xref: /openbmc/qemu/migration/multifd.c (revision 136cb9cc)
1 /*
2  * Multifd common code
3  *
4  * Copyright (c) 2019-2020 Red Hat Inc
5  *
6  * Authors:
7  *  Juan Quintela <quintela@redhat.com>
8  *
9  * This work is licensed under the terms of the GNU GPL, version 2 or later.
10  * See the COPYING file in the top-level directory.
11  */
12 
13 #include "qemu/osdep.h"
14 #include "qemu/rcu.h"
15 #include "exec/target_page.h"
16 #include "sysemu/sysemu.h"
17 #include "exec/ramblock.h"
18 #include "qemu/error-report.h"
19 #include "qapi/error.h"
20 #include "ram.h"
21 #include "migration.h"
22 #include "migration-stats.h"
23 #include "socket.h"
24 #include "tls.h"
25 #include "qemu-file.h"
26 #include "trace.h"
27 #include "multifd.h"
28 #include "threadinfo.h"
29 #include "options.h"
30 #include "qemu/yank.h"
31 #include "io/channel-socket.h"
32 #include "yank_functions.h"
33 
34 /* Multiple fd's */
35 
36 #define MULTIFD_MAGIC 0x11223344U
37 #define MULTIFD_VERSION 1
38 
39 typedef struct {
40     uint32_t magic;
41     uint32_t version;
42     unsigned char uuid[16]; /* QemuUUID */
43     uint8_t id;
44     uint8_t unused1[7];     /* Reserved for future use */
45     uint64_t unused2[4];    /* Reserved for future use */
46 } __attribute__((packed)) MultiFDInit_t;
47 
48 /* Multifd without compression */
49 
50 /**
51  * nocomp_send_setup: setup send side
52  *
53  * For no compression this function does nothing.
54  *
55  * Returns 0 for success or -1 for error
56  *
57  * @p: Params for the channel that we are using
58  * @errp: pointer to an error
59  */
60 static int nocomp_send_setup(MultiFDSendParams *p, Error **errp)
61 {
62     return 0;
63 }
64 
65 /**
66  * nocomp_send_cleanup: cleanup send side
67  *
68  * For no compression this function does nothing.
69  *
70  * @p: Params for the channel that we are using
71  * @errp: pointer to an error
72  */
73 static void nocomp_send_cleanup(MultiFDSendParams *p, Error **errp)
74 {
75     return;
76 }
77 
78 /**
79  * nocomp_send_prepare: prepare date to be able to send
80  *
81  * For no compression we just have to calculate the size of the
82  * packet.
83  *
84  * Returns 0 for success or -1 for error
85  *
86  * @p: Params for the channel that we are using
87  * @errp: pointer to an error
88  */
89 static int nocomp_send_prepare(MultiFDSendParams *p, Error **errp)
90 {
91     MultiFDPages_t *pages = p->pages;
92 
93     for (int i = 0; i < p->normal_num; i++) {
94         p->iov[p->iovs_num].iov_base = pages->block->host + p->normal[i];
95         p->iov[p->iovs_num].iov_len = p->page_size;
96         p->iovs_num++;
97     }
98 
99     p->next_packet_size = p->normal_num * p->page_size;
100     p->flags |= MULTIFD_FLAG_NOCOMP;
101     return 0;
102 }
103 
104 /**
105  * nocomp_recv_setup: setup receive side
106  *
107  * For no compression this function does nothing.
108  *
109  * Returns 0 for success or -1 for error
110  *
111  * @p: Params for the channel that we are using
112  * @errp: pointer to an error
113  */
114 static int nocomp_recv_setup(MultiFDRecvParams *p, Error **errp)
115 {
116     return 0;
117 }
118 
119 /**
120  * nocomp_recv_cleanup: setup receive side
121  *
122  * For no compression this function does nothing.
123  *
124  * @p: Params for the channel that we are using
125  */
126 static void nocomp_recv_cleanup(MultiFDRecvParams *p)
127 {
128 }
129 
130 /**
131  * nocomp_recv_pages: read the data from the channel into actual pages
132  *
133  * For no compression we just need to read things into the correct place.
