xref: /spdk/lib/blobfs/blobfs.c (revision 6951b97964ca0d206b77ecc74486f8691fc52067)
1 /*-
2  *   BSD LICENSE
3  *
4  *   Copyright (c) Intel Corporation.
5  *   All rights reserved.
6  *
7  *   Redistribution and use in source and binary forms, with or without
8  *   modification, are permitted provided that the following conditions
9  *   are met:
10  *
11  *     * Redistributions of source code must retain the above copyright
12  *       notice, this list of conditions and the following disclaimer.
13  *     * Redistributions in binary form must reproduce the above copyright
14  *       notice, this list of conditions and the following disclaimer in
15  *       the documentation and/or other materials provided with the
16  *       distribution.
17  *     * Neither the name of Intel Corporation nor the names of its
18  *       contributors may be used to endorse or promote products derived
19  *       from this software without specific prior written permission.
20  *
21  *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24  *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25  *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26  *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27  *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28  *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29  *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30  *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31  *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
32  */
33 
34 #include "spdk/stdinc.h"
35 
36 #include "spdk/blobfs.h"
37 #include "spdk/conf.h"
38 #include "blobfs_internal.h"
39 
40 #include "spdk/queue.h"
41 #include "spdk/io_channel.h"
42 #include "spdk/assert.h"
43 #include "spdk/env.h"
44 #include "spdk/util.h"
45 #include "spdk_internal/log.h"
46 
47 #define BLOBFS_TRACE(file, str, args...) \
48 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s " str, file->name, ##args)
49 
50 #define BLOBFS_TRACE_RW(file, str, args...) \
51 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS_RW, "file=%s " str, file->name, ##args)
52 
53 #define BLOBFS_CACHE_SIZE (4ULL * 1024 * 1024 * 1024)
54 #define SPDK_BLOBFS_OPTS_CLUSTER_SZ (1024 * 1024)
55 
56 static uint64_t g_fs_cache_size = BLOBFS_CACHE_SIZE;
57 static struct spdk_mempool *g_cache_pool;
58 static TAILQ_HEAD(, spdk_file) g_caches;
59 static int g_fs_count = 0;
60 static pthread_mutex_t g_cache_init_lock = PTHREAD_MUTEX_INITIALIZER;
61 static pthread_spinlock_t g_caches_lock;
62 
63 static void
64 __sem_post(void *arg, int bserrno)
65 {
66 	sem_t *sem = arg;
67 
68 	sem_post(sem);
69 }
70 
71 void
72 spdk_cache_buffer_free(struct cache_buffer *cache_buffer)
73 {
74 	spdk_mempool_put(g_cache_pool, cache_buffer->buf);
75 	free(cache_buffer);
76 }
77 
78 #define CACHE_READAHEAD_THRESHOLD	(128 * 1024)
79 
80 struct spdk_file {
81 	struct spdk_filesystem	*fs;
82 	struct spdk_blob	*blob;
83 	char			*name;
84 	uint64_t		length;
85 	bool                    is_deleted;
86 	bool			open_for_writing;
87 	uint64_t		length_flushed;
88 	uint64_t		append_pos;
89 	uint64_t		seq_byte_count;
90 	uint64_t		next_seq_offset;
91 	uint32_t		priority;
92 	TAILQ_ENTRY(spdk_file)	tailq;
93 	spdk_blob_id		blobid;
94 	uint32_t		ref_count;
95 	pthread_spinlock_t	lock;
96 	struct cache_buffer	*last;
97 	struct cache_tree	*tree;
98 	TAILQ_HEAD(open_requests_head, spdk_fs_request) open_requests;
99 	TAILQ_HEAD(sync_requests_head, spdk_fs_request) sync_requests;
100 	TAILQ_ENTRY(spdk_file)	cache_tailq;
101 };
102 
103 struct spdk_deleted_file {
104 	spdk_blob_id	id;
105 	TAILQ_ENTRY(spdk_deleted_file)	tailq;
106 };
107 
108 struct spdk_filesystem {
109 	struct spdk_blob_store	*bs;
110 	TAILQ_HEAD(, spdk_file)	files;
111 	struct spdk_bs_opts	bs_opts;
112 	struct spdk_bs_dev	*bdev;
113 	fs_send_request_fn	send_request;
114 
115 	struct {
116 		uint32_t		max_ops;
117 		struct spdk_io_channel	*sync_io_channel;
118 		struct spdk_fs_channel	*sync_fs_channel;
119 	} sync_target;
120 
121 	struct {
122 		uint32_t		max_ops;
123 		struct spdk_io_channel	*md_io_channel;
124 		struct spdk_fs_channel	*md_fs_channel;
125 	} md_target;
126 
127 	struct {
128 		uint32_t		max_ops;
129 	} io_target;
130 };
131 
132 struct spdk_fs_cb_args {
133 	union {
134 		spdk_fs_op_with_handle_complete		fs_op_with_handle;
135 		spdk_fs_op_complete			fs_op;
136 		spdk_file_op_with_handle_complete	file_op_with_handle;
137 		spdk_file_op_complete			file_op;
138 		spdk_file_stat_op_complete		stat_op;
139 	} fn;
140 	void *arg;
141 	sem_t *sem;
142 	struct spdk_filesystem *fs;
143 	struct spdk_file *file;
144 	int rc;
145 	bool from_request;
146 	union {
147 		struct {
148 			TAILQ_HEAD(, spdk_deleted_file)	deleted_files;
149 		} fs_load;
150 		struct {
151 			uint64_t	length;
152 		} truncate;
153 		struct {
154 			struct spdk_io_channel	*channel;
155 			void		*user_buf;
156 			void		*pin_buf;
157 			int		is_read;
158 			off_t		offset;
159 			size_t		length;
160 			uint64_t	start_page;
161 			uint64_t	num_pages;
162 			uint32_t	blocklen;
163 		} rw;
164 		struct {
165 			const char	*old_name;
166 			const char	*new_name;
167 		} rename;
168 		struct {
169 			struct cache_buffer	*cache_buffer;
170 			uint64_t		length;
171 		} flush;
172 		struct {
173 			struct cache_buffer	*cache_buffer;
174 			uint64_t		length;
175 			uint64_t		offset;
176 		} readahead;
177 		struct {
178 			uint64_t			offset;
179 			TAILQ_ENTRY(spdk_fs_request)	tailq;
180 			bool				xattr_in_progress;
181 		} sync;
182 		struct {
183 			uint32_t			num_clusters;
184 		} resize;
185 		struct {
186 			const char	*name;
187 			uint32_t	flags;
188 			TAILQ_ENTRY(spdk_fs_request)	tailq;
189 		} open;
190 		struct {
191 			const char	*name;
192 		} create;
193 		struct {
194 			const char	*name;
195 		} delete;
196 		struct {
197 			const char	*name;
198 		} stat;
199 	} op;
200 };
201 
202 static void cache_free_buffers(struct spdk_file *file);
203 
204 void
205 spdk_fs_opts_init(struct spdk_blobfs_opts *opts)
206 {
207 	opts->cluster_sz = SPDK_BLOBFS_OPTS_CLUSTER_SZ;
208 }
209 
210 static void
211 __initialize_cache(void)
212 {
213 	assert(g_cache_pool == NULL);
214 
215 	g_cache_pool = spdk_mempool_create("spdk_fs_cache",
216 					   g_fs_cache_size / CACHE_BUFFER_SIZE,
217 					   CACHE_BUFFER_SIZE,
218 					   SPDK_MEMPOOL_DEFAULT_CACHE_SIZE,
219 					   SPDK_ENV_SOCKET_ID_ANY);
220 	TAILQ_INIT(&g_caches);
221 	pthread_spin_init(&g_caches_lock, 0);
