block/rbd.c

   1 /*
   2  * QEMU Block driver for RADOS (Ceph)
   3  *
   4  * Copyright (C) 2010-2011 Christian Brunner <chb@muc.de>,
   5  *                         Josh Durgin <josh.durgin@dreamhost.com>
   6  *
   7  * This work is licensed under the terms of the GNU GPL, version 2.  See
   8  * the COPYING file in the top-level directory.
   9  *
  10  * Contributions after 2012-01-13 are licensed under the terms of the
  11  * GNU GPL, version 2 or (at your option) any later version.
  12  */
  13
  14 #include "qemu/osdep.h"
  15
  16 #include <rbd/librbd.h>
  17 #include "qapi/error.h"
  18 #include "qemu/error-report.h"
  19 #include "qemu/option.h"
  20 #include "block/block_int.h"
  21 #include "crypto/secret.h"
  22 #include "qemu/cutils.h"
  23 #include "qapi/qmp/qstring.h"
  24 #include "qapi/qmp/qdict.h"
  25 #include "qapi/qmp/qjson.h"
  26 #include "qapi/qmp/qlist.h"
  27
  28 /*
  29  * When specifying the image filename use:
  30  *
  31  * rbd:poolname/devicename[@snapshotname][:option1=value1[:option2=value2...]]
  32  *
  33  * poolname must be the name of an existing rados pool.
  34  *
  35  * devicename is the name of the rbd image.
  36  *
  37  * Each option given is used to configure rados, and may be any valid
  38  * Ceph option, "id", or "conf".
  39  *
  40  * The "id" option indicates what user we should authenticate as to
  41  * the Ceph cluster.  If it is excluded we will use the Ceph default
  42  * (normally 'admin').
  43  *
  44  * The "conf" option specifies a Ceph configuration file to read.  If
  45  * it is not specified, we will read from the default Ceph locations
  46  * (e.g., /etc/ceph/ceph.conf).  To avoid reading _any_ configuration
  47  * file, specify conf=/dev/null.
  48  *
  49  * Configuration values containing :, @, or = can be escaped with a
  50  * leading "\".
  51  */
  52
  53 /* rbd_aio_discard added in 0.1.2 */
  54 #if LIBRBD_VERSION_CODE >= LIBRBD_VERSION(0, 1, 2)
  55 #define LIBRBD_SUPPORTS_DISCARD
  56 #else
  57 #undef LIBRBD_SUPPORTS_DISCARD
  58 #endif
  59
  60 #define OBJ_MAX_SIZE (1UL << OBJ_DEFAULT_OBJ_ORDER)
  61
  62 #define RBD_MAX_SNAPS 100
  63
  64 /* The LIBRBD_SUPPORTS_IOVEC is defined in librbd.h */
  65 #ifdef LIBRBD_SUPPORTS_IOVEC
  66 #define LIBRBD_USE_IOVEC 1
  67 #else
  68 #define LIBRBD_USE_IOVEC 0
  69 #endif
  70
  71 typedef enum {
  72     RBD_AIO_READ,
  73     RBD_AIO_WRITE,
  74     RBD_AIO_DISCARD,
  75     RBD_AIO_FLUSH
  76 } RBDAIOCmd;
  77
  78 typedef struct RBDAIOCB {
  79     BlockAIOCB common;
  80     int64_t ret;
  81     QEMUIOVector *qiov;
  82     char *bounce;
  83     RBDAIOCmd cmd;
  84     int error;
  85     struct BDRVRBDState *s;
  86 } RBDAIOCB;
  87
  88 typedef struct RADOSCB {
  89     RBDAIOCB *acb;
  90     struct BDRVRBDState *s;
  91     int64_t size;
  92     char *buf;
  93     int64_t ret;
  94 } RADOSCB;
  95
  96 typedef struct BDRVRBDState {
  97     rados_t cluster;
  98     rados_ioctx_t io_ctx;
  99     rbd_image_t image;
 100     char *image_name;
 101     char *snap;
 102 } BDRVRBDState;
 103
 104 static char *qemu_rbd_next_tok(char *src, char delim, char **p)
 105 {
 106     char *end;
 107
 108     *p = NULL;
 109
 110     for (end = src; *end; ++end) {
 111         if (*end == delim) {
 112             break;
 113         }
 114         if (*end == '\\' && end[1] != '\0') {
 115             end++;
 116         }
 117     }
 118     if (*end == delim) {
 119         *p = end + 1;
 120         *end = '\0';
 121     }
 122     return src;
 123 }
 124
 125 static void qemu_rbd_unescape(char *src)
 126 {
 127     char *p;
 128
 129     for (p = src; *src; ++src, ++p) {
 130         if (*src == '\\' && src[1] != '\0') {
 131             src++;
 132         }
 133         *p = *src;
 134     }
 135     *p = '\0';
 136 }
 137
 138 static void qemu_rbd_parse_filename(const char *filename, QDict *options,
 139                                     Error **errp)
 140 {
 141     const char *start;
 142     char *p, *buf;
 143     QList *keypairs = NULL;
 144     char *found_str;
 145
 146     if (!strstart(filename, "rbd:", &start)) {
 147         error_setg(errp, "File name must start with 'rbd:'");
 148         return;
 149     }
 150
 151     buf = g_strdup(start);
 152     p = buf;
 153
 154     found_str = qemu_rbd_next_tok(p, '/', &p);
 155     if (!p) {
 156         error_setg(errp, "Pool name is required");
 157         goto done;
 158     }
 159     qemu_rbd_unescape(found_str);
 160     qdict_put_str(options, "pool", found_str);
 161
 162     if (strchr(p, '@')) {
 163         found_str = qemu_rbd_next_tok(p, '@', &p);
 164         qemu_rbd_unescape(found_str);
 165         qdict_put_str(options, "image", found_str);
 166
 167         found_str = qemu_rbd_next_tok(p, ':', &p);
