summaryrefslogtreecommitdiff
path: root/engines/nbd.c
blob: f8583812e5d43a1f2afdb24175c33989205ba5d4 (plain)
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
/*
 * NBD engine
 *
 * IO engine that talks to an NBD server.
 *
 * Copyright (C) 2019 Red Hat Inc.
 * Written by Richard W.M. Jones <rjones@redhat.com>
 *
 */

#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <errno.h>

#include <libnbd.h>

#include "../fio.h"
#include "../optgroup.h"

/* Actually this differs across servers, but for nbdkit ... */
#define NBD_MAX_REQUEST_SIZE (64 * 1024 * 1024)

/* Storage for the NBD handle. */
struct nbd_data {
	struct nbd_handle *nbd;
	int debug;

	/* The list of completed io_u structs. */
	struct io_u **completed;
	size_t nr_completed;
};

/* Options. */
struct nbd_options {
	void *padding;
	char *uri;
};

static struct fio_option options[] = {
	{
		.name	= "uri",
		.lname	= "NBD URI",
		.help	= "Name of NBD URI",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NBD,
		.type	= FIO_OPT_STR_STORE,
		.off1	= offsetof(struct nbd_options, uri),
	},
	{
		.name	= NULL,
	},
};

/* Alocates nbd_data. */
static int nbd_setup(struct thread_data *td)
{
	struct nbd_data *nbd_data;
	struct nbd_options *o = td->eo;
	struct fio_file *f;
	int r;
	int64_t size;

	nbd_data = calloc(1, sizeof(*nbd_data));
	if (!nbd_data) {
		td_verror(td, errno, "calloc");
		return 1;
	}
	td->io_ops_data = nbd_data;

	/* Pretend to deal with files.	See engines/rbd.c */
	if (!td->files_index) {
		add_file(td, "nbd", 0, 0);
		td->o.nr_files = td->o.nr_files ? : 1;
		td->o.open_files++;
	}
	f = td->files[0];

	nbd_data->nbd = nbd_create();
	if (!nbd_data->nbd) {
		log_err("fio: nbd_create: %s\n", nbd_get_error());
		return 1;
	}

	/* Get the debug flag which can be set through LIBNBD_DEBUG=1. */
	nbd_data->debug = nbd_get_debug(nbd_data->nbd);

	/* Connect synchronously here so we can check for the size and
	 * in future other properties of the server.
	 */
	if (!o->uri) {
		log_err("fio: nbd: uri parameter was not specified\n");
		return 1;
	}
	r = nbd_connect_uri(nbd_data->nbd, o->uri);
	if (r == -1) {
		log_err("fio: nbd_connect_uri: %s\n", nbd_get_error());
		return 1;
	}
	size = nbd_get_size(nbd_data->nbd);
	if (size == -1) {
		log_err("fio: nbd_get_size: %s\n", nbd_get_error());
		return 1;
	}

	f->real_file_size = size;

	nbd_close (nbd_data->nbd);
	nbd_data->nbd = NULL;

	return 0;
}

/* Closes socket and frees nbd_data -- the opposite of nbd_setup. */
static void nbd_cleanup(struct thread_data *td)
{
	struct nbd_data *nbd_data = td->io_ops_data;

	if (nbd_data) {
		if (nbd_data->nbd)
			nbd_close(nbd_data->nbd);
		free(nbd_data);
	}
}

/* Connect to the server from each thread. */
static int nbd_init(struct thread_data *td)
{
	struct nbd_options *o = td->eo;
	struct nbd_data *nbd_data = td->io_ops_data;
	int r;

	if (!o->uri) {
		log_err("fio: nbd: uri parameter was not specified\n");
		return 1;
	}

	nbd_data->nbd = nbd_create();
	if (!nbd_data->nbd) {
		log_err("fio: nbd_create: %s\n", nbd_get_error());
		return 1;
	}
	/* This is actually a synchronous connect and handshake. */
	r = nbd_connect_uri(nbd_data->nbd, o->uri);
	if (r == -1) {
		log_err("fio: nbd_connect_uri: %s\n", nbd_get_error());
		return 1;
	}

	log_info("fio: connected to NBD server\n");
	return 0;
}

/* A command in flight has been completed. */
static int cmd_completed (unsigned valid_flag, void *vp, int *error)
{
	struct io_u *io_u;
	struct nbd_data *nbd_data;
	struct io_u **completed;

	if (!(valid_flag & LIBNBD_CALLBACK_VALID))
		return 0;

	io_u = vp;
	nbd_data = io_u->engine_data;

	if (nbd_data->debug)
		log_info("fio: nbd: command completed\n");

	if (*error != 0)
		io_u->error = *error;
	else
		io_u->error = 0;

