xref: /qemu/block/nbd.c (revision ebda3036)
1 /*
2  * QEMU Block driver for NBD
3  *
4  * Copyright (c) 2019 Virtuozzo International GmbH.
5  * Copyright Red Hat
6  * Copyright (C) 2008 Bull S.A.S.
7  *     Author: Laurent Vivier <Laurent.Vivier@bull.net>
8  *
9  * Some parts:
10  *    Copyright (C) 2007 Anthony Liguori <anthony@codemonkey.ws>
11  *
12  * Permission is hereby granted, free of charge, to any person obtaining a copy
13  * of this software and associated documentation files (the "Software"), to deal
14  * in the Software without restriction, including without limitation the rights
15  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
16  * copies of the Software, and to permit persons to whom the Software is
17  * furnished to do so, subject to the following conditions:
18  *
19  * The above copyright notice and this permission notice shall be included in
20  * all copies or substantial portions of the Software.
21  *
22  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
23  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
24  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
25  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
26  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
27  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
28  * THE SOFTWARE.
29  */
30 
31 #include "qemu/osdep.h"
32 
33 #include "trace.h"
34 #include "qemu/uri.h"
35 #include "qemu/option.h"
36 #include "qemu/cutils.h"
37 #include "qemu/main-loop.h"
38 
39 #include "qapi/qapi-visit-sockets.h"
40 #include "qapi/qmp/qstring.h"
41 #include "qapi/clone-visitor.h"
42 
43 #include "block/qdict.h"
44 #include "block/nbd.h"
45 #include "block/block_int.h"
46 #include "block/coroutines.h"
47 
48 #include "qemu/yank.h"
49 
50 #define EN_OPTSTR ":exportname="
51 #define MAX_NBD_REQUESTS    16
52 
53 #define COOKIE_TO_INDEX(cookie) ((cookie) - 1)
54 #define INDEX_TO_COOKIE(index)  ((index) + 1)
55 
56 typedef struct {
57     Coroutine *coroutine;
58     uint64_t offset;        /* original offset of the request */
59     bool receiving;         /* sleeping in the yield in nbd_receive_replies */
60 } NBDClientRequest;
61 
62 typedef enum NBDClientState {
63     NBD_CLIENT_CONNECTING_WAIT,
64     NBD_CLIENT_CONNECTING_NOWAIT,
65     NBD_CLIENT_CONNECTED,
66     NBD_CLIENT_QUIT
67 } NBDClientState;
68 
69 typedef struct BDRVNBDState {
70     QIOChannel *ioc; /* The current I/O channel */
71     NBDExportInfo info;
72 
73     /*
74      * Protects state, free_sema, in_flight, requests[].coroutine,
75      * reconnect_delay_timer.
76      */
77     QemuMutex requests_lock;
78     NBDClientState state;
79     CoQueue free_sema;
80     unsigned in_flight;
81     NBDClientRequest requests[MAX_NBD_REQUESTS];
82     QEMUTimer *reconnect_delay_timer;
83 
84     /* Protects sending data on the socket.  */
85     CoMutex send_mutex;
86 
87     /*
88      * Protects receiving reply headers from the socket, as well as the
89      * fields reply and requests[].receiving
90      */
91     CoMutex receive_mutex;
92     NBDReply reply;
93 
94     QEMUTimer *open_timer;
95 
96     BlockDriverState *bs;
97 
98     /* Connection parameters */
99     uint32_t reconnect_delay;
100     uint32_t open_timeout;
101     SocketAddress *saddr;
102     char *export;
103     char *tlscredsid;
104     QCryptoTLSCreds *tlscreds;
105     char *tlshostname;
106     char *x_dirty_bitmap;
107     bool alloc_depth;
108 
109     NBDClientConnection *conn;
110 } BDRVNBDState;
111 
112 static void nbd_yank(void *opaque);
113 
114 static void nbd_clear_bdrvstate(BlockDriverState *bs)
115 {
116     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
117 
118     nbd_client_connection_release(s->conn);
119     s->conn = NULL;
120 
121     yank_unregister_instance(BLOCKDEV_YANK_INSTANCE(bs->node_name));
122 
123     /* Must not leave timers behind that would access freed data */
124     assert(!s->reconnect_delay_timer);
125     assert(!s->open_timer);
126 
127     object_unref(OBJECT(s->tlscreds));
128     qapi_free_SocketAddress(s->saddr);
129     s->saddr = NULL;
130     g_free(s->export);
131     s->export = NULL;
132     g_free(s->tlscredsid);
133     s->tlscredsid = NULL;
134     g_free(s->tlshostname);
135     s->tlshostname = NULL;
136     g_free(s->x_dirty_bitmap);
137     s->x_dirty_bitmap = NULL;
138 }
139 
140 /* Called with s->receive_mutex taken.  */
141 static bool coroutine_fn nbd_recv_coroutine_wake_one(NBDClientRequest *req)
142 {
143     if (req->receiving) {
144         req->receiving = false;
145         aio_co_wake(req->coroutine);
146         return true;
147     }
148 
149     return false;
150 }
151 
152 static void coroutine_fn nbd_recv_coroutines_wake(BDRVNBDState *s)
153 {
154     int i;
155 
156     QEMU_LOCK_GUARD(&s->receive_mutex);
157     for (i = 0; i < MAX_NBD_REQUESTS; i++) {
158         if (nbd_recv_coroutine_wake_one(&s->requests[i])) {
159             return;
160         }
161     }
162 }
163 
164 /* Called with s->requests_lock held.  */
165 static void coroutine_fn nbd_channel_error_locked(BDRVNBDState *s, int ret)
166 {
167     if (s->state == NBD_CLIENT_CONNECTED) {
168         qio_channel_shutdown(s->ioc, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
169     }
170 
171     if (ret == -EIO) {
172         if (s->state == NBD_CLIENT_CONNECTED) {
173             s->state = s->reconnect_delay ? NBD_CLIENT_CONNECTING_WAIT :
174                                             NBD_CLIENT_CONNECTING_NOWAIT;
175         }
176     } else {
177         s->state = NBD_CLIENT_QUIT;
178     }
179 }
180 
181 static void coroutine_fn nbd_channel_error(BDRVNBDState *s, int ret)
182 {
183     QEMU_LOCK_GUARD(&s->requests_lock);
184     nbd_channel_error_locked(s, ret);
185 }
186 
187 static void reconnect_delay_timer_del(BDRVNBDState *s)
188 {
189     if (s->reconnect_delay_timer) {
190         timer_free(s->reconnect_delay_timer);
191         s->reconnect_delay_timer = NULL;
192     }
193 }
194 
195 static void reconnect_delay_timer_cb(void *opaque)
196 {
197     BDRVNBDState *s = opaque;
198 
199     reconnect_delay_timer_del(s);
200     WITH_QEMU_LOCK_GUARD(&s->requests_lock) {
201         if (s->state != NBD_CLIENT_CONNECTING_WAIT) {
202             return;
203         }
204         s->state = NBD_CLIENT_CONNECTING_NOWAIT;
205     }
206     nbd_co_establish_connection_cancel(s->conn);
207 }
208 
209 static void reconnect_delay_timer_init(BDRVNBDState *s, uint64_t expire_time_ns)
210 {
211     assert(!s->reconnect_delay_timer);
212     s->reconnect_delay_timer = aio_timer_new(bdrv_get_aio_context(s->bs),
213                                              QEMU_CLOCK_REALTIME,
214                                              SCALE_NS,
215                                              reconnect_delay_timer_cb, s);
216     timer_mod(s->reconnect_delay_timer, expire_time_ns);
217 }
218 
219 static void nbd_teardown_connection(BlockDriverState *bs)
220 {
221     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
222 
223     assert(!s->in_flight);
224 
225     if (s->ioc) {
226         qio_channel_shutdown(s->ioc, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
227         yank_unregister_function(BLOCKDEV_YANK_INSTANCE(s->bs->node_name),
228                                  nbd_yank, s->bs);
229         object_unref(OBJECT(s->ioc));
230         s->ioc = NULL;
231     }
232 
233     WITH_QEMU_LOCK_GUARD(&s->requests_lock) {
234         s->state = NBD_CLIENT_QUIT;
235     }
236 }
237 
238 static void open_timer_del(BDRVNBDState *s)
239 {
240     if (s->open_timer) {
241         timer_free(s->open_timer);
242         s->open_timer = NULL;
243     }
244 }
245 
246 static void open_timer_cb(void *opaque)
247 {
248     BDRVNBDState *s = opaque;
249 
250     nbd_co_establish_connection_cancel(s->conn);
251     open_timer_del(s);
252 }
253 
254 static void open_timer_init(BDRVNBDState *s, uint64_t expire_time_ns)
255 {
256     assert(!s->open_timer);
257     s->open_timer = aio_timer_new(bdrv_get_aio_context(s->bs),
258                                   QEMU_CLOCK_REALTIME,
259                                   SCALE_NS,
260                                   open_timer_cb, s);
261     timer_mod(s->open_timer, expire_time_ns);
262 }
263 
264 static bool nbd_client_will_reconnect(BDRVNBDState *s)
265 {
266     /*
267      * Called only after a socket error, so this is not performance sensitive.
268      */
269     QEMU_LOCK_GUARD(&s->requests_lock);
270     return s->state == NBD_CLIENT_CONNECTING_WAIT;
271 }
272 
273 /*
274  * Update @bs with information learned during a completed negotiation process.
275  * Return failure if the server's advertised options are incompatible with the
276  * client's needs.
