xref: /openbmc/qemu/block/rbd.c (revision b45c03f5)
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 <inttypes.h>
15 
16 #include "qemu-common.h"
17 #include "qemu/error-report.h"
18 #include "block/block_int.h"
19 
20 #include <rbd/librbd.h>
21 
22 /*
23  * When specifying the image filename use:
24  *
25  * rbd:poolname/devicename[@snapshotname][:option1=value1[:option2=value2...]]
26  *
27  * poolname must be the name of an existing rados pool.
28  *
29  * devicename is the name of the rbd image.
30  *
31  * Each option given is used to configure rados, and may be any valid
32  * Ceph option, "id", or "conf".
33  *
34  * The "id" option indicates what user we should authenticate as to
35  * the Ceph cluster.  If it is excluded we will use the Ceph default
36  * (normally 'admin').
37  *
38  * The "conf" option specifies a Ceph configuration file to read.  If
39  * it is not specified, we will read from the default Ceph locations
40  * (e.g., /etc/ceph/ceph.conf).  To avoid reading _any_ configuration
41  * file, specify conf=/dev/null.
42  *
43  * Configuration values containing :, @, or = can be escaped with a
44  * leading "\".
45  */
46 
47 /* rbd_aio_discard added in 0.1.2 */
48 #if LIBRBD_VERSION_CODE >= LIBRBD_VERSION(0, 1, 2)
49 #define LIBRBD_SUPPORTS_DISCARD
50 #else
51 #undef LIBRBD_SUPPORTS_DISCARD
52 #endif
53 
54 #define OBJ_MAX_SIZE (1UL << OBJ_DEFAULT_OBJ_ORDER)
55 
56 #define RBD_MAX_CONF_NAME_SIZE 128
57 #define RBD_MAX_CONF_VAL_SIZE 512
58 #define RBD_MAX_CONF_SIZE 1024
59 #define RBD_MAX_POOL_NAME_SIZE 128
60 #define RBD_MAX_SNAP_NAME_SIZE 128
61 #define RBD_MAX_SNAPS 100
62 
63 typedef enum {
64     RBD_AIO_READ,
65     RBD_AIO_WRITE,
66     RBD_AIO_DISCARD,
67     RBD_AIO_FLUSH
68 } RBDAIOCmd;
69 
70 typedef struct RBDAIOCB {
71     BlockAIOCB common;
72     QEMUBH *bh;
73     int64_t ret;
74     QEMUIOVector *qiov;
75     char *bounce;
76     RBDAIOCmd cmd;
77     int error;
78     struct BDRVRBDState *s;
79 } RBDAIOCB;
80 
81 typedef struct RADOSCB {
82     RBDAIOCB *acb;
83     struct BDRVRBDState *s;
84     int64_t size;
85     char *buf;
86     int64_t ret;
87 } RADOSCB;
88 
89 typedef struct BDRVRBDState {
90     rados_t cluster;
91     rados_ioctx_t io_ctx;
92     rbd_image_t image;
93     char name[RBD_MAX_IMAGE_NAME_SIZE];
94     char *snap;
95 } BDRVRBDState;
96 
97 static int qemu_rbd_next_tok(char *dst, int dst_len,
98                              char *src, char delim,
99                              const char *name,
100                              char **p, Error **errp)
101 {
102     int l;
103     char *end;
104 
105     *p = NULL;
106 
107     if (delim != '\0') {
108         for (end = src; *end; ++end) {
109             if (*end == delim) {
110                 break;
111             }
112             if (*end == '\\' && end[1] != '\0') {
113                 end++;
114             }
115         }
116         if (*end == delim) {
117             *p = end + 1;
118             *end = '\0';
119         }
120     }
121     l = strlen(src);
122     if (l >= dst_len) {
123         error_setg(errp, "%s too long", name);
124         return -EINVAL;
125     } else if (l == 0) {
126         error_setg(errp, "%s too short", name);
127         return -EINVAL;
128     }
129 
130     pstrcpy(dst, dst_len, src);
131 
132     return 0;
133 }
134 
135 static void qemu_rbd_unescape(char *src)
136 {
137     char *p;
138 
139     for (p = src; *src; ++src, ++p) {
140         if (*src == '\\' && src[1] != '\0') {
141             src++;
142         }
143         *p = *src;
144     }
145     *p = '\0';
146 }
147 
148 static int qemu_rbd_parsename(const char *filename,
