1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
|
/*
* Copyright (c) 2020 Nutanix Inc. All rights reserved.
*
* Authors: Thanos Makatos <thanos@nutanix.com>
* Swapnil Ingle <swapnil.ingle@nutanix.com>
* Felipe Franciosi <felipe@nutanix.com>
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Nutanix nor the names of its contributors may be
* used to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY
* DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
* DAMAGE.
*
*/
#define _GNU_SOURCE
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <json.h>
#include <stdio.h>
#include <string.h>
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/un.h>
#include "libvfio-user.h"
#include "migration.h"
#include "private.h"
#include "tran_sock.h"
// FIXME: is this the value we want?
#define SERVER_MAX_FDS 8
static inline int
recv_blocking(int sock, void *buf, size_t len, int flags)
{
int f = fcntl(sock, F_GETFL, 0);
int ret, fret;
fret = fcntl(sock, F_SETFL, f & ~O_NONBLOCK);
assert(fret != -1);
// FIXME: be more explicit about EOF
ret = recv(sock, buf, len, flags);
fret = fcntl(sock, F_SETFL, f);
assert(fret != -1);
return ret;
}
static int
init_sock(vfu_ctx_t *vfu_ctx)
{
struct sockaddr_un addr = { .sun_family = AF_UNIX };
int ret, unix_sock;
mode_t mode;
assert(vfu_ctx != NULL);
/* FIXME SPDK can't easily run as non-root */
mode = umask(0000);
if ((unix_sock = socket(AF_UNIX, SOCK_STREAM, 0)) == -1) {
ret = -errno;
goto out;
}
if (vfu_ctx->flags & LIBVFIO_USER_FLAG_ATTACH_NB) {
ret = fcntl(unix_sock, F_SETFL,
fcntl(unix_sock, F_GETFL, 0) | O_NONBLOCK);
if (ret < 0) {
ret = -errno;
goto out;
}
vfu_ctx->sock_flags = MSG_DONTWAIT | MSG_WAITALL;
} else {
vfu_ctx->sock_flags = 0;
}
ret = snprintf(addr.sun_path, sizeof addr.sun_path, "%s", vfu_ctx->uuid);
if (ret >= (int)sizeof addr.sun_path) {
ret = -ENAMETOOLONG;
}
if (ret < 0) {
goto out;
}
/* start listening business */
ret = bind(unix_sock, (struct sockaddr*)&addr, sizeof(addr));
if (ret < 0) {
ret = errno;
goto out;
}
ret = listen(unix_sock, 0);
if (ret < 0) {
ret = -errno;
}
out:
umask(mode);
if (ret != 0) {
if (unix_sock >= 0) {
close(unix_sock);
}
return ret;
}
return unix_sock;
}
int
vfu_send_iovec(int sock, uint16_t msg_id, bool is_reply,
enum vfio_user_command cmd,
struct iovec *iovecs, size_t nr_iovecs,
int *fds, int count, int err)
{
int ret;
struct vfio_user_header hdr = {.msg_id = msg_id};
struct msghdr msg;
size_t i;
if (nr_iovecs == 0) {
iovecs = alloca(sizeof(*iovecs));
nr_iovecs = 1;
}
memset(&msg, 0, sizeof(msg));
if (is_reply) {
// FIXME: SPEC: should the reply include the command? I'd say yes?
hdr.flags.type = VFIO_USER_F_TYPE_REPLY;
if (err != 0) {
hdr.flags.error = 1U;
hdr.error_no = err;
}
} else {
hdr.cmd = cmd;
hdr.flags.type = VFIO_USER_F_TYPE_COMMAND;
}
iovecs[0].iov_base = &hdr;
iovecs[0].iov_len = sizeof(hdr);
for (i = 0; i < nr_iovecs; i++) {
hdr.msg_size += iovecs[i].iov_len;
}
msg.msg_iovlen = nr_iovecs;
msg.msg_iov = iovecs;
if (fds != NULL) {
size_t size = count * sizeof *fds;
char *buf = alloca(CMSG_SPACE(size));
msg.msg_control = buf;
msg.msg_controllen = CMSG_SPACE(size);
struct cmsghdr * cmsg = CMSG_FIRSTHDR(&msg);
cmsg->cmsg_level = SOL_SOCKET;
cmsg->cmsg_type = SCM_RIGHTS;
cmsg->cmsg_len = CMSG_LEN(size);
memcpy(CMSG_DATA(cmsg), fds, size);
}
// FIXME: this doesn't check the entire data was sent?
ret = sendmsg(sock, &msg, 0);
if (ret == -1) {
return -errno;
}
return 0;
}
int
vfu_send(int sock, uint16_t msg_id, bool is_reply,
enum vfio_user_command cmd,
void *data, size_t data_len)
{
/* [0] is for the header. */
struct iovec iovecs[2] = {
[1] = {
.iov_base = data,
.iov_len = data_len
}
};
return vfu_send_iovec(sock, msg_id, is_reply, cmd, iovecs,
ARRAY_SIZE(iovecs), NULL, 0, 0);
}
int
vfu_send_error(int sock, uint16_t msg_id,
enum vfio_user_command cmd,
int error)
{
return vfu_send_iovec(sock, msg_id, true, cmd, NULL, 0, NULL, 0, error);
}
/*
* Receive a vfio-user message. If "len" is set to non-zero, the message should
* include data of that length, which is stored in the pre-allocated "data"
* pointer.
