[fio.git] / engines / rados.c

/*
 *  Ceph Rados engine
 *
 * IO engine using Ceph's RADOS interface to test low-level performance of
 * Ceph OSDs.
 *
 */

#include <rados/librados.h>
#include <pthread.h>
#include "fio.h"
#include "../optgroup.h"

struct rados_data {
	rados_t cluster;
	rados_ioctx_t io_ctx;
	struct io_u **aio_events;
	bool connected;
	pthread_mutex_t completed_lock;
	pthread_cond_t completed_more_io;
	struct flist_head completed_operations;
	uint64_t ops_scheduled;
	uint64_t ops_completed;
};

struct fio_rados_iou {
	struct flist_head list;
	struct thread_data *td;
	struct io_u *io_u;
	rados_completion_t completion;
	rados_write_op_t write_op;
};

/* fio configuration options read from the job file */
struct rados_options {
	void *pad;
	char *cluster_name;
	char *pool_name;
	char *client_name;
	int busy_poll;
	int touch_objects;
};

static struct fio_option options[] = {
	{
		.name     = "clustername",
		.lname    = "ceph cluster name",
		.type     = FIO_OPT_STR_STORE,
		.help     = "Cluster name for ceph",
		.off1     = offsetof(struct rados_options, cluster_name),
		.category = FIO_OPT_C_ENGINE,
		.group    = FIO_OPT_G_RBD,
	},
	{
		.name     = "pool",
		.lname    = "pool name to use",
		.type     = FIO_OPT_STR_STORE,
		.help     = "Ceph pool name to benchmark against",
		.off1     = offsetof(struct rados_options, pool_name),
		.category = FIO_OPT_C_ENGINE,
		.group    = FIO_OPT_G_RBD,
	},
	{
		.name     = "clientname",
		.lname    = "rados engine clientname",
		.type     = FIO_OPT_STR_STORE,
		.help     = "Name of the ceph client to access RADOS engine",
		.off1     = offsetof(struct rados_options, client_name),
		.category = FIO_OPT_C_ENGINE,
		.group    = FIO_OPT_G_RBD,
	},
	{
		.name     = "busy_poll",
		.lname    = "busy poll mode",
		.type     = FIO_OPT_BOOL,
		.help     = "Busy poll for completions instead of sleeping",
		.off1     = offsetof(struct rados_options, busy_poll),
		.def	  = "0",
		.category = FIO_OPT_C_ENGINE,
		.group    = FIO_OPT_G_RBD,
	},
	{
		.name     = "touch_objects",
		.lname    = "touch objects on start",
		.type     = FIO_OPT_BOOL,
		.help     = "Touch (create) objects on start",
		.off1     = offsetof(struct rados_options, touch_objects),
		.def	  = "1",
		.category = FIO_OPT_C_ENGINE,
		.group    = FIO_OPT_G_RBD,
	},
	{
		.name     = NULL,
	},
};

static int _fio_setup_rados_data(struct thread_data *td,
				struct rados_data **rados_data_ptr)
{
	struct rados_data *rados;

	if (td->io_ops_data)
		return 0;

	rados = calloc(1, sizeof(struct rados_data));
	if (!rados)
		goto failed;

	rados->connected = false;

	rados->aio_events = calloc(td->o.iodepth, sizeof(struct io_u *));
	if (!rados->aio_events)
		goto failed;
	pthread_mutex_init(&rados->completed_lock, NULL);
	pthread_cond_init(&rados->completed_more_io, NULL);
	INIT_FLIST_HEAD(&rados->completed_operations);
	rados->ops_scheduled = 0;
	rados->ops_completed = 0;
	*rados_data_ptr = rados;
	return 0;

failed:
	if (rados) {
		if (rados->aio_events)
			free(rados->aio_events);
		free(rados);
	}
	return 1;
}

static void _fio_rados_rm_objects(struct thread_data *td, struct rados_data *rados)
{
	size_t i;
	for (i = 0; i < td->o.nr_files; i++) {
		struct fio_file *f = td->files[i];
		rados_remove(rados->io_ctx, f->file_name);
	}
}

static int _fio_rados_connect(struct thread_data *td)
{
	struct rados_data *rados = td->io_ops_data;
	struct rados_options *o = td->eo;
	int r;
	const uint64_t file_size =
		td->o.size / (td->o.nr_files ? td->o.nr_files : 1u);
	struct fio_file *f;
	uint32_t i;

	if (o->cluster_name) {
		char *client_name = NULL;