149                               char *pool, int pool_len,
150                               char *snap, int snap_len,
151                               char *name, int name_len,
152                               char *conf, int conf_len,
153                               Error **errp)
154 {
155     const char *start;
156     char *p, *buf;
157     int ret;
158 
159     if (!strstart(filename, "rbd:", &start)) {
160         error_setg(errp, "File name must start with 'rbd:'");
161         return -EINVAL;
162     }
163 
164     buf = g_strdup(start);
165     p = buf;
166     *snap = '\0';
167     *conf = '\0';
168 
169     ret = qemu_rbd_next_tok(pool, pool_len, p,
170                             '/', "pool name", &p, errp);
171     if (ret < 0 || !p) {
172         ret = -EINVAL;
173         goto done;
174     }
175     qemu_rbd_unescape(pool);
176 
177     if (strchr(p, '@')) {
178         ret = qemu_rbd_next_tok(name, name_len, p,
179                                 '@', "object name", &p, errp);
180         if (ret < 0) {
181             goto done;
182         }
183         ret = qemu_rbd_next_tok(snap, snap_len, p,
184                                 ':', "snap name", &p, errp);
185         qemu_rbd_unescape(snap);
186     } else {
187         ret = qemu_rbd_next_tok(name, name_len, p,
188                                 ':', "object name", &p, errp);
189     }
190     qemu_rbd_unescape(name);
191     if (ret < 0 || !p) {
192         goto done;
193     }
194 
195     ret = qemu_rbd_next_tok(conf, conf_len, p,
196                             '\0', "configuration", &p, errp);
197 
198 done:
199     g_free(buf);
200     return ret;
201 }
202 
203 static char *qemu_rbd_parse_clientname(const char *conf, char *clientname)
204 {
205     const char *p = conf;
206 
207     while (*p) {
208         int len;
209         const char *end = strchr(p, ':');
210 
211         if (end) {
212             len = end - p;
213         } else {
214             len = strlen(p);
215         }
216 
217         if (strncmp(p, "id=", 3) == 0) {
218             len -= 3;
219             strncpy(clientname, p + 3, len);
220             clientname[len] = '\0';
221             return clientname;
222         }
223         if (end == NULL) {
224             break;
225         }
226         p = end + 1;
227     }
228     return NULL;
229 }
230 
231 static int qemu_rbd_set_conf(rados_t cluster, const char *conf,
232                              bool only_read_conf_file,
233                              Error **errp)
234 {
235     char *p, *buf;
236     char name[RBD_MAX_CONF_NAME_SIZE];
237     char value[RBD_MAX_CONF_VAL_SIZE];
238     int ret = 0;
239 
240     buf = g_strdup(conf);
241     p = buf;
242 
243     while (p) {
244         ret = qemu_rbd_next_tok(name, sizeof(name), p,
245                                 '=', "conf option name", &p, errp);
246         if (ret < 0) {
247             break;
248         }
249         qemu_rbd_unescape(name);
250 
251         if (!p) {
252             error_setg(errp, "conf option %s has no value", name);
253             ret = -EINVAL;
254             break;
255         }
256 
257         ret = qemu_rbd_next_tok(value, sizeof(value), p,
258                                 ':', "conf option value", &p, errp);
259         if (ret < 0) {
260             break;
261         }
262         qemu_rbd_unescape(value);
263 
264         if (strcmp(name, "conf") == 0) {
265             /* read the conf file alone, so it doesn't override more
266                specific settings for a particular device */
267             if (only_read_conf_file) {
268                 ret = rados_conf_read_file(cluster, value);
