[fio.git] / engines / pmemblk.c

/*
 * pmemblk: IO engine that uses NVML libpmemblk to read and write data
 *
 * Copyright (C) 2016 Hewlett Packard Enterprise Development LP
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License,
 * version 2 as published by the Free Software Foundation..
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the Free
 * Software Foundation, Inc., 59 Temple Place, Suite 330,
 * Boston, MA 02111-1307 USA
 */

/*
 * pmemblk engine
 *
 * IO engine that uses libpmemblk to read and write data
 *
 * To use:
 *   ioengine=pmemblk
 *
 * Other relevant settings:
 *   thread=1   REQUIRED
 *   iodepth=1
 *   direct=1
 *   unlink=1
 *   filename=/mnt/pmem0/fiotestfile,BSIZE,FSIZEMiB
 *
 *   thread must be set to 1 for pmemblk as multiple processes cannot
 *     open the same block pool file.
 *
 *   iodepth should be set to 1 as pmemblk is always synchronous.
 *   Use numjobs to scale up.
 *
 *   direct=1 is implied as pmemblk is always direct. A warning message
 *   is printed if this is not specified.
 *
 *   unlink=1 removes the block pool file after testing, and is optional.
 *
 *   The pmem device must have a DAX-capable filesystem and be mounted
 *   with DAX enabled.  filename must point to a file on that filesystem.
 *
 *   Example:
 *     mkfs.xfs /dev/pmem0
 *     mkdir /mnt/pmem0
 *     mount -o dax /dev/pmem0 /mnt/pmem0
 *
 *   When specifying the filename, if the block pool file does not already
 *   exist, then the pmemblk engine creates the pool file if you specify
 *   the block and file sizes.  BSIZE is the block size in bytes.
 *   FSIZEMB is the pool file size in MiB.
 *
 *   See examples/pmemblk.fio for more.
 *
 */

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/uio.h>
#include <errno.h>
#include <assert.h>
#include <string.h>
#include <libpmem.h>
#include <libpmemblk.h>

#include "../fio.h"

/*
 * libpmemblk
 */
typedef struct fio_pmemblk_file *fio_pmemblk_file_t;

struct fio_pmemblk_file {
	fio_pmemblk_file_t pmb_next;
	char *pmb_filename;
	uint64_t pmb_refcnt;
	PMEMblkpool *pmb_pool;
	size_t pmb_bsize;
	size_t pmb_nblocks;
};
#define FIOFILEPMBSET(_f, _v)  do {                 \
	(_f)->engine_data = (uint64_t)(uintptr_t)(_v);  \
} while(0)
#define FIOFILEPMBGET(_f)  ((fio_pmemblk_file_t)((_f)->engine_data))

static fio_pmemblk_file_t Cache;

static pthread_mutex_t CacheLock = PTHREAD_MUTEX_INITIALIZER;

#define PMB_CREATE   (0x0001)	/* should create file */

fio_pmemblk_file_t fio_pmemblk_cache_lookup(const char *filename)
{
	fio_pmemblk_file_t i;

	for (i = Cache; i != NULL; i = i->pmb_next)
		if (!strcmp(filename, i->pmb_filename))
			return i;

	return NULL;
}

static void fio_pmemblk_cache_insert(fio_pmemblk_file_t pmb)
{
	pmb->pmb_next = Cache;
	Cache = pmb;
}

static void fio_pmemblk_cache_remove(fio_pmemblk_file_t pmb)
{
	fio_pmemblk_file_t i;

	if (pmb == Cache) {
		Cache = Cache->pmb_next;
		pmb->pmb_next = NULL;
		return;
	}

	for (i = Cache; i != NULL; i = i->pmb_next)
		if (pmb == i->pmb_next) {
			i->pmb_next = i->pmb_next->pmb_next;
			pmb->pmb_next = NULL;
			return;
		}
}