 168         qemu_rbd_unescape(found_str);
 169         qdict_put_str(options, "snapshot", found_str);
 170     } else {
 171         found_str = qemu_rbd_next_tok(p, ':', &p);
 172         qemu_rbd_unescape(found_str);
 173         qdict_put_str(options, "image", found_str);
 174     }
 175     if (!p) {
 176         goto done;
 177     }
 178
 179     /* The following are essentially all key/value pairs, and we treat
 180      * 'id' and 'conf' a bit special.  Key/value pairs may be in any order. */
 181     while (p) {
 182         char *name, *value;
 183         name = qemu_rbd_next_tok(p, '=', &p);
 184         if (!p) {
 185             error_setg(errp, "conf option %s has no value", name);
 186             break;
 187         }
 188
 189         qemu_rbd_unescape(name);
 190
 191         value = qemu_rbd_next_tok(p, ':', &p);
 192         qemu_rbd_unescape(value);
 193
 194         if (!strcmp(name, "conf")) {
 195             qdict_put_str(options, "conf", value);
 196         } else if (!strcmp(name, "id")) {
 197             qdict_put_str(options, "user", value);
 198         } else {
 199             /*
 200              * We pass these internally to qemu_rbd_set_keypairs(), so
 201              * we can get away with the simpler list of [ "key1",
 202              * "value1", "key2", "value2" ] rather than a raw dict
 203              * { "key1": "value1", "key2": "value2" } where we can't
 204              * guarantee order, or even a more correct but complex
 205              * [ { "key1": "value1" }, { "key2": "value2" } ]
 206              */
 207             if (!keypairs) {
 208                 keypairs = qlist_new();
 209             }
 210             qlist_append_str(keypairs, name);
 211             qlist_append_str(keypairs, value);
 212         }
 213     }
 214
 215     if (keypairs) {
 216         qdict_put(options, "=keyvalue-pairs",
 217                   qobject_to_json(QOBJECT(keypairs)));
 218     }
 219
 220 done:
 221     g_free(buf);
 222     QDECREF(keypairs);
 223     return;
 224 }
 225
 226
 227 static int qemu_rbd_set_auth(rados_t cluster, const char *secretid,
 228                              Error **errp)
 229 {
 230     if (secretid == 0) {
 231         return 0;
 232     }
 233
 234     gchar *secret = qcrypto_secret_lookup_as_base64(secretid,
 235                                                     errp);
 236     if (!secret) {
 237         return -1;
 238     }
 239
 240     rados_conf_set(cluster, "key", secret);
 241     g_free(secret);
 242
 243     return 0;
 244 }
 245
 246 static int qemu_rbd_set_keypairs(rados_t cluster, const char *keypairs_json,
 247                                  Error **errp)
 248 {
 249     QList *keypairs;
 250     QString *name;
 251     QString *value;
 252     const char *key;
 253     size_t remaining;
 254     int ret = 0;
 255
 256     if (!keypairs_json) {
 257         return ret;
 258     }
 259     keypairs = qobject_to_qlist(qobject_from_json(keypairs_json,
 260                                                   &error_abort));
 261     remaining = qlist_size(keypairs) / 2;
 262     assert(remaining);
 263
 264     while (remaining--) {
 265         name = qobject_to_qstring(qlist_pop(keypairs));
 266         value = qobject_to_qstring(qlist_pop(keypairs));
 267         assert(name && value);
 268         key = qstring_get_str(name);
 269
 270         ret = rados_conf_set(cluster, key, qstring_get_str(value));
 271         QDECREF(name);
 272         QDECREF(value);
 273         if (ret < 0) {
 274             error_setg_errno(errp, -ret, "invalid conf option %s", key);
 275             ret = -EINVAL;
 276             break;
 277         }
 278     }
 279
 280     QDECREF(keypairs);
 281     return ret;
 282 }
 283
 284 static void qemu_rbd_memset(RADOSCB *rcb, int64_t offs)
 285 {
 286     if (LIBRBD_USE_IOVEC) {
 287         RBDAIOCB *acb = rcb->acb;
 288         iov_memset(acb->qiov->iov, acb->qiov->niov, offs, 0,
 289                    acb->qiov->size - offs);
 290     } else {
 291         memset(rcb->buf + offs, 0, rcb->size - offs);
 292     }
 293 }
 294
 295 static QemuOptsList runtime_opts = {
 296     .name = "rbd",
 297     .head = QTAILQ_HEAD_INITIALIZER(runtime_opts.head),
 298     .desc = {
 299         {
 300             .name = "pool",
 301             .type = QEMU_OPT_STRING,
 302             .help = "Rados pool name",
 303         },
 304         {
 305             .name = "image",
 306             .type = QEMU_OPT_STRING,
 307             .help = "Image name in the pool",
 308         },
 309         {
 310             .name = "conf",
 311             .type = QEMU_OPT_STRING,
 312             .help = "Rados config file location",
 313         },
 314         {
 315             .name = "snapshot",