269                 if (ret < 0) {
270                     error_setg(errp, "error reading conf file %s", value);
271                     break;
272                 }
273             }
274         } else if (strcmp(name, "id") == 0) {
275             /* ignore, this is parsed by qemu_rbd_parse_clientname() */
276         } else if (!only_read_conf_file) {
277             ret = rados_conf_set(cluster, name, value);
278             if (ret < 0) {
279                 error_setg(errp, "invalid conf option %s", name);
280                 ret = -EINVAL;
281                 break;
282             }
283         }
284     }
285 
286     g_free(buf);
287     return ret;
288 }
289 
290 static int qemu_rbd_create(const char *filename, QemuOpts *opts, Error **errp)
291 {
292     Error *local_err = NULL;
293     int64_t bytes = 0;
294     int64_t objsize;
295     int obj_order = 0;
296     char pool[RBD_MAX_POOL_NAME_SIZE];
297     char name[RBD_MAX_IMAGE_NAME_SIZE];
298     char snap_buf[RBD_MAX_SNAP_NAME_SIZE];
299     char conf[RBD_MAX_CONF_SIZE];
300     char clientname_buf[RBD_MAX_CONF_SIZE];
301     char *clientname;
302     rados_t cluster;
303     rados_ioctx_t io_ctx;
304     int ret;
305 
306     if (qemu_rbd_parsename(filename, pool, sizeof(pool),
307                            snap_buf, sizeof(snap_buf),
308                            name, sizeof(name),
309                            conf, sizeof(conf), &local_err) < 0) {
310         error_propagate(errp, local_err);
311         return -EINVAL;
312     }
313 
314     /* Read out options */
315     bytes = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
316                      BDRV_SECTOR_SIZE);
317     objsize = qemu_opt_get_size_del(opts, BLOCK_OPT_CLUSTER_SIZE, 0);
318     if (objsize) {
319         if ((objsize - 1) & objsize) {    /* not a power of 2? */
320             error_setg(errp, "obj size needs to be power of 2");
321             return -EINVAL;
322         }
323         if (objsize < 4096) {
324             error_setg(errp, "obj size too small");
325             return -EINVAL;
326         }
327         obj_order = ctz32(objsize);
328     }
329 
330     clientname = qemu_rbd_parse_clientname(conf, clientname_buf);
331     if (rados_create(&cluster, clientname) < 0) {
332         error_setg(errp, "error initializing");
333         return -EIO;
334     }
335 
336     if (strstr(conf, "conf=") == NULL) {
337         /* try default location, but ignore failure */
338         rados_conf_read_file(cluster, NULL);
339     } else if (conf[0] != '\0' &&
340                qemu_rbd_set_conf(cluster, conf, true, &local_err) < 0) {
341         rados_shutdown(cluster);
342         error_propagate(errp, local_err);
343         return -EIO;
344     }
345 
346     if (conf[0] != '\0' &&
347         qemu_rbd_set_conf(cluster, conf, false, &local_err) < 0) {
348         rados_shutdown(cluster);
349         error_propagate(errp, local_err);
350         return -EIO;
351     }
352 
353     if (rados_connect(cluster) < 0) {
354         error_setg(errp, "error connecting");
355         rados_shutdown(cluster);
356         return -EIO;
357     }
358 
359     if (rados_ioctx_create(cluster, pool, &io_ctx) < 0) {
360         error_setg(errp, "error opening pool %s", pool);
361         rados_shutdown(cluster);
362         return -EIO;
363     }
364 
365     ret = rbd_create(io_ctx, name, bytes, &obj_order);
366     rados_ioctx_destroy(io_ctx);
367     rados_shutdown(cluster);
368 
369     return ret;
370 }
371 
372 /*
373  * This aio completion is being called from rbd_finish_bh() and runs in qemu
374  * BH context.