/*
 * to control block size and gross file size at the libpmemblk
 * level, we allow the block size and file size to be appended
 * to the file name:
 *
 *   path[,bsize,fsizemib]
 *
 * note that we do not use the fio option "filesize" to dictate
 * the file size because we can only give libpmemblk the gross
 * file size, which is different from the net or usable file
 * size (which is probably what fio wants).
 *
 * the final path without the parameters is returned in ppath.
 * the block size and file size are returned in pbsize and fsize.
 *
 * note that the user specifies the file size in MiB, but
 * we return bytes from here.
 */
static void pmb_parse_path(const char *pathspec, char **ppath, uint64_t *pbsize,
			   uint64_t *pfsize)
{
	char *path;
	char *s;
	uint64_t bsize;
	uint64_t fsizemib;

	path = strdup(pathspec);
	if (!path) {
		*ppath = NULL;
		return;
	}

	/* extract sizes, if given */
	s = strrchr(path, ',');
	if (s && (fsizemib = strtoull(s + 1, NULL, 10))) {
		*s = 0;
		s = strrchr(path, ',');
		if (s && (bsize = strtoull(s + 1, NULL, 10))) {
			*s = 0;
			*ppath = path;
			*pbsize = bsize;
			*pfsize = fsizemib << 20;
			return;
		}
	}

	/* size specs not found */
	strcpy(path, pathspec);
	*ppath = path;
	*pbsize = 0;
	*pfsize = 0;
}

static fio_pmemblk_file_t pmb_open(const char *pathspec, int flags)
{
	fio_pmemblk_file_t pmb;
	char *path = NULL;
	uint64_t bsize = 0;
	uint64_t fsize = 0;

	pmb_parse_path(pathspec, &path, &bsize, &fsize);
	if (!path)
		return NULL;

	pthread_mutex_lock(&CacheLock);

	pmb = fio_pmemblk_cache_lookup(path);
	if (!pmb) {
		pmb = malloc(sizeof(*pmb));
		if (!pmb)
			goto error;

		/* try opening existing first, create it if needed */
		pmb->pmb_pool = pmemblk_open(path, bsize);
		if (!pmb->pmb_pool && (errno == ENOENT) &&
		    (flags & PMB_CREATE) && (0 < fsize) && (0 < bsize)) {
			pmb->pmb_pool =
			    pmemblk_create(path, bsize, fsize, 0644);
		}
		if (!pmb->pmb_pool) {
			log_err("pmemblk: unable to open pmemblk pool file %s (%s)\n",
			     path, strerror(errno));
			goto error;
		}

		pmb->pmb_filename = path;
		pmb->pmb_next = NULL;
		pmb->pmb_refcnt = 0;
		pmb->pmb_bsize = pmemblk_bsize(pmb->pmb_pool);
		pmb->pmb_nblocks = pmemblk_nblock(pmb->pmb_pool);

		fio_pmemblk_cache_insert(pmb);
	}

	pmb->pmb_refcnt += 1;

	pthread_mutex_unlock(&CacheLock);

	return pmb;

error:
	if (pmb) {
		if (pmb->pmb_pool)
			pmemblk_close(pmb->pmb_pool);
		pmb->pmb_pool = NULL;
		pmb->pmb_filename = NULL;
		free(pmb);
	}
	if (path)
		free(path);

	pthread_mutex_unlock(&CacheLock);
	return NULL;
}

static void pmb_close(fio_pmemblk_file_t pmb, const bool keep)
{
	pthread_mutex_lock(&CacheLock);

	pmb->pmb_refcnt--;

	if (!keep && !pmb->pmb_refcnt) {
		pmemblk_close(pmb->pmb_pool);
		pmb->pmb_pool = NULL;
		free(pmb->pmb_filename);
		pmb->pmb_filename = NULL;
		fio_pmemblk_cache_remove(pmb);
		free(pmb);
	}

	pthread_mutex_unlock(&CacheLock);
}

static int pmb_get_flags(struct thread_data *td, uint64_t *pflags)
{
	static int thread_warned = 0;
	static int odirect_warned = 0;

	uint64_t flags = 0;

	if (!td->o.use_thread) {
		if (!thread_warned) {
			thread_warned = 1;
			log_err("pmemblk: must set thread=1 for pmemblk engine\n");
		}
		return 1;
	}

	if (!td->o.odirect && !odirect_warned) {
		odirect_warned = 1;
		log_info("pmemblk: direct == 0, but pmemblk is always direct\n");
	}

	if (td->o.allow_create)
		flags |= PMB_CREATE;

