeout)
 105 {
 106     AioHandler *node;
 107     int i, ret = 0;
 108     struct epoll_event events[128];
 109
 110     assert(npfd == 1);
 111     assert(pfds[0].fd == ctx->epollfd);
 112     if (timeout > 0) {
 113         ret = qemu_poll_ns(pfds, npfd, timeout);
 114     }
 115     if (timeout <= 0 || ret > 0) {
 116         ret = epoll_wait(ctx->epollfd, events,
 117                          sizeof(events) / sizeof(events[0]),
 118                          timeout);
 119         if (ret <= 0) {
 120             goto out;
 121         }
 122         for (i = 0; i < ret; i++) {
 123             int ev = events[i].events;
 124             node = events[i].data.ptr;
 125             node->pfd.revents = (ev & EPOLLIN ? G_IO_IN : 0) |
 126                 (ev & EPOLLOUT ? G_IO_OUT : 0) |
 127                 (ev & EPOLLHUP ? G_IO_HUP : 0) |
 128                 (ev & EPOLLERR ? G_IO_ERR : 0);
 129         }
 130     }
 131 out:
 132     return ret;
 133 }
 134
 135 static bool aio_epoll_enabled(AioContext *ctx)
 136 {
 137     /* Fall back to ppoll when external clients are disabled. */
 138     return !aio_external_disabled(ctx) && ctx->epoll_enabled;
 139 }
 140
 141 static bool aio_epoll_check_poll(AioContext *ctx, GPollFD *pfds,
 142                                  unsigned npfd, int64_t timeout)
 143 {
 144     if (!ctx->epoll_available) {
 145         return false;
 146     }
 147     if (aio_epoll_enabled(ctx)) {
 148         return true;
 149     }
 150     if (npfd >= EPOLL_ENABLE_THRESHOLD) {
 151         if (aio_epoll_try_enable(ctx)) {
 152             return true;
 153         } else {
 154             aio_epoll_disable(ctx);
 155         }
 156     }
 157     return false;
 158 }
 159
 160 #else
 161
 162 static void aio_epoll_update(AioContext *ctx, AioHandler *node, bool is_new)
 163 {
 164 }
 165
 166 static int aio_epoll(AioContext *ctx, GPollFD *pfds,
 167                      unsigned npfd, int64_t timeout)
 168 {
 169     assert(false);
 170 }
 171
 172 static bool aio_epoll_enabled(AioContext *ctx)
 173 {
 174     return false;
 175 }
 176
 177 static bool aio_epoll_check_poll(AioContext *ctx, GPollFD *pfds,
 178                           unsigned npfd, int64_t timeout)
 179 {
 180     return false;
 181 }
 182
 183 #endif
 184
 185 static AioHandler *find_aio_handler(AioContext *ctx, int fd)
 186 {
 187     AioHandler *node;
 188
 189     QLIST_FOREACH(node, &ctx->aio_handlers, node) {
 190         if (node->pfd.fd == fd)
 191             if (!node->deleted)
 192                 return node;
 193     }
 194
 195     return NULL;
 196 }
 197
 198 void aio_set_fd_handler(AioContext *ctx,
 199                         int fd,
 200                         bool is_external,
 201                         IOHandler *io_read,
 202                         IOHandler *io_write,
 203                         void *opaque)
 204 {
 205     AioHandler *node;
 206     bool is_new = false;
 207     bool deleted = false;
 208
 209     node = find_aio_handler(ctx, fd);
 210
 211     /* Are we deleting the fd handler? */
 212     if (!io_read && !io_write) {
 213         if (node == NULL) {
 214             return;
 215         }
 216
 217         g_source_remove_poll(&ctx->source, &node->pfd);
 218
 219         /* If the lock is held, just mark the node as deleted */
 220         if (ctx->walking_handlers) {
 221             node->deleted = 1;
 222             node->pfd.revents = 0;
 223         } else {
 224             /* Otherwise, delete it for real.  We can't just mark it as
 225              * deleted because deleted nodes are only cleaned up after
 226              * releasing the walking_handlers lock.