375  */
376 static void qemu_rbd_complete_aio(RADOSCB *rcb)
377 {
378     RBDAIOCB *acb = rcb->acb;
379     int64_t r;
380 
381     r = rcb->ret;
382 
383     if (acb->cmd != RBD_AIO_READ) {
384         if (r < 0) {
385             acb->ret = r;
386             acb->error = 1;
387         } else if (!acb->error) {
388             acb->ret = rcb->size;
389         }
390     } else {
391         if (r < 0) {
392             memset(rcb->buf, 0, rcb->size);
393             acb->ret = r;
394             acb->error = 1;
395         } else if (r < rcb->size) {
396             memset(rcb->buf + r, 0, rcb->size - r);
397             if (!acb->error) {
398                 acb->ret = rcb->size;
399             }
400         } else if (!acb->error) {
401             acb->ret = r;
402         }
403     }
404 
405     g_free(rcb);
406 
407     if (acb->cmd == RBD_AIO_READ) {
408         qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
409     }
410     qemu_vfree(acb->bounce);
411     acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
412 
413     qemu_aio_unref(acb);
414 }
415 
416 /* TODO Convert to fine grained options */
417 static QemuOptsList runtime_opts = {
418     .name = "rbd",
419     .head = QTAILQ_HEAD_INITIALIZER(runtime_opts.head),
420     .desc = {
421         {
422             .name = "filename",
423             .type = QEMU_OPT_STRING,
424             .help = "Specification of the rbd image",
425         },
426         { /* end of list */ }
427     },
428 };
429 
430 static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags,
431                          Error **errp)
432 {
433     BDRVRBDState *s = bs->opaque;
434     char pool[RBD_MAX_POOL_NAME_SIZE];
435     char snap_buf[RBD_MAX_SNAP_NAME_SIZE];
436     char conf[RBD_MAX_CONF_SIZE];
437     char clientname_buf[RBD_MAX_CONF_SIZE];
438     char *clientname;
439     QemuOpts *opts;
440     Error *local_err = NULL;
441     const char *filename;
442     int r;
443 
444     opts = qemu_opts_create(&runtime_opts, NULL, 0, &error_abort);
445     qemu_opts_absorb_qdict(opts, options, &local_err);
446     if (local_err) {
447         error_propagate(errp, local_err);
448         qemu_opts_del(opts);
449         return -EINVAL;
450     }
451 
452     filename = qemu_opt_get(opts, "filename");
453 
454     if (qemu_rbd_parsename(filename, pool, sizeof(pool),
455                            snap_buf, sizeof(snap_buf),
456                            s->name, sizeof(s->name),
457                            conf, sizeof(conf), errp) < 0) {
458         r = -EINVAL;
459         goto failed_opts;
460     }
461 
462     clientname = qemu_rbd_parse_clientname(conf, clientname_buf);
463     r = rados_create(&s->cluster, clientname);
464     if (r < 0) {
465         error_setg(errp, "error initializing");
466         goto failed_opts;
467     }
468 
469     s->snap = NULL;
470     if (snap_buf[0] != '\0') {
471         s->snap = g_strdup(snap_buf);
472     }
473 
474     if (strstr(conf, "conf=") == NULL) {
475         /* try default location, but ignore failure */
476         rados_conf_read_file(s->cluster, NULL);
477     } else if (conf[0] != '\0') {
478         r = qemu_rbd_set_conf(s->cluster, conf, true, errp);
479         if (r < 0) {
480             goto failed_shutdown;
481         }
482     }
483 
484     if (conf[0] != '\0') {
485         r = qemu_rbd_set_conf(s->cluster, conf, false, errp);
486         if (r < 0) {
487             goto failed_shutdown;
488         }
489     }
490 
491     /*
492      * Fallback to more conservative semantics if setting cache
493      * options fails. Ignore errors from setting rbd_cache because the
494      * only possible error is that the option does not exist, and
495      * librbd defaults to no caching. If write through caching cannot
496      * be set up, fall back to no caching.