 316             .type = QEMU_OPT_STRING,
 317             .help = "Ceph snapshot name",
 318         },
 319         {
 320             /* maps to 'id' in rados_create() */
 321             .name = "user",
 322             .type = QEMU_OPT_STRING,
 323             .help = "Rados id name",
 324         },
 325         /*
 326          * server.* extracted manually, see qemu_rbd_mon_host()
 327          */
 328         {
 329             .name = "password-secret",
 330             .type = QEMU_OPT_STRING,
 331             .help = "ID of secret providing the password",
 332         },
 333
 334         /*
 335          * Keys for qemu_rbd_parse_filename(), not in the QAPI schema
 336          */
 337         {
 338             /*
 339              * HACK: name starts with '=' so that qemu_opts_parse()
 340              * can't set it
 341              */
 342             .name = "=keyvalue-pairs",
 343             .type = QEMU_OPT_STRING,
 344             .help = "Legacy rados key/value option parameters",
 345         },
 346         {
 347             .name = "filename",
 348             .type = QEMU_OPT_STRING,
 349         },
 350         { /* end of list */ }
 351     },
 352 };
 353
 354 static int qemu_rbd_create(const char *filename, QemuOpts *opts, Error **errp)
 355 {
 356     Error *local_err = NULL;
 357     int64_t bytes = 0;
 358     int64_t objsize;
 359     int obj_order = 0;
 360     const char *pool, *image_name, *conf, *user, *keypairs;
 361     const char *secretid;
 362     rados_t cluster;
 363     rados_ioctx_t io_ctx;
 364     QDict *options = NULL;
 365     int ret = 0;
 366
 367     secretid = qemu_opt_get(opts, "password-secret");
 368
 369     /* Read out options */
 370     bytes = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
 371                      BDRV_SECTOR_SIZE);
 372     objsize = qemu_opt_get_size_del(opts, BLOCK_OPT_CLUSTER_SIZE, 0);
 373     if (objsize) {
 374         if ((objsize - 1) & objsize) {    /* not a power of 2? */
 375             error_setg(errp, "obj size needs to be power of 2");
 376             ret = -EINVAL;
 377             goto exit;
 378         }
 379         if (objsize < 4096) {
 380             error_setg(errp, "obj size too small");
 381             ret = -EINVAL;
 382             goto exit;
 383         }
 384         obj_order = ctz32(objsize);
 385     }
 386
 387     options = qdict_new();
 388     qemu_rbd_parse_filename(filename, options, &local_err);
 389     if (local_err) {
 390         ret = -EINVAL;
 391         error_propagate(errp, local_err);
 392         goto exit;
 393     }
 394
 395     /*
 396      * Caution: while qdict_get_try_str() is fine, getting non-string
 397      * types would require more care.  When @options come from -blockdev
 398      * or blockdev_add, its members are typed according to the QAPI
 399      * schema, but when they come from -drive, they're all QString.
 400      */
 401     pool       = qdict_get_try_str(options, "pool");
 402     conf       = qdict_get_try_str(options, "conf");
 403     user       = qdict_get_try_str(options, "user");
 404     image_name = qdict_get_try_str(options, "image");
 405     keypairs   = qdict_get_try_str(options, "=keyvalue-pairs");
 406
 407     ret = rados_create(&cluster, user);
 408     if (ret < 0) {
 409         error_setg_errno(errp, -ret, "error initializing");
 410         goto exit;
 411     }
 412
 413     /* try default location when conf=NULL, but ignore failure */
 414     ret = rados_conf_read_file(cluster, conf);
 415     if (conf && ret < 0) {
 416         error_setg_errno(errp, -ret, "error reading conf file %s", conf);
 417         ret = -EIO;
 418         goto shutdown;
 419     }
 420
 421     ret = qemu_rbd_set_keypairs(cluster, keypairs, errp);
 422     if (ret < 0) {
 423         ret = -EIO;
 424         goto shutdown;
 425     }
 426
 427     if (qemu_rbd_set_auth(cluster, secretid, errp) < 0) {
 428         ret = -EIO;
 429         goto shutdown;
 430     }
 431
 432     ret = rados_connect(cluster);
 433     if (ret < 0) {
 434         error_setg_errno(errp, -ret, "error connecting");
 435         goto shutdown;
 436     }
 437
 438     ret = rados_ioctx_create(cluster, pool, &io_ctx);
 439     if (ret < 0) {
 440         error_setg_errno(errp, -ret, "error opening pool %s", pool);
 441         goto shutdown;
 442     }
 443
 444     ret = rbd_create(io_ctx, image_name, bytes, &obj_order);
 445     if (ret < 0) {
 446         error_setg_errno(errp, -ret, "error rbd create");
 447     }
 448
 449     rados_ioctx_destroy(io_ctx);
 450
 451 shutdown:
 452     rados_shutdown(cluster);
 453
 454 exit:
 455     QDECREF(options);
 456     return ret;
 457 }
 458
 459 /*
 460  * This aio completion is being called from rbd_finish_bh() and runs in qemu
 461  * BH context.