 227              */
 228             QLIST_REMOVE(node, node);
 229             deleted = true;
 230         }
 231     } else {
 232         if (node == NULL) {
 233             /* Alloc and insert if it's not already there */
 234             node = g_new0(AioHandler, 1);
 235             node->pfd.fd = fd;
 236             QLIST_INSERT_HEAD(&ctx->aio_handlers, node, node);
 237
 238             g_source_add_poll(&ctx->source, &node->pfd);
 239             is_new = true;
 240         }
 241         /* Update handler with latest information */
 242         node->io_read = io_read;
 243         node->io_write = io_write;
 244         node->opaque = opaque;
 245         node->is_external = is_external;
 246
 247         node->pfd.events = (io_read ? G_IO_IN | G_IO_HUP | G_IO_ERR : 0);
 248         node->pfd.events |= (io_write ? G_IO_OUT | G_IO_ERR : 0);
 249     }
 250
 251     aio_epoll_update(ctx, node, is_new);
 252     aio_notify(ctx);
 253     if (deleted) {
 254         g_free(node);
 255     }
 256 }
 257
 258 void aio_set_event_notifier(AioContext *ctx,
 259                             EventNotifier *notifier,
 260                             bool is_external,
 261                             EventNotifierHandler *io_read)
 262 {
 263     aio_set_fd_handler(ctx, event_notifier_get_fd(notifier),
 264                        is_external, (IOHandler *)io_read, NULL, notifier);
 265 }
 266
 267 bool aio_prepare(AioContext *ctx)
 268 {
 269     return false;
 270 }
 271
 272 bool aio_pending(AioContext *ctx)
 273 {
 274     AioHandler *node;
 275
 276     QLIST_FOREACH(node, &ctx->aio_handlers, node) {
 277         int revents;
 278
 279         revents = node->pfd.revents & node->pfd.events;
 280         if (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR) && node->io_read &&
 281             aio_node_check(ctx, node->is_external)) {
 282             return true;
 283         }
 284         if (revents & (G_IO_OUT | G_IO_ERR) && node->io_write &&
 285             aio_node_check(ctx, node->is_external)) {
 286             return true;
 287         }
 288     }
 289
 290     return false;
 291 }
 292
 293 bool aio_dispatch(AioContext *ctx)
 294 {
 295     AioHandler *node;
 296     bool progress = false;
 297
 298     /*
 299      * If there are callbacks left that have been queued, we need to call them.
 300      * Do not call select in this case, because it is possible that the caller
 301      * does not need a complete flush (as is the case for aio_poll loops).
 302      */
 303     if (aio_bh_poll(ctx)) {
 304         progress = true;
 305     }
 306
 307     /*
 308      * We have to walk very carefully in case aio_set_fd_handler is
 309      * called while we're walking.
 310      */
 311     node = QLIST_FIRST(&ctx->aio_handlers);
 312     while (node) {
 313         AioHandler *tmp;
 314         int revents;
 315
 316         ctx->walking_handlers++;
 317
 318         revents = node->pfd.revents & node->pfd.events;
 319         node->pfd.revents = 0;
 320
 321         if (!node->deleted &&
 322             (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
 323             aio_node_check(ctx, node->is_external) &&
 324             node->io_read) {
 325             node->io_read(node->opaque);
 326
 327             /* aio_notify() does not count as progress */
 328             if (node->opaque != &ctx->notifier) {
 329                 progress = true;
 330             }
 331         }
 332         if (!node->deleted &&
 333             (revents & (G_IO_OUT | G_IO_ERR)) &&
 334             aio_node_check(ctx, node->is_external) &&
 335             node->io_write) {
 336             node->io_write(node->opaque);
 337             progress = true;
 338         }
 339
 340         tmp = node;
 341         node = QLIST_NEXT(node, node);
 342
 343         ctx->walking_handlers--;
 344
 345         if (!ctx->walking_handlers && tmp->deleted) {
 346             QLIST_REMOVE(tmp, node);
 347             g_free(tmp);
 348         }
 349     }
 350
 351     /* Run our timers */
 352     progress |= timerlistgroup_run_timers(&ctx->tlg);
 353
 354     return progress;
 355 }
 356
 357 /* These thread-local variables are used only in a small part of aio_poll
 358  * around the call to the poll() system call.  In particular they are not
 359  * used while aio_poll is performing callbacks, which makes it much easier
 360  * to think about reentrancy!
 361  *
 362  * Stack-allocated arrays would be perfect but they have size limitations;
 363  * heap allocation is expensive enough that we want to reuse arrays across
 364  * calls to aio_poll().  And because poll() has to be called without holding
 365  * any lock, the arrays cannot be stored in AioContext.  Thread-local data
 366  * has none of the disadvantages of these three options.