	(*pflags) = flags;
	return 0;
}

static int fio_pmemblk_open_file(struct thread_data *td, struct fio_file *f)
{
	uint64_t flags = 0;
	fio_pmemblk_file_t pmb;

	if (pmb_get_flags(td, &flags))
		return 1;

	pmb = pmb_open(f->file_name, flags);
	if (!pmb)
		return 1;

	FIOFILEPMBSET(f, pmb);
	return 0;
}

static int fio_pmemblk_close_file(struct thread_data fio_unused *td,
				  struct fio_file *f)
{
	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);

	if (pmb)
		pmb_close(pmb, false);

	FIOFILEPMBSET(f, NULL);
	return 0;
}

static int fio_pmemblk_get_file_size(struct thread_data *td, struct fio_file *f)
{
	uint64_t flags = 0;
	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);

	if (fio_file_size_known(f))
		return 0;

	if (!pmb) {
		if (pmb_get_flags(td, &flags))
			return 1;
		pmb = pmb_open(f->file_name, flags);
		if (!pmb)
			return 1;
	}

	f->real_file_size = pmb->pmb_bsize * pmb->pmb_nblocks;

	fio_file_set_size_known(f);

	if (!FIOFILEPMBGET(f))
		pmb_close(pmb, true);

	return 0;
}

static int fio_pmemblk_queue(struct thread_data *td, struct io_u *io_u)
{
	struct fio_file *f = io_u->file;
	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);

	unsigned long long off;
	unsigned long len;
	void *buf;

	fio_ro_check(td, io_u);

	switch (io_u->ddir) {
	case DDIR_READ:
	case DDIR_WRITE:
		off = io_u->offset;
		len = io_u->xfer_buflen;

		io_u->error = EINVAL;
		if (off % pmb->pmb_bsize)
			break;
		if (len % pmb->pmb_bsize)
			break;
		if ((off + len) / pmb->pmb_bsize > pmb->pmb_nblocks)
			break;

		io_u->error = 0;
		buf = io_u->xfer_buf;
		off /= pmb->pmb_bsize;
		len /= pmb->pmb_bsize;
		while (0 < len) {
			if (io_u->ddir == DDIR_READ &&
			   0 != pmemblk_read(pmb->pmb_pool, buf, off)) {
				io_u->error = errno;
				break;
			} else if (0 != pmemblk_write(pmb->pmb_pool, buf, off)) {
				io_u->error = errno;
				break;
			}
			buf += pmb->pmb_bsize;
			off++;
			len--;
		}
		off *= pmb->pmb_bsize;
		len *= pmb->pmb_bsize;
		io_u->resid = io_u->xfer_buflen - (off - io_u->offset);
		break;
	case DDIR_SYNC:
	case DDIR_DATASYNC:
	case DDIR_SYNC_FILE_RANGE:
		/* we're always sync'd */
		io_u->error = 0;
		break;
	default:
		io_u->error = EINVAL;
		break;
	}

	return FIO_Q_COMPLETED;
}

static int fio_pmemblk_unlink_file(struct thread_data *td, struct fio_file *f)
{
	char *path = NULL;
	uint64_t bsize = 0;
	uint64_t fsize = 0;