	/* Add this completion to the list so it can be picked up
	 * later by ->event.
	 */
	completed = realloc(nbd_data->completed,
			    sizeof(struct io_u *) *
			    (nbd_data->nr_completed+1));
	if (completed == NULL) {
		io_u->error = errno;
		return 0;
	}

	nbd_data->completed = completed;
	nbd_data->completed[nbd_data->nr_completed] = io_u;
	nbd_data->nr_completed++;

	return 0;
}

/* Begin read or write request. */
static enum fio_q_status nbd_queue(struct thread_data *td,
				   struct io_u *io_u)
{
	struct nbd_data *nbd_data = td->io_ops_data;
	int r;

	fio_ro_check(td, io_u);

	io_u->engine_data = nbd_data;

	if (io_u->ddir == DDIR_WRITE || io_u->ddir == DDIR_READ)
		assert(io_u->xfer_buflen <= NBD_MAX_REQUEST_SIZE);

	switch (io_u->ddir) {
	case DDIR_READ:
		r = nbd_aio_pread_callback(nbd_data->nbd,
					   io_u->xfer_buf, io_u->xfer_buflen,
					   io_u->offset,
					   cmd_completed, io_u,
					   0);
		break;
	case DDIR_WRITE:
		r = nbd_aio_pwrite_callback(nbd_data->nbd,
					    io_u->xfer_buf, io_u->xfer_buflen,
					    io_u->offset,
					    cmd_completed, io_u,
					    0);
		break;
	case DDIR_TRIM:
		r = nbd_aio_trim_callback(nbd_data->nbd, io_u->xfer_buflen,
					  io_u->offset,
					  cmd_completed, io_u,
					  0);
		break;
	case DDIR_SYNC:
		/* XXX We could probably also handle
		 * DDIR_SYNC_FILE_RANGE with a bit of effort.
		 */
		r = nbd_aio_flush_callback(nbd_data->nbd,
					   cmd_completed, io_u,
					   0);
		break;
	default:
		io_u->error = EINVAL;
		return FIO_Q_COMPLETED;
	}

	if (r == -1) {
		/* errno is optional information on libnbd error path;
		 * if it's 0, set it to a default value
		 */
		io_u->error = nbd_get_errno();
		if (io_u->error == 0)
			io_u->error = EIO;
		return FIO_Q_COMPLETED;
	}

	if (nbd_data->debug)
		log_info("fio: nbd: command issued\n");
	io_u->error = 0;
	return FIO_Q_QUEUED;
}

static unsigned retire_commands(struct nbd_handle *nbd)
{
	int64_t cookie;
	unsigned r = 0;

	while ((cookie = nbd_aio_peek_command_completed(nbd)) > 0) {
		/* Ignore the return value.  cmd_completed has already
		 * checked for an error and set io_u->error.  We only
		 * have to call this to retire the command.
		 */
		nbd_aio_command_completed(nbd, cookie);
		r++;
	}

	if (nbd_get_debug(nbd))
		log_info("fio: nbd: %u commands retired\n", r);
	return r;
}

static int nbd_getevents(struct thread_data *td, unsigned int min,
			 unsigned int max, const struct timespec *t)
{
	struct nbd_data *nbd_data = td->io_ops_data;
	int r;
	unsigned events = 0;
	int timeout;

	/* XXX This handling of timeout is wrong because it will wait
	 * for up to loop iterations * timeout.
	 */
	timeout = !t ? -1 : t->tv_sec * 1000 + t->tv_nsec / 1000000;

	while (events < min) {
		r = nbd_poll(nbd_data->nbd, timeout);
		if (r == -1) {
			/* error in poll */
			log_err("fio: nbd_poll: %s\n", nbd_get_error());
			return -1;
		}
		else {
			/* poll made progress */
			events += retire_commands(nbd_data->nbd);
		}
	}

	return events;
}

static struct io_u *nbd_event(struct thread_data *td, int event)
{
	struct nbd_data *nbd_data = td->io_ops_data;

	if (nbd_data->nr_completed == 0)
		return NULL;

	/* XXX We ignore the event number and assume fio calls us
	 * exactly once for [0..nr_events-1].
	 */
	nbd_data->nr_completed--;
	return nbd_data->completed[nbd_data->nr_completed];
}

static int nbd_io_u_init(struct thread_data *td, struct io_u *io_u)
{
	io_u->engine_data = NULL;
	return 0;
}

static void nbd_io_u_free(struct thread_data *td, struct io_u *io_u)
{
	/* Nothing needs to be done. */
}

static int nbd_open_file(struct thread_data *td, struct fio_file *f)
{
	return 0;
}

static int nbd_invalidate(struct thread_data *td, struct fio_file *f)
{
	return 0;
}

static struct ioengine_ops ioengine = {
	.name			= "nbd",
	.version		= FIO_IOOPS_VERSION,
	.options		= options,
	.option_struct_size	= sizeof(struct nbd_options),
	.flags			= FIO_DISKLESSIO | FIO_NOEXTEND,

	.setup			= nbd_setup,
	.init			= nbd_init,
	.cleanup		= nbd_cleanup,
	.queue			= nbd_queue,
	.getevents		= nbd_getevents,
	.event			= nbd_event,
	.io_u_init		= nbd_io_u_init,
	.io_u_free		= nbd_io_u_free,

	.open_file		= nbd_open_file,
	.invalidate		= nbd_invalidate,
};

static void fio_init fio_nbd_register(void)
{
	register_ioengine(&ioengine);
}

static void fio_exit fio_nbd_unregister(void)
{
	unregister_ioengine(&ioengine);
}