 462  */
 463 static void qemu_rbd_complete_aio(RADOSCB *rcb)
 464 {
 465     RBDAIOCB *acb = rcb->acb;
 466     int64_t r;
 467
 468     r = rcb->ret;
 469
 470     if (acb->cmd != RBD_AIO_READ) {
 471         if (r < 0) {
 472             acb->ret = r;
 473             acb->error = 1;
 474         } else if (!acb->error) {
 475             acb->ret = rcb->size;
 476         }
 477     } else {
 478         if (r < 0) {
 479             qemu_rbd_memset(rcb, 0);
 480             acb->ret = r;
 481             acb->error = 1;
 482         } else if (r < rcb->size) {
 483             qemu_rbd_memset(rcb, r);
 484             if (!acb->error) {
 485                 acb->ret = rcb->size;
 486             }
 487         } else if (!acb->error) {
 488             acb->ret = r;
 489         }
 490     }
 491
 492     g_free(rcb);
 493
 494     if (!LIBRBD_USE_IOVEC) {
 495         if (acb->cmd == RBD_AIO_READ) {
 496             qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
 497         }
 498         qemu_vfree(acb->bounce);
 499     }
 500
 501     acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
 502
 503     qemu_aio_unref(acb);
 504 }
 505
 506 static char *qemu_rbd_mon_host(QDict *options, Error **errp)
 507 {
 508     const char **vals = g_new(const char *, qdict_size(options) + 1);
 509     char keybuf[32];
 510     const char *host, *port;
 511     char *rados_str;
 512     int i;
 513
 514     for (i = 0;; i++) {
 515         sprintf(keybuf, "server.%d.host", i);
 516         host = qdict_get_try_str(options, keybuf);
 517         qdict_del(options, keybuf);
 518         sprintf(keybuf, "server.%d.port", i);
 519         port = qdict_get_try_str(options, keybuf);
 520         qdict_del(options, keybuf);
 521         if (!host && !port) {
 522             break;
 523         }
 524         if (!host) {
 525             error_setg(errp, "Parameter server.%d.host is missing", i);
 526             rados_str = NULL;
 527             goto out;
 528         }
 529
 530         if (strchr(host, ':')) {
 531             vals[i] = port ? g_strdup_printf("[%s]:%s", host, port)
 532                 : g_strdup_printf("[%s]", host);
 533         } else {
 534             vals[i] = port ? g_strdup_printf("%s:%s", host, port)
 535                 : g_strdup(host);
 536         }
 537     }
 538     vals[i] = NULL;
 539
 540     rados_str = i ? g_strjoinv(";", (char **)vals) : NULL;
 541 out:
 542     g_strfreev((char **)vals);
 543     return rados_str;
 544 }
 545
 546 static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags,
 547                          Error **errp)
 548 {
 549     BDRVRBDState *s = bs->opaque;
 550     const char *pool, *snap, *conf, *user, *image_name, *keypairs;
 551     const char *secretid, *filename;
 552     QemuOpts *opts;
 553     Error *local_err = NULL;
 554     char *mon_host = NULL;
 555     int r;
 556
 557     /* If we are given a filename, parse the filename, with precedence given to
 558      * filename encoded options */
 559     filename = qdict_get_try_str(options, "filename");
 560     if (filename) {
 561         warn_report("'filename' option specified. "
 562                     "This is an unsupported option, and may be deprecated "
 563                     "in the future");
 564         qemu_rbd_parse_filename(filename, options, &local_err);
 565         if (local_err) {
 566             r = -EINVAL;
 567             error_propagate(errp, local_err);
 568             goto exit;
 569         }
 570     }
 571
 572     opts = qemu_opts_create(&runtime_opts, NULL, 0, &error_abort);
 573     qemu_opts_absorb_qdict(opts, options, &local_err);
 574     if (local_err) {
 575         error_propagate(errp, local_err);
 576         r = -EINVAL;
 577         goto failed_opts;
 578     }
 579
 580     mon_host = qemu_rbd_mon_host(options, &local_err);
 581     if (local_err) {
 582         error_propagate(errp, local_err);
 583         r = -EINVAL;
 584         goto failed_opts;
 585     }
 586
 587     secretid = qemu_opt_get(opts, "password-secret");
 588
 589     pool           = qemu_opt_get(opts, "pool");
 590     conf           = qemu_opt_get(opts, "conf");
 591     snap           = qemu_opt_get(opts, "snapshot");
 592     user           = qemu_opt_get(opts, "user");
 593     image_name     = qemu_opt_get(opts, "image");
 594     keypairs       = qemu_opt_get(opts, "=keyvalue-pairs");
 595
 596     if (!pool || !image_name) {
 597         error_setg(errp, "Parameters 'pool' and 'image' are required");
 598         r = -EINVAL;
 599         goto failed_opts;
 600     }
 601
 602     r = rados_create(&s->cluster, user);
 603     if (r < 0) {
 604         error_setg_errno(errp, -r, "error initializing");
 605         goto failed_opts;
 606     }
 607
 608     s->snap = g_strdup(snap);
 609     s->image_name = g_strdup(image_name);
 610
 611     /* try default location when conf=NULL, but ignore failure */
 612     r = rados_conf_read_file(s->cluster, conf);
 613     if (conf && r < 0) {
 614         error_setg_errno(errp, -r, "error reading conf file %s", conf);
 615         goto failed_shutdown;
 616     }
 617
 618     r = qemu_rbd_set_keypairs(s->cluster, keypairs, errp);
 619     if (r < 0) {
 620         goto failed_shutdown;
 621     }
 622
 623     if (mon_host) {
 624         r = rados_conf_set(s->cluster, "mon_host", mon_host);
 625         if (r < 0) {
 626             goto failed_shutdown;
 627         }
 628     }
 629
 630     if (qemu_rbd_set_auth(s->cluster, secretid, errp) < 0) {
 631         r = -EIO;
 632         goto failed_shutdown;
 633     }
 634
 635     /*
 636      * Fallback to more conservative semantics if setting cache
 637      * options fails. Ignore errors from setting rbd_cache because the
 638      * only possible error is that the option does not exist, and
 639      * librbd defaults to no caching. If write through caching cannot
 640      * be set up, fall back to no caching.