	/*
	 * we need our own unlink in case the user has specified
	 * the block and file sizes in the path name.  we parse
	 * the file_name to determine the file name we actually used.
	 */

	pmb_parse_path(f->file_name, &path, &bsize, &fsize);
	if (!path)
		return ENOENT;

	unlink(path);
	free(path);
	return 0;
}

static struct ioengine_ops ioengine = {
	.name = "pmemblk",
	.version = FIO_IOOPS_VERSION,
	.queue = fio_pmemblk_queue,
	.open_file = fio_pmemblk_open_file,
	.close_file = fio_pmemblk_close_file,
	.get_file_size = fio_pmemblk_get_file_size,
	.unlink_file = fio_pmemblk_unlink_file,
	.flags = FIO_SYNCIO | FIO_DISKLESSIO | FIO_NOEXTEND | FIO_NODISKUTIL,
};

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

static void fio_exit fio_pmemblk_unregister(void)
{
	unregister_ioengine(&ioengine);
}
Commit	Line	Data
43f3cec2 BB	1	/*
	2	* pmemblk: IO engine that uses NVML libpmemblk to read and write data
	3	*
	4	* Copyright (C) 2016 Hewlett Packard Enterprise Development LP
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License,
	8	* version 2 as published by the Free Software Foundation..
	9	*
	10	* This program is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	13	* GNU General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU General Public
	16	* License along with this program; if not, write to the Free
	17	* Software Foundation, Inc., 59 Temple Place, Suite 330,
	18	* Boston, MA 02111-1307 USA
	19	*/
	20
43f3cec2 BB	21	/*
	22	* pmemblk engine
	23	*
	24	* IO engine that uses libpmemblk to read and write data
	25	*
f1d85c6e BB	26	* To use:
f1d85c6e BB	27	* ioengine=pmemblk
43f3cec2 BB	28	*
43f3cec2 BB	29	* Other relevant settings:
0cbb3f53	30	* thread=1 REQUIRED
43f3cec2 BB	31	* iodepth=1
43f3cec2 BB	32	* direct=1
43f3cec2	33	* unlink=1
0cbb3f53	34	* filename=/mnt/pmem0/fiotestfile,BSIZE,FSIZEMiB
43f3cec2 BB	35	*
	36	* thread must be set to 1 for pmemblk as multiple processes cannot
	37	* open the same block pool file.
	38	*
	39	* iodepth should be set to 1 as pmemblk is always synchronous.
	40	* Use numjobs to scale up.
	41	*
0cbb3f53 RE	42	* direct=1 is implied as pmemblk is always direct. A warning message
	43	* is printed if this is not specified.
	44	*
	45	* unlink=1 removes the block pool file after testing, and is optional.
	46	*
	47	* The pmem device must have a DAX-capable filesystem and be mounted
	48	* with DAX enabled. filename must point to a file on that filesystem.
43f3cec2	49	*
0cbb3f53 RE	50	* Example:
	51	* mkfs.xfs /dev/pmem0
	52	* mkdir /mnt/pmem0
	53	* mount -o dax /dev/pmem0 /mnt/pmem0
43f3cec2 BB	54	*
43f3cec2 BB	55	* When specifying the filename, if the block pool file does not already
0cbb3f53	56	* exist, then the pmemblk engine creates the pool file if you specify
43f3cec2	57	* the block and file sizes. BSIZE is the block size in bytes.