134  *
135  * Returns 0 for success or -1 for error
136  *
137  * @p: Params for the channel that we are using
138  * @errp: pointer to an error
139  */
140 static int nocomp_recv_pages(MultiFDRecvParams *p, Error **errp)
141 {
142     uint32_t flags = p->flags & MULTIFD_FLAG_COMPRESSION_MASK;
143 
144     if (flags != MULTIFD_FLAG_NOCOMP) {
145         error_setg(errp, "multifd %u: flags received %x flags expected %x",
146                    p->id, flags, MULTIFD_FLAG_NOCOMP);
147         return -1;
148     }
149     for (int i = 0; i < p->normal_num; i++) {
150         p->iov[i].iov_base = p->host + p->normal[i];
151         p->iov[i].iov_len = p->page_size;
152     }
153     return qio_channel_readv_all(p->c, p->iov, p->normal_num, errp);
154 }
155 
156 static MultiFDMethods multifd_nocomp_ops = {
157     .send_setup = nocomp_send_setup,
158     .send_cleanup = nocomp_send_cleanup,
159     .send_prepare = nocomp_send_prepare,
160     .recv_setup = nocomp_recv_setup,
161     .recv_cleanup = nocomp_recv_cleanup,
162     .recv_pages = nocomp_recv_pages
163 };
164 
165 static MultiFDMethods *multifd_ops[MULTIFD_COMPRESSION__MAX] = {
166     [MULTIFD_COMPRESSION_NONE] = &multifd_nocomp_ops,
167 };
168 
169 void multifd_register_ops(int method, MultiFDMethods *ops)
170 {
171     assert(0 < method && method < MULTIFD_COMPRESSION__MAX);
172     multifd_ops[method] = ops;
173 }
174 
175 static int multifd_send_initial_packet(MultiFDSendParams *p, Error **errp)
176 {
177     MultiFDInit_t msg = {};
178     size_t size = sizeof(msg);
179     int ret;
180 
181     msg.magic = cpu_to_be32(MULTIFD_MAGIC);
182     msg.version = cpu_to_be32(MULTIFD_VERSION);
183     msg.id = p->id;
184     memcpy(msg.uuid, &qemu_uuid.data, sizeof(msg.uuid));
185 
186     ret = qio_channel_write_all(p->c, (char *)&msg, size, errp);
187     if (ret != 0) {
188         return -1;
189     }
190     stat64_add(&mig_stats.multifd_bytes, size);
191     stat64_add(&mig_stats.transferred, size);
192     return 0;
193 }
194 
195 static int multifd_recv_initial_packet(QIOChannel *c, Error **errp)
196 {
197     MultiFDInit_t msg;
198     int ret;
199 
200     ret = qio_channel_read_all(c, (char *)&msg, sizeof(msg), errp);
201     if (ret != 0) {
202         return -1;
203     }
204 
205     msg.magic = be32_to_cpu(msg.magic);
206     msg.version = be32_to_cpu(msg.version);
207 
208     if (msg.magic != MULTIFD_MAGIC) {
209         error_setg(errp, "multifd: received packet magic %x "
210                    "expected %x", msg.magic, MULTIFD_MAGIC);
211         return -1;
212     }
213 
214     if (msg.version != MULTIFD_VERSION) {
215         error_setg(errp, "multifd: received packet version %u "
216                    "expected %u", msg.version, MULTIFD_VERSION);
217         return -1;
218     }
219 
220     if (memcmp(msg.uuid, &qemu_uuid, sizeof(qemu_uuid))) {
221         char *uuid = qemu_uuid_unparse_strdup(&qemu_uuid);
222         char *msg_uuid = qemu_uuid_unparse_strdup((const QemuUUID *)msg.uuid);
223 
224         error_setg(errp, "multifd: received uuid '%s' and expected "
225                    "uuid '%s' for channel %hhd", msg_uuid, uuid, msg.id);
226         g_free(uuid);
227         g_free(msg_uuid);
228         return -1;
229     }
230 
231     if (msg.id > migrate_multifd_channels()) {
232         error_setg(errp, "multifd: received channel version %u "
233                    "expected %u", msg.version, MULTIFD_VERSION);
234         return -1;
235     }
236 
237     return msg.id;
238 }
239 
240 static MultiFDPages_t *multifd_pages_init(size_t size)
241 {
242     MultiFDPages_t *pages = g_new0(MultiFDPages_t, 1);
243 
244     pages->allocated = size;
245     pages->offset = g_new0(ram_addr_t, size);
246 
247     return pages;
248 }
249 
250 static void multifd_pages_clear(MultiFDPages_t *pages)
251 {
252     pages->num = 0;
253     pages->allocated = 0;
254     pages->packet_num = 0;
255     pages->block = NULL;
256     g_free(pages->offset);
257     pages->offset = NULL;
258     g_free(pages);
259 }
260 
261 static void multifd_send_fill_packet(MultiFDSendParams *p)
262 {
263     MultiFDPacket_t *packet = p->packet;
264     int i;
265 
266     packet->flags = cpu_to_be32(p->flags);
267     packet->pages_alloc = cpu_to_be32(p->pages->allocated);
268     packet->normal_pages = cpu_to_be32(p->normal_num);
269     packet->next_packet_size = cpu_to_be32(p->next_packet_size);
270     packet->packet_num = cpu_to_be64(p->packet_num);
271 
272     if (p->pages->block) {
273         strncpy(packet->ramblock, p->pages->block->idstr, 256);
274     }
275 
276     for (i = 0; i < p->normal_num; i++) {
277         /* there are architectures where ram_addr_t is 32 bit */
278         uint64_t temp = p->normal[i];
279 
280         packet->offset[i] = cpu_to_be64(temp);
281     }
282 }
283 
284 static int multifd_recv_unfill_packet(MultiFDRecvParams *p, Error **errp)
285 {
286     MultiFDPacket_t *packet = p->packet;
287     int i;
288 
289     packet->magic = be32_to_cpu(packet->magic);
290     if (packet->magic != MULTIFD_MAGIC) {
291         error_setg(errp, "multifd: received packet "
292                    "magic %x and expected magic %x",
293                    packet->magic, MULTIFD_MAGIC);
294         return -1;
295     }
296 
297     packet->version = be32_to_cpu(packet->version);
298     if (packet->version != MULTIFD_VERSION) {
299         error_setg(errp, "multifd: received packet "
300                    "version %u and expected version %u",
301                    packet->version, MULTIFD_VERSION);
302         return -1;
303     }
304 
305     p->flags = be32_to_cpu(packet->flags);
306 
307     packet->pages_alloc = be32_to_cpu(packet->pages_alloc);
308     /*
309      * If we received a packet that is 100 times bigger than expected
310      * just stop migration.  It is a magic number.