		/*
		* If we specify cluser name, the rados_create2
		* will not assume 'client.'. name is considered
		* as a full type.id namestr
		*/
		if (o->client_name) {
			if (!index(o->client_name, '.')) {
				client_name = calloc(1, strlen("client.") +
					strlen(o->client_name) + 1);
				strcat(client_name, "client.");
				strcat(client_name, o->client_name);
			} else {
				client_name = o->client_name;
			}
		}

		r = rados_create2(&rados->cluster, o->cluster_name,
			client_name, 0);

		if (client_name && !index(o->client_name, '.'))
			free(client_name);
	} else
		r = rados_create(&rados->cluster, o->client_name);

	if (o->pool_name == NULL) {
		log_err("rados pool name must be provided.\n");
		goto failed_early;
	}

	if (r < 0) {
		log_err("rados_create failed.\n");
		goto failed_early;
	}

	r = rados_conf_read_file(rados->cluster, NULL);
	if (r < 0) {
		log_err("rados_conf_read_file failed.\n");
		goto failed_early;
	}

	r = rados_connect(rados->cluster);
	if (r < 0) {
		log_err("rados_connect failed.\n");
		goto failed_early;
	}

	r = rados_ioctx_create(rados->cluster, o->pool_name, &rados->io_ctx);
	if (r < 0) {
		log_err("rados_ioctx_create failed.\n");
		goto failed_shutdown;
	}

	for (i = 0; i < td->o.nr_files; i++) {
		f = td->files[i];
		f->real_file_size = file_size;
		if (o->touch_objects) {
			r = rados_write(rados->io_ctx, f->file_name, "", 0, 0);
			if (r < 0) {
				goto failed_obj_create;
			}
		}
	}
	return 0;

failed_obj_create:
	_fio_rados_rm_objects(td, rados);
	rados_ioctx_destroy(rados->io_ctx);
	rados->io_ctx = NULL;
failed_shutdown:
	rados_shutdown(rados->cluster);
	rados->cluster = NULL;
failed_early:
	return 1;
}

static void _fio_rados_disconnect(struct rados_data *rados)
{
	if (!rados)
		return;

	if (rados->io_ctx) {
		rados_ioctx_destroy(rados->io_ctx);
		rados->io_ctx = NULL;
	}

	if (rados->cluster) {
		rados_shutdown(rados->cluster);
		rados->cluster = NULL;
	}
}

static void fio_rados_cleanup(struct thread_data *td)
{
	struct rados_data *rados = td->io_ops_data;
	if (rados) {
		pthread_mutex_lock(&rados->completed_lock);
		while (rados->ops_scheduled != rados->ops_completed)
			pthread_cond_wait(&rados->completed_more_io, &rados->completed_lock);
		pthread_mutex_unlock(&rados->completed_lock);
		_fio_rados_rm_objects(td, rados);
		_fio_rados_disconnect(rados);
		free(rados->aio_events);
		free(rados);
	}
}

static void complete_callback(rados_completion_t cb, void *arg)
{
	struct fio_rados_iou *fri = (struct fio_rados_iou *)arg;
	struct rados_data *rados = fri->td->io_ops_data;
	assert(fri->completion);
	assert(rados_aio_is_complete(fri->completion));
	pthread_mutex_lock(&rados->completed_lock);
	flist_add_tail(&fri->list, &rados->completed_operations);
	rados->ops_completed++;
	pthread_mutex_unlock(&rados->completed_lock);
	pthread_cond_signal(&rados->completed_more_io);
}

static enum fio_q_status fio_rados_queue(struct thread_data *td,
					 struct io_u *io_u)
{
	struct rados_data *rados = td->io_ops_data;
	struct fio_rados_iou *fri = io_u->engine_data;
	char *object = io_u->file->file_name;
	int r = -1;

	fio_ro_check(td, io_u);