277  */
278 static int nbd_handle_updated_info(BlockDriverState *bs, Error **errp)
279 {
280     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
281     int ret;
282 
283     if (s->x_dirty_bitmap) {
284         if (!s->info.base_allocation) {
285             error_setg(errp, "requested x-dirty-bitmap %s not found",
286                        s->x_dirty_bitmap);
287             return -EINVAL;
288         }
289         if (strcmp(s->x_dirty_bitmap, "qemu:allocation-depth") == 0) {
290             s->alloc_depth = true;
291         }
292     }
293 
294     if (s->info.flags & NBD_FLAG_READ_ONLY) {
295         ret = bdrv_apply_auto_read_only(bs, "NBD export is read-only", errp);
296         if (ret < 0) {
297             return ret;
298         }
299     }
300 
301     if (s->info.flags & NBD_FLAG_SEND_FUA) {
302         bs->supported_write_flags = BDRV_REQ_FUA;
303         bs->supported_zero_flags |= BDRV_REQ_FUA;
304     }
305 
306     if (s->info.flags & NBD_FLAG_SEND_WRITE_ZEROES) {
307         bs->supported_zero_flags |= BDRV_REQ_MAY_UNMAP;
308         if (s->info.flags & NBD_FLAG_SEND_FAST_ZERO) {
309             bs->supported_zero_flags |= BDRV_REQ_NO_FALLBACK;
310         }
311     }
312 
313     trace_nbd_client_handshake_success(s->export);
314 
315     return 0;
316 }
317 
318 int coroutine_fn nbd_co_do_establish_connection(BlockDriverState *bs,
319                                                 bool blocking, Error **errp)
320 {
321     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
322     int ret;
323     IO_CODE();
324 
325     assert_bdrv_graph_readable();
326     assert(!s->ioc);
327 
328     s->ioc = nbd_co_establish_connection(s->conn, &s->info, blocking, errp);
329     if (!s->ioc) {
330         return -ECONNREFUSED;
331     }
332 
333     yank_register_function(BLOCKDEV_YANK_INSTANCE(s->bs->node_name), nbd_yank,
334                            bs);
335 
336     ret = nbd_handle_updated_info(s->bs, NULL);
337     if (ret < 0) {
338         /*
339          * We have connected, but must fail for other reasons.
340          * Send NBD_CMD_DISC as a courtesy to the server.
341          */
342         NBDRequest request = { .type = NBD_CMD_DISC };
343 
344         nbd_send_request(s->ioc, &request);
345 
346         yank_unregister_function(BLOCKDEV_YANK_INSTANCE(s->bs->node_name),
347                                  nbd_yank, bs);
348         object_unref(OBJECT(s->ioc));
349         s->ioc = NULL;
350 
351         return ret;
352     }
353 
354     qio_channel_set_blocking(s->ioc, false, NULL);
355     qio_channel_attach_aio_context(s->ioc, bdrv_get_aio_context(bs));
356 
357     /* successfully connected */
358     WITH_QEMU_LOCK_GUARD(&s->requests_lock) {
359         s->state = NBD_CLIENT_CONNECTED;
360     }
361 
362     return 0;
363 }
364 
365 /* Called with s->requests_lock held.  */
366 static bool nbd_client_connecting(BDRVNBDState *s)
367 {
368     return s->state == NBD_CLIENT_CONNECTING_WAIT ||
369         s->state == NBD_CLIENT_CONNECTING_NOWAIT;
370 }
371 
372 /* Called with s->requests_lock taken.  */
373 static void coroutine_fn GRAPH_RDLOCK nbd_reconnect_attempt(BDRVNBDState *s)
374 {
375     int ret;
376     bool blocking = s->state == NBD_CLIENT_CONNECTING_WAIT;
377 
378     /*
379      * Now we are sure that nobody is accessing the channel, and no one will
380      * try until we set the state to CONNECTED.
381      */
382     assert(nbd_client_connecting(s));
383     assert(s->in_flight == 1);
384 
385     trace_nbd_reconnect_attempt(s->bs->in_flight);
386 
387     if (blocking && !s->reconnect_delay_timer) {
388         /*
389          * It's the first reconnect attempt after switching to
390          * NBD_CLIENT_CONNECTING_WAIT
391          */
392         g_assert(s->reconnect_delay);
393         reconnect_delay_timer_init(s,
394             qemu_clock_get_ns(QEMU_CLOCK_REALTIME) +
395             s->reconnect_delay * NANOSECONDS_PER_SECOND);
396     }
397 
398     /* Finalize previous connection if any */
399     if (s->ioc) {
400         qio_channel_detach_aio_context(s->ioc);
401         yank_unregister_function(BLOCKDEV_YANK_INSTANCE(s->bs->node_name),
402                                  nbd_yank, s->bs);
403         object_unref(OBJECT(s->ioc));
404         s->ioc = NULL;
405     }
406 
407     qemu_mutex_unlock(&s->requests_lock);
408     ret = nbd_co_do_establish_connection(s->bs, blocking, NULL);
409     trace_nbd_reconnect_attempt_result(ret, s->bs->in_flight);
410     qemu_mutex_lock(&s->requests_lock);
411 
412     /*
413      * The reconnect attempt is done (maybe successfully, maybe not), so
414      * we no longer need this timer.  Delete it so it will not outlive
415      * this I/O request (so draining removes all timers).
416      */
417     reconnect_delay_timer_del(s);
418 }
419 
420 static coroutine_fn int nbd_receive_replies(BDRVNBDState *s, uint64_t cookie)
421 {
422     int ret;
423     uint64_t ind = COOKIE_TO_INDEX(cookie), ind2;
424     QEMU_LOCK_GUARD(&s->receive_mutex);
425 
426     while (true) {
427         if (s->reply.cookie == cookie) {
428             /* We are done */
429             return 0;
430         }
431 
432         if (s->reply.cookie != 0) {
433             /*
434              * Some other request is being handled now. It should already be
435              * woken by whoever set s->reply.cookie (or never wait in this
436              * yield). So, we should not wake it here.
437              */
438             ind2 = COOKIE_TO_INDEX(s->reply.cookie);
439             assert(!s->requests[ind2].receiving);
440 
441             s->requests[ind].receiving = true;
442             qemu_co_mutex_unlock(&s->receive_mutex);
443 
444             qemu_coroutine_yield();
445             /*
446              * We may be woken for 2 reasons:
447              * 1. From this function, executing in parallel coroutine, when our
448              *    cookie is received.
449              * 2. From nbd_co_receive_one_chunk(), when previous request is
450              *    finished and s->reply.cookie set to 0.
451              * Anyway, it's OK to lock the mutex and go to the next iteration.
452              */
453 
454             qemu_co_mutex_lock(&s->receive_mutex);
455             assert(!s->requests[ind].receiving);
456             continue;
457         }
458 
459         /* We are under mutex and cookie is 0. We have to do the dirty work. */
460         assert(s->reply.cookie == 0);
461         ret = nbd_receive_reply(s->bs, s->ioc, &s->reply, NULL);
462         if (ret <= 0) {
463             ret = ret ? ret : -EIO;
464             nbd_channel_error(s, ret);
465             return ret;
466         }
467         if (nbd_reply_is_structured(&s->reply) && !s->info.structured_reply) {
468             nbd_channel_error(s, -EINVAL);
469             return -EINVAL;
470         }
471         ind2 = COOKIE_TO_INDEX(s->reply.cookie);
472         if (ind2 >= MAX_NBD_REQUESTS || !s->requests[ind2].coroutine) {
473             nbd_channel_error(s, -EINVAL);
474             return -EINVAL;
475         }
476         if (s->reply.cookie == cookie) {
477             /* We are done */
478             return 0;
479         }
480         nbd_recv_coroutine_wake_one(&s->requests[ind2]);
481     }
482 }
483 
484 static int coroutine_fn GRAPH_RDLOCK
485 nbd_co_send_request(BlockDriverState *bs, NBDRequest *request,
486                     QEMUIOVector *qiov)
487 {
488     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
489     int rc, i = -1;
490 
491     qemu_mutex_lock(&s->requests_lock);
492     while (s->in_flight == MAX_NBD_REQUESTS ||
493            (s->state != NBD_CLIENT_CONNECTED && s->in_flight > 0)) {
494         qemu_co_queue_wait(&s->free_sema, &s->requests_lock);
495     }
496 
497     s->in_flight++;
498     if (s->state != NBD_CLIENT_CONNECTED) {
499         if (nbd_client_connecting(s)) {
500             nbd_reconnect_attempt(s);
501             qemu_co_queue_restart_all(&s->free_sema);
502         }
503         if (s->state != NBD_CLIENT_CONNECTED) {
504             rc = -EIO;
505             goto err;
506         }
507     }
508 
509     for (i = 0; i < MAX_NBD_REQUESTS; i++) {
510         if (s->requests[i].coroutine == NULL) {
511             break;
512         }
513     }
514 
515     assert(i < MAX_NBD_REQUESTS);
516     s->requests[i].coroutine = qemu_coroutine_self();
517     s->requests[i].offset = request->from;
518     s->requests[i].receiving = false;
519     qemu_mutex_unlock(&s->requests_lock);
520 
521     qemu_co_mutex_lock(&s->send_mutex);
522     request->cookie = INDEX_TO_COOKIE(i);
523 
524     assert(s->ioc);
525 
526     if (qiov) {
527         qio_channel_set_cork(s->ioc, true);
528         rc = nbd_send_request(s->ioc, request);
529         if (rc >= 0 && qio_channel_writev_all(s->ioc, qiov->iov, qiov->niov,
530                                               NULL) < 0) {
531             rc = -EIO;
532         }
533         qio_channel_set_cork(s->ioc, false);
534     } else {
535         rc = nbd_send_request(s->ioc, request);
536     }
537     qemu_co_mutex_unlock(&s->send_mutex);
538 
539     if (rc < 0) {
540         qemu_mutex_lock(&s->requests_lock);
541 err:
542         nbd_channel_error_locked(s, rc);
543         if (i != -1) {
544             s->requests[i].coroutine = NULL;
545         }
546         s->in_flight--;
547         qemu_co_queue_next(&s->free_sema);
548         qemu_mutex_unlock(&s->requests_lock);
549     }
550     return rc;
551 }
552 
553 static inline uint16_t payload_advance16(uint8_t **payload)
554 {
555     *payload += 2;
556     return lduw_be_p(*payload - 2);
557 }
558 
559 static inline uint32_t payload_advance32(uint8_t **payload)
560 {
561     *payload += 4;
562     return ldl_be_p(*payload - 4);
563 }
564 
565 static inline uint64_t payload_advance64(uint8_t **payload)
566 {
567     *payload += 8;
568     return ldq_be_p(*payload - 8);
569 }
570 
571 static int nbd_parse_offset_hole_payload(BDRVNBDState *s,
572                                          NBDStructuredReplyChunk *chunk,
573                                          uint8_t *payload, uint64_t orig_offset,
574                                          QEMUIOVector *qiov, Error **errp)
575 {
576     uint64_t offset;
577     uint32_t hole_size;
578 
579     if (chunk->length != sizeof(offset) + sizeof(hole_size)) {
580         error_setg(errp, "Protocol error: invalid payload for "
581                          "NBD_REPLY_TYPE_OFFSET_HOLE");
582         return -EINVAL;
583     }
584 
585     offset = payload_advance64(&payload);
586     hole_size = payload_advance32(&payload);
587 
588     if (!hole_size || offset < orig_offset || hole_size > qiov->size ||
589         offset > orig_offset + qiov->size - hole_size) {
590         error_setg(errp, "Protocol error: server sent chunk exceeding requested"
591                          " region");
592         return -EINVAL;
593     }
594     if (s->info.min_block &&
595         !QEMU_IS_ALIGNED(hole_size, s->info.min_block)) {
596         trace_nbd_structured_read_compliance("hole");
597     }
598 
599     qemu_iovec_memset(qiov, offset - orig_offset, 0, hole_size);
600 
601     return 0;
602 }
603 
604 /*
605  * nbd_parse_blockstatus_payload
606  * Based on our request, we expect only one extent in reply, for the
607  * base:allocation context.