0cbb3f53	58	* FSIZEMB is the pool file size in MiB.
43f3cec2	59	*
f1d85c6e BB	60	* See examples/pmemblk.fio for more.
f1d85c6e BB	61	*
43f3cec2 BB	62	*/
	63
	64	#include <stdio.h>
	65	#include <stdlib.h>
	66	#include <unistd.h>
	67	#include <sys/uio.h>
	68	#include <errno.h>
	69	#include <assert.h>
43f3cec2	70	#include <string.h>
cf8775b8 RE	71	#include <libpmem.h>
cf8775b8 RE	72	#include <libpmemblk.h>
43f3cec2 BB	73
	74	#include "../fio.h"
	75
43f3cec2 BB	76	/*
	77	* libpmemblk
	78	*/
c086b314	79	typedef struct fio_pmemblk_file *fio_pmemblk_file_t;
630431cc	80
43f3cec2	81	struct fio_pmemblk_file {
c086b314 JA	82	fio_pmemblk_file_t pmb_next;
	83	char *pmb_filename;
	84	uint64_t pmb_refcnt;
	85	PMEMblkpool *pmb_pool;
	86	size_t pmb_bsize;
	87	size_t pmb_nblocks;
43f3cec2 BB	88	};
	89	#define FIOFILEPMBSET(_f, _v) do { \
	90	(_f)->engine_data = (uint64_t)(uintptr_t)(_v); \
	91	} while(0)
	92	#define FIOFILEPMBGET(_f) ((fio_pmemblk_file_t)((_f)->engine_data))
	93
630431cc	94	static fio_pmemblk_file_t Cache;
43f3cec2	95
c086b314	96	static pthread_mutex_t CacheLock = PTHREAD_MUTEX_INITIALIZER;
43f3cec2	97
c086b314	98	#define PMB_CREATE (0x0001) /* should create file */
43f3cec2	99
c086b314	100	fio_pmemblk_file_t fio_pmemblk_cache_lookup(const char *filename)
43f3cec2	101	{
c086b314	102	fio_pmemblk_file_t i;
43f3cec2 BB	103
43f3cec2 BB	104	for (i = Cache; i != NULL; i = i->pmb_next)
630431cc	105	if (!strcmp(filename, i->pmb_filename))
43f3cec2 BB	106	return i;
	107
	108	return NULL;
630431cc	109	}
43f3cec2	110
c086b314	111	static void fio_pmemblk_cache_insert(fio_pmemblk_file_t pmb)
43f3cec2 BB	112	{
	113	pmb->pmb_next = Cache;
	114	Cache = pmb;
630431cc	115	}
43f3cec2	116
c086b314	117	static void fio_pmemblk_cache_remove(fio_pmemblk_file_t pmb)
43f3cec2	118	{
c086b314	119	fio_pmemblk_file_t i;
43f3cec2 BB	120
	121	if (pmb == Cache) {
	122	Cache = Cache->pmb_next;
	123	pmb->pmb_next = NULL;
	124	return;
	125	}
	126
	127	for (i = Cache; i != NULL; i = i->pmb_next)
	128	if (pmb == i->pmb_next) {
	129	i->pmb_next = i->pmb_next->pmb_next;
	130	pmb->pmb_next = NULL;
	131	return;
	132	}
630431cc	133	}
43f3cec2 BB	134
	135	/*
	136	* to control block size and gross file size at the libpmemblk
	137	* level, we allow the block size and file size to be appended
	138	* to the file name:
	139	*
0cbb3f53	140	* path[,bsize,fsizemib]
43f3cec2 BB	141	*
	142	* note that we do not use the fio option "filesize" to dictate
	143	* the file size because we can only give libpmemblk the gross
	144	* file size, which is different from the net or usable file
	145	* size (which is probably what fio wants).
	146	*
	147	* the final path without the parameters is returned in ppath.
	148	* the block size and file size are returned in pbsize and fsize.
	149	*
0cbb3f53	150	* note that the user specifies the file size in MiB, but
43f3cec2 BB	151	* we return bytes from here.