	if (io_u->ddir == DDIR_WRITE) {
		 r = rados_aio_create_completion(fri, complete_callback,
			NULL, &fri->completion);
		if (r < 0) {
			log_err("rados_aio_create_completion failed.\n");
			goto failed;
		}

		r = rados_aio_write(rados->io_ctx, object, fri->completion,
			io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
		if (r < 0) {
			log_err("rados_write failed.\n");
			goto failed_comp;
		}
		rados->ops_scheduled++;
		return FIO_Q_QUEUED;
	} else if (io_u->ddir == DDIR_READ) {
		r = rados_aio_create_completion(fri, complete_callback,
			NULL, &fri->completion);
		if (r < 0) {
			log_err("rados_aio_create_completion failed.\n");
			goto failed;
		}
		r = rados_aio_read(rados->io_ctx, object, fri->completion,
			io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
		if (r < 0) {
			log_err("rados_aio_read failed.\n");
			goto failed_comp;
		}
		rados->ops_scheduled++;
		return FIO_Q_QUEUED;
	} else if (io_u->ddir == DDIR_TRIM) {
		r = rados_aio_create_completion(fri, complete_callback,
			NULL , &fri->completion);
		if (r < 0) {
			log_err("rados_aio_create_completion failed.\n");
			goto failed;
		}
		fri->write_op = rados_create_write_op();
		if (fri->write_op == NULL) {
			log_err("rados_create_write_op failed.\n");
			goto failed_comp;
		}
		rados_write_op_zero(fri->write_op, io_u->offset,
			io_u->xfer_buflen);
		r = rados_aio_write_op_operate(fri->write_op, rados->io_ctx,
			fri->completion, object, NULL, 0);
		if (r < 0) {
			log_err("rados_aio_write_op_operate failed.\n");
			goto failed_write_op;
		}
		rados->ops_scheduled++;
		return FIO_Q_QUEUED;
	 }

	log_err("WARNING: Only DDIR_READ, DDIR_WRITE and DDIR_TRIM are supported!");

failed_write_op:
	rados_release_write_op(fri->write_op);
failed_comp:
	rados_aio_release(fri->completion);
failed:
	io_u->error = -r;
	td_verror(td, io_u->error, "xfer");
	return FIO_Q_COMPLETED;
}

static struct io_u *fio_rados_event(struct thread_data *td, int event)
{
	struct rados_data *rados = td->io_ops_data;
	return rados->aio_events[event];
}

int fio_rados_getevents(struct thread_data *td, unsigned int min,
	unsigned int max, const struct timespec *t)
{
	struct rados_data *rados = td->io_ops_data;
	unsigned int events = 0;
	struct fio_rados_iou *fri;

	pthread_mutex_lock(&rados->completed_lock);
	while (events < min) {
		while (flist_empty(&rados->completed_operations)) {
			pthread_cond_wait(&rados->completed_more_io, &rados->completed_lock);
		}
		assert(!flist_empty(&rados->completed_operations));
		
		fri = flist_first_entry(&rados->completed_operations, struct fio_rados_iou, list);
		assert(fri->completion);
		assert(rados_aio_is_complete(fri->completion));
		if (fri->write_op != NULL) {
			rados_release_write_op(fri->write_op);
			fri->write_op = NULL;
		}
		rados_aio_release(fri->completion);
		fri->completion = NULL;

		rados->aio_events[events] = fri->io_u;
		events ++;
		flist_del(&fri->list);
		if (events >= max) break;
	}
	pthread_mutex_unlock(&rados->completed_lock);
	return events;
}

static int fio_rados_setup(struct thread_data *td)
{
	struct rados_data *rados = NULL;
	int r;
	/* allocate engine specific structure to deal with librados. */
	r = _fio_setup_rados_data(td, &rados);
	if (r) {
		log_err("fio_setup_rados_data failed.\n");
		goto cleanup;
	}
	td->io_ops_data = rados;

	/* Force single process mode.
	*/
	td->o.use_thread = 1;