608  */
609 static int nbd_parse_blockstatus_payload(BDRVNBDState *s,
610                                          NBDStructuredReplyChunk *chunk,
611                                          uint8_t *payload, uint64_t orig_length,
612                                          NBDExtent *extent, Error **errp)
613 {
614     uint32_t context_id;
615 
616     /* The server succeeded, so it must have sent [at least] one extent */
617     if (chunk->length < sizeof(context_id) + sizeof(*extent)) {
618         error_setg(errp, "Protocol error: invalid payload for "
619                          "NBD_REPLY_TYPE_BLOCK_STATUS");
620         return -EINVAL;
621     }
622 
623     context_id = payload_advance32(&payload);
624     if (s->info.context_id != context_id) {
625         error_setg(errp, "Protocol error: unexpected context id %d for "
626                          "NBD_REPLY_TYPE_BLOCK_STATUS, when negotiated context "
627                          "id is %d", context_id,
628                          s->info.context_id);
629         return -EINVAL;
630     }
631 
632     extent->length = payload_advance32(&payload);
633     extent->flags = payload_advance32(&payload);
634 
635     if (extent->length == 0) {
636         error_setg(errp, "Protocol error: server sent status chunk with "
637                    "zero length");
638         return -EINVAL;
639     }
640 
641     /*
642      * A server sending unaligned block status is in violation of the
643      * protocol, but as qemu-nbd 3.1 is such a server (at least for
644      * POSIX files that are not a multiple of 512 bytes, since qemu
645      * rounds files up to 512-byte multiples but lseek(SEEK_HOLE)
646      * still sees an implicit hole beyond the real EOF), it's nicer to
647      * work around the misbehaving server. If the request included
648      * more than the final unaligned block, truncate it back to an
649      * aligned result; if the request was only the final block, round
650      * up to the full block and change the status to fully-allocated
651      * (always a safe status, even if it loses information).
652      */
653     if (s->info.min_block && !QEMU_IS_ALIGNED(extent->length,
654                                                    s->info.min_block)) {
655         trace_nbd_parse_blockstatus_compliance("extent length is unaligned");
656         if (extent->length > s->info.min_block) {
657             extent->length = QEMU_ALIGN_DOWN(extent->length,
658                                              s->info.min_block);
659         } else {
660             extent->length = s->info.min_block;
661             extent->flags = 0;
662         }
663     }
664 
665     /*
666      * We used NBD_CMD_FLAG_REQ_ONE, so the server should not have
667      * sent us any more than one extent, nor should it have included
668      * status beyond our request in that extent. However, it's easy
669      * enough to ignore the server's noncompliance without killing the
670      * connection; just ignore trailing extents, and clamp things to
671      * the length of our request.
672      */
673     if (chunk->length > sizeof(context_id) + sizeof(*extent)) {
674         trace_nbd_parse_blockstatus_compliance("more than one extent");
675     }
676     if (extent->length > orig_length) {
677         extent->length = orig_length;
678         trace_nbd_parse_blockstatus_compliance("extent length too large");
679     }
680 
681     /*
682      * HACK: if we are using x-dirty-bitmaps to access
683      * qemu:allocation-depth, treat all depths > 2 the same as 2,
684      * since nbd_client_co_block_status is only expecting the low two
685      * bits to be set.
686      */
687     if (s->alloc_depth && extent->flags > 2) {
688         extent->flags = 2;
689     }
690 
691     return 0;
692 }
693 
694 /*
695  * nbd_parse_error_payload
696  * on success @errp contains message describing nbd error reply
697  */
698 static int nbd_parse_error_payload(NBDStructuredReplyChunk *chunk,
699                                    uint8_t *payload, int *request_ret,
700                                    Error **errp)
701 {
702     uint32_t error;
703     uint16_t message_size;
704 
705     assert(chunk->type & (1 << 15));
706 
707     if (chunk->length < sizeof(error) + sizeof(message_size)) {
708         error_setg(errp,
709                    "Protocol error: invalid payload for structured error");
710         return -EINVAL;
711     }
712 
713     error = nbd_errno_to_system_errno(payload_advance32(&payload));
714     if (error == 0) {
715         error_setg(errp, "Protocol error: server sent structured error chunk "
716                          "with error = 0");
717         return -EINVAL;
718     }
719 
720     *request_ret = -error;
721     message_size = payload_advance16(&payload);
722 
723     if (message_size > chunk->length - sizeof(error) - sizeof(message_size)) {
724         error_setg(errp, "Protocol error: server sent structured error chunk "
725                          "with incorrect message size");
726         return -EINVAL;
727     }
728 
729     /* TODO: Add a trace point to mention the server complaint */
730 
731     /* TODO handle ERROR_OFFSET */
732 
733     return 0;
734 }
735 
736 static int coroutine_fn
737 nbd_co_receive_offset_data_payload(BDRVNBDState *s, uint64_t orig_offset,
738                                    QEMUIOVector *qiov, Error **errp)
739 {
740     QEMUIOVector sub_qiov;
741     uint64_t offset;
742     size_t data_size;
743     int ret;
744     NBDStructuredReplyChunk *chunk = &s->reply.structured;
745 
746     assert(nbd_reply_is_structured(&s->reply));
747 
748     /* The NBD spec requires at least one byte of payload */
749     if (chunk->length <= sizeof(offset)) {
750         error_setg(errp, "Protocol error: invalid payload for "
751                          "NBD_REPLY_TYPE_OFFSET_DATA");
752         return -EINVAL;
753     }
754 
755     if (nbd_read64(s->ioc, &offset, "OFFSET_DATA offset", errp) < 0) {
756         return -EIO;
757     }
758 
759     data_size = chunk->length - sizeof(offset);
760     assert(data_size);
761     if (offset < orig_offset || data_size > qiov->size ||
762         offset > orig_offset + qiov->size - data_size) {
763         error_setg(errp, "Protocol error: server sent chunk exceeding requested"
764                          " region");
765         return -EINVAL;
766     }
767     if (s->info.min_block && !QEMU_IS_ALIGNED(data_size, s->info.min_block)) {
768         trace_nbd_structured_read_compliance("data");
769     }
770 
771     qemu_iovec_init(&sub_qiov, qiov->niov);
772     qemu_iovec_concat(&sub_qiov, qiov, offset - orig_offset, data_size);
773     ret = qio_channel_readv_all(s->ioc, sub_qiov.iov, sub_qiov.niov, errp);
774     qemu_iovec_destroy(&sub_qiov);
775 
776     return ret < 0 ? -EIO : 0;
777 }
778 
779 #define NBD_MAX_MALLOC_PAYLOAD 1000
780 static coroutine_fn int nbd_co_receive_structured_payload(
781         BDRVNBDState *s, void **payload, Error **errp)
782 {
783     int ret;
784     uint32_t len;
785 
786     assert(nbd_reply_is_structured(&s->reply));
787 
788     len = s->reply.structured.length;
789 
790     if (len == 0) {
791         return 0;
792     }
793 
794     if (payload == NULL) {
795         error_setg(errp, "Unexpected structured payload");
796         return -EINVAL;
797     }
798 
799     if (len > NBD_MAX_MALLOC_PAYLOAD) {
800         error_setg(errp, "Payload too large");
801         return -EINVAL;
802     }
803 
804     *payload = g_new(char, len);
805     ret = nbd_read(s->ioc, *payload, len, "structured payload", errp);
806     if (ret < 0) {
807         g_free(*payload);
808         *payload = NULL;
809         return ret;
810     }
811 
812     return 0;
813 }
814 
815 /*
816  * nbd_co_do_receive_one_chunk
817  * for simple reply:
818  *   set request_ret to received reply error
819  *   if qiov is not NULL: read payload to @qiov
820  * for structured reply chunk:
821  *   if error chunk: read payload, set @request_ret, do not set @payload
822  *   else if offset_data chunk: read payload data to @qiov, do not set @payload
823  *   else: read payload to @payload
824  *
825  * If function fails, @errp contains corresponding error message, and the
826  * connection with the server is suspect.  If it returns 0, then the
827  * transaction succeeded (although @request_ret may be a negative errno
828  * corresponding to the server's error reply), and errp is unchanged.