222 }
223 
224 static void
225 __free_cache(void)
226 {
227 	assert(g_cache_pool != NULL);
228 
229 	spdk_mempool_free(g_cache_pool);
230 	g_cache_pool = NULL;
231 }
232 
233 static uint64_t
234 __file_get_blob_size(struct spdk_file *file)
235 {
236 	uint64_t cluster_sz;
237 
238 	cluster_sz = file->fs->bs_opts.cluster_sz;
239 	return cluster_sz * spdk_blob_get_num_clusters(file->blob);
240 }
241 
242 struct spdk_fs_request {
243 	struct spdk_fs_cb_args		args;
244 	TAILQ_ENTRY(spdk_fs_request)	link;
245 	struct spdk_fs_channel		*channel;
246 };
247 
248 struct spdk_fs_channel {
249 	struct spdk_fs_request		*req_mem;
250 	TAILQ_HEAD(, spdk_fs_request)	reqs;
251 	sem_t				sem;
252 	struct spdk_filesystem		*fs;
253 	struct spdk_io_channel		*bs_channel;
254 	fs_send_request_fn		send_request;
255 	bool				sync;
256 	pthread_spinlock_t		lock;
257 };
258 
259 static struct spdk_fs_request *
260 alloc_fs_request(struct spdk_fs_channel *channel)
261 {
262 	struct spdk_fs_request *req;
263 
264 	if (channel->sync) {
265 		pthread_spin_lock(&channel->lock);
266 	}
267 
268 	req = TAILQ_FIRST(&channel->reqs);
269 	if (req) {
270 		TAILQ_REMOVE(&channel->reqs, req, link);
271 	}
272 
273 	if (channel->sync) {
274 		pthread_spin_unlock(&channel->lock);
275 	}
276 
277 	if (req == NULL) {
278 		return NULL;
279 	}
280 	memset(req, 0, sizeof(*req));
281 	req->channel = channel;
282 	req->args.from_request = true;
283 
284 	return req;
285 }
286 
287 static void
288 free_fs_request(struct spdk_fs_request *req)
289 {
290 	struct spdk_fs_channel *channel = req->channel;
291 
292 	if (channel->sync) {
293 		pthread_spin_lock(&channel->lock);
294 	}
295 
296 	TAILQ_INSERT_HEAD(&req->channel->reqs, req, link);
297 
298 	if (channel->sync) {
299 		pthread_spin_unlock(&channel->lock);
300 	}
301 }
302 
303 static int
304 _spdk_fs_channel_create(struct spdk_filesystem *fs, struct spdk_fs_channel *channel,
305 			uint32_t max_ops)
306 {
307 	uint32_t i;
308 
309 	channel->req_mem = calloc(max_ops, sizeof(struct spdk_fs_request));
310 	if (!channel->req_mem) {
311 		return -1;
312 	}
313 
314 	TAILQ_INIT(&channel->reqs);
315 	sem_init(&channel->sem, 0, 0);
316 
317 	for (i = 0; i < max_ops; i++) {
318 		TAILQ_INSERT_TAIL(&channel->reqs, &channel->req_mem[i], link);
319 	}
320 
321 	channel->fs = fs;
322 
323 	return 0;
324 }
325 
326 static int
327 _spdk_fs_md_channel_create(void *io_device, void *ctx_buf)
328 {
329 	struct spdk_filesystem		*fs;
330 	struct spdk_fs_channel		*channel = ctx_buf;
331 
332 	fs = SPDK_CONTAINEROF(io_device, struct spdk_filesystem, md_target);
333 
334 	return _spdk_fs_channel_create(fs, channel, fs->md_target.max_ops);
335 }
336 
337 static int
338 _spdk_fs_sync_channel_create(void *io_device, void *ctx_buf)
339 {
340 	struct spdk_filesystem		*fs;
341 	struct spdk_fs_channel		*channel = ctx_buf;
342 
343 	fs = SPDK_CONTAINEROF(io_device, struct spdk_filesystem, sync_target);
344 
345 	return _spdk_fs_channel_create(fs, channel, fs->sync_target.max_ops);
346 }
347 
348 static int
349 _spdk_fs_io_channel_create(void *io_device, void *ctx_buf)
350 {
351 	struct spdk_filesystem		*fs;
352 	struct spdk_fs_channel		*channel = ctx_buf;
353 
354 	fs = SPDK_CONTAINEROF(io_device, struct spdk_filesystem, io_target);
355 
356 	return _spdk_fs_channel_create(fs, channel, fs->io_target.max_ops);
357 }
358 
359 static void
360 _spdk_fs_channel_destroy(void *io_device, void *ctx_buf)
361 {
362 	struct spdk_fs_channel *channel = ctx_buf;
363 
364 	free(channel->req_mem);
365 	if (channel->bs_channel != NULL) {
366 		spdk_bs_free_io_channel(channel->bs_channel);
367 	}
368 }
369 
370 static void
371 __send_request_direct(fs_request_fn fn, void *arg)
372 {
373 	fn(arg);
374 }
375 
376 static void
377 common_fs_bs_init(struct spdk_filesystem *fs, struct spdk_blob_store *bs)
378 {
379 	fs->bs = bs;
380 	fs->bs_opts.cluster_sz = spdk_bs_get_cluster_size(bs);
381 	fs->md_target.md_fs_channel->bs_channel = spdk_bs_alloc_io_channel(fs->bs);
382 	fs->md_target.md_fs_channel->send_request = __send_request_direct;
383 	fs->sync_target.sync_fs_channel->bs_channel = spdk_bs_alloc_io_channel(fs->bs);
384 	fs->sync_target.sync_fs_channel->send_request = __send_request_direct;
385 
386 	pthread_mutex_lock(&g_cache_init_lock);
387 	if (g_fs_count == 0) {
388 		__initialize_cache();
389 	}
390 	g_fs_count++;
391 	pthread_mutex_unlock(&g_cache_init_lock);
392 }
393 
394 static void
395 init_cb(void *ctx, struct spdk_blob_store *bs, int bserrno)
396 {
397 	struct spdk_fs_request *req = ctx;
398 	struct spdk_fs_cb_args *args = &req->args;
399 	struct spdk_filesystem *fs = args->fs;
400 
401 	if (bserrno == 0) {
402 		common_fs_bs_init(fs, bs);
403 	} else {
404 		free(fs);
405 		fs = NULL;
406 	}
407 
408 	args->fn.fs_op_with_handle(args->arg, fs, bserrno);
409 	free_fs_request(req);
410 }
411 
412 static void
413 fs_conf_parse(void)
414 {
415 	struct spdk_conf_section *sp;
416 
417 	sp = spdk_conf_find_section(NULL, "Blobfs");
418 	if (sp == NULL) {
419 		g_fs_cache_buffer_shift = CACHE_BUFFER_SHIFT_DEFAULT;
420 		return;
421 	}
422 
423 	g_fs_cache_buffer_shift = spdk_conf_section_get_intval(sp, "CacheBufferShift");
424 	if (g_fs_cache_buffer_shift <= 0) {
425 		g_fs_cache_buffer_shift = CACHE_BUFFER_SHIFT_DEFAULT;
426 	}
427 }
428 
429 static struct spdk_filesystem *
430 fs_alloc(struct spdk_bs_dev *dev, fs_send_request_fn send_request_fn)
431 {
432 	struct spdk_filesystem *fs;
433 
434 	fs = calloc(1, sizeof(*fs));
435 	if (fs == NULL) {
436 		return NULL;
437 	}
438 
439 	fs->bdev = dev;
440 	fs->send_request = send_request_fn;
441 	TAILQ_INIT(&fs->files);
442 
443 	fs->md_target.max_ops = 512;
444 	spdk_io_device_register(&fs->md_target, _spdk_fs_md_channel_create, _spdk_fs_channel_destroy,
445 				sizeof(struct spdk_fs_channel));
446 	fs->md_target.md_io_channel = spdk_get_io_channel(&fs->md_target);
447 	fs->md_target.md_fs_channel = spdk_io_channel_get_ctx(fs->md_target.md_io_channel);
448 
449 	fs->sync_target.max_ops = 512;
450 	spdk_io_device_register(&fs->sync_target, _spdk_fs_sync_channel_create, _spdk_fs_channel_destroy,
451 				sizeof(struct spdk_fs_channel));
452 	fs->sync_target.sync_io_channel = spdk_get_io_channel(&fs->sync_target);
453 	fs->sync_target.sync_fs_channel = spdk_io_channel_get_ctx(fs->sync_target.sync_io_channel);
454 
455 	fs->io_target.max_ops = 512;
456 	spdk_io_device_register(&fs->io_target, _spdk_fs_io_channel_create, _spdk_fs_channel_destroy,