 641      */
 642     if (flags & BDRV_O_NOCACHE) {
 643         rados_conf_set(s->cluster, "rbd_cache", "false");
 644     } else {
 645         rados_conf_set(s->cluster, "rbd_cache", "true");
 646     }
 647
 648     r = rados_connect(s->cluster);
 649     if (r < 0) {
 650         error_setg_errno(errp, -r, "error connecting");
 651         goto failed_shutdown;
 652     }
 653
 654     r = rados_ioctx_create(s->cluster, pool, &s->io_ctx);
 655     if (r < 0) {
 656         error_setg_errno(errp, -r, "error opening pool %s", pool);
 657         goto failed_shutdown;
 658     }
 659
 660     /* rbd_open is always r/w */
 661     r = rbd_open(s->io_ctx, s->image_name, &s->image, s->snap);
 662     if (r < 0) {
 663         error_setg_errno(errp, -r, "error reading header from %s",
 664                          s->image_name);
 665         goto failed_open;
 666     }
 667
 668     /* If we are using an rbd snapshot, we must be r/o, otherwise
 669      * leave as-is */
 670     if (s->snap != NULL) {
 671         if (!bdrv_is_read_only(bs)) {
 672             error_report("Opening rbd snapshots without an explicit "
 673                          "read-only=on option is deprecated. Future versions "
 674                          "will refuse to open the image instead of "
 675                          "automatically marking the image read-only.");
 676             r = bdrv_set_read_only(bs, true, &local_err);
 677             if (r < 0) {
 678                 error_propagate(errp, local_err);
 679                 goto failed_open;
 680             }
 681         }
 682     }
 683
 684     qemu_opts_del(opts);
 685     return 0;
 686
 687 failed_open:
 688     rados_ioctx_destroy(s->io_ctx);
 689 failed_shutdown:
 690     rados_shutdown(s->cluster);
 691     g_free(s->snap);
 692     g_free(s->image_name);
 693 failed_opts:
 694     qemu_opts_del(opts);
 695     g_free(mon_host);
 696 exit:
 697     return r;
 698 }
 699
 700
 701 /* Since RBD is currently always opened R/W via the API,
 702  * we just need to check if we are using a snapshot or not, in
 703  * order to determine if we will allow it to be R/W */
 704 static int qemu_rbd_reopen_prepare(BDRVReopenState *state,
 705                                    BlockReopenQueue *queue, Error **errp)
 706 {
 707     BDRVRBDState *s = state->bs->opaque;
 708     int ret = 0;
 709
 710     if (s->snap && state->flags & BDRV_O_RDWR) {
 711         error_setg(errp,
 712                    "Cannot change node '%s' to r/w when using RBD snapshot",
 713                    bdrv_get_device_or_node_name(state->bs));
 714         ret = -EINVAL;
 715     }
 716
 717     return ret;
 718 }
 719
 720 static void qemu_rbd_close(BlockDriverState *bs)
 721 {
 722     BDRVRBDState *s = bs->opaque;
 723
 724     rbd_close(s->image);
 725     rados_ioctx_destroy(s->io_ctx);
 726     g_free(s->snap);
 727     g_free(s->image_name);
 728     rados_shutdown(s->cluster);
 729 }
 730
 731 static const AIOCBInfo rbd_aiocb_info = {
 732     .aiocb_size = sizeof(RBDAIOCB),
 733 };
 734
 735 static void rbd_finish_bh(void *opaque)
 736 {
 737     RADOSCB *rcb = opaque;
 738     qemu_rbd_complete_aio(rcb);
 739 }
 740
 741 /*
 742  * This is the callback function for rbd_aio_read and _write
 743  *
 744  * Note: this function is being called from a non qemu thread so
 745  * we need to be careful about what we do here. Generally we only
 746  * schedule a BH, and do the rest of the io completion handling
 747  * from rbd_finish_bh() which runs in a qemu context.