	/* connect in the main thread to determine to determine
	* the size of the given RADOS block device. And disconnect
	* later on.
	*/
	r = _fio_rados_connect(td);
	if (r) {
		log_err("fio_rados_connect failed.\n");
		goto cleanup;
	}
	rados->connected = true;

	return 0;
cleanup:
	fio_rados_cleanup(td);
	return r;
}

/* open/invalidate are noops. we set the FIO_DISKLESSIO flag in ioengine_ops to
   prevent fio from creating the files
*/
static int fio_rados_open(struct thread_data *td, struct fio_file *f)
{
	return 0;
}
static int fio_rados_invalidate(struct thread_data *td, struct fio_file *f)
{
	return 0;
}

static void fio_rados_io_u_free(struct thread_data *td, struct io_u *io_u)
{
	struct fio_rados_iou *fri = io_u->engine_data;

	if (fri) {
		io_u->engine_data = NULL;
		fri->td = NULL;
		if (fri->completion)
			rados_aio_release(fri->completion);
		if (fri->write_op)
			rados_release_write_op(fri->write_op);
		free(fri);
	}
}

static int fio_rados_io_u_init(struct thread_data *td, struct io_u *io_u)
{
	struct fio_rados_iou *fri;
	fri = calloc(1, sizeof(*fri));
	fri->io_u = io_u;
	fri->td = td;
	INIT_FLIST_HEAD(&fri->list);
	io_u->engine_data = fri;
	return 0;
}

/* ioengine_ops for get_ioengine() */
FIO_STATIC struct ioengine_ops ioengine = {
	.name = "rados",
	.version		= FIO_IOOPS_VERSION,
	.flags			= FIO_DISKLESSIO,
	.setup			= fio_rados_setup,
	.queue			= fio_rados_queue,
	.getevents		= fio_rados_getevents,
	.event			= fio_rados_event,
	.cleanup		= fio_rados_cleanup,
	.open_file		= fio_rados_open,
	.invalidate		= fio_rados_invalidate,
	.options		= options,
	.io_u_init		= fio_rados_io_u_init,
	.io_u_free		= fio_rados_io_u_free,
	.option_struct_size	= sizeof(struct rados_options),
};