43f3cec2 BB	152	*/
630431cc JA	153	static void pmb_parse_path(const char pathspec, char ppath, uint64_t pbsize,
630431cc JA	154	uint64_t *pfsize)
43f3cec2	155	{
c086b314 JA	156	char *path;
	157	char *s;
	158	uint64_t bsize;
0cbb3f53	159	uint64_t fsizemib;
43f3cec2 BB	160
43f3cec2 BB	161	path = strdup(pathspec);
630431cc	162	if (!path) {
43f3cec2 BB	163	*ppath = NULL;
	164	return;
	165	}
	166
	167	/* extract sizes, if given */
	168	s = strrchr(path, ',');
0cbb3f53	169	if (s && (fsizemib = strtoull(s + 1, NULL, 10))) {
43f3cec2 BB	170	*s = 0;
43f3cec2 BB	171	s = strrchr(path, ',');
c086b314	172	if (s && (bsize = strtoull(s + 1, NULL, 10))) {
43f3cec2	173	*s = 0;
c086b314	174	*ppath = path;
43f3cec2	175	*pbsize = bsize;
0cbb3f53	176	*pfsize = fsizemib << 20;
43f3cec2 BB	177	return;
	178	}
	179	}
	180
	181	/* size specs not found */
	182	strcpy(path, pathspec);
c086b314	183	*ppath = path;
43f3cec2 BB	184	*pbsize = 0;
43f3cec2 BB	185	*pfsize = 0;
630431cc	186	}
43f3cec2	187
630431cc	188	static fio_pmemblk_file_t pmb_open(const char *pathspec, int flags)
43f3cec2	189	{
c086b314 JA	190	fio_pmemblk_file_t pmb;
	191	char *path = NULL;
	192	uint64_t bsize = 0;
	193	uint64_t fsize = 0;
43f3cec2 BB	194
43f3cec2 BB	195	pmb_parse_path(pathspec, &path, &bsize, &fsize);
630431cc	196	if (!path)
43f3cec2 BB	197	return NULL;
43f3cec2 BB	198
de88999f	199	pthread_mutex_lock(&CacheLock);
43f3cec2 BB	200
43f3cec2 BB	201	pmb = fio_pmemblk_cache_lookup(path);
630431cc	202	if (!pmb) {
43f3cec2	203	pmb = malloc(sizeof(*pmb));
630431cc	204	if (!pmb)
43f3cec2 BB	205	goto error;
	206
	207	/* try opening existing first, create it if needed */
	208	pmb->pmb_pool = pmemblk_open(path, bsize);
630431cc	209	if (!pmb->pmb_pool && (errno == ENOENT) &&
c086b314 JA	210	(flags & PMB_CREATE) && (0 < fsize) && (0 < bsize)) {
	211	pmb->pmb_pool =
	212	pmemblk_create(path, bsize, fsize, 0644);
43f3cec2	213	}
630431cc	214	if (!pmb->pmb_pool) {
a0998600 RE	215	log_err("pmemblk: unable to open pmemblk pool file %s (%s)\n",
a0998600 RE	216	path, strerror(errno));
43f3cec2 BB	217	goto error;
	218	}
	219
	220	pmb->pmb_filename = path;
c086b314 JA	221	pmb->pmb_next = NULL;
	222	pmb->pmb_refcnt = 0;
	223	pmb->pmb_bsize = pmemblk_bsize(pmb->pmb_pool);
	224	pmb->pmb_nblocks = pmemblk_nblock(pmb->pmb_pool);
43f3cec2 BB	225
	226	fio_pmemblk_cache_insert(pmb);
	227	}
	228
	229	pmb->pmb_refcnt += 1;
	230
de88999f	231	pthread_mutex_unlock(&CacheLock);
c086b314	232
43f3cec2 BB	233	return pmb;
	234
	235	error:
630431cc JA	236	if (pmb) {
630431cc JA	237	if (pmb->pmb_pool)
43f3cec2	238	pmemblk_close(pmb->pmb_pool);
c086b314	239	pmb->pmb_pool = NULL;
43f3cec2 BB	240	pmb->pmb_filename = NULL;
	241	free(pmb);
	242	}
630431cc	243	if (path)
43f3cec2	244	free(path);
de88999f JA	245
de88999f JA	246	pthread_mutex_unlock(&CacheLock);
43f3cec2	247	return NULL;
630431cc	248	}