311      */
312     if (packet->pages_alloc > p->page_count) {
313         error_setg(errp, "multifd: received packet "
314                    "with size %u and expected a size of %u",
315                    packet->pages_alloc, p->page_count) ;
316         return -1;
317     }
318 
319     p->normal_num = be32_to_cpu(packet->normal_pages);
320     if (p->normal_num > packet->pages_alloc) {
321         error_setg(errp, "multifd: received packet "
322                    "with %u pages and expected maximum pages are %u",
323                    p->normal_num, packet->pages_alloc) ;
324         return -1;
325     }
326 
327     p->next_packet_size = be32_to_cpu(packet->next_packet_size);
328     p->packet_num = be64_to_cpu(packet->packet_num);
329 
330     if (p->normal_num == 0) {
331         return 0;
332     }
333 
334     /* make sure that ramblock is 0 terminated */
335     packet->ramblock[255] = 0;
336     p->block = qemu_ram_block_by_name(packet->ramblock);
337     if (!p->block) {
338         error_setg(errp, "multifd: unknown ram block %s",
339                    packet->ramblock);
340         return -1;
341     }
342 
343     p->host = p->block->host;
344     for (i = 0; i < p->normal_num; i++) {
345         uint64_t offset = be64_to_cpu(packet->offset[i]);
346 
347         if (offset > (p->block->used_length - p->page_size)) {
348             error_setg(errp, "multifd: offset too long %" PRIu64
349                        " (max " RAM_ADDR_FMT ")",
350                        offset, p->block->used_length);
351             return -1;
352         }
353         p->normal[i] = offset;
354     }
355 
356     return 0;
357 }
358 
359 struct {
360     MultiFDSendParams *params;
361     /* array of pages to sent */
362     MultiFDPages_t *pages;
363     /* global number of generated multifd packets */
364     uint64_t packet_num;
365     /* send channels ready */
366     QemuSemaphore channels_ready;
367     /*
368      * Have we already run terminate threads.  There is a race when it
369      * happens that we got one error while we are exiting.
370      * We will use atomic operations.  Only valid values are 0 and 1.
371      */
372     int exiting;
373     /* multifd ops */
374     MultiFDMethods *ops;
375 } *multifd_send_state;
376 
377 /*
378  * How we use multifd_send_state->pages and channel->pages?
379  *
380  * We create a pages for each channel, and a main one.  Each time that
381  * we need to send a batch of pages we interchange the ones between
382  * multifd_send_state and the channel that is sending it.  There are
383  * two reasons for that:
384  *    - to not have to do so many mallocs during migration
385  *    - to make easier to know what to free at the end of migration
386  *
387  * This way we always know who is the owner of each "pages" struct,
388  * and we don't need any locking.  It belongs to the migration thread
389  * or to the channel thread.  Switching is safe because the migration
390  * thread is using the channel mutex when changing it, and the channel
391  * have to had finish with its own, otherwise pending_job can't be
392  * false.
393  */
394 
395 static int multifd_send_pages(QEMUFile *f)
396 {
397     int i;
398     static int next_channel;
399     MultiFDSendParams *p = NULL; /* make happy gcc */
400     MultiFDPages_t *pages = multifd_send_state->pages;
401 
402     if (qatomic_read(&multifd_send_state->exiting)) {
403         return -1;
404     }
405 
406     qemu_sem_wait(&multifd_send_state->channels_ready);
407     /*
408      * next_channel can remain from a previous migration that was
409      * using more channels, so ensure it doesn't overflow if the
410      * limit is lower now.
411      */
412     next_channel %= migrate_multifd_channels();
413     for (i = next_channel;; i = (i + 1) % migrate_multifd_channels()) {
414         p = &multifd_send_state->params[i];
415 
416         qemu_mutex_lock(&p->mutex);
417         if (p->quit) {
418             error_report("%s: channel %d has already quit!", __func__, i);
419             qemu_mutex_unlock(&p->mutex);
420             return -1;
421         }
422         if (!p->pending_job) {
423             p->pending_job++;
424             next_channel = (i + 1) % migrate_multifd_channels();
425             break;
426         }
427         qemu_mutex_unlock(&p->mutex);
428     }
429     assert(!p->pages->num);
430     assert(!p->pages->block);
431 
432     p->packet_num = multifd_send_state->packet_num++;
433     multifd_send_state->pages = p->pages;
434     p->pages = pages;
435     qemu_mutex_unlock(&p->mutex);
436     qemu_sem_post(&p->sem);
437 
438     return 1;
439 }
440 
441 int multifd_queue_page(QEMUFile *f, RAMBlock *block, ram_addr_t offset)
442 {
443     MultiFDPages_t *pages = multifd_send_state->pages;
444     bool changed = false;
445 
446     if (!pages->block) {
447         pages->block = block;
448     }
449 
450     if (pages->block == block) {
451         pages->offset[pages->num] = offset;
452         pages->num++;
453 
454         if (pages->num < pages->allocated) {
455             return 1;
456         }
457     } else {
458         changed = true;
459     }
460 
461     if (multifd_send_pages(f) < 0) {
462         return -1;
463     }
464 
465     if (changed) {
466         return multifd_queue_page(f, block, offset);
467     }
468 
469     return 1;
470 }
471 
472 static void multifd_send_terminate_threads(Error *err)
473 {
474     int i;
475 
476     trace_multifd_send_terminate_threads(err != NULL);
477 
478     if (err) {
479         MigrationState *s = migrate_get_current();
480         migrate_set_error(s, err);
481         if (s->state == MIGRATION_STATUS_SETUP ||
482             s->state == MIGRATION_STATUS_PRE_SWITCHOVER ||
483             s->state == MIGRATION_STATUS_DEVICE ||
484             s->state == MIGRATION_STATUS_ACTIVE) {
485             migrate_set_state(&s->state, s->state,
486                               MIGRATION_STATUS_FAILED);
487         }
488     }
489 
490     /*
491      * We don't want to exit each threads twice.  Depending on where
492      * we get the error, or if there are two independent errors in two
493      * threads at the same time, we can end calling this function
494      * twice.