*
* FIXME: in general, sort out negative err returns - they should only be used
* when we're going to return > 0 on success, and even then "errno" might be
* better.
*/
int
vfu_recv(int sock, struct vfio_user_header *hdr, bool is_reply,
uint16_t *msg_id, void *data, size_t *len)
{
int ret;
/* FIXME if ret == -1 then fcntl can overwrite recv's errno */
ret = recv_blocking(sock, hdr, sizeof(*hdr), 0);
if (ret == -1) {
return -errno;
}
if (ret < (int)sizeof(*hdr)) {
return -EINVAL;
}
if (is_reply) {
if (hdr->msg_id != *msg_id) {
return -EINVAL;
}
if (hdr->flags.type != VFIO_USER_F_TYPE_REPLY) {
return -EINVAL;
}
if (hdr->flags.error == 1U) {
if (hdr->error_no <= 0) {
hdr->error_no = EINVAL;
}
return -hdr->error_no;
}
} else {
if (hdr->flags.type != VFIO_USER_F_TYPE_COMMAND) {
return -EINVAL;
}
*msg_id = hdr->msg_id;
}
if (len != NULL && *len > 0 && hdr->msg_size > sizeof *hdr) {
ret = recv_blocking(sock, data, MIN(hdr->msg_size - sizeof *hdr, *len),
0);
if (ret < 0) {
return ret;
}
if (*len != (size_t)ret) { /* FIXME we should allow receiving less */
return -EINVAL;
}
*len = ret;
}
return 0;
}
/*
* Like vfu_recv(), but will automatically allocate reply data.
*
* FIXME: this does an unconstrained alloc of client-supplied data.
*/
int
vfu_recv_alloc(int sock, struct vfio_user_header *hdr, bool is_reply,
uint16_t *msg_id, void **datap, size_t *lenp)
{
void *data;
size_t len;
int ret;
ret = vfu_recv(sock, hdr, is_reply, msg_id, NULL, NULL);
if (ret != 0) {
return ret;
}
assert(hdr->msg_size >= sizeof (*hdr));
len = hdr->msg_size - sizeof (*hdr);
if (len == 0) {
*datap = NULL;
*lenp = 0;
return 0;
}
data = calloc(1, len);
if (data == NULL) {
return -errno;
}
ret = recv_blocking(sock, data, len, 0);
if (ret < 0) {
free(data);
return -errno;
}
if (len != (size_t)ret) {
free(data);
return -EINVAL;
}
*datap = data;
*lenp = len;
return 0;
}
/*
* FIXME: all these send/recv handlers need to be made robust against async
* messages.
*/
int
vfu_msg_iovec(int sock, uint16_t msg_id, enum vfio_user_command cmd,
struct iovec *iovecs, size_t nr_iovecs,
int *send_fds, size_t fd_count,
struct vfio_user_header *hdr,
void *recv_data, size_t recv_len)
{
int ret = vfu_send_iovec(sock, msg_id, false, cmd, iovecs, nr_iovecs,
send_fds, fd_count, 0);
if (ret < 0) {
return ret;
}
if (hdr == NULL) {
hdr = alloca(sizeof *hdr);
}
return vfu_recv(sock, hdr, true, &msg_id, recv_data, &recv_len);
}
int
vfu_msg(int sock, uint16_t msg_id, enum vfio_user_command cmd,
void *send_data, size_t send_len,
struct vfio_user_header *hdr,
void *recv_data, size_t recv_len)
{
/* [0] is for the header. */
struct iovec iovecs[2] = {
[1] = {
.iov_base = send_data,
.iov_len = send_len
}
};
return vfu_msg_iovec(sock, msg_id, cmd, iovecs, ARRAY_SIZE(iovecs),
NULL, 0, hdr, recv_data, recv_len);
}
/*
* Expected JSON is of the form:
*
* {
* "capabilities": {
* "max_fds": 32,
* "migration": {
* "pgsize": 4096
* }
* }
* }
*
* with everything being optional. Note that json_object_get_uint64() is only
* available in newer library versions, so we don't use it.