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

static void fio_exit fio_rados_unregister(void)
{
	unregister_ioengine(&ioengine);
}
Commit	Line	Data
d5f9b0ea IF	1	/*
	2	* Ceph Rados engine
	3	*
	4	* IO engine using Ceph's RADOS interface to test low-level performance of
	5	* Ceph OSDs.
	6	*
	7	*/
	8
	9	#include <rados/librados.h>
	10	#include <pthread.h>
	11	#include "fio.h"
	12	#include "../optgroup.h"
	13
a21dafb0	14	struct rados_data {
1e30d8d0 AK	15	rados_t cluster;
	16	rados_ioctx_t io_ctx;
	17	struct io_u **aio_events;
	18	bool connected;
	19	pthread_mutex_t completed_lock;
95625c6d	20	pthread_cond_t completed_more_io;
1e30d8d0 AK	21	struct flist_head completed_operations;
	22	uint64_t ops_scheduled;
	23	uint64_t ops_completed;
a21dafb0 AK	24	};
a21dafb0 AK	25
d5f9b0ea	26	struct fio_rados_iou {
a21dafb0	27	struct flist_head list;
d5f9b0ea IF	28	struct thread_data *td;
	29	struct io_u *io_u;
	30	rados_completion_t completion;
	31	rados_write_op_t write_op;
	32	};
	33
d5f9b0ea IF	34	/* fio configuration options read from the job file */
	35	struct rados_options {
	36	void *pad;
	37	char *cluster_name;
	38	char *pool_name;
	39	char *client_name;
	40	int busy_poll;
2b728756	41	int touch_objects;
d5f9b0ea IF	42	};
	43
	44	static struct fio_option options[] = {
	45	{
	46	.name = "clustername",
	47	.lname = "ceph cluster name",
	48	.type = FIO_OPT_STR_STORE,
	49	.help = "Cluster name for ceph",
	50	.off1 = offsetof(struct rados_options, cluster_name),
	51	.category = FIO_OPT_C_ENGINE,
	52	.group = FIO_OPT_G_RBD,
	53	},
	54	{
	55	.name = "pool",
	56	.lname = "pool name to use",
	57	.type = FIO_OPT_STR_STORE,
	58	.help = "Ceph pool name to benchmark against",
	59	.off1 = offsetof(struct rados_options, pool_name),
	60	.category = FIO_OPT_C_ENGINE,
	61	.group = FIO_OPT_G_RBD,
	62	},
	63	{
	64	.name = "clientname",
	65	.lname = "rados engine clientname",
	66	.type = FIO_OPT_STR_STORE,
	67	.help = "Name of the ceph client to access RADOS engine",
	68	.off1 = offsetof(struct rados_options, client_name),
	69	.category = FIO_OPT_C_ENGINE,
	70	.group = FIO_OPT_G_RBD,
	71	},
	72	{
	73	.name = "busy_poll",
	74	.lname = "busy poll mode",
	75	.type = FIO_OPT_BOOL,
	76	.help = "Busy poll for completions instead of sleeping",
	77	.off1 = offsetof(struct rados_options, busy_poll),
	78	.def = "0",
	79	.category = FIO_OPT_C_ENGINE,
	80	.group = FIO_OPT_G_RBD,
	81	},
2b728756 AK	82	{
	83	.name = "touch_objects",
	84	.lname = "touch objects on start",
	85	.type = FIO_OPT_BOOL,
	86	.help = "Touch (create) objects on start",
	87	.off1 = offsetof(struct rados_options, touch_objects),
	88	.def = "1",
	89	.category = FIO_OPT_C_ENGINE,
	90	.group = FIO_OPT_G_RBD,
	91	},
d5f9b0ea IF	92	{
	93	.name = NULL,
	94	},
	95	};
	96
	97	static int _fio_setup_rados_data(struct thread_data *td,
	98	struct rados_data **rados_data_ptr)
	99	{
	100	struct rados_data *rados;
	101
	102	if (td->io_ops_data)
	103	return 0;
	104
	105	rados = calloc(1, sizeof(struct rados_data));
	106	if (!rados)
	107	goto failed;
	108
	109	rados->connected = false;
	110
	111	rados->aio_events = calloc(td->o.iodepth, sizeof(struct io_u *));
	112	if (!rados->aio_events)
	113	goto failed;
a21dafb0 AK	114	pthread_mutex_init(&rados->completed_lock, NULL);
	115	pthread_cond_init(&rados->completed_more_io, NULL);
	116	INIT_FLIST_HEAD(&rados->completed_operations);
1e30d8d0 AK	117	rados->ops_scheduled = 0;
1e30d8d0 AK	118	rados->ops_completed = 0;
d5f9b0ea IF	119	*rados_data_ptr = rados;
	120	return 0;
	121
	122	failed:
	123	if (rados) {
d5f9b0ea IF	124	if (rados->aio_events)
	125	free(rados->aio_events);