497      */
498     if (flags & BDRV_O_NOCACHE) {
499         rados_conf_set(s->cluster, "rbd_cache", "false");
500     } else {
501         rados_conf_set(s->cluster, "rbd_cache", "true");
502     }
503 
504     r = rados_connect(s->cluster);
505     if (r < 0) {
506         error_setg(errp, "error connecting");
507         goto failed_shutdown;
508     }
509 
510     r = rados_ioctx_create(s->cluster, pool, &s->io_ctx);
511     if (r < 0) {
512         error_setg(errp, "error opening pool %s", pool);
513         goto failed_shutdown;
514     }
515 
516     r = rbd_open(s->io_ctx, s->name, &s->image, s->snap);
517     if (r < 0) {
518         error_setg(errp, "error reading header from %s", s->name);
519         goto failed_open;
520     }
521 
522     bs->read_only = (s->snap != NULL);
523 
524     qemu_opts_del(opts);
525     return 0;
526 
527 failed_open:
528     rados_ioctx_destroy(s->io_ctx);
529 failed_shutdown:
530     rados_shutdown(s->cluster);
531     g_free(s->snap);
532 failed_opts:
533     qemu_opts_del(opts);
534     return r;
535 }
536 
537 static void qemu_rbd_close(BlockDriverState *bs)
538 {
539     BDRVRBDState *s = bs->opaque;
540 
541     rbd_close(s->image);
542     rados_ioctx_destroy(s->io_ctx);
543     g_free(s->snap);
544     rados_shutdown(s->cluster);
545 }
546 
547 static const AIOCBInfo rbd_aiocb_info = {
548     .aiocb_size = sizeof(RBDAIOCB),
549 };
550 
551 static void rbd_finish_bh(void *opaque)
552 {
553     RADOSCB *rcb = opaque;
554     qemu_bh_delete(rcb->acb->bh);
555     qemu_rbd_complete_aio(rcb);
556 }
557 
558 /*
559  * This is the callback function for rbd_aio_read and _write
560  *
561  * Note: this function is being called from a non qemu thread so
562  * we need to be careful about what we do here. Generally we only
563  * schedule a BH, and do the rest of the io completion handling
564  * from rbd_finish_bh() which runs in a qemu context.
565  */
566 static void rbd_finish_aiocb(rbd_completion_t c, RADOSCB *rcb)
567 {
568     RBDAIOCB *acb = rcb->acb;
569 
570     rcb->ret = rbd_aio_get_return_value(c);
571     rbd_aio_release(c);
572 
573     acb->bh = aio_bh_new(bdrv_get_aio_context(acb->common.bs),
574                          rbd_finish_bh, rcb);
575     qemu_bh_schedule(acb->bh);
576 }
577 
578 static int rbd_aio_discard_wrapper(rbd_image_t image,
579                                    uint64_t off,
580                                    uint64_t len,
581                                    rbd_completion_t comp)
582 {
583 #ifdef LIBRBD_SUPPORTS_DISCARD
584     return rbd_aio_discard(image, off, len, comp);
585 #else
586     return -ENOTSUP;
587 #endif
588 }
589 
590 static int rbd_aio_flush_wrapper(rbd_image_t image,
591                                  rbd_completion_t comp)
592 {
593 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
594     return rbd_aio_flush(image, comp);
595 #else
596     return -ENOTSUP;
597 #endif
598 }
599 
600 static BlockAIOCB *rbd_start_aio(BlockDriverState *bs,
601                                  int64_t sector_num,
602                                  QEMUIOVector *qiov,
603                                  int nb_sectors,
604                                  BlockCompletionFunc *cb,
605                                  void *opaque,
606                                  RBDAIOCmd cmd)
607 {
608     RBDAIOCB *acb;
609     RADOSCB *rcb = NULL;
610     rbd_completion_t c;
611     int64_t off, size;
612     char *buf;
613     int r;
614 
615     BDRVRBDState *s = bs->opaque;
616 
617     acb = qemu_aio_get(&rbd_aiocb_info, bs, cb, opaque);