457 				sizeof(struct spdk_fs_channel));
458 
459 	return fs;
460 }
461 
462 void
463 spdk_fs_init(struct spdk_bs_dev *dev, struct spdk_blobfs_opts *opt,
464 	     fs_send_request_fn send_request_fn,
465 	     spdk_fs_op_with_handle_complete cb_fn, void *cb_arg)
466 {
467 	struct spdk_filesystem *fs;
468 	struct spdk_fs_request *req;
469 	struct spdk_fs_cb_args *args;
470 	struct spdk_bs_opts opts = {};
471 
472 	fs = fs_alloc(dev, send_request_fn);
473 	if (fs == NULL) {
474 		cb_fn(cb_arg, NULL, -ENOMEM);
475 		return;
476 	}
477 
478 	fs_conf_parse();
479 
480 	req = alloc_fs_request(fs->md_target.md_fs_channel);
481 	if (req == NULL) {
482 		spdk_put_io_channel(fs->md_target.md_io_channel);
483 		spdk_io_device_unregister(&fs->md_target, NULL);
484 		spdk_put_io_channel(fs->sync_target.sync_io_channel);
485 		spdk_io_device_unregister(&fs->sync_target, NULL);
486 		spdk_io_device_unregister(&fs->io_target, NULL);
487 		free(fs);
488 		cb_fn(cb_arg, NULL, -ENOMEM);
489 		return;
490 	}
491 
492 	args = &req->args;
493 	args->fn.fs_op_with_handle = cb_fn;
494 	args->arg = cb_arg;
495 	args->fs = fs;
496 
497 	spdk_bs_opts_init(&opts);
498 	strncpy(opts.bstype.bstype, "BLOBFS", SPDK_BLOBSTORE_TYPE_LENGTH);
499 	if (opt) {
500 		opts.cluster_sz = opt->cluster_sz;
501 	}
502 	spdk_bs_init(dev, &opts, init_cb, req);
503 }
504 
505 static struct spdk_file *
506 file_alloc(struct spdk_filesystem *fs)
507 {
508 	struct spdk_file *file;
509 
510 	file = calloc(1, sizeof(*file));
511 	if (file == NULL) {
512 		return NULL;
513 	}
514 
515 	file->tree = calloc(1, sizeof(*file->tree));
516 	if (file->tree == NULL) {
517 		free(file);
518 		return NULL;
519 	}
520 
521 	file->fs = fs;
522 	TAILQ_INIT(&file->open_requests);
523 	TAILQ_INIT(&file->sync_requests);
524 	pthread_spin_init(&file->lock, 0);
525 	TAILQ_INSERT_TAIL(&fs->files, file, tailq);
526 	file->priority = SPDK_FILE_PRIORITY_LOW;
527 	return file;
528 }
529 
530 static void iter_delete_cb(void *ctx, int bserrno);
531 
532 static int
533 _handle_deleted_files(struct spdk_fs_request *req)
534 {
535 	struct spdk_fs_cb_args *args = &req->args;
536 	struct spdk_filesystem *fs = args->fs;
537 
538 	if (!TAILQ_EMPTY(&args->op.fs_load.deleted_files)) {
539 		struct spdk_deleted_file *deleted_file;
540 
541 		deleted_file = TAILQ_FIRST(&args->op.fs_load.deleted_files);
542 		TAILQ_REMOVE(&args->op.fs_load.deleted_files, deleted_file, tailq);
543 		spdk_bs_delete_blob(fs->bs, deleted_file->id, iter_delete_cb, req);
544 		free(deleted_file);
545 		return 0;
546 	}
547 
548 	return 1;
549 }
550 
551 static void
552 iter_delete_cb(void *ctx, int bserrno)
553 {
554 	struct spdk_fs_request *req = ctx;
555 	struct spdk_fs_cb_args *args = &req->args;
556 	struct spdk_filesystem *fs = args->fs;
557 
558 	if (_handle_deleted_files(req) == 0) {
559 		return;
560 	}
561 
562 	args->fn.fs_op_with_handle(args->arg, fs, 0);
563 	free_fs_request(req);
564 
565 }
566 
567 static void
568 iter_cb(void *ctx, struct spdk_blob *blob, int rc)
569 {
570 	struct spdk_fs_request *req = ctx;
571 	struct spdk_fs_cb_args *args = &req->args;
572 	struct spdk_filesystem *fs = args->fs;
573 	uint64_t *length;
574 	const char *name;
575 	uint32_t *is_deleted;
576 	size_t value_len;
577 
578 	if (rc == -ENOENT) {
579 		/* Finished iterating */
580 		if (_handle_deleted_files(req) == 0) {
581 			return;
582 		}
583 		args->fn.fs_op_with_handle(args->arg, fs, 0);
584 		free_fs_request(req);
585 		return;
586 	} else if (rc < 0) {
587 		args->fn.fs_op_with_handle(args->arg, fs, rc);
588 		free_fs_request(req);
589 		return;
590 	}
591 
592 	rc = spdk_blob_get_xattr_value(blob, "name", (const void **)&name, &value_len);
593 	if (rc < 0) {
594 		args->fn.fs_op_with_handle(args->arg, fs, rc);
595 		free_fs_request(req);
596 		return;
597 	}
598 
599 	rc = spdk_blob_get_xattr_value(blob, "length", (const void **)&length, &value_len);
600 	if (rc < 0) {
601 		args->fn.fs_op_with_handle(args->arg, fs, rc);
602 		free_fs_request(req);
603 		return;
604 	}
605 
606 	assert(value_len == 8);
607 
608 	/* This file could be deleted last time without close it, then app crashed, so we delete it now */
609 	rc = spdk_blob_get_xattr_value(blob, "is_deleted", (const void **)&is_deleted, &value_len);
610 	if (rc < 0) {
611 		struct spdk_file *f;
612 
613 		f = file_alloc(fs);
614 		if (f == NULL) {
615 			args->fn.fs_op_with_handle(args->arg, fs, -ENOMEM);
616 			free_fs_request(req);
617 			return;
618 		}
619 
620 		f->name = strdup(name);
621 		f->blobid = spdk_blob_get_id(blob);
622 		f->length = *length;
623 		f->length_flushed = *length;
624 		f->append_pos = *length;
625 		SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "added file %s length=%ju\n", f->name, f->length);
626 	} else {
627 		struct spdk_deleted_file *deleted_file;
628 
629 		deleted_file = calloc(1, sizeof(*deleted_file));
630 		if (deleted_file == NULL) {
631 			args->fn.fs_op_with_handle(args->arg, fs, -ENOMEM);
632 			free_fs_request(req);
633 			return;
634 		}
635 		deleted_file->id = spdk_blob_get_id(blob);
636 		TAILQ_INSERT_TAIL(&args->op.fs_load.deleted_files, deleted_file, tailq);
637 	}
638 
639 	spdk_bs_iter_next(fs->bs, blob, iter_cb, req);
640 }
641 
642 static void
643 load_cb(void *ctx, struct spdk_blob_store *bs, int bserrno)
644 {
645 	struct spdk_fs_request *req = ctx;
646 	struct spdk_fs_cb_args *args = &req->args;
647 	struct spdk_filesystem *fs = args->fs;
648 	struct spdk_bs_type bstype;
649 	static const struct spdk_bs_type blobfs_type = {"BLOBFS"};
650 	static const struct spdk_bs_type zeros;
651 
652 	if (bserrno != 0) {
653 		args->fn.fs_op_with_handle(args->arg, NULL, bserrno);
654 		free_fs_request(req);
655 		free(fs);
656 		return;
657 	}
658 
659 	bstype = spdk_bs_get_bstype(bs);
660 
661 	if (!memcmp(&bstype, &zeros, sizeof(bstype))) {
662 		SPDK_DEBUGLOG(SPDK_LOG_BLOB, "assigning bstype\n");
663 		spdk_bs_set_bstype(bs, blobfs_type);
664 	} else if (memcmp(&bstype, &blobfs_type, sizeof(bstype))) {
665 		SPDK_DEBUGLOG(SPDK_LOG_BLOB, "not blobfs\n");
666 		SPDK_TRACEDUMP(SPDK_LOG_BLOB, "bstype", &bstype, sizeof(bstype));
667 		args->fn.fs_op_with_handle(args->arg, NULL, bserrno);
668 		free_fs_request(req);
669 		free(fs);
670 		return;
671 	}
672 
673 	common_fs_bs_init(fs, bs);
674 	spdk_bs_iter_first(fs->bs, iter_cb, req);
675 }
676 
677 void
678 spdk_fs_load(struct spdk_bs_dev *dev, fs_send_request_fn send_request_fn,