829  */
830 static coroutine_fn int nbd_co_do_receive_one_chunk(
831         BDRVNBDState *s, uint64_t cookie, bool only_structured,
832         int *request_ret, QEMUIOVector *qiov, void **payload, Error **errp)
833 {
834     int ret;
835     int i = COOKIE_TO_INDEX(cookie);
836     void *local_payload = NULL;
837     NBDStructuredReplyChunk *chunk;
838 
839     if (payload) {
840         *payload = NULL;
841     }
842     *request_ret = 0;
843 
844     ret = nbd_receive_replies(s, cookie);
845     if (ret < 0) {
846         error_setg(errp, "Connection closed");
847         return -EIO;
848     }
849     assert(s->ioc);
850 
851     assert(s->reply.cookie == cookie);
852 
853     if (nbd_reply_is_simple(&s->reply)) {
854         if (only_structured) {
855             error_setg(errp, "Protocol error: simple reply when structured "
856                              "reply chunk was expected");
857             return -EINVAL;
858         }
859 
860         *request_ret = -nbd_errno_to_system_errno(s->reply.simple.error);
861         if (*request_ret < 0 || !qiov) {
862             return 0;
863         }
864 
865         return qio_channel_readv_all(s->ioc, qiov->iov, qiov->niov,
866                                      errp) < 0 ? -EIO : 0;
867     }
868 
869     /* handle structured reply chunk */
870     assert(s->info.structured_reply);
871     chunk = &s->reply.structured;
872 
873     if (chunk->type == NBD_REPLY_TYPE_NONE) {
874         if (!(chunk->flags & NBD_REPLY_FLAG_DONE)) {
875             error_setg(errp, "Protocol error: NBD_REPLY_TYPE_NONE chunk without"
876                        " NBD_REPLY_FLAG_DONE flag set");
877             return -EINVAL;
878         }
879         if (chunk->length) {
880             error_setg(errp, "Protocol error: NBD_REPLY_TYPE_NONE chunk with"
881                        " nonzero length");
882             return -EINVAL;
883         }
884         return 0;
885     }
886 
887     if (chunk->type == NBD_REPLY_TYPE_OFFSET_DATA) {
888         if (!qiov) {
889             error_setg(errp, "Unexpected NBD_REPLY_TYPE_OFFSET_DATA chunk");
890             return -EINVAL;
891         }
892 
893         return nbd_co_receive_offset_data_payload(s, s->requests[i].offset,
894                                                   qiov, errp);
895     }
896 
897     if (nbd_reply_type_is_error(chunk->type)) {
898         payload = &local_payload;
899     }
900 
901     ret = nbd_co_receive_structured_payload(s, payload, errp);
902     if (ret < 0) {
903         return ret;
904     }
905 
906     if (nbd_reply_type_is_error(chunk->type)) {
907         ret = nbd_parse_error_payload(chunk, local_payload, request_ret, errp);
908         g_free(local_payload);
909         return ret;
910     }
911 
912     return 0;
913 }
914 
915 /*
916  * nbd_co_receive_one_chunk
917  * Read reply, wake up connection_co and set s->quit if needed.
918  * Return value is a fatal error code or normal nbd reply error code
919  */
920 static coroutine_fn int nbd_co_receive_one_chunk(
921         BDRVNBDState *s, uint64_t cookie, bool only_structured,
922         int *request_ret, QEMUIOVector *qiov, NBDReply *reply, void **payload,
923         Error **errp)
924 {
925     int ret = nbd_co_do_receive_one_chunk(s, cookie, only_structured,
926                                           request_ret, qiov, payload, errp);
927 
928     if (ret < 0) {
929         memset(reply, 0, sizeof(*reply));
930         nbd_channel_error(s, ret);
931     } else {
932         /* For assert at loop start in nbd_connection_entry */
933         *reply = s->reply;
934     }
935     s->reply.cookie = 0;
936 
937     nbd_recv_coroutines_wake(s);
938 
939     return ret;
940 }
941 
942 typedef struct NBDReplyChunkIter {
943     int ret;
944     int request_ret;
945     Error *err;
946     bool done, only_structured;
947 } NBDReplyChunkIter;
948 
949 static void nbd_iter_channel_error(NBDReplyChunkIter *iter,
950                                    int ret, Error **local_err)
951 {
952     assert(local_err && *local_err);
953     assert(ret < 0);
954 
955     if (!iter->ret) {
956         iter->ret = ret;
957         error_propagate(&iter->err, *local_err);
958     } else {
959         error_free(*local_err);
960     }
961 
962     *local_err = NULL;
963 }
964 
965 static void nbd_iter_request_error(NBDReplyChunkIter *iter, int ret)
966 {
967     assert(ret < 0);
968 
969     if (!iter->request_ret) {
970         iter->request_ret = ret;
971     }
972 }
973 
974 /*
975  * NBD_FOREACH_REPLY_CHUNK
976  * The pointer stored in @payload requires g_free() to free it.
977  */
978 #define NBD_FOREACH_REPLY_CHUNK(s, iter, cookie, structured, \
979                                 qiov, reply, payload) \
980     for (iter = (NBDReplyChunkIter) { .only_structured = structured }; \
981          nbd_reply_chunk_iter_receive(s, &iter, cookie, qiov, reply, payload);)
982 
983 /*
984  * nbd_reply_chunk_iter_receive
985  * The pointer stored in @payload requires g_free() to free it.
986  */
987 static bool coroutine_fn nbd_reply_chunk_iter_receive(BDRVNBDState *s,
988                                                       NBDReplyChunkIter *iter,
989                                                       uint64_t cookie,
990                                                       QEMUIOVector *qiov,
991                                                       NBDReply *reply,
992                                                       void **payload)
993 {
994     int ret, request_ret;
995     NBDReply local_reply;
996     NBDStructuredReplyChunk *chunk;
997     Error *local_err = NULL;
998 
999     if (iter->done) {
1000         /* Previous iteration was last. */
1001         goto break_loop;
1002     }
1003 
1004     if (reply == NULL) {
1005         reply = &local_reply;
1006     }
1007 
1008     ret = nbd_co_receive_one_chunk(s, cookie, iter->only_structured,
1009                                    &request_ret, qiov, reply, payload,
1010                                    &local_err);
1011     if (ret < 0) {
1012         nbd_iter_channel_error(iter, ret, &local_err);
1013     } else if (request_ret < 0) {
1014         nbd_iter_request_error(iter, request_ret);
1015     }
1016 
1017     /* Do not execute the body of NBD_FOREACH_REPLY_CHUNK for simple reply. */
1018     if (nbd_reply_is_simple(reply) || iter->ret < 0) {
1019         goto break_loop;
1020     }
1021 
1022     chunk = &reply->structured;
1023     iter->only_structured = true;
1024 
1025     if (chunk->type == NBD_REPLY_TYPE_NONE) {
1026         /* NBD_REPLY_FLAG_DONE is already checked in nbd_co_receive_one_chunk */
1027         assert(chunk->flags & NBD_REPLY_FLAG_DONE);
1028         goto break_loop;
1029     }
1030 
1031     if (chunk->flags & NBD_REPLY_FLAG_DONE) {
1032         /* This iteration is last. */
1033         iter->done = true;
1034     }
1035 
1036     /* Execute the loop body */
1037     return true;
1038 
1039 break_loop:
1040     qemu_mutex_lock(&s->requests_lock);
1041     s->requests[COOKIE_TO_INDEX(cookie)].coroutine = NULL;
1042     s->in_flight--;
1043     qemu_co_queue_next(&s->free_sema);
1044     qemu_mutex_unlock(&s->requests_lock);
1045 
1046     return false;
1047 }
1048 
1049 static int coroutine_fn
1050 nbd_co_receive_return_code(BDRVNBDState *s, uint64_t cookie,
1051                            int *request_ret, Error **errp)
1052 {
1053     NBDReplyChunkIter iter;
1054 
1055     NBD_FOREACH_REPLY_CHUNK(s, iter, cookie, false, NULL, NULL, NULL) {
1056         /* nbd_reply_chunk_iter_receive does all the work */
1057     }
1058 
1059     error_propagate(errp, iter.err);
1060     *request_ret = iter.request_ret;
1061     return iter.ret;
1062 }
1063 
1064 static int coroutine_fn
1065 nbd_co_receive_cmdread_reply(BDRVNBDState *s, uint64_t cookie,
1066                              uint64_t offset, QEMUIOVector *qiov,
1067                              int *request_ret, Error **errp)
1068 {
1069     NBDReplyChunkIter iter;
1070     NBDReply reply;
1071     void *payload = NULL;
1072     Error *local_err = NULL;
1073 
1074     NBD_FOREACH_REPLY_CHUNK(s, iter, cookie, s->info.structured_reply,
1075                             qiov, &reply, &payload)
1076     {
1077         int ret;
1078         NBDStructuredReplyChunk *chunk = &reply.structured;
1079 
1080         assert(nbd_reply_is_structured(&reply));
1081 
1082         switch (chunk->type) {
1083         case NBD_REPLY_TYPE_OFFSET_DATA:
1084             /*
1085              * special cased in nbd_co_receive_one_chunk, data is already
1086              * in qiov
1087              */
1088             break;
1089         case NBD_REPLY_TYPE_OFFSET_HOLE:
1090             ret = nbd_parse_offset_hole_payload(s, &reply.structured, payload,
1091                                                 offset, qiov, &local_err);
1092             if (ret < 0) {
1093                 nbd_channel_error(s, ret);
1094                 nbd_iter_channel_error(&iter, ret, &local_err);
1095             }
1096             break;
1097         default:
1098             if (!nbd_reply_type_is_error(chunk->type)) {
1099                 /* not allowed reply type */
1100                 nbd_channel_error(s, -EINVAL);
1101                 error_setg(&local_err,
1102                            "Unexpected reply type: %d (%s) for CMD_READ",
1103                            chunk->type, nbd_reply_type_lookup(chunk->type));
1104                 nbd_iter_channel_error(&iter, -EINVAL, &local_err);
1105             }
1106         }
1107 
1108         g_free(payload);
1109         payload = NULL;
1110     }
1111 
1112     error_propagate(errp, iter.err);
1113     *request_ret = iter.request_ret;
1114     return iter.ret;
1115 }
1116 
1117 static int coroutine_fn
1118 nbd_co_receive_blockstatus_reply(BDRVNBDState *s, uint64_t cookie,
1119                                  uint64_t length, NBDExtent *extent,
1120                                  int *request_ret, Error **errp)
1121 {
1122     NBDReplyChunkIter iter;