 748  */
 749 static void rbd_finish_aiocb(rbd_completion_t c, RADOSCB *rcb)
 750 {
 751     RBDAIOCB *acb = rcb->acb;
 752
 753     rcb->ret = rbd_aio_get_return_value(c);
 754     rbd_aio_release(c);
 755
 756     aio_bh_schedule_oneshot(bdrv_get_aio_context(acb->common.bs),
 757                             rbd_finish_bh, rcb);
 758 }
 759
 760 static int rbd_aio_discard_wrapper(rbd_image_t image,
 761                                    uint64_t off,
 762                                    uint64_t len,
 763                                    rbd_completion_t comp)
 764 {
 765 #ifdef LIBRBD_SUPPORTS_DISCARD
 766     return rbd_aio_discard(image, off, len, comp);
 767 #else
 768     return -ENOTSUP;
 769 #endif
 770 }
 771
 772 static int rbd_aio_flush_wrapper(rbd_image_t image,
 773                                  rbd_completion_t comp)
 774 {
 775 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
 776     return rbd_aio_flush(image, comp);
 777 #else
 778     return -ENOTSUP;
 779 #endif
 780 }
 781
 782 static BlockAIOCB *rbd_start_aio(BlockDriverState *bs,
 783                                  int64_t off,
 784                                  QEMUIOVector *qiov,
 785                                  int64_t size,
 786                                  BlockCompletionFunc *cb,
 787                                  void *opaque,
 788                                  RBDAIOCmd cmd)
 789 {
 790     RBDAIOCB *acb;
 791     RADOSCB *rcb = NULL;
 792     rbd_completion_t c;
 793     int r;
 794
 795     BDRVRBDState *s = bs->opaque;
 796
 797     acb = qemu_aio_get(&rbd_aiocb_info, bs, cb, opaque);
 798     acb->cmd = cmd;
 799     acb->qiov = qiov;
 800     assert(!qiov || qiov->size == size);
 801
 802     rcb = g_new(RADOSCB, 1);
 803
 804     if (!LIBRBD_USE_IOVEC) {
 805         if (cmd == RBD_AIO_DISCARD || cmd == RBD_AIO_FLUSH) {
 806             acb->bounce = NULL;
 807         } else {
 808             acb->bounce = qemu_try_blockalign(bs, qiov->size);
 809             if (acb->bounce == NULL) {
 810                 goto failed;
 811             }
 812         }
 813         if (cmd == RBD_AIO_WRITE) {
 814             qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
 815         }
 816         rcb->buf = acb->bounce;
 817     }
 818
 819     acb->ret = 0;
 820     acb->error = 0;
 821     acb->s = s;
 822
 823     rcb->acb = acb;
 824     rcb->s = acb->s;
 825     rcb->size = size;
 826     r = rbd_aio_create_completion(rcb, (rbd_callback_t) rbd_finish_aiocb, &c);
 827     if (r < 0) {
 828         goto failed;
 829     }
 830
 831     switch (cmd) {
 832     case RBD_AIO_WRITE:
 833 #ifdef LIBRBD_SUPPORTS_IOVEC
 834             r = rbd_aio_writev(s->image, qiov->iov, qiov->niov, off, c);
 835 #else
 836             r = rbd_aio_write(s->image, off, size, rcb->buf, c);
 837 #endif
 838         break;
 839     case RBD_AIO_READ:
 840 #ifdef LIBRBD_SUPPORTS_IOVEC
 841             r = rbd_aio_readv(s->image, qiov->iov, qiov->niov, off, c);
 842 #else
 843             r = rbd_aio_read(s->image, off, size, rcb->buf, c);
 844 #endif
 845         break;
 846     case RBD_AIO_DISCARD:
 847         r = rbd_aio_discard_wrapper(s->image, off, size, c);
 848         break;
 849     case RBD_AIO_FLUSH:
 850         r = rbd_aio_flush_wrapper(s->image, c);
 851         break;
 852     default:
 853         r = -EINVAL;
 854     }
 855
 856     if (r < 0) {
 857         goto failed_completion;
 858     }
 859     return &acb->common;
 860
 861 failed_completion:
 862     rbd_aio_release(c);
 863 failed:
 864     g_free(rcb);
 865     if (!LIBRBD_USE_IOVEC) {
 866         qemu_vfree(acb->bounce);
 867     }
 868
 869     qemu_aio_unref(acb);
 870     return NULL;
 871 }
 872
 873 static BlockAIOCB *qemu_rbd_aio_readv(BlockDriverState *bs,
 874                                       int64_t sector_num,
 875                                       QEMUIOVector *qiov,
 876                                       int nb_sectors,
 877                                       BlockCompletionFunc *cb,
 878                                       void *opaque)
 879 {
 880     return rbd_start_aio(bs, sector_num << BDRV_SECTOR_BITS, qiov,
 881                          (int64_t) nb_sectors << BDRV_SECTOR_BITS, cb, opaque,
 882                          RBD_AIO_READ);
 883 }
 884
 885 static BlockAIOCB *qemu_rbd_aio_writev(BlockDriverState *bs,
 886                                        int64_t sector_num,
 887                                        QEMUIOVector *qiov,
 888                                        int nb_sectors,