679 	     spdk_fs_op_with_handle_complete cb_fn, void *cb_arg)
680 {
681 	struct spdk_filesystem *fs;
682 	struct spdk_fs_cb_args *args;
683 	struct spdk_fs_request *req;
684 	struct spdk_bs_opts opts = {};
685 
686 	fs = fs_alloc(dev, send_request_fn);
687 	if (fs == NULL) {
688 		cb_fn(cb_arg, NULL, -ENOMEM);
689 		return;
690 	}
691 
692 	fs_conf_parse();
693 
694 	req = alloc_fs_request(fs->md_target.md_fs_channel);
695 	if (req == NULL) {
696 		spdk_put_io_channel(fs->md_target.md_io_channel);
697 		spdk_io_device_unregister(&fs->md_target, NULL);
698 		spdk_put_io_channel(fs->sync_target.sync_io_channel);
699 		spdk_io_device_unregister(&fs->sync_target, NULL);
700 		spdk_io_device_unregister(&fs->io_target, NULL);
701 		free(fs);
702 		cb_fn(cb_arg, NULL, -ENOMEM);
703 		return;
704 	}
705 
706 	args = &req->args;
707 	args->fn.fs_op_with_handle = cb_fn;
708 	args->arg = cb_arg;
709 	args->fs = fs;
710 	TAILQ_INIT(&args->op.fs_load.deleted_files);
711 
712 	spdk_bs_opts_init(&opts);
713 
714 	spdk_bs_load(dev, &opts, load_cb, req);
715 }
716 
717 static void
718 unload_cb(void *ctx, int bserrno)
719 {
720 	struct spdk_fs_request *req = ctx;
721 	struct spdk_fs_cb_args *args = &req->args;
722 	struct spdk_filesystem *fs = args->fs;
723 
724 	pthread_mutex_lock(&g_cache_init_lock);
725 	g_fs_count--;
726 	if (g_fs_count == 0) {
727 		__free_cache();
728 	}
729 	pthread_mutex_unlock(&g_cache_init_lock);
730 
731 	args->fn.fs_op(args->arg, bserrno);
732 	free(req);
733 
734 	spdk_io_device_unregister(&fs->io_target, NULL);
735 	spdk_io_device_unregister(&fs->sync_target, NULL);
736 	spdk_io_device_unregister(&fs->md_target, NULL);
737 
738 	free(fs);
739 }
740 
741 void
742 spdk_fs_unload(struct spdk_filesystem *fs, spdk_fs_op_complete cb_fn, void *cb_arg)
743 {
744 	struct spdk_fs_request *req;
745 	struct spdk_fs_cb_args *args;
746 
747 	/*
748 	 * We must free the md_channel before unloading the blobstore, so just
749 	 *  allocate this request from the general heap.
750 	 */
751 	req = calloc(1, sizeof(*req));
752 	if (req == NULL) {
753 		cb_fn(cb_arg, -ENOMEM);
754 		return;
755 	}
756 
757 	args = &req->args;
758 	args->fn.fs_op = cb_fn;
759 	args->arg = cb_arg;
760 	args->fs = fs;
761 
762 	spdk_fs_free_io_channel(fs->md_target.md_io_channel);
763 	spdk_fs_free_io_channel(fs->sync_target.sync_io_channel);
764 	spdk_bs_unload(fs->bs, unload_cb, req);
765 }
766 
767 static struct spdk_file *
768 fs_find_file(struct spdk_filesystem *fs, const char *name)
769 {
770 	struct spdk_file *file;
771 
772 	TAILQ_FOREACH(file, &fs->files, tailq) {
773 		if (!strncmp(name, file->name, SPDK_FILE_NAME_MAX)) {
774 			return file;
775 		}
776 	}
777 
778 	return NULL;
779 }
780 
781 void
782 spdk_fs_file_stat_async(struct spdk_filesystem *fs, const char *name,
783 			spdk_file_stat_op_complete cb_fn, void *cb_arg)
784 {
785 	struct spdk_file_stat stat;
786 	struct spdk_file *f = NULL;
787 
788 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
789 		cb_fn(cb_arg, NULL, -ENAMETOOLONG);
790 		return;
791 	}
792 
793 	f = fs_find_file(fs, name);
794 	if (f != NULL) {
795 		stat.blobid = f->blobid;
796 		stat.size = f->length;
797 		cb_fn(cb_arg, &stat, 0);
798 		return;
799 	}
800 
801 	cb_fn(cb_arg, NULL, -ENOENT);
802 }
803 
804 static void
805 __copy_stat(void *arg, struct spdk_file_stat *stat, int fserrno)
806 {
807 	struct spdk_fs_request *req = arg;
808 	struct spdk_fs_cb_args *args = &req->args;
809 
810 	args->rc = fserrno;
811 	if (fserrno == 0) {
812 		memcpy(args->arg, stat, sizeof(*stat));
813 	}
814 	sem_post(args->sem);
815 }
816 
817 static void
818 __file_stat(void *arg)
819 {
820 	struct spdk_fs_request *req = arg;
821 	struct spdk_fs_cb_args *args = &req->args;
822 
823 	spdk_fs_file_stat_async(args->fs, args->op.stat.name,
824 				args->fn.stat_op, req);
825 }
826 
827 int
828 spdk_fs_file_stat(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
829 		  const char *name, struct spdk_file_stat *stat)
830 {
831 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
832 	struct spdk_fs_request *req;
833 	int rc;
834 
835 	req = alloc_fs_request(channel);
836 	assert(req != NULL);
837 
838 	req->args.fs = fs;
839 	req->args.op.stat.name = name;
840 	req->args.fn.stat_op = __copy_stat;
841 	req->args.arg = stat;
842 	req->args.sem = &channel->sem;
843 	channel->send_request(__file_stat, req);
844 	sem_wait(&channel->sem);
845 
846 	rc = req->args.rc;
847 	free_fs_request(req);
848 
849 	return rc;
850 }
851 
852 static void
853 fs_create_blob_close_cb(void *ctx, int bserrno)
854 {
855 	struct spdk_fs_request *req = ctx;
856 	struct spdk_fs_cb_args *args = &req->args;
857 
858 	args->fn.file_op(args->arg, bserrno);
859 	free_fs_request(req);
860 }
861 
862 static void
863 fs_create_blob_open_cb(void *ctx, struct spdk_blob *blob, int bserrno)
864 {
865 	struct spdk_fs_request *req = ctx;
866 	struct spdk_fs_cb_args *args = &req->args;
867 	struct spdk_file *f = args->file;
868 	uint64_t length = 0;
869 
870 	spdk_blob_resize(blob, 1);
871 	spdk_blob_set_xattr(blob, "name", f->name, strlen(f->name) + 1);
872 	spdk_blob_set_xattr(blob, "length", &length, sizeof(length));
873 
874 	spdk_blob_close(blob, fs_create_blob_close_cb, args);
875 }
876 
877 static void
878 fs_create_blob_create_cb(void *ctx, spdk_blob_id blobid, int bserrno)
879 {
880 	struct spdk_fs_request *req = ctx;
881 	struct spdk_fs_cb_args *args = &req->args;
882 	struct spdk_file *f = args->file;
883 
884 	f->blobid = blobid;
885 	spdk_bs_open_blob(f->fs->bs, blobid, fs_create_blob_open_cb, req);
886 }
887 
888 void
889 spdk_fs_create_file_async(struct spdk_filesystem *fs, const char *name,
890 			  spdk_file_op_complete cb_fn, void *cb_arg)
891 {
892 	struct spdk_file *file;
893 	struct spdk_fs_request *req;
894 	struct spdk_fs_cb_args *args;
895 
896 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
897 		cb_fn(cb_arg, -ENAMETOOLONG);
898 		return;
899 	}
900 
901 	file = fs_find_file(fs, name);
902 	if (file != NULL) {
903 		cb_fn(cb_arg, -EEXIST);
904 		return;
905 	}
906 
907 	file = file_alloc(fs);
908 	if (file == NULL) {
909 		cb_fn(cb_arg, -ENOMEM);
910 		return;
911 	}
912 
913 	req = alloc_fs_request(fs->md_target.md_fs_channel);
914 	if (req == NULL) {
915 		cb_fn(cb_arg, -ENOMEM);
916 		return;
917 	}
918 
919 	args = &req->args;
920 	args->file = file;
921 	args->fn.file_op = cb_fn;
922 	args->arg = cb_arg;