495      */
496     if (qatomic_xchg(&multifd_send_state->exiting, 1)) {
497         return;
498     }
499 
500     for (i = 0; i < migrate_multifd_channels(); i++) {
501         MultiFDSendParams *p = &multifd_send_state->params[i];
502 
503         qemu_mutex_lock(&p->mutex);
504         p->quit = true;
505         qemu_sem_post(&p->sem);
506         if (p->c) {
507             qio_channel_shutdown(p->c, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
508         }
509         qemu_mutex_unlock(&p->mutex);
510     }
511 }
512 
513 void multifd_save_cleanup(void)
514 {
515     int i;
516 
517     if (!migrate_multifd()) {
518         return;
519     }
520     multifd_send_terminate_threads(NULL);
521     for (i = 0; i < migrate_multifd_channels(); i++) {
522         MultiFDSendParams *p = &multifd_send_state->params[i];
523 
524         if (p->running) {
525             qemu_thread_join(&p->thread);
526         }
527     }
528     for (i = 0; i < migrate_multifd_channels(); i++) {
529         MultiFDSendParams *p = &multifd_send_state->params[i];
530         Error *local_err = NULL;
531 
532         if (p->registered_yank) {
533             migration_ioc_unregister_yank(p->c);
534         }
535         socket_send_channel_destroy(p->c);
536         p->c = NULL;
537         qemu_mutex_destroy(&p->mutex);
538         qemu_sem_destroy(&p->sem);
539         qemu_sem_destroy(&p->sem_sync);
540         g_free(p->name);
541         p->name = NULL;
542         multifd_pages_clear(p->pages);
543         p->pages = NULL;
544         p->packet_len = 0;
545         g_free(p->packet);
546         p->packet = NULL;
547         g_free(p->iov);
548         p->iov = NULL;
549         g_free(p->normal);
550         p->normal = NULL;
551         multifd_send_state->ops->send_cleanup(p, &local_err);
552         if (local_err) {
553             migrate_set_error(migrate_get_current(), local_err);
554             error_free(local_err);
555         }
556     }
557     qemu_sem_destroy(&multifd_send_state->channels_ready);
558     g_free(multifd_send_state->params);
559     multifd_send_state->params = NULL;
560     multifd_pages_clear(multifd_send_state->pages);
561     multifd_send_state->pages = NULL;
562     g_free(multifd_send_state);
563     multifd_send_state = NULL;
564 }
565 
566 static int multifd_zero_copy_flush(QIOChannel *c)
567 {
568     int ret;
569     Error *err = NULL;
570 
571     ret = qio_channel_flush(c, &err);
572     if (ret < 0) {
573         error_report_err(err);
574         return -1;
575     }
576     if (ret == 1) {
577         stat64_add(&mig_stats.dirty_sync_missed_zero_copy, 1);
578     }
579 
580     return ret;
581 }
582 
583 int multifd_send_sync_main(QEMUFile *f)
584 {
585     int i;
586     bool flush_zero_copy;
587 
588     if (!migrate_multifd()) {
589         return 0;
590     }
591     if (multifd_send_state->pages->num) {
592         if (multifd_send_pages(f) < 0) {
593             error_report("%s: multifd_send_pages fail", __func__);
594             return -1;
595         }
596     }
597 
598     /*
599      * When using zero-copy, it's necessary to flush the pages before any of
600      * the pages can be sent again, so we'll make sure the new version of the
601      * pages will always arrive _later_ than the old pages.
602      *
603      * Currently we achieve this by flushing the zero-page requested writes
604      * per ram iteration, but in the future we could potentially optimize it
605      * to be less frequent, e.g. only after we finished one whole scanning of
606      * all the dirty bitmaps.