1123     NBDReply reply;
1124     void *payload = NULL;
1125     Error *local_err = NULL;
1126     bool received = false;
1127 
1128     assert(!extent->length);
1129     NBD_FOREACH_REPLY_CHUNK(s, iter, cookie, false, NULL, &reply, &payload) {
1130         int ret;
1131         NBDStructuredReplyChunk *chunk = &reply.structured;
1132 
1133         assert(nbd_reply_is_structured(&reply));
1134 
1135         switch (chunk->type) {
1136         case NBD_REPLY_TYPE_BLOCK_STATUS:
1137             if (received) {
1138                 nbd_channel_error(s, -EINVAL);
1139                 error_setg(&local_err, "Several BLOCK_STATUS chunks in reply");
1140                 nbd_iter_channel_error(&iter, -EINVAL, &local_err);
1141             }
1142             received = true;
1143 
1144             ret = nbd_parse_blockstatus_payload(s, &reply.structured,
1145                                                 payload, length, extent,
1146                                                 &local_err);
1147             if (ret < 0) {
1148                 nbd_channel_error(s, ret);
1149                 nbd_iter_channel_error(&iter, ret, &local_err);
1150             }
1151             break;
1152         default:
1153             if (!nbd_reply_type_is_error(chunk->type)) {
1154                 nbd_channel_error(s, -EINVAL);
1155                 error_setg(&local_err,
1156                            "Unexpected reply type: %d (%s) "
1157                            "for CMD_BLOCK_STATUS",
1158                            chunk->type, nbd_reply_type_lookup(chunk->type));
1159                 nbd_iter_channel_error(&iter, -EINVAL, &local_err);
1160             }
1161         }
1162 
1163         g_free(payload);
1164         payload = NULL;
1165     }
1166 
1167     if (!extent->length && !iter.request_ret) {
1168         error_setg(&local_err, "Server did not reply with any status extents");
1169         nbd_iter_channel_error(&iter, -EIO, &local_err);
1170     }
1171 
1172     error_propagate(errp, iter.err);
1173     *request_ret = iter.request_ret;
1174     return iter.ret;
1175 }
1176 
1177 static int coroutine_fn GRAPH_RDLOCK
1178 nbd_co_request(BlockDriverState *bs, NBDRequest *request,
1179                QEMUIOVector *write_qiov)
1180 {
1181     int ret, request_ret;
1182     Error *local_err = NULL;
1183     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1184 
1185     assert(request->type != NBD_CMD_READ);
1186     if (write_qiov) {
1187         assert(request->type == NBD_CMD_WRITE);
1188         assert(request->len == iov_size(write_qiov->iov, write_qiov->niov));
1189     } else {
1190         assert(request->type != NBD_CMD_WRITE);
1191     }
1192 
1193     do {
1194         ret = nbd_co_send_request(bs, request, write_qiov);
1195         if (ret < 0) {
1196             continue;
1197         }
1198 
1199         ret = nbd_co_receive_return_code(s, request->cookie,
1200                                          &request_ret, &local_err);
1201         if (local_err) {
1202             trace_nbd_co_request_fail(request->from, request->len,
1203                                       request->cookie, request->flags,
1204                                       request->type,
1205                                       nbd_cmd_lookup(request->type),
1206                                       ret, error_get_pretty(local_err));
1207             error_free(local_err);
1208             local_err = NULL;
1209         }
1210     } while (ret < 0 && nbd_client_will_reconnect(s));
1211 
1212     return ret ? ret : request_ret;
1213 }
1214 
1215 static int coroutine_fn GRAPH_RDLOCK
1216 nbd_client_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes,
1217                      QEMUIOVector *qiov, BdrvRequestFlags flags)
1218 {
1219     int ret, request_ret;
1220     Error *local_err = NULL;
1221     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1222     NBDRequest request = {
1223         .type = NBD_CMD_READ,
1224         .from = offset,
1225         .len = bytes,
1226     };
1227 
1228     assert(bytes <= NBD_MAX_BUFFER_SIZE);
1229 
1230     if (!bytes) {
1231         return 0;
1232     }
1233     /*
1234      * Work around the fact that the block layer doesn't do
1235      * byte-accurate sizing yet - if the read exceeds the server's
1236      * advertised size because the block layer rounded size up, then
1237      * truncate the request to the server and tail-pad with zero.
1238      */
1239     if (offset >= s->info.size) {
1240         assert(bytes < BDRV_SECTOR_SIZE);
1241         qemu_iovec_memset(qiov, 0, 0, bytes);
1242         return 0;
1243     }
1244     if (offset + bytes > s->info.size) {
1245         uint64_t slop = offset + bytes - s->info.size;
1246 
1247         assert(slop < BDRV_SECTOR_SIZE);
1248         qemu_iovec_memset(qiov, bytes - slop, 0, slop);
1249         request.len -= slop;
1250     }
1251 
1252     do {
1253         ret = nbd_co_send_request(bs, &request, NULL);
1254         if (ret < 0) {
1255             continue;
1256         }
1257 
1258         ret = nbd_co_receive_cmdread_reply(s, request.cookie, offset, qiov,
1259                                            &request_ret, &local_err);
1260         if (local_err) {
1261             trace_nbd_co_request_fail(request.from, request.len, request.cookie,
1262                                       request.flags, request.type,
1263                                       nbd_cmd_lookup(request.type),
1264                                       ret, error_get_pretty(local_err));
1265             error_free(local_err);
1266             local_err = NULL;
1267         }
1268     } while (ret < 0 && nbd_client_will_reconnect(s));
1269 
1270     return ret ? ret : request_ret;
1271 }
1272 
1273 static int coroutine_fn GRAPH_RDLOCK
1274 nbd_client_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes,
1275                       QEMUIOVector *qiov, BdrvRequestFlags flags)
1276 {
1277     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1278     NBDRequest request = {
1279         .type = NBD_CMD_WRITE,
1280         .from = offset,
1281         .len = bytes,
1282     };
1283 
1284     assert(!(s->info.flags & NBD_FLAG_READ_ONLY));
1285     if (flags & BDRV_REQ_FUA) {
1286         assert(s->info.flags & NBD_FLAG_SEND_FUA);
1287         request.flags |= NBD_CMD_FLAG_FUA;
1288     }
1289 
1290     assert(bytes <= NBD_MAX_BUFFER_SIZE);
1291 
1292     if (!bytes) {
1293         return 0;
1294     }
1295     return nbd_co_request(bs, &request, qiov);
1296 }
1297 
1298 static int coroutine_fn GRAPH_RDLOCK
1299 nbd_client_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes,
1300                             BdrvRequestFlags flags)
1301 {
1302     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1303     NBDRequest request = {
1304         .type = NBD_CMD_WRITE_ZEROES,
1305         .from = offset,
1306         .len = bytes,  /* .len is uint32_t actually */
1307     };
1308 
1309     assert(bytes <= UINT32_MAX); /* rely on max_pwrite_zeroes */
1310 
1311     assert(!(s->info.flags & NBD_FLAG_READ_ONLY));
1312     if (!(s->info.flags & NBD_FLAG_SEND_WRITE_ZEROES)) {
1313         return -ENOTSUP;
1314     }
1315 
1316     if (flags & BDRV_REQ_FUA) {
1317         assert(s->info.flags & NBD_FLAG_SEND_FUA);
1318         request.flags |= NBD_CMD_FLAG_FUA;
1319     }
1320     if (!(flags & BDRV_REQ_MAY_UNMAP)) {
1321         request.flags |= NBD_CMD_FLAG_NO_HOLE;
1322     }
1323     if (flags & BDRV_REQ_NO_FALLBACK) {
1324         assert(s->info.flags & NBD_FLAG_SEND_FAST_ZERO);
1325         request.flags |= NBD_CMD_FLAG_FAST_ZERO;
1326     }
1327 
1328     if (!bytes) {
1329         return 0;
1330     }
1331     return nbd_co_request(bs, &request, NULL);
1332 }
1333 
1334 static int coroutine_fn GRAPH_RDLOCK nbd_client_co_flush(BlockDriverState *bs)
1335 {
1336     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1337     NBDRequest request = { .type = NBD_CMD_FLUSH };
1338 
1339     if (!(s->info.flags & NBD_FLAG_SEND_FLUSH)) {
1340         return 0;
1341     }
1342 
1343     request.from = 0;
1344     request.len = 0;
1345 
1346     return nbd_co_request(bs, &request, NULL);
1347 }
1348 
1349 static int coroutine_fn GRAPH_RDLOCK
1350 nbd_client_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes)
1351 {
1352     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1353     NBDRequest request = {
1354         .type = NBD_CMD_TRIM,
1355         .from = offset,
1356         .len = bytes, /* len is uint32_t */
1357     };
1358 
1359     assert(bytes <= UINT32_MAX); /* rely on max_pdiscard */
1360 
1361     assert(!(s->info.flags & NBD_FLAG_READ_ONLY));
1362     if (!(s->info.flags & NBD_FLAG_SEND_TRIM) || !bytes) {
1363         return 0;
1364     }
1365 
1366     return nbd_co_request(bs, &request, NULL);
1367 }
1368 
1369 static int coroutine_fn GRAPH_RDLOCK nbd_client_co_block_status(
1370         BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes,
1371         int64_t *pnum, int64_t *map, BlockDriverState **file)
1372 {
1373     int ret, request_ret;
1374     NBDExtent extent = { 0 };
1375     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1376     Error *local_err = NULL;
1377 
1378     NBDRequest request = {
1379         .type = NBD_CMD_BLOCK_STATUS,
1380         .from = offset,
1381         .len = MIN(QEMU_ALIGN_DOWN(INT_MAX, bs->bl.request_alignment),
1382                    MIN(bytes, s->info.size - offset)),
1383         .flags = NBD_CMD_FLAG_REQ_ONE,
1384     };
1385 
1386     if (!s->info.base_allocation) {
1387         *pnum = bytes;
1388         *map = offset;
1389         *file = bs;
1390         return BDRV_BLOCK_DATA | BDRV_BLOCK_OFFSET_VALID;
1391     }
1392 
1393     /*
1394      * Work around the fact that the block layer doesn't do
1395      * byte-accurate sizing yet - if the status request exceeds the
1396      * server's advertised size because the block layer rounded size
1397      * up, we truncated the request to the server (above), or are
1398      * called on just the hole.