	126	free(rados);
	127	}
	128	return 1;
	129	}
	130
21f277b8	131	static void _fio_rados_rm_objects(struct thread_data td, struct rados_data rados)
d5f9b0ea IF	132	{
d5f9b0ea IF	133	size_t i;
21f277b8 AK	134	for (i = 0; i < td->o.nr_files; i++) {
	135	struct fio_file *f = td->files[i];
	136	rados_remove(rados->io_ctx, f->file_name);
d5f9b0ea IF	137	}
	138	}
	139
	140	static int _fio_rados_connect(struct thread_data *td)
	141	{
	142	struct rados_data *rados = td->io_ops_data;
	143	struct rados_options *o = td->eo;
	144	int r;
	145	const uint64_t file_size =
	146	td->o.size / (td->o.nr_files ? td->o.nr_files : 1u);
	147	struct fio_file *f;
	148	uint32_t i;
d5f9b0ea IF	149
	150	if (o->cluster_name) {
	151	char *client_name = NULL;
	152
	153	/*
	154	* If we specify cluser name, the rados_create2
	155	* will not assume 'client.'. name is considered
	156	* as a full type.id namestr
	157	*/
	158	if (o->client_name) {
	159	if (!index(o->client_name, '.')) {
	160	client_name = calloc(1, strlen("client.") +
	161	strlen(o->client_name) + 1);
	162	strcat(client_name, "client.");
	163	strcat(client_name, o->client_name);
	164	} else {
	165	client_name = o->client_name;
	166	}
	167	}
	168
	169	r = rados_create2(&rados->cluster, o->cluster_name,
	170	client_name, 0);
	171
	172	if (client_name && !index(o->client_name, '.'))
	173	free(client_name);
	174	} else
	175	r = rados_create(&rados->cluster, o->client_name);
	176
21f277b8 AK	177	if (o->pool_name == NULL) {
	178	log_err("rados pool name must be provided.\n");
	179	goto failed_early;
	180	}
	181
d5f9b0ea IF	182	if (r < 0) {
	183	log_err("rados_create failed.\n");
	184	goto failed_early;
	185	}
	186
	187	r = rados_conf_read_file(rados->cluster, NULL);
	188	if (r < 0) {
	189	log_err("rados_conf_read_file failed.\n");
	190	goto failed_early;
	191	}
	192
	193	r = rados_connect(rados->cluster);
	194	if (r < 0) {
	195	log_err("rados_connect failed.\n");
	196	goto failed_early;
	197	}
	198
	199	r = rados_ioctx_create(rados->cluster, o->pool_name, &rados->io_ctx);
	200	if (r < 0) {
	201	log_err("rados_ioctx_create failed.\n");
	202	goto failed_shutdown;
	203	}
	204
21f277b8	205	for (i = 0; i < td->o.nr_files; i++) {
d5f9b0ea IF	206	f = td->files[i];
d5f9b0ea IF	207	f->real_file_size = file_size;
2b728756 AK	208	if (o->touch_objects) {
	209	r = rados_write(rados->io_ctx, f->file_name, "", 0, 0);
	210	if (r < 0) {
	211	goto failed_obj_create;
	212	}
d5f9b0ea IF	213	}
d5f9b0ea IF	214	}
21f277b8	215	return 0;
d5f9b0ea IF	216
d5f9b0ea IF	217	failed_obj_create:
21f277b8	218	_fio_rados_rm_objects(td, rados);
d5f9b0ea IF	219	rados_ioctx_destroy(rados->io_ctx);
	220	rados->io_ctx = NULL;
	221	failed_shutdown:
	222	rados_shutdown(rados->cluster);
	223	rados->cluster = NULL;
	224	failed_early:
	225	return 1;
	226	}
	227
	228	static void _fio_rados_disconnect(struct rados_data *rados)
	229	{
	230	if (!rados)
	231	return;
	232
d5f9b0ea IF	233	if (rados->io_ctx) {
	234	rados_ioctx_destroy(rados->io_ctx);
	235	rados->io_ctx = NULL;
	236	}
	237
	238	if (rados->cluster) {
	239	rados_shutdown(rados->cluster);
	240	rados->cluster = NULL;
	241	}
	242	}
	243
	244	static void fio_rados_cleanup(struct thread_data *td)
	245	{
	246	struct rados_data *rados = td->io_ops_data;
d5f9b0ea	247	if (rados) {
1e30d8d0 AK	248	pthread_mutex_lock(&rados->completed_lock);
	249	while (rados->ops_scheduled != rados->ops_completed)
	250	pthread_cond_wait(&rados->completed_more_io, &rados->completed_lock);
	251	pthread_mutex_unlock(&rados->completed_lock);
21f277b8	252	_fio_rados_rm_objects(td, rados);
d5f9b0ea	253	_fio_rados_disconnect(rados);