618     acb->cmd = cmd;
619     acb->qiov = qiov;
620     if (cmd == RBD_AIO_DISCARD || cmd == RBD_AIO_FLUSH) {
621         acb->bounce = NULL;
622     } else {
623         acb->bounce = qemu_try_blockalign(bs, qiov->size);
624         if (acb->bounce == NULL) {
625             goto failed;
626         }
627     }
628     acb->ret = 0;
629     acb->error = 0;
630     acb->s = s;
631     acb->bh = NULL;
632 
633     if (cmd == RBD_AIO_WRITE) {
634         qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
635     }
636 
637     buf = acb->bounce;
638 
639     off = sector_num * BDRV_SECTOR_SIZE;
640     size = nb_sectors * BDRV_SECTOR_SIZE;
641 
642     rcb = g_new(RADOSCB, 1);
643     rcb->acb = acb;
644     rcb->buf = buf;
645     rcb->s = acb->s;
646     rcb->size = size;
647     r = rbd_aio_create_completion(rcb, (rbd_callback_t) rbd_finish_aiocb, &c);
648     if (r < 0) {
649         goto failed;
650     }
651 
652     switch (cmd) {
653     case RBD_AIO_WRITE:
654         r = rbd_aio_write(s->image, off, size, buf, c);
655         break;
656     case RBD_AIO_READ:
657         r = rbd_aio_read(s->image, off, size, buf, c);
658         break;
659     case RBD_AIO_DISCARD:
660         r = rbd_aio_discard_wrapper(s->image, off, size, c);
661         break;
662     case RBD_AIO_FLUSH:
663         r = rbd_aio_flush_wrapper(s->image, c);
664         break;
665     default:
666         r = -EINVAL;
667     }
668 
669     if (r < 0) {
670         goto failed_completion;
671     }
672 
673     return &acb->common;
674 
675 failed_completion:
676     rbd_aio_release(c);
677 failed:
678     g_free(rcb);
679     qemu_vfree(acb->bounce);
680     qemu_aio_unref(acb);
681     return NULL;
682 }
683 
684 static BlockAIOCB *qemu_rbd_aio_readv(BlockDriverState *bs,
685                                       int64_t sector_num,
686                                       QEMUIOVector *qiov,
687                                       int nb_sectors,
688                                       BlockCompletionFunc *cb,
689                                       void *opaque)
690 {
691     return rbd_start_aio(bs, sector_num, qiov, nb_sectors, cb, opaque,
692                          RBD_AIO_READ);
693 }
694 
695 static BlockAIOCB *qemu_rbd_aio_writev(BlockDriverState *bs,
696                                        int64_t sector_num,
697                                        QEMUIOVector *qiov,
698                                        int nb_sectors,
699                                        BlockCompletionFunc *cb,
700                                        void *opaque)
701 {
702     return rbd_start_aio(bs, sector_num, qiov, nb_sectors, cb, opaque,
703                          RBD_AIO_WRITE);
704 }
705 
706 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
707 static BlockAIOCB *qemu_rbd_aio_flush(BlockDriverState *bs,
708                                       BlockCompletionFunc *cb,
709                                       void *opaque)
710 {
711     return rbd_start_aio(bs, 0, NULL, 0, cb, opaque, RBD_AIO_FLUSH);
712 }
713 
714 #else
715 
716 static int qemu_rbd_co_flush(BlockDriverState *bs)
717 {
718 #if LIBRBD_VERSION_CODE >= LIBRBD_VERSION(0, 1, 1)
719     /* rbd_flush added in 0.1.1 */
720     BDRVRBDState *s = bs->opaque;
721     return rbd_flush(s->image);
722 #else
723     return 0;
724 #endif
725 }
726 #endif
727 
728 static int qemu_rbd_getinfo(BlockDriverState *bs, BlockDriverInfo *bdi)
729 {
730     BDRVRBDState *s = bs->opaque;
731     rbd_image_info_t info;
732     int r;