607      */
608 
609     flush_zero_copy = migrate_zero_copy_send();
610 
611     for (i = 0; i < migrate_multifd_channels(); i++) {
612         MultiFDSendParams *p = &multifd_send_state->params[i];
613 
614         trace_multifd_send_sync_main_signal(p->id);
615 
616         qemu_mutex_lock(&p->mutex);
617 
618         if (p->quit) {
619             error_report("%s: channel %d has already quit", __func__, i);
620             qemu_mutex_unlock(&p->mutex);
621             return -1;
622         }
623 
624         p->packet_num = multifd_send_state->packet_num++;
625         p->flags |= MULTIFD_FLAG_SYNC;
626         p->pending_job++;
627         qemu_mutex_unlock(&p->mutex);
628         qemu_sem_post(&p->sem);
629     }
630     for (i = 0; i < migrate_multifd_channels(); i++) {
631         MultiFDSendParams *p = &multifd_send_state->params[i];
632 
633         qemu_sem_wait(&multifd_send_state->channels_ready);
634         trace_multifd_send_sync_main_wait(p->id);
635         qemu_sem_wait(&p->sem_sync);
636 
637         if (flush_zero_copy && p->c && (multifd_zero_copy_flush(p->c) < 0)) {
638             return -1;
639         }
640     }
641     trace_multifd_send_sync_main(multifd_send_state->packet_num);
642 
643     return 0;
644 }
645 
646 static void *multifd_send_thread(void *opaque)
647 {
648     MultiFDSendParams *p = opaque;
649     MigrationThread *thread = NULL;
650     Error *local_err = NULL;
651     int ret = 0;
652     bool use_zero_copy_send = migrate_zero_copy_send();
653 
654     thread = migration_threads_add(p->name, qemu_get_thread_id());
655 
656     trace_multifd_send_thread_start(p->id);
657     rcu_register_thread();
658 
659     if (multifd_send_initial_packet(p, &local_err) < 0) {
660         ret = -1;
661         goto out;
662     }
663     /* initial packet */
664     p->num_packets = 1;
665 
666     while (true) {
667         qemu_sem_post(&multifd_send_state->channels_ready);
668         qemu_sem_wait(&p->sem);
669 
670         if (qatomic_read(&multifd_send_state->exiting)) {
671             break;
672         }
673         qemu_mutex_lock(&p->mutex);
674 
675         if (p->pending_job) {
676             uint64_t packet_num = p->packet_num;
677             uint32_t flags;
678             p->normal_num = 0;
679 
680             if (use_zero_copy_send) {
681                 p->iovs_num = 0;
682             } else {
683                 p->iovs_num = 1;
684             }
685 
686             for (int i = 0; i < p->pages->num; i++) {
687                 p->normal[p->normal_num] = p->pages->offset[i];
688                 p->normal_num++;
689             }
690 
691             if (p->normal_num) {
692                 ret = multifd_send_state->ops->send_prepare(p, &local_err);
693                 if (ret != 0) {
694                     qemu_mutex_unlock(&p->mutex);
695                     break;
696                 }
697             }
698             multifd_send_fill_packet(p);
699             flags = p->flags;
700             p->flags = 0;
701             p->num_packets++;
702             p->total_normal_pages += p->normal_num;
703             p->pages->num = 0;
704             p->pages->block = NULL;
705             qemu_mutex_unlock(&p->mutex);
706 
707             trace_multifd_send(p->id, packet_num, p->normal_num, flags,
708                                p->next_packet_size);
709 
710             if (use_zero_copy_send) {
711                 /* Send header first, without zerocopy */
712                 ret = qio_channel_write_all(p->c, (void *)p->packet,
713                                             p->packet_len, &local_err);
714                 if (ret != 0) {
715                     break;
716                 }
717                 stat64_add(&mig_stats.multifd_bytes, p->packet_len);
718                 stat64_add(&mig_stats.transferred, p->packet_len);
719             } else {
720                 /* Send header using the same writev call */
721                 p->iov[0].iov_len = p->packet_len;
722                 p->iov[0].iov_base = p->packet;
723             }
724 
725             ret = qio_channel_writev_full_all(p->c, p->iov, p->iovs_num, NULL,
726                                               0, p->write_flags, &local_err);
727             if (ret != 0) {
728                 break;
729             }
730 
731             stat64_add(&mig_stats.multifd_bytes, p->next_packet_size);
732             stat64_add(&mig_stats.transferred, p->next_packet_size);
733             qemu_mutex_lock(&p->mutex);
734             p->pending_job--;
735             qemu_mutex_unlock(&p->mutex);
736 
737             if (flags & MULTIFD_FLAG_SYNC) {
738                 qemu_sem_post(&p->sem_sync);
739             }
740         } else if (p->quit) {
741             qemu_mutex_unlock(&p->mutex);
742             break;
743         } else {
744             qemu_mutex_unlock(&p->mutex);
745             /* sometimes there are spurious wakeups */
746         }
747     }
748 
749 out:
750     if (local_err) {
751         trace_multifd_send_error(p->id);
752         multifd_send_terminate_threads(local_err);
753         error_free(local_err);
754     }
755 
756     /*
757      * Error happen, I will exit, but I can't just leave, tell
758      * who pay attention to me.
759      */
760     if (ret != 0) {
761         qemu_sem_post(&p->sem_sync);
762         qemu_sem_post(&multifd_send_state->channels_ready);
763     }
764 
765     qemu_mutex_lock(&p->mutex);
766     p->running = false;
767     qemu_mutex_unlock(&p->mutex);
768 
769     rcu_unregister_thread();
770     migration_threads_remove(thread);
771     trace_multifd_send_thread_end(p->id, p->num_packets, p->total_normal_pages);
772 
773     return NULL;
774 }
775 
776 static bool multifd_channel_connect(MultiFDSendParams *p,
777                                     QIOChannel *ioc,
778                                     Error *error);
779 
780 static void multifd_tls_outgoing_handshake(QIOTask *task,
781                                            gpointer opaque)
782 {
783     MultiFDSendParams *p = opaque;
784     QIOChannel *ioc = QIO_CHANNEL(qio_task_get_source(task));
785     Error *err = NULL;
786 
787     if (qio_task_propagate_error(task, &err)) {
788         trace_multifd_tls_outgoing_handshake_error(ioc, error_get_pretty(err));
789     } else {
790         trace_multifd_tls_outgoing_handshake_complete(ioc);
791     }
792 
793     if (!multifd_channel_connect(p, ioc, err)) {
794         /*
795          * Error happen, mark multifd_send_thread status as 'quit' although it
796          * is not created, and then tell who pay attention to me.