d5f9b0ea IF	254	free(rados->aio_events);
	255	free(rados);
	256	}
	257	}
	258
a21dafb0 AK	259	static void complete_callback(rados_completion_t cb, void *arg)
	260	{
	261	struct fio_rados_iou fri = (struct fio_rados_iou )arg;
	262	struct rados_data *rados = fri->td->io_ops_data;
	263	assert(fri->completion);
	264	assert(rados_aio_is_complete(fri->completion));
	265	pthread_mutex_lock(&rados->completed_lock);
	266	flist_add_tail(&fri->list, &rados->completed_operations);
1e30d8d0	267	rados->ops_completed++;
a21dafb0 AK	268	pthread_mutex_unlock(&rados->completed_lock);
	269	pthread_cond_signal(&rados->completed_more_io);
	270	}
	271
2e4ef4fb JA	272	static enum fio_q_status fio_rados_queue(struct thread_data *td,
2e4ef4fb JA	273	struct io_u *io_u)
d5f9b0ea IF	274	{
	275	struct rados_data *rados = td->io_ops_data;
	276	struct fio_rados_iou *fri = io_u->engine_data;
21f277b8	277	char *object = io_u->file->file_name;
d5f9b0ea IF	278	int r = -1;
	279
	280	fio_ro_check(td, io_u);
	281
	282	if (io_u->ddir == DDIR_WRITE) {
a21dafb0	283	r = rados_aio_create_completion(fri, complete_callback,
d5f9b0ea IF	284	NULL, &fri->completion);
	285	if (r < 0) {
	286	log_err("rados_aio_create_completion failed.\n");
	287	goto failed;
	288	}
	289
	290	r = rados_aio_write(rados->io_ctx, object, fri->completion,
	291	io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
	292	if (r < 0) {
	293	log_err("rados_write failed.\n");
	294	goto failed_comp;
	295	}
1e30d8d0	296	rados->ops_scheduled++;
d5f9b0ea IF	297	return FIO_Q_QUEUED;
d5f9b0ea IF	298	} else if (io_u->ddir == DDIR_READ) {
a21dafb0	299	r = rados_aio_create_completion(fri, complete_callback,
d5f9b0ea IF	300	NULL, &fri->completion);
	301	if (r < 0) {
	302	log_err("rados_aio_create_completion failed.\n");
	303	goto failed;
	304	}
	305	r = rados_aio_read(rados->io_ctx, object, fri->completion,
	306	io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
	307	if (r < 0) {
	308	log_err("rados_aio_read failed.\n");
	309	goto failed_comp;
	310	}
1e30d8d0	311	rados->ops_scheduled++;
d5f9b0ea IF	312	return FIO_Q_QUEUED;
d5f9b0ea IF	313	} else if (io_u->ddir == DDIR_TRIM) {
a21dafb0	314	r = rados_aio_create_completion(fri, complete_callback,
d5f9b0ea IF	315	NULL , &fri->completion);
	316	if (r < 0) {
	317	log_err("rados_aio_create_completion failed.\n");
	318	goto failed;
	319	}
	320	fri->write_op = rados_create_write_op();
	321	if (fri->write_op == NULL) {
	322	log_err("rados_create_write_op failed.\n");
	323	goto failed_comp;
	324	}
	325	rados_write_op_zero(fri->write_op, io_u->offset,
	326	io_u->xfer_buflen);
	327	r = rados_aio_write_op_operate(fri->write_op, rados->io_ctx,
	328	fri->completion, object, NULL, 0);
	329	if (r < 0) {
	330	log_err("rados_aio_write_op_operate failed.\n");
	331	goto failed_write_op;
	332	}
1e30d8d0	333	rados->ops_scheduled++;
d5f9b0ea IF	334	return FIO_Q_QUEUED;
	335	}
	336
	337	log_err("WARNING: Only DDIR_READ, DDIR_WRITE and DDIR_TRIM are supported!");
	338
	339	failed_write_op:
	340	rados_release_write_op(fri->write_op);
	341	failed_comp:
	342	rados_aio_release(fri->completion);
	343	failed:
	344	io_u->error = -r;
	345	td_verror(td, io_u->error, "xfer");
	346	return FIO_Q_COMPLETED;
	347	}
	348
	349	static struct io_u fio_rados_event(struct thread_data td, int event)
	350	{
	351	struct rados_data *rados = td->io_ops_data;
	352	return rados->aio_events[event];
	353	}
	354
	355	int fio_rados_getevents(struct thread_data *td, unsigned int min,
	356	unsigned int max, const struct timespec *t)
	357	{
	358	struct rados_data *rados = td->io_ops_data;
d5f9b0ea	359	unsigned int events = 0;