923 
924 	file->name = strdup(name);
925 	spdk_bs_create_blob(fs->bs, fs_create_blob_create_cb, args);
926 }
927 
928 static void
929 __fs_create_file_done(void *arg, int fserrno)
930 {
931 	struct spdk_fs_request *req = arg;
932 	struct spdk_fs_cb_args *args = &req->args;
933 
934 	args->rc = fserrno;
935 	sem_post(args->sem);
936 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.create.name);
937 }
938 
939 static void
940 __fs_create_file(void *arg)
941 {
942 	struct spdk_fs_request *req = arg;
943 	struct spdk_fs_cb_args *args = &req->args;
944 
945 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.create.name);
946 	spdk_fs_create_file_async(args->fs, args->op.create.name, __fs_create_file_done, req);
947 }
948 
949 int
950 spdk_fs_create_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel, const char *name)
951 {
952 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
953 	struct spdk_fs_request *req;
954 	struct spdk_fs_cb_args *args;
955 	int rc;
956 
957 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
958 
959 	req = alloc_fs_request(channel);
960 	assert(req != NULL);
961 
962 	args = &req->args;
963 	args->fs = fs;
964 	args->op.create.name = name;
965 	args->sem = &channel->sem;
966 	fs->send_request(__fs_create_file, req);
967 	sem_wait(&channel->sem);
968 	rc = args->rc;
969 	free_fs_request(req);
970 
971 	return rc;
972 }
973 
974 static void
975 fs_open_blob_done(void *ctx, struct spdk_blob *blob, int bserrno)
976 {
977 	struct spdk_fs_request *req = ctx;
978 	struct spdk_fs_cb_args *args = &req->args;
979 	struct spdk_file *f = args->file;
980 
981 	f->blob = blob;
982 	while (!TAILQ_EMPTY(&f->open_requests)) {
983 		req = TAILQ_FIRST(&f->open_requests);
984 		args = &req->args;
985 		TAILQ_REMOVE(&f->open_requests, req, args.op.open.tailq);
986 		args->fn.file_op_with_handle(args->arg, f, bserrno);
987 		free_fs_request(req);
988 	}
989 }
990 
991 static void
992 fs_open_blob_create_cb(void *ctx, int bserrno)
993 {
994 	struct spdk_fs_request *req = ctx;
995 	struct spdk_fs_cb_args *args = &req->args;
996 	struct spdk_file *file = args->file;
997 	struct spdk_filesystem *fs = args->fs;
998 
999 	if (file == NULL) {
1000 		/*
1001 		 * This is from an open with CREATE flag - the file
1002 		 *  is now created so look it up in the file list for this
1003 		 *  filesystem.
1004 		 */
1005 		file = fs_find_file(fs, args->op.open.name);
1006 		assert(file != NULL);
1007 		args->file = file;
1008 	}
1009 
1010 	file->ref_count++;
1011 	TAILQ_INSERT_TAIL(&file->open_requests, req, args.op.open.tailq);
1012 	if (file->ref_count == 1) {
1013 		assert(file->blob == NULL);
1014 		spdk_bs_open_blob(fs->bs, file->blobid, fs_open_blob_done, req);
1015 	} else if (file->blob != NULL) {
1016 		fs_open_blob_done(req, file->blob, 0);
1017 	} else {
1018 		/*
1019 		 * The blob open for this file is in progress due to a previous
1020 		 *  open request.  When that open completes, it will invoke the
1021 		 *  open callback for this request.
1022 		 */
1023 	}
1024 }
1025 
1026 void
1027 spdk_fs_open_file_async(struct spdk_filesystem *fs, const char *name, uint32_t flags,
1028 			spdk_file_op_with_handle_complete cb_fn, void *cb_arg)
1029 {
1030 	struct spdk_file *f = NULL;
1031 	struct spdk_fs_request *req;
1032 	struct spdk_fs_cb_args *args;
1033 
1034 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1035 		cb_fn(cb_arg, NULL, -ENAMETOOLONG);
1036 		return;
1037 	}
1038 
1039 	f = fs_find_file(fs, name);
1040 	if (f == NULL && !(flags & SPDK_BLOBFS_OPEN_CREATE)) {
1041 		cb_fn(cb_arg, NULL, -ENOENT);
1042 		return;
1043 	}
1044 
1045 	if (f != NULL && f->is_deleted == true) {
1046 		cb_fn(cb_arg, NULL, -ENOENT);
1047 		return;
1048 	}
1049 
1050 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1051 	if (req == NULL) {
1052 		cb_fn(cb_arg, NULL, -ENOMEM);
1053 		return;
1054 	}
1055 
1056 	args = &req->args;
1057 	args->fn.file_op_with_handle = cb_fn;
1058 	args->arg = cb_arg;
1059 	args->file = f;
1060 	args->fs = fs;
1061 	args->op.open.name = name;
1062 
1063 	if (f == NULL) {
1064 		spdk_fs_create_file_async(fs, name, fs_open_blob_create_cb, req);
1065 	} else {
1066 		fs_open_blob_create_cb(req, 0);
1067 	}
1068 }
1069 
1070 static void
1071 __fs_open_file_done(void *arg, struct spdk_file *file, int bserrno)
1072 {
1073 	struct spdk_fs_request *req = arg;
1074 	struct spdk_fs_cb_args *args = &req->args;
1075 
1076 	args->file = file;
1077 	args->rc = bserrno;
1078 	sem_post(args->sem);
1079 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.open.name);
1080 }
1081 
1082 static void
1083 __fs_open_file(void *arg)
1084 {
1085 	struct spdk_fs_request *req = arg;
1086 	struct spdk_fs_cb_args *args = &req->args;
1087 
1088 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.open.name);
1089 	spdk_fs_open_file_async(args->fs, args->op.open.name, args->op.open.flags,
1090 				__fs_open_file_done, req);
1091 }
1092 
1093 int
1094 spdk_fs_open_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1095 		  const char *name, uint32_t flags, struct spdk_file **file)
1096 {
1097 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1098 	struct spdk_fs_request *req;
1099 	struct spdk_fs_cb_args *args;
1100 	int rc;
1101 
1102 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
1103 
1104 	req = alloc_fs_request(channel);
1105 	assert(req != NULL);
1106 
1107 	args = &req->args;
1108 	args->fs = fs;
1109 	args->op.open.name = name;
1110 	args->op.open.flags = flags;
1111 	args->sem = &channel->sem;
1112 	fs->send_request(__fs_open_file, req);
1113 	sem_wait(&channel->sem);
1114 	rc = args->rc;
1115 	if (rc == 0) {
1116 		*file = args->file;
1117 	} else {
1118 		*file = NULL;
1119 	}
1120 	free_fs_request(req);
1121 
1122 	return rc;
1123 }
1124 
1125 static void
1126 fs_rename_blob_close_cb(void *ctx, int bserrno)
1127 {
1128 	struct spdk_fs_request *req = ctx;
1129 	struct spdk_fs_cb_args *args = &req->args;
1130 
1131 	args->fn.fs_op(args->arg, bserrno);
1132 	free_fs_request(req);
1133 }
1134 
1135 static void
1136 fs_rename_blob_open_cb(void *ctx, struct spdk_blob *blob, int bserrno)
1137 {
1138 	struct spdk_fs_request *req = ctx;
1139 	struct spdk_fs_cb_args *args = &req->args;
1140 	const char *new_name = args->op.rename.new_name;
1141 
1142 	spdk_blob_set_xattr(blob, "name", new_name, strlen(new_name) + 1);
1143 	spdk_blob_close(blob, fs_rename_blob_close_cb, req);
1144 }
1145 
1146 static void