 367  */
 368 static __thread GPollFD *pollfds;
 369 static __thread AioHandler **nodes;
 370 static __thread unsigned npfd, nalloc;
 371 static __thread Notifier pollfds_cleanup_notifier;
 372
 373 static void pollfds_cleanup(Notifier *n, void *unused)
 374 {
 375     g_assert(npfd == 0);
 376     g_free(pollfds);
 377     g_free(nodes);
 378     nalloc = 0;
 379 }
 380
 381 static void add_pollfd(AioHandler *node)
 382 {
 383     if (npfd == nalloc) {
 384         if (nalloc == 0) {
 385             pollfds_cleanup_notifier.notify = pollfds_cleanup;
 386             qemu_thread_atexit_add(&pollfds_cleanup_notifier);
 387             nalloc = 8;
 388         } else {
 389             g_assert(nalloc <= INT_MAX);
 390             nalloc *= 2;
 391         }
 392         pollfds = g_renew(GPollFD, pollfds, nalloc);
 393         nodes = g_renew(AioHandler *, nodes, nalloc);
 394     }
 395     nodes[npfd] = node;
 396     pollfds[npfd] = (GPollFD) {
 397         .fd = node->pfd.fd,
 398         .events = node->pfd.events,
 399     };
 400     npfd++;
 401 }
 402
 403 bool aio_poll(AioContext *ctx, bool blocking)
 404 {
 405     AioHandler *node;
 406     int i, ret;
 407     bool progress;
 408     int64_t timeout;
 409
 410     aio_context_acquire(ctx);
 411     progress = false;
 412
 413     /* aio_notify can avoid the expensive event_notifier_set if
 414      * everything (file descriptors, bottom halves, timers) will
 415      * be re-evaluated before the next blocking poll().  This is
 416      * already true when aio_poll is called with blocking == false;
 417      * if blocking == true, it is only true after poll() returns,
 418      * so disable the optimization now.
 419      */
 420     if (blocking) {
 421         atomic_add(&ctx->notify_me, 2);
 422     }
 423
 424     ctx->walking_handlers++;
 425
 426     assert(npfd == 0);
 427
 428     /* fill pollfds */
 429
 430     if (!aio_epoll_enabled(ctx)) {
 431         QLIST_FOREACH(node, &ctx->aio_handlers, node) {
 432             if (!node->deleted && node->pfd.events
 433                 && aio_node_check(ctx, node->is_external)) {
 434                 add_pollfd(node);
 435             }
 436         }
 437     }
 438
 439     timeout = blocking ? aio_compute_timeout(ctx) : 0;
 440
 441     /* wait until next event */
 442     if (timeout) {
 443         aio_context_release(ctx);
 444     }
 445     if (aio_epoll_check_poll(ctx, pollfds, npfd, timeout)) {
 446         AioHandler epoll_handler;
 447
 448         epoll_handler.pfd.fd = ctx->epollfd;
 449         epoll_handler.pfd.events = G_IO_IN | G_IO_OUT | G_IO_HUP | G_IO_ERR;
 450         npfd = 0;
 451         add_pollfd(&epoll_handler);
 452         ret = aio_epoll(ctx, pollfds, npfd, timeout);
 453     } else  {
 454         ret = qemu_poll_ns(pollfds, npfd, timeout);
 455     }
 456     if (blocking) {
 457         atomic_sub(&ctx->notify_me, 2);
 458     }
 459     if (timeout) {
 460         aio_context_acquire(ctx);
 461     }
 462
 463     aio_notify_accept(ctx);
 464
 465     /* if we have any readable fds, dispatch event */
 466     if (ret > 0) {
 467         for (i = 0; i < npfd; i++) {
 468             nodes[i]->pfd.revents = pollfds[i].revents;
 469         }
 470     }
 471
 472     npfd = 0;
 473     ctx->walking_handlers--;
 474
 475     /* Run dispatch even if there were no readable fds to run timers */
 476     if (aio_dispatch(ctx)) {
 477         progress = true;
 478     }
 479
 480     aio_context_release(ctx);
 481
 482     return progress;
 483 }
 484
 485 void aio_context_setup(AioContext *ctx)
 486 {
 487 #ifdef CONFIG_EPOLL_CREATE1
 488     assert(!ctx->epollfd);
 489     ctx->epollfd = epoll_create1(EPOLL_CLOEXEC);
 490     if (ctx->epollfd == -1) {
 491         fprintf(stderr, "Failed to create epoll instance: %s", strerror(errno));
 492         ctx->epoll_available = false;
 493     } else {
 494         ctx->epoll_available = true;
 495     }
 496 #endif
 497 }