43f3cec2	249
630431cc	250	static void pmb_close(fio_pmemblk_file_t pmb, const bool keep)
43f3cec2	251	{
de88999f	252	pthread_mutex_lock(&CacheLock);
43f3cec2 BB	253
	254	pmb->pmb_refcnt--;
	255
630431cc	256	if (!keep && !pmb->pmb_refcnt) {
43f3cec2 BB	257	pmemblk_close(pmb->pmb_pool);
	258	pmb->pmb_pool = NULL;
	259	free(pmb->pmb_filename);
	260	pmb->pmb_filename = NULL;
	261	fio_pmemblk_cache_remove(pmb);
	262	free(pmb);
	263	}
	264
de88999f	265	pthread_mutex_unlock(&CacheLock);
630431cc	266	}
43f3cec2	267
630431cc	268	static int pmb_get_flags(struct thread_data td, uint64_t pflags)
43f3cec2	269	{
c086b314	270	static int thread_warned = 0;
43f3cec2 BB	271	static int odirect_warned = 0;
43f3cec2 BB	272
c086b314	273	uint64_t flags = 0;
43f3cec2 BB	274
	275	if (!td->o.use_thread) {
	276	if (!thread_warned) {
	277	thread_warned = 1;
a0998600	278	log_err("pmemblk: must set thread=1 for pmemblk engine\n");
43f3cec2 BB	279	}
	280	return 1;
	281	}
	282
	283	if (!td->o.odirect && !odirect_warned) {
	284	odirect_warned = 1;
a0998600	285	log_info("pmemblk: direct == 0, but pmemblk is always direct\n");
43f3cec2 BB	286	}
	287
	288	if (td->o.allow_create)
	289	flags \|= PMB_CREATE;
	290
	291	(*pflags) = flags;
	292	return 0;
630431cc	293	}
43f3cec2	294
c086b314	295	static int fio_pmemblk_open_file(struct thread_data td, struct fio_file f)
43f3cec2	296	{
c086b314 JA	297	uint64_t flags = 0;
c086b314 JA	298	fio_pmemblk_file_t pmb;
43f3cec2	299
630431cc	300	if (pmb_get_flags(td, &flags))
43f3cec2 BB	301	return 1;
	302
	303	pmb = pmb_open(f->file_name, flags);
630431cc	304	if (!pmb)
43f3cec2 BB	305	return 1;
	306
	307	FIOFILEPMBSET(f, pmb);
43f3cec2	308	return 0;
630431cc	309	}
43f3cec2	310
630431cc JA	311	static int fio_pmemblk_close_file(struct thread_data fio_unused *td,
630431cc JA	312	struct fio_file *f)
43f3cec2	313	{
c086b314	314	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);
43f3cec2 BB	315
43f3cec2 BB	316	if (pmb)
630431cc	317	pmb_close(pmb, false);
43f3cec2 BB	318
43f3cec2 BB	319	FIOFILEPMBSET(f, NULL);
43f3cec2	320	return 0;
630431cc	321	}
43f3cec2	322
c086b314	323	static int fio_pmemblk_get_file_size(struct thread_data td, struct fio_file f)
43f3cec2	324	{
c086b314 JA	325	uint64_t flags = 0;
c086b314 JA	326	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);
43f3cec2 BB	327
	328	if (fio_file_size_known(f))
	329	return 0;
	330
630431cc JA	331	if (!pmb) {
630431cc JA	332	if (pmb_get_flags(td, &flags))
43f3cec2 BB	333	return 1;
43f3cec2 BB	334	pmb = pmb_open(f->file_name, flags);
630431cc	335	if (!pmb)
43f3cec2 BB	336	return 1;
	337	}
	338
	339	f->real_file_size = pmb->pmb_bsize * pmb->pmb_nblocks;
	340
	341	fio_file_set_size_known(f);
	342
630431cc JA	343	if (!FIOFILEPMBGET(f))
630431cc JA	344	pmb_close(pmb, true);
43f3cec2 BB	345
43f3cec2 BB	346	return 0;
630431cc	347	}
43f3cec2	348