1147 __spdk_fs_md_rename_file(struct spdk_fs_request *req)
1148 {
1149 	struct spdk_fs_cb_args *args = &req->args;
1150 	struct spdk_file *f;
1151 
1152 	f = fs_find_file(args->fs, args->op.rename.old_name);
1153 	if (f == NULL) {
1154 		args->fn.fs_op(args->arg, -ENOENT);
1155 		free_fs_request(req);
1156 		return;
1157 	}
1158 
1159 	free(f->name);
1160 	f->name = strdup(args->op.rename.new_name);
1161 	args->file = f;
1162 	spdk_bs_open_blob(args->fs->bs, f->blobid, fs_rename_blob_open_cb, req);
1163 }
1164 
1165 static void
1166 fs_rename_delete_done(void *arg, int fserrno)
1167 {
1168 	__spdk_fs_md_rename_file(arg);
1169 }
1170 
1171 void
1172 spdk_fs_rename_file_async(struct spdk_filesystem *fs,
1173 			  const char *old_name, const char *new_name,
1174 			  spdk_file_op_complete cb_fn, void *cb_arg)
1175 {
1176 	struct spdk_file *f;
1177 	struct spdk_fs_request *req;
1178 	struct spdk_fs_cb_args *args;
1179 
1180 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "old=%s new=%s\n", old_name, new_name);
1181 	if (strnlen(new_name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1182 		cb_fn(cb_arg, -ENAMETOOLONG);
1183 		return;
1184 	}
1185 
1186 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1187 	if (req == NULL) {
1188 		cb_fn(cb_arg, -ENOMEM);
1189 		return;
1190 	}
1191 
1192 	args = &req->args;
1193 	args->fn.fs_op = cb_fn;
1194 	args->fs = fs;
1195 	args->arg = cb_arg;
1196 	args->op.rename.old_name = old_name;
1197 	args->op.rename.new_name = new_name;
1198 
1199 	f = fs_find_file(fs, new_name);
1200 	if (f == NULL) {
1201 		__spdk_fs_md_rename_file(req);
1202 		return;
1203 	}
1204 
1205 	/*
1206 	 * The rename overwrites an existing file.  So delete the existing file, then
1207 	 *  do the actual rename.
1208 	 */
1209 	spdk_fs_delete_file_async(fs, new_name, fs_rename_delete_done, req);
1210 }
1211 
1212 static void
1213 __fs_rename_file_done(void *arg, int fserrno)
1214 {
1215 	struct spdk_fs_request *req = arg;
1216 	struct spdk_fs_cb_args *args = &req->args;
1217 
1218 	args->rc = fserrno;
1219 	sem_post(args->sem);
1220 }
1221 
1222 static void
1223 __fs_rename_file(void *arg)
1224 {
1225 	struct spdk_fs_request *req = arg;
1226 	struct spdk_fs_cb_args *args = &req->args;
1227 
1228 	spdk_fs_rename_file_async(args->fs, args->op.rename.old_name, args->op.rename.new_name,
1229 				  __fs_rename_file_done, req);
1230 }
1231 
1232 int
1233 spdk_fs_rename_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1234 		    const char *old_name, const char *new_name)
1235 {
1236 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1237 	struct spdk_fs_request *req;
1238 	struct spdk_fs_cb_args *args;
1239 	int rc;
1240 
1241 	req = alloc_fs_request(channel);
1242 	assert(req != NULL);
1243 
1244 	args = &req->args;
1245 
1246 	args->fs = fs;
1247 	args->op.rename.old_name = old_name;
1248 	args->op.rename.new_name = new_name;
1249 	args->sem = &channel->sem;
1250 	fs->send_request(__fs_rename_file, req);
1251 	sem_wait(&channel->sem);
1252 	rc = args->rc;
1253 	free_fs_request(req);
1254 	return rc;
1255 }
1256 
1257 static void
1258 blob_delete_cb(void *ctx, int bserrno)
1259 {
1260 	struct spdk_fs_request *req = ctx;
1261 	struct spdk_fs_cb_args *args = &req->args;
1262 
1263 	args->fn.file_op(args->arg, bserrno);
1264 	free_fs_request(req);
1265 }
1266 
1267 void
1268 spdk_fs_delete_file_async(struct spdk_filesystem *fs, const char *name,
1269 			  spdk_file_op_complete cb_fn, void *cb_arg)
1270 {
1271 	struct spdk_file *f;
1272 	spdk_blob_id blobid;
1273 	struct spdk_fs_request *req;
1274 	struct spdk_fs_cb_args *args;
1275 
1276 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
1277 
1278 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1279 		cb_fn(cb_arg, -ENAMETOOLONG);
1280 		return;
1281 	}
1282 
1283 	f = fs_find_file(fs, name);
1284 	if (f == NULL) {
1285 		cb_fn(cb_arg, -ENOENT);
1286 		return;
1287 	}
1288 
1289 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1290 	if (req == NULL) {
1291 		cb_fn(cb_arg, -ENOMEM);
1292 		return;
1293 	}
1294 
1295 	args = &req->args;
1296 	args->fn.file_op = cb_fn;
1297 	args->arg = cb_arg;
1298 
1299 	if (f->ref_count > 0) {
1300 		/* If the ref > 0, we mark the file as deleted and delete it when we close it. */
1301 		f->is_deleted = true;
1302 		spdk_blob_set_xattr(f->blob, "is_deleted", &f->is_deleted, sizeof(bool));
1303 		spdk_blob_sync_md(f->blob, blob_delete_cb, args);
1304 		return;
1305 	}
1306 
1307 	TAILQ_REMOVE(&fs->files, f, tailq);
1308 
1309 	cache_free_buffers(f);
1310 
1311 	blobid = f->blobid;
1312 
1313 	free(f->name);
1314 	free(f->tree);
1315 	free(f);
1316 
1317 	spdk_bs_delete_blob(fs->bs, blobid, blob_delete_cb, req);
1318 }
1319 
1320 static void
1321 __fs_delete_file_done(void *arg, int fserrno)
1322 {
1323 	struct spdk_fs_request *req = arg;
1324 	struct spdk_fs_cb_args *args = &req->args;
1325 
1326 	args->rc = fserrno;
1327 	sem_post(args->sem);
1328 }
1329 
1330 static void
1331 __fs_delete_file(void *arg)
1332 {
1333 	struct spdk_fs_request *req = arg;
1334 	struct spdk_fs_cb_args *args = &req->args;
1335 
1336 	spdk_fs_delete_file_async(args->fs, args->op.delete.name, __fs_delete_file_done, req);
1337 }
1338 
1339 int
1340 spdk_fs_delete_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1341 		    const char *name)
1342 {
1343 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1344 	struct spdk_fs_request *req;
1345 	struct spdk_fs_cb_args *args;
1346 	int rc;
1347 
1348 	req = alloc_fs_request(channel);
1349 	assert(req != NULL);
1350 
1351 	args = &req->args;
1352 	args->fs = fs;
1353 	args->op.delete.name = name;
1354 	args->sem = &channel->sem;
1355 	fs->send_request(__fs_delete_file, req);
1356 	sem_wait(&channel->sem);
1357 	rc = args->rc;
1358 	free_fs_request(req);
1359 
1360 	return rc;
1361 }
1362 
1363 spdk_fs_iter
1364 spdk_fs_iter_first(struct spdk_filesystem *fs)
1365 {
1366 	struct spdk_file *f;
1367 
1368 	f = TAILQ_FIRST(&fs->files);
1369 	return f;
1370 }
1371 
1372 spdk_fs_iter
1373 spdk_fs_iter_next(spdk_fs_iter iter)
1374 {
1375 	struct spdk_file *f = iter;
1376 
1377 	if (f == NULL) {
1378 		return NULL;
1379 	}
1380 
1381 	f = TAILQ_NEXT(f, tailq);
1382 	return f;
1383 }
1384 
1385 const char *
1386 spdk_file_get_name(struct spdk_file *file)
1387 {
1388 	return file->name;
1389 }
1390 
1391 uint64_t
1392 spdk_file_get_length(struct spdk_file *file)
1393 {
1394 	assert(file != NULL);