1399      */
1400     if (offset >= s->info.size) {
1401         *pnum = bytes;
1402         assert(bytes < BDRV_SECTOR_SIZE);
1403         /* Intentionally don't report offset_valid for the hole */
1404         return BDRV_BLOCK_ZERO;
1405     }
1406 
1407     if (s->info.min_block) {
1408         assert(QEMU_IS_ALIGNED(request.len, s->info.min_block));
1409     }
1410     do {
1411         ret = nbd_co_send_request(bs, &request, NULL);
1412         if (ret < 0) {
1413             continue;
1414         }
1415 
1416         ret = nbd_co_receive_blockstatus_reply(s, request.cookie, bytes,
1417                                                &extent, &request_ret,
1418                                                &local_err);
1419         if (local_err) {
1420             trace_nbd_co_request_fail(request.from, request.len, request.cookie,
1421                                       request.flags, request.type,
1422                                       nbd_cmd_lookup(request.type),
1423                                       ret, error_get_pretty(local_err));
1424             error_free(local_err);
1425             local_err = NULL;
1426         }
1427     } while (ret < 0 && nbd_client_will_reconnect(s));
1428 
1429     if (ret < 0 || request_ret < 0) {
1430         return ret ? ret : request_ret;
1431     }
1432 
1433     assert(extent.length);
1434     *pnum = extent.length;
1435     *map = offset;
1436     *file = bs;
1437     return (extent.flags & NBD_STATE_HOLE ? 0 : BDRV_BLOCK_DATA) |
1438         (extent.flags & NBD_STATE_ZERO ? BDRV_BLOCK_ZERO : 0) |
1439         BDRV_BLOCK_OFFSET_VALID;
1440 }
1441 
1442 static int nbd_client_reopen_prepare(BDRVReopenState *state,
1443                                      BlockReopenQueue *queue, Error **errp)
1444 {
1445     BDRVNBDState *s = (BDRVNBDState *)state->bs->opaque;
1446 
1447     if ((state->flags & BDRV_O_RDWR) && (s->info.flags & NBD_FLAG_READ_ONLY)) {
1448         error_setg(errp, "Can't reopen read-only NBD mount as read/write");
1449         return -EACCES;
1450     }
1451     return 0;
1452 }
1453 
1454 static void nbd_yank(void *opaque)
1455 {
1456     BlockDriverState *bs = opaque;
1457     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1458 
1459     QEMU_LOCK_GUARD(&s->requests_lock);
1460     qio_channel_shutdown(s->ioc, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
1461     s->state = NBD_CLIENT_QUIT;
1462 }
1463 
1464 static void nbd_client_close(BlockDriverState *bs)
1465 {
1466     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1467     NBDRequest request = { .type = NBD_CMD_DISC };
1468 
1469     if (s->ioc) {
1470         nbd_send_request(s->ioc, &request);
1471     }
1472 
1473     nbd_teardown_connection(bs);
1474 }
1475 
1476 
1477 /*
1478  * Parse nbd_open options
1479  */
1480 
1481 static int nbd_parse_uri(const char *filename, QDict *options)
1482 {
1483     URI *uri;
1484     const char *p;
1485     QueryParams *qp = NULL;
1486     int ret = 0;
1487     bool is_unix;
1488 
1489     uri = uri_parse(filename);
1490     if (!uri) {
1491         return -EINVAL;
1492     }
1493 
1494     /* transport */
1495     if (!g_strcmp0(uri->scheme, "nbd")) {
1496         is_unix = false;
1497     } else if (!g_strcmp0(uri->scheme, "nbd+tcp")) {
1498         is_unix = false;
1499     } else if (!g_strcmp0(uri->scheme, "nbd+unix")) {
1500         is_unix = true;
1501     } else {
1502         ret = -EINVAL;
1503         goto out;
1504     }
1505 
1506     p = uri->path ? uri->path : "";
1507     if (p[0] == '/') {
1508         p++;
1509     }
1510     if (p[0]) {
1511         qdict_put_str(options, "export", p);
1512     }
1513 
1514     qp = query_params_parse(uri->query);
1515     if (qp->n > 1 || (is_unix && !qp->n) || (!is_unix && qp->n)) {
1516         ret = -EINVAL;
1517         goto out;
1518     }
1519 
1520     if (is_unix) {
1521         /* nbd+unix:///export?socket=path */
1522         if (uri->server || uri->port || strcmp(qp->p[0].name, "socket")) {
1523             ret = -EINVAL;
1524             goto out;
1525         }
1526         qdict_put_str(options, "server.type", "unix");
1527         qdict_put_str(options, "server.path", qp->p[0].value);
1528     } else {
1529         QString *host;
1530         char *port_str;
1531 
1532         /* nbd[+tcp]://host[:port]/export */
1533         if (!uri->server) {
1534             ret = -EINVAL;
1535             goto out;
1536         }
1537 
1538         /* strip braces from literal IPv6 address */
1539         if (uri->server[0] == '[') {
1540             host = qstring_from_substr(uri->server, 1,
1541                                        strlen(uri->server) - 1);
1542         } else {
1543             host = qstring_from_str(uri->server);
1544         }
1545 
1546         qdict_put_str(options, "server.type", "inet");
1547         qdict_put(options, "server.host", host);
1548 
1549         port_str = g_strdup_printf("%d", uri->port ?: NBD_DEFAULT_PORT);
1550         qdict_put_str(options, "server.port", port_str);
1551         g_free(port_str);
1552     }
1553 
1554 out:
1555     if (qp) {
1556         query_params_free(qp);
1557     }
1558     uri_free(uri);
1559     return ret;
1560 }
1561 
1562 static bool nbd_has_filename_options_conflict(QDict *options, Error **errp)
1563 {
1564     const QDictEntry *e;
1565 
1566     for (e = qdict_first(options); e; e = qdict_next(options, e)) {
1567         if (!strcmp(e->key, "host") ||
1568             !strcmp(e->key, "port") ||
1569             !strcmp(e->key, "path") ||
1570             !strcmp(e->key, "export") ||
1571             strstart(e->key, "server.", NULL))
1572         {
1573             error_setg(errp, "Option '%s' cannot be used with a file name",
1574                        e->key);
1575             return true;
1576         }
1577     }
1578 
1579     return false;
1580 }
1581 
1582 static void nbd_parse_filename(const char *filename, QDict *options,
1583                                Error **errp)
1584 {
1585     g_autofree char *file = NULL;
1586     char *export_name;
1587     const char *host_spec;
1588     const char *unixpath;
1589 
1590     if (nbd_has_filename_options_conflict(options, errp)) {
1591         return;
1592     }
1593 
1594     if (strstr(filename, "://")) {
1595         int ret = nbd_parse_uri(filename, options);
1596         if (ret < 0) {
1597             error_setg(errp, "No valid URL specified");
1598         }
1599         return;
1600     }
1601 
1602     file = g_strdup(filename);
1603 
1604     export_name = strstr(file, EN_OPTSTR);
1605     if (export_name) {
1606         if (export_name[strlen(EN_OPTSTR)] == 0) {
1607             return;
1608         }
1609         export_name[0] = 0; /* truncate 'file' */
1610         export_name += strlen(EN_OPTSTR);
1611 
1612         qdict_put_str(options, "export", export_name);
1613     }
1614 
1615     /* extract the host_spec - fail if it's not nbd:... */
1616     if (!strstart(file, "nbd:", &host_spec)) {
1617         error_setg(errp, "File name string for NBD must start with 'nbd:'");
1618         return;
1619     }
1620 
1621     if (!*host_spec) {
1622         return;
1623     }
1624 
1625     /* are we a UNIX or TCP socket? */
1626     if (strstart(host_spec, "unix:", &unixpath)) {
1627         qdict_put_str(options, "server.type", "unix");
1628         qdict_put_str(options, "server.path", unixpath);
1629     } else {
1630         InetSocketAddress *addr = g_new(InetSocketAddress, 1);
1631 
1632         if (inet_parse(addr, host_spec, errp)) {
1633             goto out_inet;
1634         }
1635 
1636         qdict_put_str(options, "server.type", "inet");
1637         qdict_put_str(options, "server.host", addr->host);
1638         qdict_put_str(options, "server.port", addr->port);
1639     out_inet:
1640         qapi_free_InetSocketAddress(addr);
1641     }
1642 }
1643 
1644 static bool nbd_process_legacy_socket_options(QDict *output_options,
1645                                               QemuOpts *legacy_opts,
1646                                               Error **errp)
1647 {
1648     const char *path = qemu_opt_get(legacy_opts, "path");
1649     const char *host = qemu_opt_get(legacy_opts, "host");
1650     const char *port = qemu_opt_get(legacy_opts, "port");
1651     const QDictEntry *e;
1652 
1653     if (!path && !host && !port) {
1654         return true;
1655     }
1656 
1657     for (e = qdict_first(output_options); e; e = qdict_next(output_options, e))
1658     {
1659         if (strstart(e->key, "server.", NULL)) {
1660             error_setg(errp, "Cannot use 'server' and path/host/port at the "
1661                        "same time");
1662             return false;
1663         }
1664     }
1665 
1666     if (path && host) {
1667         error_setg(errp, "path and host may not be used at the same time");
1668         return false;
1669     } else if (path) {
1670         if (port) {
1671             error_setg(errp, "port may not be used without host");
1672             return false;
1673         }
1674 
1675         qdict_put_str(output_options, "server.type", "unix");
1676         qdict_put_str(output_options, "server.path", path);
1677     } else if (host) {
1678         qdict_put_str(output_options, "server.type", "inet");
1679         qdict_put_str(output_options, "server.host", host);
1680         qdict_put_str(output_options, "server.port",
1681                       port ?: stringify(NBD_DEFAULT_PORT));
1682     }
1683 
1684     return true;
1685 }
1686 
1687 static SocketAddress *nbd_config(BDRVNBDState *s, QDict *options,
1688                                  Error **errp)
1689 {
1690     SocketAddress *saddr = NULL;
1691     QDict *addr = NULL;
1692     Visitor *iv = NULL;
1693 
1694     qdict_extract_subqdict(options, &addr, "server.");
1695     if (!qdict_size(addr)) {
1696         error_setg(errp, "NBD server address missing");