733 
734     r = rbd_stat(s->image, &info, sizeof(info));
735     if (r < 0) {
736         return r;
737     }
738 
739     bdi->cluster_size = info.obj_size;
740     return 0;
741 }
742 
743 static int64_t qemu_rbd_getlength(BlockDriverState *bs)
744 {
745     BDRVRBDState *s = bs->opaque;
746     rbd_image_info_t info;
747     int r;
748 
749     r = rbd_stat(s->image, &info, sizeof(info));
750     if (r < 0) {
751         return r;
752     }
753 
754     return info.size;
755 }
756 
757 static int qemu_rbd_truncate(BlockDriverState *bs, int64_t offset)
758 {
759     BDRVRBDState *s = bs->opaque;
760     int r;
761 
762     r = rbd_resize(s->image, offset);
763     if (r < 0) {
764         return r;
765     }
766 
767     return 0;
768 }
769 
770 static int qemu_rbd_snap_create(BlockDriverState *bs,
771                                 QEMUSnapshotInfo *sn_info)
772 {
773     BDRVRBDState *s = bs->opaque;
774     int r;
775 
776     if (sn_info->name[0] == '\0') {
777         return -EINVAL; /* we need a name for rbd snapshots */
778     }
779 
780     /*
781      * rbd snapshots are using the name as the user controlled unique identifier
782      * we can't use the rbd snapid for that purpose, as it can't be set
783      */
784     if (sn_info->id_str[0] != '\0' &&
785         strcmp(sn_info->id_str, sn_info->name) != 0) {
786         return -EINVAL;
787     }
788 
789     if (strlen(sn_info->name) >= sizeof(sn_info->id_str)) {
790         return -ERANGE;
791     }
792 
793     r = rbd_snap_create(s->image, sn_info->name);
794     if (r < 0) {
795         error_report("failed to create snap: %s", strerror(-r));
796         return r;
797     }
798 
799     return 0;
800 }
801 
802 static int qemu_rbd_snap_remove(BlockDriverState *bs,
803                                 const char *snapshot_id,
804                                 const char *snapshot_name,
805                                 Error **errp)
806 {
807     BDRVRBDState *s = bs->opaque;
808     int r;
809 
810     if (!snapshot_name) {
811         error_setg(errp, "rbd need a valid snapshot name");
812         return -EINVAL;
813     }
814 
815     /* If snapshot_id is specified, it must be equal to name, see
816        qemu_rbd_snap_list() */
817     if (snapshot_id && strcmp(snapshot_id, snapshot_name)) {
818         error_setg(errp,
819                    "rbd do not support snapshot id, it should be NULL or "
820                    "equal to snapshot name");
821         return -EINVAL;
822     }
823 
824     r = rbd_snap_remove(s->image, snapshot_name);
825     if (r < 0) {
826         error_setg_errno(errp, -r, "Failed to remove the snapshot");
827     }
828     return r;
829 }
830 
831 static int qemu_rbd_snap_rollback(BlockDriverState *bs,
832                                   const char *snapshot_name)
833 {
834     BDRVRBDState *s = bs->opaque;
835     int r;
836 
837     r = rbd_snap_rollback(s->image, snapshot_name);
838     return r;
839 }
840 
841 static int qemu_rbd_snap_list(BlockDriverState *bs,
842                               QEMUSnapshotInfo **psn_tab)
843 {
844     BDRVRBDState *s = bs->opaque;
845     QEMUSnapshotInfo *sn_info, *sn_tab = NULL;
846     int i, snap_count;
847     rbd_snap_info_t *snaps;
848     int max_snaps = RBD_MAX_SNAPS;
849 
850     do {
851         snaps = g_new(rbd_snap_info_t, max_snaps);
852         snap_count = rbd_snap_list(s->image, snaps, &max_snaps);
853         if (snap_count <= 0) {
854             g_free(snaps);
855         }
856     } while (snap_count == -ERANGE);
857 
858     if (snap_count <= 0) {