1395 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s length=0x%jx\n", file->name, file->length);
1396 	return file->length;
1397 }
1398 
1399 static void
1400 fs_truncate_complete_cb(void *ctx, int bserrno)
1401 {
1402 	struct spdk_fs_request *req = ctx;
1403 	struct spdk_fs_cb_args *args = &req->args;
1404 
1405 	args->fn.file_op(args->arg, bserrno);
1406 	free_fs_request(req);
1407 }
1408 
1409 static uint64_t
1410 __bytes_to_clusters(uint64_t length, uint64_t cluster_sz)
1411 {
1412 	return (length + cluster_sz - 1) / cluster_sz;
1413 }
1414 
1415 void
1416 spdk_file_truncate_async(struct spdk_file *file, uint64_t length,
1417 			 spdk_file_op_complete cb_fn, void *cb_arg)
1418 {
1419 	struct spdk_filesystem *fs;
1420 	size_t num_clusters;
1421 	struct spdk_fs_request *req;
1422 	struct spdk_fs_cb_args *args;
1423 
1424 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s old=0x%jx new=0x%jx\n", file->name, file->length, length);
1425 	if (length == file->length) {
1426 		cb_fn(cb_arg, 0);
1427 		return;
1428 	}
1429 
1430 	req = alloc_fs_request(file->fs->md_target.md_fs_channel);
1431 	if (req == NULL) {
1432 		cb_fn(cb_arg, -ENOMEM);
1433 		return;
1434 	}
1435 
1436 	args = &req->args;
1437 	args->fn.file_op = cb_fn;
1438 	args->arg = cb_arg;
1439 	args->file = file;
1440 	fs = file->fs;
1441 
1442 	num_clusters = __bytes_to_clusters(length, fs->bs_opts.cluster_sz);
1443 
1444 	spdk_blob_resize(file->blob, num_clusters);
1445 	spdk_blob_set_xattr(file->blob, "length", &length, sizeof(length));
1446 
1447 	file->length = length;
1448 	if (file->append_pos > file->length) {
1449 		file->append_pos = file->length;
1450 	}
1451 
1452 	spdk_blob_sync_md(file->blob, fs_truncate_complete_cb, args);
1453 }
1454 
1455 static void
1456 __truncate(void *arg)
1457 {
1458 	struct spdk_fs_request *req = arg;
1459 	struct spdk_fs_cb_args *args = &req->args;
1460 
1461 	spdk_file_truncate_async(args->file, args->op.truncate.length,
1462 				 args->fn.file_op, args->arg);
1463 }
1464 
1465 void
1466 spdk_file_truncate(struct spdk_file *file, struct spdk_io_channel *_channel,
1467 		   uint64_t length)
1468 {
1469 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1470 	struct spdk_fs_request *req;
1471 	struct spdk_fs_cb_args *args;
1472 
1473 	req = alloc_fs_request(channel);
1474 	assert(req != NULL);
1475 
1476 	args = &req->args;
1477 
1478 	args->file = file;
1479 	args->op.truncate.length = length;
1480 	args->fn.file_op = __sem_post;
1481 	args->arg = &channel->sem;
1482 
1483 	channel->send_request(__truncate, req);
1484 	sem_wait(&channel->sem);
1485 	free_fs_request(req);
1486 }
1487 
1488 static void
1489 __rw_done(void *ctx, int bserrno)
1490 {
1491 	struct spdk_fs_request *req = ctx;
1492 	struct spdk_fs_cb_args *args = &req->args;
1493 
1494 	spdk_dma_free(args->op.rw.pin_buf);
1495 	args->fn.file_op(args->arg, bserrno);
1496 	free_fs_request(req);
1497 }
1498 
1499 static void
1500 __read_done(void *ctx, int bserrno)
1501 {
1502 	struct spdk_fs_request *req = ctx;
1503 	struct spdk_fs_cb_args *args = &req->args;
1504 
1505 	if (args->op.rw.is_read) {
1506 		memcpy(args->op.rw.user_buf,
1507 		       args->op.rw.pin_buf + (args->op.rw.offset & 0xFFF),
1508 		       args->op.rw.length);
1509 		__rw_done(req, 0);
1510 	} else {
1511 		memcpy(args->op.rw.pin_buf + (args->op.rw.offset & 0xFFF),
1512 		       args->op.rw.user_buf,
1513 		       args->op.rw.length);
1514 		spdk_bs_io_write_blob(args->file->blob, args->op.rw.channel,
1515 				      args->op.rw.pin_buf,
1516 				      args->op.rw.start_page, args->op.rw.num_pages,
1517 				      __rw_done, req);
1518 	}
1519 }
1520 
1521 static void
1522 __do_blob_read(void *ctx, int fserrno)
1523 {
1524 	struct spdk_fs_request *req = ctx;
1525 	struct spdk_fs_cb_args *args = &req->args;
1526 
1527 	spdk_bs_io_read_blob(args->file->blob, args->op.rw.channel,
1528 			     args->op.rw.pin_buf,
1529 			     args->op.rw.start_page, args->op.rw.num_pages,
1530 			     __read_done, req);
1531 }
1532 
1533 static void
1534 __get_page_parameters(struct spdk_file *file, uint64_t offset, uint64_t length,
1535 		      uint64_t *start_page, uint32_t *page_size, uint64_t *num_pages)
1536 {
1537 	uint64_t end_page;
1538 
1539 	*page_size = spdk_bs_get_page_size(file->fs->bs);
1540 	*start_page = offset / *page_size;
1541 	end_page = (offset + length - 1) / *page_size;
1542 	*num_pages = (end_page - *start_page + 1);
1543 }
1544 
1545 static void
1546 __readwrite(struct spdk_file *file, struct spdk_io_channel *_channel,
1547 	    void *payload, uint64_t offset, uint64_t length,
1548 	    spdk_file_op_complete cb_fn, void *cb_arg, int is_read)
1549 {
1550 	struct spdk_fs_request *req;
1551 	struct spdk_fs_cb_args *args;
1552 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1553 	uint64_t start_page, num_pages, pin_buf_length;
1554 	uint32_t page_size;
1555 
1556 	if (is_read && offset + length > file->length) {
1557 		cb_fn(cb_arg, -EINVAL);
1558 		return;
1559 	}
1560 
1561 	req = alloc_fs_request(channel);
1562 	if (req == NULL) {
1563 		cb_fn(cb_arg, -ENOMEM);
1564 		return;
1565 	}
1566 
1567 	args = &req->args;
1568 	args->fn.file_op = cb_fn;
1569 	args->arg = cb_arg;
1570 	args->file = file;
1571 	args->op.rw.channel = channel->bs_channel;
1572 	args->op.rw.user_buf = payload;
1573 	args->op.rw.is_read = is_read;
1574 	args->op.rw.offset = offset;
1575 	args->op.rw.length = length;
1576 
1577 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
1578 	pin_buf_length = num_pages * page_size;
1579 	args->op.rw.pin_buf = spdk_dma_malloc(pin_buf_length, 4096, NULL);
1580 
1581 	args->op.rw.start_page = start_page;
1582 	args->op.rw.num_pages = num_pages;
1583 
1584 	if (!is_read && file->length < offset + length) {
1585 		spdk_file_truncate_async(file, offset + length, __do_blob_read, req);
1586 	} else {
1587 		__do_blob_read(req, 0);
1588 	}
1589 }
1590 
1591 void
1592 spdk_file_write_async(struct spdk_file *file, struct spdk_io_channel *channel,
1593 		      void *payload, uint64_t offset, uint64_t length,
1594 		      spdk_file_op_complete cb_fn, void *cb_arg)
1595 {
1596 	__readwrite(file, channel, payload, offset, length, cb_fn, cb_arg, 0);
1597 }
1598 
1599 void
1600 spdk_file_read_async(struct spdk_file *file, struct spdk_io_channel *channel,
1601 		     void *payload, uint64_t offset, uint64_t length,
1602 		     spdk_file_op_complete cb_fn, void *cb_arg)
1603 {