 889                                        BlockCompletionFunc *cb,
 890                                        void *opaque)
 891 {
 892     return rbd_start_aio(bs, sector_num << BDRV_SECTOR_BITS, qiov,
 893                          (int64_t) nb_sectors << BDRV_SECTOR_BITS, cb, opaque,
 894                          RBD_AIO_WRITE);
 895 }
 896
 897 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
 898 static BlockAIOCB *qemu_rbd_aio_flush(BlockDriverState *bs,
 899                                       BlockCompletionFunc *cb,
 900                                       void *opaque)
 901 {
 902     return rbd_start_aio(bs, 0, NULL, 0, cb, opaque, RBD_AIO_FLUSH);
 903 }
 904
 905 #else
 906
 907 static int qemu_rbd_co_flush(BlockDriverState *bs)
 908 {
 909 #if LIBRBD_VERSION_CODE >= LIBRBD_VERSION(0, 1, 1)
 910     /* rbd_flush added in 0.1.1 */
 911     BDRVRBDState *s = bs->opaque;
 912     return rbd_flush(s->image);
 913 #else
 914     return 0;
 915 #endif
 916 }
 917 #endif
 918
 919 static int qemu_rbd_getinfo(BlockDriverState *bs, BlockDriverInfo *bdi)
 920 {
 921     BDRVRBDState *s = bs->opaque;
 922     rbd_image_info_t info;
 923     int r;
 924
 925     r = rbd_stat(s->image, &info, sizeof(info));
 926     if (r < 0) {
 927         return r;
 928     }
 929
 930     bdi->cluster_size = info.obj_size;
 931     return 0;
 932 }
 933
 934 static int64_t qemu_rbd_getlength(BlockDriverState *bs)
 935 {
 936     BDRVRBDState *s = bs->opaque;
 937     rbd_image_info_t info;
 938     int r;
 939
 940     r = rbd_stat(s->image, &info, sizeof(info));
 941     if (r < 0) {
 942         return r;
 943     }
 944
 945     return info.size;
 946 }
 947
 948 static int qemu_rbd_truncate(BlockDriverState *bs, int64_t offset,
 949                              PreallocMode prealloc, Error **errp)
 950 {
 951     BDRVRBDState *s = bs->opaque;
 952     int r;
 953
 954     if (prealloc != PREALLOC_MODE_OFF) {
 955         error_setg(errp, "Unsupported preallocation mode '%s'",
 956                    PreallocMode_str(prealloc));
 957         return -ENOTSUP;
 958     }
 959
 960     r = rbd_resize(s->image, offset);
 961     if (r < 0) {
 962         error_setg_errno(errp, -r, "Failed to resize file");
 963         return r;
 964     }
 965
 966     return 0;
 967 }
 968
 969 static int qemu_rbd_snap_create(BlockDriverState *bs,
 970                                 QEMUSnapshotInfo *sn_info)
 971 {
 972     BDRVRBDState *s = bs->opaque;
 973     int r;
 974
 975     if (sn_info->name[0] == '\0') {
 976         return -EINVAL; /* we need a name for rbd snapshots */
 977     }
 978
 979     /*
 980      * rbd snapshots are using the name as the user controlled unique identifier
 981      * we can't use the rbd snapid for that purpose, as it can't be set
 982      */
 983     if (sn_info->id_str[0] != '\0' &&
 984         strcmp(sn_info->id_str, sn_info->name) != 0) {
 985         return -EINVAL;
 986     }
 987
 988     if (strlen(sn_info->name) >= sizeof(sn_info->id_str)) {
 989         return -ERANGE;
 990     }
 991
 992     r = rbd_snap_create(s->image, sn_info->name);
 993     if (r < 0) {
 994         error_report("failed to create snap: %s", strerror(-r));
 995         return r;
 996     }
 997
 998     return 0;
 999 }
1000
1001 static int qemu_rbd_snap_remove(BlockDriverState *bs,
1002                                 const char *snapshot_id,
1003                                 const char *snapshot_name,
1004                                 Error **errp)
1005 {
1006     BDRVRBDState *s = bs->opaque;
1007     int r;
1008
1009     if (!snapshot_name) {
1010         error_setg(errp, "rbd need a valid snapshot name");
1011         return -EINVAL;
1012     }
1013
1014     /* If snapshot_id is specified, it must be equal to name, see
1015        qemu_rbd_snap_list() */
1016     if (snapshot_id && strcmp(snapshot_id, snapshot_name)) {
1017         error_setg(errp,
1018                    "rbd do not support snapshot id, it should be NULL or "
1019                    "equal to snapshot name");
1020         return -EINVAL;
1021     }
1022
1023     r = rbd_snap_remove(s->image, snapshot_name);
1024     if (r < 0) {
1025         error_setg_errno(errp, -r, "Failed to remove the snapshot");
1026     }
1027     return r;
1028 }
1029
1030 static int qemu_rbd_snap_rollback(BlockDriverState *bs,
1031                                   const char *snapshot_name)
1032 {
1033     BDRVRBDState *s = bs->opaque;
1034
1035     return rbd_snap_rollback(s->image, snapshot_name);
1036 }
1037