797          */
798         p->quit = true;
799         qemu_sem_post(&multifd_send_state->channels_ready);
800         qemu_sem_post(&p->sem_sync);
801     }
802 }
803 
804 static void *multifd_tls_handshake_thread(void *opaque)
805 {
806     MultiFDSendParams *p = opaque;
807     QIOChannelTLS *tioc = QIO_CHANNEL_TLS(p->c);
808 
809     qio_channel_tls_handshake(tioc,
810                               multifd_tls_outgoing_handshake,
811                               p,
812                               NULL,
813                               NULL);
814     return NULL;
815 }
816 
817 static void multifd_tls_channel_connect(MultiFDSendParams *p,
818                                         QIOChannel *ioc,
819                                         Error **errp)
820 {
821     MigrationState *s = migrate_get_current();
822     const char *hostname = s->hostname;
823     QIOChannelTLS *tioc;
824 
825     tioc = migration_tls_client_create(ioc, hostname, errp);
826     if (!tioc) {
827         return;
828     }
829 
830     object_unref(OBJECT(ioc));
831     trace_multifd_tls_outgoing_handshake_start(ioc, tioc, hostname);
832     qio_channel_set_name(QIO_CHANNEL(tioc), "multifd-tls-outgoing");
833     p->c = QIO_CHANNEL(tioc);
834     qemu_thread_create(&p->thread, "multifd-tls-handshake-worker",
835                        multifd_tls_handshake_thread, p,
836                        QEMU_THREAD_JOINABLE);
837 }
838 
839 static bool multifd_channel_connect(MultiFDSendParams *p,
840                                     QIOChannel *ioc,
841                                     Error *error)
842 {
843     trace_multifd_set_outgoing_channel(
844         ioc, object_get_typename(OBJECT(ioc)),
845         migrate_get_current()->hostname, error);
846 
847     if (error) {
848         return false;
849     }
850     if (migrate_channel_requires_tls_upgrade(ioc)) {
851         multifd_tls_channel_connect(p, ioc, &error);
852         if (!error) {
853             /*
854              * tls_channel_connect will call back to this
855              * function after the TLS handshake,
856              * so we mustn't call multifd_send_thread until then
857              */
858             return true;
859         } else {
860             return false;
861         }
862     } else {
863         migration_ioc_register_yank(ioc);
864         p->registered_yank = true;
865         p->c = ioc;
866         qemu_thread_create(&p->thread, p->name, multifd_send_thread, p,
867                            QEMU_THREAD_JOINABLE);
868     }
869     return true;
870 }
871 
872 static void multifd_new_send_channel_cleanup(MultiFDSendParams *p,
873                                              QIOChannel *ioc, Error *err)
874 {
875      migrate_set_error(migrate_get_current(), err);
876      /* Error happen, we need to tell who pay attention to me */
877      qemu_sem_post(&multifd_send_state->channels_ready);
878      qemu_sem_post(&p->sem_sync);
879      /*
880       * Although multifd_send_thread is not created, but main migration
881       * thread need to judge whether it is running, so we need to mark
882       * its status.
883       */
884      p->quit = true;
885      object_unref(OBJECT(ioc));
886      error_free(err);
887 }
888 
889 static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
890 {
891     MultiFDSendParams *p = opaque;
892     QIOChannel *sioc = QIO_CHANNEL(qio_task_get_source(task));
893     Error *local_err = NULL;
894 
895     trace_multifd_new_send_channel_async(p->id);
896     if (!qio_task_propagate_error(task, &local_err)) {
897         p->c = sioc;
898         qio_channel_set_delay(p->c, false);
899         p->running = true;
900         if (multifd_channel_connect(p, sioc, local_err)) {
901             return;
902         }
903     }
904 
905     multifd_new_send_channel_cleanup(p, sioc, local_err);
906 }
907 
908 int multifd_save_setup(Error **errp)
909 {
910     int thread_count;
911     uint32_t page_count = MULTIFD_PACKET_SIZE / qemu_target_page_size();
912     uint8_t i;
913 
914     if (!migrate_multifd()) {
915         return 0;
916     }
917 
918     thread_count = migrate_multifd_channels();
919     multifd_send_state = g_malloc0(sizeof(*multifd_send_state));
920     multifd_send_state->params = g_new0(MultiFDSendParams, thread_count);
921     multifd_send_state->pages = multifd_pages_init(page_count);
922     qemu_sem_init(&multifd_send_state->channels_ready, 0);
923     qatomic_set(&multifd_send_state->exiting, 0);
924     multifd_send_state->ops = multifd_ops[migrate_multifd_compression()];
925 
926     for (i = 0; i < thread_count; i++) {
927         MultiFDSendParams *p = &multifd_send_state->params[i];
928 
929         qemu_mutex_init(&p->mutex);
930         qemu_sem_init(&p->sem, 0);
931         qemu_sem_init(&p->sem_sync, 0);
932         p->quit = false;
933         p->pending_job = 0;
934         p->id = i;
935         p->pages = multifd_pages_init(page_count);
936         p->packet_len = sizeof(MultiFDPacket_t)
937                       + sizeof(uint64_t) * page_count;
938         p->packet = g_malloc0(p->packet_len);
939         p->packet->magic = cpu_to_be32(MULTIFD_MAGIC);
940         p->packet->version = cpu_to_be32(MULTIFD_VERSION);
941         p->name = g_strdup_printf("multifdsend_%d", i);
942         /* We need one extra place for the packet header */
943         p->iov = g_new0(struct iovec, page_count + 1);