1697         goto done;
1698     }
1699 
1700     iv = qobject_input_visitor_new_flat_confused(addr, errp);
1701     if (!iv) {
1702         goto done;
1703     }
1704 
1705     if (!visit_type_SocketAddress(iv, NULL, &saddr, errp)) {
1706         goto done;
1707     }
1708 
1709     if (socket_address_parse_named_fd(saddr, errp) < 0) {
1710         qapi_free_SocketAddress(saddr);
1711         saddr = NULL;
1712         goto done;
1713     }
1714 
1715 done:
1716     qobject_unref(addr);
1717     visit_free(iv);
1718     return saddr;
1719 }
1720 
1721 static QCryptoTLSCreds *nbd_get_tls_creds(const char *id, Error **errp)
1722 {
1723     Object *obj;
1724     QCryptoTLSCreds *creds;
1725 
1726     obj = object_resolve_path_component(
1727         object_get_objects_root(), id);
1728     if (!obj) {
1729         error_setg(errp, "No TLS credentials with id '%s'",
1730                    id);
1731         return NULL;
1732     }
1733     creds = (QCryptoTLSCreds *)
1734         object_dynamic_cast(obj, TYPE_QCRYPTO_TLS_CREDS);
1735     if (!creds) {
1736         error_setg(errp, "Object with id '%s' is not TLS credentials",
1737                    id);
1738         return NULL;
1739     }
1740 
1741     if (!qcrypto_tls_creds_check_endpoint(creds,
1742                                           QCRYPTO_TLS_CREDS_ENDPOINT_CLIENT,
1743                                           errp)) {
1744         return NULL;
1745     }
1746     object_ref(obj);
1747     return creds;
1748 }
1749 
1750 
1751 static QemuOptsList nbd_runtime_opts = {
1752     .name = "nbd",
1753     .head = QTAILQ_HEAD_INITIALIZER(nbd_runtime_opts.head),
1754     .desc = {
1755         {
1756             .name = "host",
1757             .type = QEMU_OPT_STRING,
1758             .help = "TCP host to connect to",
1759         },
1760         {
1761             .name = "port",
1762             .type = QEMU_OPT_STRING,
1763             .help = "TCP port to connect to",
1764         },
1765         {
1766             .name = "path",
1767             .type = QEMU_OPT_STRING,
1768             .help = "Unix socket path to connect to",
1769         },
1770         {
1771             .name = "export",
1772             .type = QEMU_OPT_STRING,
1773             .help = "Name of the NBD export to open",
1774         },
1775         {
1776             .name = "tls-creds",
1777             .type = QEMU_OPT_STRING,
1778             .help = "ID of the TLS credentials to use",
1779         },
1780         {
1781             .name = "tls-hostname",
1782             .type = QEMU_OPT_STRING,
1783             .help = "Override hostname for validating TLS x509 certificate",
1784         },
1785         {
1786             .name = "x-dirty-bitmap",
1787             .type = QEMU_OPT_STRING,
1788             .help = "experimental: expose named dirty bitmap in place of "
1789                     "block status",
1790         },
1791         {
1792             .name = "reconnect-delay",
1793             .type = QEMU_OPT_NUMBER,
1794             .help = "On an unexpected disconnect, the nbd client tries to "
1795                     "connect again until succeeding or encountering a serious "
1796                     "error.  During the first @reconnect-delay seconds, all "
1797                     "requests are paused and will be rerun on a successful "
1798                     "reconnect. After that time, any delayed requests and all "
1799                     "future requests before a successful reconnect will "
1800                     "immediately fail. Default 0",
1801         },
1802         {
1803             .name = "open-timeout",
1804             .type = QEMU_OPT_NUMBER,
1805             .help = "In seconds. If zero, the nbd driver tries the connection "
1806                     "only once, and fails to open if the connection fails. "
1807                     "If non-zero, the nbd driver will repeat connection "
1808                     "attempts until successful or until @open-timeout seconds "
1809                     "have elapsed. Default 0",
1810         },
1811         { /* end of list */ }
1812     },
1813 };
1814 
1815 static int nbd_process_options(BlockDriverState *bs, QDict *options,
1816                                Error **errp)
1817 {
1818     BDRVNBDState *s = bs->opaque;
1819     QemuOpts *opts;
1820     int ret = -EINVAL;
1821 
1822     opts = qemu_opts_create(&nbd_runtime_opts, NULL, 0, &error_abort);
1823     if (!qemu_opts_absorb_qdict(opts, options, errp)) {
1824         goto error;
1825     }
1826 
1827     /* Translate @host, @port, and @path to a SocketAddress */
1828     if (!nbd_process_legacy_socket_options(options, opts, errp)) {
1829         goto error;
1830     }
1831 
1832     /* Pop the config into our state object. Exit if invalid. */
1833     s->saddr = nbd_config(s, options, errp);
1834     if (!s->saddr) {
1835         goto error;
1836     }
1837 
1838     s->export = g_strdup(qemu_opt_get(opts, "export"));
1839     if (s->export && strlen(s->export) > NBD_MAX_STRING_SIZE) {
1840         error_setg(errp, "export name too long to send to server");
1841         goto error;
1842     }
1843 
1844     s->tlscredsid = g_strdup(qemu_opt_get(opts, "tls-creds"));
1845     if (s->tlscredsid) {
1846         s->tlscreds = nbd_get_tls_creds(s->tlscredsid, errp);
1847         if (!s->tlscreds) {
1848             goto error;
1849         }
1850 
1851         s->tlshostname = g_strdup(qemu_opt_get(opts, "tls-hostname"));
1852         if (!s->tlshostname &&
1853             s->saddr->type == SOCKET_ADDRESS_TYPE_INET) {
1854             s->tlshostname = g_strdup(s->saddr->u.inet.host);
1855         }
1856     }
1857 
1858     s->x_dirty_bitmap = g_strdup(qemu_opt_get(opts, "x-dirty-bitmap"));
1859     if (s->x_dirty_bitmap && strlen(s->x_dirty_bitmap) > NBD_MAX_STRING_SIZE) {
1860         error_setg(errp, "x-dirty-bitmap query too long to send to server");
1861         goto error;
1862     }
1863 
1864     s->reconnect_delay = qemu_opt_get_number(opts, "reconnect-delay", 0);
1865     s->open_timeout = qemu_opt_get_number(opts, "open-timeout", 0);
1866 
1867     ret = 0;
1868 
1869  error:
1870     qemu_opts_del(opts);
1871     return ret;
1872 }
1873 
1874 static int nbd_open(BlockDriverState *bs, QDict *options, int flags,
1875                     Error **errp)
1876 {
1877     int ret;
1878     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1879 
1880     s->bs = bs;
1881     qemu_mutex_init(&s->requests_lock);
1882     qemu_co_queue_init(&s->free_sema);
1883     qemu_co_mutex_init(&s->send_mutex);
1884     qemu_co_mutex_init(&s->receive_mutex);
1885 
1886     if (!yank_register_instance(BLOCKDEV_YANK_INSTANCE(bs->node_name), errp)) {
1887         return -EEXIST;
1888     }
1889 
1890     ret = nbd_process_options(bs, options, errp);
1891     if (ret < 0) {
1892         goto fail;
1893     }
1894 
1895     s->conn = nbd_client_connection_new(s->saddr, true, s->export,
1896                                         s->x_dirty_bitmap, s->tlscreds,
1897                                         s->tlshostname);
1898 
1899     if (s->open_timeout) {
1900         nbd_client_connection_enable_retry(s->conn);
1901         open_timer_init(s, qemu_clock_get_ns(QEMU_CLOCK_REALTIME) +
1902                         s->open_timeout * NANOSECONDS_PER_SECOND);
1903     }
1904 
1905     s->state = NBD_CLIENT_CONNECTING_WAIT;
1906     ret = nbd_do_establish_connection(bs, true, errp);
1907     if (ret < 0) {
1908         goto fail;
1909     }
1910 
1911     /*
1912      * The connect attempt is done, so we no longer need this timer.
1913      * Delete it, because we do not want it to be around when this node
1914      * is drained or closed.
1915      */
1916     open_timer_del(s);
1917 
1918     nbd_client_connection_enable_retry(s->conn);
1919 
1920     return 0;
1921 
1922 fail:
1923     open_timer_del(s);
1924     nbd_clear_bdrvstate(bs);
1925     return ret;
1926 }
1927 
1928 static void nbd_refresh_limits(BlockDriverState *bs, Error **errp)
1929 {
1930     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
1931     uint32_t min = s->info.min_block;
1932     uint32_t max = MIN_NON_ZERO(NBD_MAX_BUFFER_SIZE, s->info.max_block);
1933 
1934     /*
1935      * If the server did not advertise an alignment:
1936      * - a size that is not sector-aligned implies that an alignment
1937      *   of 1 can be used to access those tail bytes
1938      * - advertisement of block status requires an alignment of 1, so
1939      *   that we don't violate block layer constraints that block
1940      *   status is always aligned (as we can't control whether the
1941      *   server will report sub-sector extents, such as a hole at EOF
1942      *   on an unaligned POSIX file)
1943      * - otherwise, assume the server is so old that we are safer avoiding
1944      *   sub-sector requests
1945      */
1946     if (!min) {
1947         min = (!QEMU_IS_ALIGNED(s->info.size, BDRV_SECTOR_SIZE) ||
1948                s->info.base_allocation) ? 1 : BDRV_SECTOR_SIZE;
1949     }
1950 
1951     bs->bl.request_alignment = min;
1952     bs->bl.max_pdiscard = QEMU_ALIGN_DOWN(INT_MAX, min);
1953     bs->bl.max_pwrite_zeroes = max;
1954     bs->bl.max_transfer = max;
1955 
1956     if (s->info.opt_block &&
1957         s->info.opt_block > bs->bl.opt_transfer) {
1958         bs->bl.opt_transfer = s->info.opt_block;
1959     }
1960 }
1961 
1962 static void nbd_close(BlockDriverState *bs)
1963 {
1964     nbd_client_close(bs);
1965     nbd_clear_bdrvstate(bs);
1966 }
1967 
1968 /*
1969  * NBD cannot truncate, but if the caller asks to truncate to the same size, or
1970  * to a smaller size with exact=false, there is no reason to fail the
1971  * operation.
1972  *
1973  * Preallocation mode is ignored since it does not seems useful to fail when
1974  * we never change anything.