859         goto done;
860     }
861 
862     sn_tab = g_new0(QEMUSnapshotInfo, snap_count);
863 
864     for (i = 0; i < snap_count; i++) {
865         const char *snap_name = snaps[i].name;
866 
867         sn_info = sn_tab + i;
868         pstrcpy(sn_info->id_str, sizeof(sn_info->id_str), snap_name);
869         pstrcpy(sn_info->name, sizeof(sn_info->name), snap_name);
870 
871         sn_info->vm_state_size = snaps[i].size;
872         sn_info->date_sec = 0;
873         sn_info->date_nsec = 0;
874         sn_info->vm_clock_nsec = 0;
875     }
876     rbd_snap_list_end(snaps);
877     g_free(snaps);
878 
879  done:
880     *psn_tab = sn_tab;
881     return snap_count;
882 }
883 
884 #ifdef LIBRBD_SUPPORTS_DISCARD
885 static BlockAIOCB* qemu_rbd_aio_discard(BlockDriverState *bs,
886                                         int64_t sector_num,
887                                         int nb_sectors,
888                                         BlockCompletionFunc *cb,
889                                         void *opaque)
890 {
891     return rbd_start_aio(bs, sector_num, NULL, nb_sectors, cb, opaque,
892                          RBD_AIO_DISCARD);
893 }
894 #endif
895 
896 #ifdef LIBRBD_SUPPORTS_INVALIDATE
897 static void qemu_rbd_invalidate_cache(BlockDriverState *bs,
898                                       Error **errp)
899 {
900     BDRVRBDState *s = bs->opaque;
901     int r = rbd_invalidate_cache(s->image);
902     if (r < 0) {
903         error_setg_errno(errp, -r, "Failed to invalidate the cache");
904     }
905 }
906 #endif
907 
908 static QemuOptsList qemu_rbd_create_opts = {
909     .name = "rbd-create-opts",
910     .head = QTAILQ_HEAD_INITIALIZER(qemu_rbd_create_opts.head),
911     .desc = {
912         {
913             .name = BLOCK_OPT_SIZE,
914             .type = QEMU_OPT_SIZE,
915             .help = "Virtual disk size"
916         },
917         {
918             .name = BLOCK_OPT_CLUSTER_SIZE,
919             .type = QEMU_OPT_SIZE,
920             .help = "RBD object size"
921         },
922         { /* end of list */ }
923     }
924 };
925 
926 static BlockDriver bdrv_rbd = {
927     .format_name        = "rbd",
928     .instance_size      = sizeof(BDRVRBDState),
929     .bdrv_needs_filename = true,
930     .bdrv_file_open     = qemu_rbd_open,
931     .bdrv_close         = qemu_rbd_close,
932     .bdrv_create        = qemu_rbd_create,
933     .bdrv_has_zero_init = bdrv_has_zero_init_1,
934     .bdrv_get_info      = qemu_rbd_getinfo,
935     .create_opts        = &qemu_rbd_create_opts,
936     .bdrv_getlength     = qemu_rbd_getlength,
937     .bdrv_truncate      = qemu_rbd_truncate,
938     .protocol_name      = "rbd",
939 
940     .bdrv_aio_readv         = qemu_rbd_aio_readv,
941     .bdrv_aio_writev        = qemu_rbd_aio_writev,
942 
943 #ifdef LIBRBD_SUPPORTS_AIO_FLUSH
944     .bdrv_aio_flush         = qemu_rbd_aio_flush,
945 #else
946     .bdrv_co_flush_to_disk  = qemu_rbd_co_flush,
947 #endif
948 
949 #ifdef LIBRBD_SUPPORTS_DISCARD
950     .bdrv_aio_discard       = qemu_rbd_aio_discard,
951 #endif
952 
953     .bdrv_snapshot_create   = qemu_rbd_snap_create,
954     .bdrv_snapshot_delete   = qemu_rbd_snap_remove,
955     .bdrv_snapshot_list     = qemu_rbd_snap_list,
956     .bdrv_snapshot_goto     = qemu_rbd_snap_rollback,
957 #ifdef LIBRBD_SUPPORTS_INVALIDATE
958     .bdrv_invalidate_cache  = qemu_rbd_invalidate_cache,
959 #endif
960 };
961 
962 static void bdrv_rbd_init(void)
963 {
964     bdrv_register(&bdrv_rbd);
965 }
966 
967 block_init(bdrv_rbd_init);
968