944         p->normal = g_new0(ram_addr_t, page_count);
945         p->page_size = qemu_target_page_size();
946         p->page_count = page_count;
947 
948         if (migrate_zero_copy_send()) {
949             p->write_flags = QIO_CHANNEL_WRITE_FLAG_ZERO_COPY;
950         } else {
951             p->write_flags = 0;
952         }
953 
954         socket_send_channel_create(multifd_new_send_channel_async, p);
955     }
956 
957     for (i = 0; i < thread_count; i++) {
958         MultiFDSendParams *p = &multifd_send_state->params[i];
959         Error *local_err = NULL;
960         int ret;
961 
962         ret = multifd_send_state->ops->send_setup(p, &local_err);
963         if (ret) {
964             error_propagate(errp, local_err);
965             return ret;
966         }
967     }
968     return 0;
969 }
970 
971 struct {
972     MultiFDRecvParams *params;
973     /* number of created threads */
974     int count;
975     /* syncs main thread and channels */
976     QemuSemaphore sem_sync;
977     /* global number of generated multifd packets */
978     uint64_t packet_num;
979     /* multifd ops */
980     MultiFDMethods *ops;
981 } *multifd_recv_state;
982 
983 static void multifd_recv_terminate_threads(Error *err)
984 {
985     int i;
986 
987     trace_multifd_recv_terminate_threads(err != NULL);
988 
989     if (err) {
990         MigrationState *s = migrate_get_current();
991         migrate_set_error(s, err);
992         if (s->state == MIGRATION_STATUS_SETUP ||
993             s->state == MIGRATION_STATUS_ACTIVE) {
994             migrate_set_state(&s->state, s->state,
995                               MIGRATION_STATUS_FAILED);
996         }
997     }
998 
999     for (i = 0; i < migrate_multifd_channels(); i++) {
1000         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1001 
1002         qemu_mutex_lock(&p->mutex);
1003         p->quit = true;
1004         /*
1005          * We could arrive here for two reasons:
1006          *  - normal quit, i.e. everything went fine, just finished
1007          *  - error quit: We close the channels so the channel threads
1008          *    finish the qio_channel_read_all_eof()
1009          */
1010         if (p->c) {
1011             qio_channel_shutdown(p->c, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
1012         }
1013         qemu_mutex_unlock(&p->mutex);
1014     }
1015 }
1016 
1017 void multifd_load_shutdown(void)
1018 {
1019     if (migrate_multifd()) {
1020         multifd_recv_terminate_threads(NULL);
1021     }
1022 }
1023 
1024 void multifd_load_cleanup(void)
1025 {
1026     int i;
1027 
1028     if (!migrate_multifd()) {
1029         return;
1030     }
1031     multifd_recv_terminate_threads(NULL);
1032     for (i = 0; i < migrate_multifd_channels(); i++) {
1033         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1034 
1035         if (p->running) {
1036             /*
1037              * multifd_recv_thread may hung at MULTIFD_FLAG_SYNC handle code,
1038              * however try to wakeup it without harm in cleanup phase.
1039              */
1040             qemu_sem_post(&p->sem_sync);
1041         }
1042 
1043         qemu_thread_join(&p->thread);
1044     }
1045     for (i = 0; i < migrate_multifd_channels(); i++) {
1046         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1047 
1048         migration_ioc_unregister_yank(p->c);
1049         object_unref(OBJECT(p->c));
1050         p->c = NULL;
1051         qemu_mutex_destroy(&p->mutex);
1052         qemu_sem_destroy(&p->sem_sync);
1053         g_free(p->name);
1054         p->name = NULL;
1055         p->packet_len = 0;
1056         g_free(p->packet);
1057         p->packet = NULL;
1058         g_free(p->iov);
1059         p->iov = NULL;
1060         g_free(p->normal);
1061         p->normal = NULL;
1062         multifd_recv_state->ops->recv_cleanup(p);
1063     }
1064     qemu_sem_destroy(&multifd_recv_state->sem_sync);
1065     g_free(multifd_recv_state->params);
1066     multifd_recv_state->params = NULL;
1067     g_free(multifd_recv_state);
1068     multifd_recv_state = NULL;
1069 }
1070 
1071 void multifd_recv_sync_main(void)
1072 {
1073     int i;
1074 
1075     if (!migrate_multifd()) {
1076         return;
1077     }
1078     for (i = 0; i < migrate_multifd_channels(); i++) {
1079         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1080 
1081         trace_multifd_recv_sync_main_wait(p->id);
1082         qemu_sem_wait(&multifd_recv_state->sem_sync);
1083     }
1084     for (i = 0; i < migrate_multifd_channels(); i++) {
1085         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1086 
1087         WITH_QEMU_LOCK_GUARD(&p->mutex) {
1088             if (multifd_recv_state->packet_num < p->packet_num) {
1089                 multifd_recv_state->packet_num = p->packet_num;
1090             }
1091         }
1092         trace_multifd_recv_sync_main_signal(p->id);
1093         qemu_sem_post(&p->sem_sync);
1094     }
1095     trace_multifd_recv_sync_main(multifd_recv_state->packet_num);
1096 }
1097 
1098 static void *multifd_recv_thread(void *opaque)
1099 {
1100     MultiFDRecvParams *p = opaque;
1101     Error *local_err = NULL;
1102     int ret;
1103 
1104     trace_multifd_recv_thread_start(p->id);
1105     rcu_register_thread();
1106 
1107     while (true) {
1108         uint32_t flags;
1109 
1110         if (p->quit) {
1111             break;
1112         }
1113 
1114         ret = qio_channel_read_all_eof(p->c, (void *)p->packet,