d5f9b0ea	360	struct fio_rados_iou *fri;
a21dafb0 AK	361
	362	pthread_mutex_lock(&rados->completed_lock);
	363	while (events < min) {
	364	while (flist_empty(&rados->completed_operations)) {
	365	pthread_cond_wait(&rados->completed_more_io, &rados->completed_lock);
d5f9b0ea	366	}
a21dafb0 AK	367	assert(!flist_empty(&rados->completed_operations));
a21dafb0 AK	368
6a3f4e6c	369	fri = flist_first_entry(&rados->completed_operations, struct fio_rados_iou, list);
a21dafb0 AK	370	assert(fri->completion);
	371	assert(rados_aio_is_complete(fri->completion));
	372	if (fri->write_op != NULL) {
	373	rados_release_write_op(fri->write_op);
	374	fri->write_op = NULL;
	375	}
	376	rados_aio_release(fri->completion);
	377	fri->completion = NULL;
	378
	379	rados->aio_events[events] = fri->io_u;
	380	events ++;
	381	flist_del(&fri->list);
	382	if (events >= max) break;
	383	}
	384	pthread_mutex_unlock(&rados->completed_lock);
	385	return events;
d5f9b0ea IF	386	}
	387
	388	static int fio_rados_setup(struct thread_data *td)
	389	{
	390	struct rados_data *rados = NULL;
	391	int r;
	392	/* allocate engine specific structure to deal with librados. */
	393	r = _fio_setup_rados_data(td, &rados);
	394	if (r) {
	395	log_err("fio_setup_rados_data failed.\n");
	396	goto cleanup;
	397	}
	398	td->io_ops_data = rados;
	399
	400	/* Force single process mode.
	401	*/
	402	td->o.use_thread = 1;
	403
	404	/* connect in the main thread to determine to determine
	405	* the size of the given RADOS block device. And disconnect
	406	* later on.
	407	*/
	408	r = _fio_rados_connect(td);
	409	if (r) {
	410	log_err("fio_rados_connect failed.\n");
	411	goto cleanup;
	412	}
	413	rados->connected = true;
	414
	415	return 0;
	416	cleanup:
	417	fio_rados_cleanup(td);
	418	return r;
	419	}
	420
	421	/* open/invalidate are noops. we set the FIO_DISKLESSIO flag in ioengine_ops to
	422	prevent fio from creating the files
	423	*/
	424	static int fio_rados_open(struct thread_data td, struct fio_file f)
	425	{
	426	return 0;
	427	}
	428	static int fio_rados_invalidate(struct thread_data td, struct fio_file f)
	429	{
	430	return 0;
	431	}
	432
	433	static void fio_rados_io_u_free(struct thread_data td, struct io_u io_u)
	434	{
	435	struct fio_rados_iou *fri = io_u->engine_data;
	436
	437	if (fri) {
	438	io_u->engine_data = NULL;
	439	fri->td = NULL;
	440	if (fri->completion)
	441	rados_aio_release(fri->completion);
	442	if (fri->write_op)
	443	rados_release_write_op(fri->write_op);
	444	free(fri);
	445	}
	446	}
	447
	448	static int fio_rados_io_u_init(struct thread_data td, struct io_u io_u)
	449	{
450	struct fio_rados_iou *fri;
451	fri = calloc(1, sizeof(*fri));
452	fri->io_u = io_u;
453	fri->td = td;
a21dafb0	454	INIT_FLIST_HEAD(&fri->list);
d5f9b0ea IF	455	io_u->engine_data = fri;
	456	return 0;
	457	}
	458
	459	/* ioengine_ops for get_ioengine() */
5a8a6a03	460	FIO_STATIC struct ioengine_ops ioengine = {
d5f9b0ea IF	461	.name = "rados",
	462	.version = FIO_IOOPS_VERSION,
	463	.flags = FIO_DISKLESSIO,
	464	.setup = fio_rados_setup,
	465	.queue = fio_rados_queue,
	466	.getevents = fio_rados_getevents,
	467	.event = fio_rados_event,
	468	.cleanup = fio_rados_cleanup,
	469	.open_file = fio_rados_open,
	470	.invalidate = fio_rados_invalidate,
	471	.options = options,
	472	.io_u_init = fio_rados_io_u_init,
	473	.io_u_free = fio_rados_io_u_free,
	474	.option_struct_size = sizeof(struct rados_options),
	475	};
	476
	477	static void fio_init fio_rados_register(void)
	478	{
	479	register_ioengine(&ioengine);
	480	}
	481
	482	static void fio_exit fio_rados_unregister(void)
	483	{
	484	unregister_ioengine(&ioengine);
	485	}