*/
int
vfu_parse_version_json(const char *json_str,
int *client_max_fdsp, size_t *pgsizep)
{
struct json_object *jo_caps = NULL;
struct json_object *jo_top = NULL;
struct json_object *jo = NULL;
int ret = -EINVAL;
if ((jo_top = json_tokener_parse(json_str)) == NULL) {
goto out;
}
if (!json_object_object_get_ex(jo_top, "capabilities", &jo_caps)) {
ret = 0;
goto out;
}
if (json_object_get_type(jo_caps) != json_type_object) {
goto out;
}
if (json_object_object_get_ex(jo_caps, "max_fds", &jo)) {
if (json_object_get_type(jo) != json_type_int) {
goto out;
}
errno = 0;
*client_max_fdsp = (int)json_object_get_int64(jo);
if (errno != 0) {
goto out;
}
}
if (json_object_object_get_ex(jo_caps, "migration", &jo)) {
struct json_object *jo2 = NULL;
if (json_object_get_type(jo) != json_type_object) {
goto out;
}
if (json_object_object_get_ex(jo, "pgsize", &jo2)) {
if (json_object_get_type(jo2) != json_type_int) {
goto out;
}
errno = 0;
*pgsizep = (size_t)json_object_get_int64(jo2);
if (errno != 0) {
goto out;
}
}
}
ret = 0;
out:
/* We just need to put our top-level object. */
json_object_put(jo_top);
return ret;
}
int
recv_version(vfu_ctx_t *vfu_ctx, int sock, uint16_t *msg_idp,
struct vfio_user_version **versionp)
{
struct vfio_user_version *cversion = NULL;
struct vfio_user_header hdr;
size_t vlen = 0;
int ret;
*versionp = NULL;
ret = vfu_recv_alloc(sock, &hdr, false, msg_idp,
(void **)&cversion, &vlen);
if (ret < 0) {
vfu_log(vfu_ctx, VFU_ERR, "failed to receive version: %s",
strerror(-ret));
return ret;
}
if (hdr.cmd != VFIO_USER_VERSION) {
vfu_log(vfu_ctx, VFU_ERR, "msg%hx: invalid cmd %hu (expected %hu)",
*msg_idp, hdr.cmd, VFIO_USER_VERSION);
ret = -EINVAL;
goto out;
}
if (vlen < sizeof (*cversion)) {
vfu_log(vfu_ctx, VFU_ERR,
"msg%hx (VFIO_USER_VERSION): invalid size %lu", *msg_idp, vlen);
ret = -EINVAL;
goto out;
}
if (cversion->major != LIB_VFIO_USER_MAJOR) {
vfu_log(vfu_ctx, VFU_ERR, "unsupported client major %hu (must be %hu)",
cversion->major, LIB_VFIO_USER_MAJOR);
ret = -ENOTSUP;
goto out;
}
vfu_ctx->client_max_fds = 1;
if (vlen > sizeof (*cversion)) {
const char *json_str = (const char *)cversion->data;
size_t len = vlen - sizeof (*cversion);
size_t pgsize = 0;
if (json_str[len - 1] != '\0') {
vfu_log(vfu_ctx, VFU_ERR, "ignoring invalid JSON from client");
ret = -EINVAL;
goto out;
}
ret = vfu_parse_version_json(json_str, &vfu_ctx->client_max_fds,
&pgsize);
if (ret < 0) {
/* No client-supplied strings in the log for release build. */
#ifdef DEBUG
vfu_log(vfu_ctx, VFU_ERR, "failed to parse client JSON \"%s\"",
json_str);
#else
vfu_log(vfu_ctx, VFU_ERR, "failed to parse client JSON");
#endif
goto out;
}
if (vfu_ctx->migration != NULL && pgsize != 0) {
ret = migration_set_pgsize(vfu_ctx->migration, pgsize);
if (ret != 0) {
vfu_log(vfu_ctx, VFU_ERR, "refusing client page size of %zu",
pgsize);
goto out;
}
}
// FIXME: is the code resilient against ->client_max_fds == 0?
if (vfu_ctx->client_max_fds < 0 ||
vfu_ctx->client_max_fds > VFIO_USER_CLIENT_MAX_FDS_LIMIT) {
vfu_log(vfu_ctx, VFU_ERR, "refusing client max_fds of %d",
vfu_ctx->client_max_fds);
ret = -EINVAL;
goto out;
}
}
out:
if (ret != 0) {
// FIXME: spec, is it OK to just have the header?