1115                                        p->packet_len, &local_err);
1116         if (ret == 0 || ret == -1) {   /* 0: EOF  -1: Error */
1117             break;
1118         }
1119 
1120         qemu_mutex_lock(&p->mutex);
1121         ret = multifd_recv_unfill_packet(p, &local_err);
1122         if (ret) {
1123             qemu_mutex_unlock(&p->mutex);
1124             break;
1125         }
1126 
1127         flags = p->flags;
1128         /* recv methods don't know how to handle the SYNC flag */
1129         p->flags &= ~MULTIFD_FLAG_SYNC;
1130         trace_multifd_recv(p->id, p->packet_num, p->normal_num, flags,
1131                            p->next_packet_size);
1132         p->num_packets++;
1133         p->total_normal_pages += p->normal_num;
1134         qemu_mutex_unlock(&p->mutex);
1135 
1136         if (p->normal_num) {
1137             ret = multifd_recv_state->ops->recv_pages(p, &local_err);
1138             if (ret != 0) {
1139                 break;
1140             }
1141         }
1142 
1143         if (flags & MULTIFD_FLAG_SYNC) {
1144             qemu_sem_post(&multifd_recv_state->sem_sync);
1145             qemu_sem_wait(&p->sem_sync);
1146         }
1147     }
1148 
1149     if (local_err) {
1150         multifd_recv_terminate_threads(local_err);
1151         error_free(local_err);
1152     }
1153     qemu_mutex_lock(&p->mutex);
1154     p->running = false;
1155     qemu_mutex_unlock(&p->mutex);
1156 
1157     rcu_unregister_thread();
1158     trace_multifd_recv_thread_end(p->id, p->num_packets, p->total_normal_pages);
1159 
1160     return NULL;
1161 }
1162 
1163 int multifd_load_setup(Error **errp)
1164 {
1165     int thread_count;
1166     uint32_t page_count = MULTIFD_PACKET_SIZE / qemu_target_page_size();
1167     uint8_t i;
1168 
1169     /*
1170      * Return successfully if multiFD recv state is already initialised
1171      * or multiFD is not enabled.
1172      */
1173     if (multifd_recv_state || !migrate_multifd()) {
1174         return 0;
1175     }
1176 
1177     thread_count = migrate_multifd_channels();
1178     multifd_recv_state = g_malloc0(sizeof(*multifd_recv_state));
1179     multifd_recv_state->params = g_new0(MultiFDRecvParams, thread_count);
1180     qatomic_set(&multifd_recv_state->count, 0);
1181     qemu_sem_init(&multifd_recv_state->sem_sync, 0);
1182     multifd_recv_state->ops = multifd_ops[migrate_multifd_compression()];
1183 
1184     for (i = 0; i < thread_count; i++) {
1185         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1186 
1187         qemu_mutex_init(&p->mutex);
1188         qemu_sem_init(&p->sem_sync, 0);
1189         p->quit = false;
1190         p->id = i;
1191         p->packet_len = sizeof(MultiFDPacket_t)
1192                       + sizeof(uint64_t) * page_count;
1193         p->packet = g_malloc0(p->packet_len);
1194         p->name = g_strdup_printf("multifdrecv_%d", i);
1195         p->iov = g_new0(struct iovec, page_count);
1196         p->normal = g_new0(ram_addr_t, page_count);
1197         p->page_count = page_count;
1198         p->page_size = qemu_target_page_size();
1199     }
1200 
1201     for (i = 0; i < thread_count; i++) {
1202         MultiFDRecvParams *p = &multifd_recv_state->params[i];
1203         Error *local_err = NULL;
1204         int ret;
1205 
1206         ret = multifd_recv_state->ops->recv_setup(p, &local_err);
1207         if (ret) {
1208             error_propagate(errp, local_err);
1209             return ret;
1210         }
1211     }
1212     return 0;
1213 }
1214 
1215 bool multifd_recv_all_channels_created(void)
1216 {
1217     int thread_count = migrate_multifd_channels();
1218 
1219     if (!migrate_multifd()) {
1220         return true;
1221     }
1222 
1223     if (!multifd_recv_state) {
1224         /* Called before any connections created */
1225         return false;
1226     }
1227 
1228     return thread_count == qatomic_read(&multifd_recv_state->count);
1229 }
1230 
1231 /*
1232  * Try to receive all multifd channels to get ready for the migration.
1233  * Sets @errp when failing to receive the current channel.
1234  */
1235 void multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
1236 {
1237     MultiFDRecvParams *p;
1238     Error *local_err = NULL;
1239     int id;
1240 
1241     id = multifd_recv_initial_packet(ioc, &local_err);
1242     if (id < 0) {
1243         multifd_recv_terminate_threads(local_err);
1244         error_propagate_prepend(errp, local_err,
1245                                 "failed to receive packet"
1246                                 " via multifd channel %d: ",
1247                                 qatomic_read(&multifd_recv_state->count));
1248         return;
1249     }
1250     trace_multifd_recv_new_channel(id);
1251 
1252     p = &multifd_recv_state->params[id];
1253     if (p->c != NULL) {
1254         error_setg(&local_err, "multifd: received id '%d' already setup'",
1255                    id);
1256         multifd_recv_terminate_threads(local_err);
1257         error_propagate(errp, local_err);
1258         return;
1259     }
1260     p->c = ioc;
1261     object_ref(OBJECT(ioc));
1262     /* initial packet */
1263     p->num_packets = 1;
1264 
1265     p->running = true;
1266     qemu_thread_create(&p->thread, p->name, multifd_recv_thread, p,
1267                        QEMU_THREAD_JOINABLE);
1268     qatomic_inc(&multifd_recv_state->count);
1269 }
1270