1975  */
1976 static int coroutine_fn nbd_co_truncate(BlockDriverState *bs, int64_t offset,
1977                                         bool exact, PreallocMode prealloc,
1978                                         BdrvRequestFlags flags, Error **errp)
1979 {
1980     BDRVNBDState *s = bs->opaque;
1981 
1982     if (offset != s->info.size && exact) {
1983         error_setg(errp, "Cannot resize NBD nodes");
1984         return -ENOTSUP;
1985     }
1986 
1987     if (offset > s->info.size) {
1988         error_setg(errp, "Cannot grow NBD nodes");
1989         return -EINVAL;
1990     }
1991 
1992     return 0;
1993 }
1994 
1995 static int64_t coroutine_fn nbd_co_getlength(BlockDriverState *bs)
1996 {
1997     BDRVNBDState *s = bs->opaque;
1998 
1999     return s->info.size;
2000 }
2001 
2002 static void nbd_refresh_filename(BlockDriverState *bs)
2003 {
2004     BDRVNBDState *s = bs->opaque;
2005     const char *host = NULL, *port = NULL, *path = NULL;
2006     size_t len = 0;
2007 
2008     if (s->saddr->type == SOCKET_ADDRESS_TYPE_INET) {
2009         const InetSocketAddress *inet = &s->saddr->u.inet;
2010         if (!inet->has_ipv4 && !inet->has_ipv6 && !inet->has_to) {
2011             host = inet->host;
2012             port = inet->port;
2013         }
2014     } else if (s->saddr->type == SOCKET_ADDRESS_TYPE_UNIX) {
2015         path = s->saddr->u.q_unix.path;
2016     } /* else can't represent as pseudo-filename */
2017 
2018     if (path && s->export) {
2019         len = snprintf(bs->exact_filename, sizeof(bs->exact_filename),
2020                        "nbd+unix:///%s?socket=%s", s->export, path);
2021     } else if (path && !s->export) {
2022         len = snprintf(bs->exact_filename, sizeof(bs->exact_filename),
2023                        "nbd+unix://?socket=%s", path);
2024     } else if (host && s->export) {
2025         len = snprintf(bs->exact_filename, sizeof(bs->exact_filename),
2026                        "nbd://%s:%s/%s", host, port, s->export);
2027     } else if (host && !s->export) {
2028         len = snprintf(bs->exact_filename, sizeof(bs->exact_filename),
2029                        "nbd://%s:%s", host, port);
2030     }
2031     if (len >= sizeof(bs->exact_filename)) {
2032         /* Name is too long to represent exactly, so leave it empty. */
2033         bs->exact_filename[0] = '\0';
2034     }
2035 }
2036 
2037 static char *nbd_dirname(BlockDriverState *bs, Error **errp)
2038 {
2039     /* The generic bdrv_dirname() implementation is able to work out some
2040      * directory name for NBD nodes, but that would be wrong. So far there is no
2041      * specification for how "export paths" would work, so NBD does not have
2042      * directory names. */
2043     error_setg(errp, "Cannot generate a base directory for NBD nodes");
2044     return NULL;
2045 }
2046 
2047 static const char *const nbd_strong_runtime_opts[] = {
2048     "path",
2049     "host",
2050     "port",
2051     "export",
2052     "tls-creds",
2053     "tls-hostname",
2054     "server.",
2055 
2056     NULL
2057 };
2058 
2059 static void nbd_cancel_in_flight(BlockDriverState *bs)
2060 {
2061     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
2062 
2063     reconnect_delay_timer_del(s);
2064 
2065     qemu_mutex_lock(&s->requests_lock);
2066     if (s->state == NBD_CLIENT_CONNECTING_WAIT) {
2067         s->state = NBD_CLIENT_CONNECTING_NOWAIT;
2068     }
2069     qemu_mutex_unlock(&s->requests_lock);
2070 
2071     nbd_co_establish_connection_cancel(s->conn);
2072 }
2073 
2074 static void nbd_attach_aio_context(BlockDriverState *bs,
2075                                    AioContext *new_context)
2076 {
2077     BDRVNBDState *s = bs->opaque;
2078 
2079     /* The open_timer is used only during nbd_open() */
2080     assert(!s->open_timer);
2081 
2082     /*
2083      * The reconnect_delay_timer is scheduled in I/O paths when the
2084      * connection is lost, to cancel the reconnection attempt after a
2085      * given time.  Once this attempt is done (successfully or not),
2086      * nbd_reconnect_attempt() ensures the timer is deleted before the
2087      * respective I/O request is resumed.
2088      * Since the AioContext can only be changed when a node is drained,
2089      * the reconnect_delay_timer cannot be active here.
2090      */
2091     assert(!s->reconnect_delay_timer);
2092 
2093     if (s->ioc) {
2094         qio_channel_attach_aio_context(s->ioc, new_context);
2095     }
2096 }
2097 
2098 static void nbd_detach_aio_context(BlockDriverState *bs)
2099 {
2100     BDRVNBDState *s = bs->opaque;
2101 
2102     assert(!s->open_timer);
2103     assert(!s->reconnect_delay_timer);
2104 
2105     if (s->ioc) {
2106         qio_channel_detach_aio_context(s->ioc);
2107     }
2108 }
2109 
2110 static BlockDriver bdrv_nbd = {
2111     .format_name                = "nbd",
2112     .protocol_name              = "nbd",
2113     .instance_size              = sizeof(BDRVNBDState),
2114     .bdrv_parse_filename        = nbd_parse_filename,
2115     .bdrv_co_create_opts        = bdrv_co_create_opts_simple,
2116     .create_opts                = &bdrv_create_opts_simple,
2117     .bdrv_file_open             = nbd_open,
2118     .bdrv_reopen_prepare        = nbd_client_reopen_prepare,
2119     .bdrv_co_preadv             = nbd_client_co_preadv,
2120     .bdrv_co_pwritev            = nbd_client_co_pwritev,
2121     .bdrv_co_pwrite_zeroes      = nbd_client_co_pwrite_zeroes,
2122     .bdrv_close                 = nbd_close,
2123     .bdrv_co_flush_to_os        = nbd_client_co_flush,
2124     .bdrv_co_pdiscard           = nbd_client_co_pdiscard,
2125     .bdrv_refresh_limits        = nbd_refresh_limits,
2126     .bdrv_co_truncate           = nbd_co_truncate,
2127     .bdrv_co_getlength          = nbd_co_getlength,
2128     .bdrv_refresh_filename      = nbd_refresh_filename,
2129     .bdrv_co_block_status       = nbd_client_co_block_status,
2130     .bdrv_dirname               = nbd_dirname,
2131     .strong_runtime_opts        = nbd_strong_runtime_opts,
2132     .bdrv_cancel_in_flight      = nbd_cancel_in_flight,
2133 
2134     .bdrv_attach_aio_context    = nbd_attach_aio_context,
2135     .bdrv_detach_aio_context    = nbd_detach_aio_context,
2136 };
2137 
2138 static BlockDriver bdrv_nbd_tcp = {
2139     .format_name                = "nbd",
2140     .protocol_name              = "nbd+tcp",
2141     .instance_size              = sizeof(BDRVNBDState),
2142     .bdrv_parse_filename        = nbd_parse_filename,
2143     .bdrv_co_create_opts        = bdrv_co_create_opts_simple,
2144     .create_opts                = &bdrv_create_opts_simple,
2145     .bdrv_file_open             = nbd_open,
2146     .bdrv_reopen_prepare        = nbd_client_reopen_prepare,
2147     .bdrv_co_preadv             = nbd_client_co_preadv,
2148     .bdrv_co_pwritev            = nbd_client_co_pwritev,
2149     .bdrv_co_pwrite_zeroes      = nbd_client_co_pwrite_zeroes,
2150     .bdrv_close                 = nbd_close,
2151     .bdrv_co_flush_to_os        = nbd_client_co_flush,
2152     .bdrv_co_pdiscard           = nbd_client_co_pdiscard,
2153     .bdrv_refresh_limits        = nbd_refresh_limits,
2154     .bdrv_co_truncate           = nbd_co_truncate,
2155     .bdrv_co_getlength          = nbd_co_getlength,
2156     .bdrv_refresh_filename      = nbd_refresh_filename,
2157     .bdrv_co_block_status       = nbd_client_co_block_status,
2158     .bdrv_dirname               = nbd_dirname,
2159     .strong_runtime_opts        = nbd_strong_runtime_opts,
2160     .bdrv_cancel_in_flight      = nbd_cancel_in_flight,
2161 
2162     .bdrv_attach_aio_context    = nbd_attach_aio_context,
2163     .bdrv_detach_aio_context    = nbd_detach_aio_context,
2164 };
2165 
2166 static BlockDriver bdrv_nbd_unix = {
2167     .format_name                = "nbd",
2168     .protocol_name              = "nbd+unix",
2169     .instance_size              = sizeof(BDRVNBDState),
2170     .bdrv_parse_filename        = nbd_parse_filename,
2171     .bdrv_co_create_opts        = bdrv_co_create_opts_simple,
2172     .create_opts                = &bdrv_create_opts_simple,
2173     .bdrv_file_open             = nbd_open,
2174     .bdrv_reopen_prepare        = nbd_client_reopen_prepare,
2175     .bdrv_co_preadv             = nbd_client_co_preadv,
2176     .bdrv_co_pwritev            = nbd_client_co_pwritev,
2177     .bdrv_co_pwrite_zeroes      = nbd_client_co_pwrite_zeroes,
2178     .bdrv_close                 = nbd_close,
2179     .bdrv_co_flush_to_os        = nbd_client_co_flush,
2180     .bdrv_co_pdiscard           = nbd_client_co_pdiscard,
2181     .bdrv_refresh_limits        = nbd_refresh_limits,
2182     .bdrv_co_truncate           = nbd_co_truncate,
2183     .bdrv_co_getlength          = nbd_co_getlength,
2184     .bdrv_refresh_filename      = nbd_refresh_filename,
2185     .bdrv_co_block_status       = nbd_client_co_block_status,
2186     .bdrv_dirname               = nbd_dirname,
2187     .strong_runtime_opts        = nbd_strong_runtime_opts,
2188     .bdrv_cancel_in_flight      = nbd_cancel_in_flight,
2189 
2190     .bdrv_attach_aio_context    = nbd_attach_aio_context,
2191     .bdrv_detach_aio_context    = nbd_detach_aio_context,
2192 };
2193 
2194 static void bdrv_nbd_init(void)
2195 {
2196     bdrv_register(&bdrv_nbd);
2197     bdrv_register(&bdrv_nbd_tcp);
2198     bdrv_register(&bdrv_nbd_unix);
2199 }
2200 
2201 block_init(bdrv_nbd_init);
2202