(void) vfu_send_error(sock, *msg_idp, hdr.cmd, ret);
free(cversion);
cversion = NULL;
}
*versionp = cversion;
return ret;
}
int
send_version(vfu_ctx_t *vfu_ctx, int sock, uint16_t msg_id,
struct vfio_user_version *cversion)
{
struct vfio_user_version sversion = { 0 };
struct iovec iovecs[3] = { { 0 } };
char server_caps[1024];
int slen;
if (vfu_ctx->migration == NULL) {
slen = snprintf(server_caps, sizeof (server_caps),
"{"
"\"capabilities\":{"
"\"max_fds\":%u"
"}"
"}", SERVER_MAX_FDS);
} else {
slen = snprintf(server_caps, sizeof (server_caps),
"{"
"\"capabilities\":{"
"\"max_fds\":%u,"
"\"migration\":{"
"\"pgsize\":%zu"
"}"
"}"
"}", SERVER_MAX_FDS, migration_get_pgsize(vfu_ctx->migration));
}
// FIXME: we should save the client minor here, and check that before trying
// to send unsupported things.
sversion.major = LIB_VFIO_USER_MAJOR;
sversion.minor = MIN(cversion->minor, LIB_VFIO_USER_MINOR);
/* [0] is for the header. */
iovecs[1].iov_base = &sversion;
iovecs[1].iov_len = sizeof (sversion);
iovecs[2].iov_base = server_caps;
/* Include the NUL. */
iovecs[2].iov_len = slen + 1;
return vfu_send_iovec(sock, msg_id, true, VFIO_USER_VERSION,
iovecs, ARRAY_SIZE(iovecs), NULL, 0, 0);
}
static int
negotiate(vfu_ctx_t *vfu_ctx, int sock)
{
struct vfio_user_version *client_version = NULL;
uint16_t msg_id = 1;
int ret;
ret = recv_version(vfu_ctx, sock, &msg_id, &client_version);
if (ret < 0) {
vfu_log(vfu_ctx, VFU_ERR, "failed to recv version: %s", strerror(-ret));
return ret;
}
ret = send_version(vfu_ctx, sock, msg_id, client_version);
free(client_version);
if (ret < 0) {
vfu_log(vfu_ctx, VFU_ERR, "failed to send version: %s", strerror(-ret));
}
return ret;
}
/**
* vfu_ctx: libvfio-user context
* FIXME: this shouldn't be happening as part of vfu_ctx_create().
*/
static int
open_sock(vfu_ctx_t *vfu_ctx)
{
int ret;
int conn_fd;
assert(vfu_ctx != NULL);
conn_fd = accept(vfu_ctx->fd, NULL, NULL);
if (conn_fd == -1) {
return conn_fd;
}
ret = negotiate(vfu_ctx, conn_fd);
if (ret < 0) {
close(conn_fd);
return ret;
}
vfu_ctx->conn_fd = conn_fd;
return conn_fd;
}
static int
close_sock(vfu_ctx_t *vfu_ctx)
{
return close(vfu_ctx->conn_fd);
}
static int
get_request_sock(vfu_ctx_t *vfu_ctx, struct vfio_user_header *hdr,
int *fds, size_t *nr_fds)
{
int ret;
struct iovec iov = {.iov_base = hdr, .iov_len = sizeof *hdr};
struct msghdr msg = {.msg_iov = &iov, .msg_iovlen = 1};
struct cmsghdr *cmsg;
msg.msg_controllen = CMSG_SPACE(sizeof(int) * *nr_fds);
msg.msg_control = alloca(msg.msg_controllen);
*nr_fds = 0;
/*
* TODO ideally we should set O_NONBLOCK on the fd so that the syscall is
* faster (?). I tried that and get short reads, so we need to store the
* partially received buffer somewhere and retry.
*/
ret = recvmsg(vfu_ctx->conn_fd, &msg, vfu_ctx->sock_flags);
if (ret == -1) {
return -errno;
}
if (msg.msg_flags & MSG_CTRUNC || msg.msg_flags & MSG_TRUNC) {
return -EFAULT;
}
for (cmsg = CMSG_FIRSTHDR(&msg); cmsg != NULL; cmsg = CMSG_NXTHDR(&msg, cmsg)) {
if (cmsg->cmsg_level != SOL_SOCKET || cmsg->cmsg_type != SCM_RIGHTS) {
continue;
}
if (cmsg->cmsg_len < CMSG_LEN(sizeof(int))) {
return -EINVAL;
}
int size = cmsg->cmsg_len - CMSG_LEN(0);
if (size % sizeof(int) != 0) {
return -EINVAL;
}
*nr_fds = (int)(size / sizeof(int));
memcpy(fds, CMSG_DATA(cmsg), *nr_fds * sizeof(int));
break;
}
return ret;
}
struct transport_ops sock_transport_ops = {
.init = init_sock,
.attach = open_sock,
.detach = close_sock,
.get_request = get_request_sock,
};
/* ex: set tabstop=4 shiftwidth=4 softtabstop=4 expandtab: */
|