c086b314	349	static int fio_pmemblk_queue(struct thread_data td, struct io_u io_u)
43f3cec2	350	{
c086b314 JA	351	struct fio_file *f = io_u->file;
c086b314 JA	352	fio_pmemblk_file_t pmb = FIOFILEPMBGET(f);
43f3cec2	353
c086b314 JA	354	unsigned long long off;
	355	unsigned long len;
	356	void *buf;
43f3cec2 BB	357
	358	fio_ro_check(td, io_u);
	359
	360	switch (io_u->ddir) {
	361	case DDIR_READ:
43f3cec2 BB	362	case DDIR_WRITE:
	363	off = io_u->offset;
	364	len = io_u->xfer_buflen;
	365
	366	io_u->error = EINVAL;
630431cc	367	if (off % pmb->pmb_bsize)
43f3cec2	368	break;
630431cc	369	if (len % pmb->pmb_bsize)
43f3cec2 BB	370	break;
	371	if ((off + len) / pmb->pmb_bsize > pmb->pmb_nblocks)
	372	break;
	373
	374	io_u->error = 0;
	375	buf = io_u->xfer_buf;
	376	off /= pmb->pmb_bsize;
	377	len /= pmb->pmb_bsize;
	378	while (0 < len) {
cf8775b8 RE	379	if (io_u->ddir == DDIR_READ &&
	380	0 != pmemblk_read(pmb->pmb_pool, buf, off)) {
	381	io_u->error = errno;
	382	break;
	383	} else if (0 != pmemblk_write(pmb->pmb_pool, buf, off)) {
43f3cec2 BB	384	io_u->error = errno;
	385	break;
	386	}
	387	buf += pmb->pmb_bsize;
	388	off++;
	389	len--;
	390	}
	391	off *= pmb->pmb_bsize;
	392	len *= pmb->pmb_bsize;
	393	io_u->resid = io_u->xfer_buflen - (off - io_u->offset);
	394	break;
	395	case DDIR_SYNC:
	396	case DDIR_DATASYNC:
	397	case DDIR_SYNC_FILE_RANGE:
	398	/* we're always sync'd */
	399	io_u->error = 0;
	400	break;
	401	default:
	402	io_u->error = EINVAL;
	403	break;
	404	}
	405
	406	return FIO_Q_COMPLETED;
630431cc	407	}
43f3cec2	408
c086b314	409	static int fio_pmemblk_unlink_file(struct thread_data td, struct fio_file f)
43f3cec2	410	{
c086b314 JA	411	char *path = NULL;
	412	uint64_t bsize = 0;
	413	uint64_t fsize = 0;
43f3cec2 BB	414
	415	/*
	416	* we need our own unlink in case the user has specified
	417	* the block and file sizes in the path name. we parse
	418	* the file_name to determine the file name we actually used.
	419	*/
	420
	421	pmb_parse_path(f->file_name, &path, &bsize, &fsize);
630431cc	422	if (!path)
2442c935	423	return ENOENT;
43f3cec2 BB	424
	425	unlink(path);
	426	free(path);
43f3cec2	427	return 0;
630431cc	428	}
43f3cec2	429
13690c10	430	static struct ioengine_ops ioengine = {
c086b314 JA	431	.name = "pmemblk",
	432	.version = FIO_IOOPS_VERSION,
	433	.queue = fio_pmemblk_queue,
	434	.open_file = fio_pmemblk_open_file,
	435	.close_file = fio_pmemblk_close_file,
	436	.get_file_size = fio_pmemblk_get_file_size,
	437	.unlink_file = fio_pmemblk_unlink_file,
	438	.flags = FIO_SYNCIO \| FIO_DISKLESSIO \| FIO_NOEXTEND \| FIO_NODISKUTIL,
43f3cec2 BB	439	};
43f3cec2 BB	440
630431cc	441	static void fio_init fio_pmemblk_register(void)
43f3cec2 BB	442	{
	443	register_ioengine(&ioengine);
	444	}
	445
630431cc	446	static void fio_exit fio_pmemblk_unregister(void)
43f3cec2 BB	447	{
	448	unregister_ioengine(&ioengine);
	449	}