1038 static int qemu_rbd_snap_list(BlockDriverState *bs,
1039                               QEMUSnapshotInfo **psn_tab)
1040 {
1041     BDRVRBDState *s = bs->opaque;
1042     QEMUSnapshotInfo *sn_info, *sn_tab = NULL;
1043     int i, snap_count;
1044     rbd_snap_info_t *snaps;
1045     int max_snaps = RBD_MAX_SNAPS;
1046
1047     do {
1048         snaps = g_new(rbd_snap_info_t, max_snaps);
1049         snap_count = rbd_snap_list(s->image, snaps, &max_snaps);
1050         if (snap_count <= 0) {
1051             g_free(snaps);
1052         }
1053     } while (snap_count == -ERANGE);
1054
1055     if (snap_count <= 0) {
1056         goto done;
1057     }
1058
1059     sn_tab = g_new0(QEMUSnapshotInfo, snap_count);
1060
1061     for (i = 0; i < snap_count; i++) {
1062         const char *snap_name = snaps[i].name;
1063
1064         sn_info = sn_tab + i;
1065         pstrcpy(sn_info->id_str, sizeof(sn_info->id_str), snap_name);
1066         pstrcpy(sn_info->name, sizeof(sn_info->name), snap_name);
1067
1068         sn_info->vm_state_size = snaps[i].size;
1069         sn_info->date_sec = 0;
1070         sn_info->date_nsec = 0;
1071         sn_info->vm_clock_nsec = 0;
1072     }
1073     rbd_snap_list_end(snaps);
1074     g_free(snaps);
1075
1076  done:
1077     *psn_tab = sn_tab;
1078     return snap_count;
1079 }
1080
1081 #ifdef LIBRBD_SUPPORTS_DISCARD
1082 static BlockAIOCB *qemu_rbd_aio_pdiscard(BlockDriverState *bs,
1083                                          int64_t offset,
1084                                          int bytes,
1085                                          BlockCompletionFunc *cb,
1086                                          void *opaque)
1087 {
1088     return rbd_start_aio(bs, offset, NULL, bytes, cb, opaque,
1089                          RBD_AIO_DISCARD);
1090 }
1091 #endif
1092
1093 #ifdef LIBRBD_SUPPORTS_INVALIDATE
1094 static void qemu_rbd_invalidate_cache(BlockDriverState *bs,
1095                                       Error **errp)
1096 {
1097     BDRVRBDState *s = bs->opaque;
1098     int r = rbd_invalidate_cache(s->image);
1099     if (r < 0) {
1100         error_setg_errno(errp, -r, "Failed to invalidate the cache");
1101     }
1102 }
1103 #endif
1104
1105 static QemuOptsList qemu_rbd_create_opts = {
1106     .name = "rbd-create-opts",
1107     .head = QTAILQ_HEAD_INITIALIZER(qemu_rbd_create_opts.head),
1108     .desc = {
1109         {
1110             .name = BLOCK_OPT_SIZE,
1111             .type = QEMU_OPT_SIZE,
1112             .help = "Virtual disk size"
1113         },
1114         {
1115             .name = BLOCK_OPT_CLUSTER_SIZE,
1116             .type = QEMU_OPT_SIZE,
1117             .help = "RBD object size"
1118         },
1119         {
1120             .name = "password-secret",
1121             .type = QEMU_OPT_STRING,
1122             .help = "ID of secret providing the password",
1123         },
1124         { /* end of list */ }
1125     }
1126 };
1127
1128 static BlockDriver bdrv_rbd = {
1129     .format_name            = "rbd",
1130     .instance_size          = sizeof(BDRVRBDState),
1131     .bdrv_parse_filename    = qemu_rbd_parse_filename,
1132     .bdrv_file_open         = qemu_rbd_open,
1133     .bdrv_close             = qemu_rbd_close,
1134     .bdrv_reopen_prepare    = qemu_rbd_reopen_prepare,
1135     .bdrv_create            = qemu_rbd_create,
1136     .bdrv_has_zero_init     = bdrv_has_zero_init_1,
1137     .bdrv_get_info          = qemu_rbd_getinfo,
1138     .create_opts            = &qemu_rbd_create_opts,
1139     .bdrv_getlength         = qemu_rbd_getlength,
1140     .bdrv_truncate          = qemu_rbd_truncate,
1141     .protocol_name          = "rbd",
1142
1143     .bdrv_aio_readv         = qemu_rbd_aio_readv,
1144     .bdrv_aio_writev        = qemu_rbd_aio_writev,
1145
1146 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
1147     .bdrv_aio_flush         = qemu_rbd_aio_flush,
1148 #else
1149     .bdrv_co_flush_to_disk  = qemu_rbd_co_flush,
1150 #endif
1151
1152 #ifdef LIBRBD_SUPPORTS_DISCARD
1153     .bdrv_aio_pdiscard      = qemu_rbd_aio_pdiscard,
1154 #endif
1155
1156     .bdrv_snapshot_create   = qemu_rbd_snap_create,
1157     .bdrv_snapshot_delete   = qemu_rbd_snap_remove,
1158     .bdrv_snapshot_list     = qemu_rbd_snap_list,
1159     .bdrv_snapshot_goto     = qemu_rbd_snap_rollback,
1160 #ifdef LIBRBD_SUPPORTS_INVALIDATE
1161     .bdrv_invalidate_cache  = qemu_rbd_invalidate_cache,
1162 #endif
1163 };
1164
1165 static void bdrv_rbd_init(void)
1166 {
1167     bdrv_register(&bdrv_rbd);
1168 }
1169
1170 block_init(bdrv_rbd_init);