1604 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s offset=%jx length=%jx\n",
1605 		      file->name, offset, length);
1606 	__readwrite(file, channel, payload, offset, length, cb_fn, cb_arg, 1);
1607 }
1608 
1609 struct spdk_io_channel *
1610 spdk_fs_alloc_io_channel(struct spdk_filesystem *fs)
1611 {
1612 	struct spdk_io_channel *io_channel;
1613 	struct spdk_fs_channel *fs_channel;
1614 
1615 	io_channel = spdk_get_io_channel(&fs->io_target);
1616 	fs_channel = spdk_io_channel_get_ctx(io_channel);
1617 	fs_channel->bs_channel = spdk_bs_alloc_io_channel(fs->bs);
1618 	fs_channel->send_request = __send_request_direct;
1619 
1620 	return io_channel;
1621 }
1622 
1623 struct spdk_io_channel *
1624 spdk_fs_alloc_io_channel_sync(struct spdk_filesystem *fs)
1625 {
1626 	struct spdk_io_channel *io_channel;
1627 	struct spdk_fs_channel *fs_channel;
1628 
1629 	io_channel = spdk_get_io_channel(&fs->io_target);
1630 	fs_channel = spdk_io_channel_get_ctx(io_channel);
1631 	fs_channel->send_request = fs->send_request;
1632 	fs_channel->sync = 1;
1633 	pthread_spin_init(&fs_channel->lock, 0);
1634 
1635 	return io_channel;
1636 }
1637 
1638 void
1639 spdk_fs_free_io_channel(struct spdk_io_channel *channel)
1640 {
1641 	spdk_put_io_channel(channel);
1642 }
1643 
1644 void
1645 spdk_fs_set_cache_size(uint64_t size_in_mb)
1646 {
1647 	g_fs_cache_size = size_in_mb * 1024 * 1024;
1648 }
1649 
1650 uint64_t
1651 spdk_fs_get_cache_size(void)
1652 {
1653 	return g_fs_cache_size / (1024 * 1024);
1654 }
1655 
1656 static void __file_flush(void *_args);
1657 
1658 static void *
1659 alloc_cache_memory_buffer(struct spdk_file *context)
1660 {
1661 	struct spdk_file *file;
1662 	void *buf;
1663 
1664 	buf = spdk_mempool_get(g_cache_pool);
1665 	if (buf != NULL) {
1666 		return buf;
1667 	}
1668 
1669 	pthread_spin_lock(&g_caches_lock);
1670 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1671 		if (!file->open_for_writing &&
1672 		    file->priority == SPDK_FILE_PRIORITY_LOW &&
1673 		    file != context) {
1674 			break;
1675 		}
1676 	}
1677 	pthread_spin_unlock(&g_caches_lock);
1678 	if (file != NULL) {
1679 		cache_free_buffers(file);
1680 		buf = spdk_mempool_get(g_cache_pool);
1681 		if (buf != NULL) {
1682 			return buf;
1683 		}
1684 	}
1685 
1686 	pthread_spin_lock(&g_caches_lock);
1687 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1688 		if (!file->open_for_writing && file != context) {
1689 			break;
1690 		}
1691 	}
1692 	pthread_spin_unlock(&g_caches_lock);
1693 	if (file != NULL) {
1694 		cache_free_buffers(file);
1695 		buf = spdk_mempool_get(g_cache_pool);
1696 		if (buf != NULL) {
1697 			return buf;
1698 		}
1699 	}
1700 
1701 	pthread_spin_lock(&g_caches_lock);
1702 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1703 		if (file != context) {
1704 			break;
1705 		}
1706 	}
1707 	pthread_spin_unlock(&g_caches_lock);
1708 	if (file != NULL) {
1709 		cache_free_buffers(file);
1710 		buf = spdk_mempool_get(g_cache_pool);
1711 		if (buf != NULL) {
1712 			return buf;
1713 		}
1714 	}
1715 
1716 	return NULL;
1717 }
1718 
1719 static struct cache_buffer *
1720 cache_insert_buffer(struct spdk_file *file, uint64_t offset)
1721 {
1722 	struct cache_buffer *buf;
1723 	int count = 0;
1724 
1725 	buf = calloc(1, sizeof(*buf));
1726 	if (buf == NULL) {
1727 		SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "calloc failed\n");
1728 		return NULL;
1729 	}
1730 
1731 	buf->buf = alloc_cache_memory_buffer(file);
1732 	while (buf->buf == NULL) {
1733 		/*
1734 		 * TODO: alloc_cache_memory_buffer() should eventually free
1735 		 *  some buffers.  Need a more sophisticated check here, instead
1736 		 *  of just bailing if 100 tries does not result in getting a
1737 		 *  free buffer.  This will involve using the sync channel's
1738 		 *  semaphore to block until a buffer becomes available.
1739 		 */
1740 		if (count++ == 100) {
1741 			SPDK_ERRLOG("could not allocate cache buffer\n");
1742 			assert(false);
1743 			free(buf);
1744 			return NULL;
1745 		}
1746 		buf->buf = alloc_cache_memory_buffer(file);
1747 	}
1748 
1749 	buf->buf_size = CACHE_BUFFER_SIZE;
1750 	buf->offset = offset;
1751 
1752 	pthread_spin_lock(&g_caches_lock);
1753 	if (file->tree->present_mask == 0) {
1754 		TAILQ_INSERT_TAIL(&g_caches, file, cache_tailq);
1755 	}
1756 	file->tree = spdk_tree_insert_buffer(file->tree, buf);
1757 	pthread_spin_unlock(&g_caches_lock);
1758 
1759 	return buf;
1760 }
1761 
1762 static struct cache_buffer *
1763 cache_append_buffer(struct spdk_file *file)
1764 {
1765 	struct cache_buffer *last;
1766 
1767 	assert(file->last == NULL || file->last->bytes_filled == file->last->buf_size);
1768 	assert((file->append_pos % CACHE_BUFFER_SIZE) == 0);
1769 
1770 	last = cache_insert_buffer(file, file->append_pos);
1771 	if (last == NULL) {
1772 		SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "cache_insert_buffer failed\n");
1773 		return NULL;
1774 	}
1775 
1776 	file->last = last;
1777 
1778 	return last;
1779 }
1780 
1781 static void
1782 __wake_caller(struct spdk_fs_cb_args *args)
1783 {
1784 	sem_post(args->sem);
1785 }
1786 
1787 static void __check_sync_reqs(struct spdk_file *file);
1788 
1789 static void
1790 __file_cache_finish_sync(struct spdk_file *file)
1791 {
1792 	struct spdk_fs_request *sync_req;
1793 	struct spdk_fs_cb_args *sync_args;
1794 
1795 	pthread_spin_lock(&file->lock);
1796 	sync_req = TAILQ_FIRST(&file->sync_requests);
1797 	sync_args = &sync_req->args;
1798 	assert(sync_args->op.sync.offset <= file->length_flushed);
1799 	BLOBFS_TRACE(file, "sync done offset=%jx\n", sync_args->op.sync.offset);
1800 	TAILQ_REMOVE(&file->sync_requests, sync_req, args.op.sync.tailq);
1801 	pthread_spin_unlock(&file->lock);
1802 
1803 	sync_args->fn.file_op(sync_args->arg, 0);
1804 	__check_sync_reqs(file);
1805 
1806 	pthread_spin_lock(&file->lock);
1807 	free_fs_request(sync_req);
1808 	pthread_spin_unlock(&file->lock);
1809 }
1810 
1811 static void
1812 __file_cache_finish_sync_bs_cb(void *ctx, int bserrno)
1813 {
1814 	struct spdk_file *file = ctx;
1815 
1816 	__file_cache_finish_sync(file);
1817 }
1818 
1819 static void
1820 __free_args(struct spdk_fs_cb_args *args)
1821 {
1822 	struct spdk_fs_request *req;
1823 
1824 	if (!args->from_request) {
1825 		free(args);
1826 	} else {
1827 		/* Depends on args being at the start of the spdk_fs_request structure. */
1828 		req = (struct spdk_fs_request *)args;
1829 		free_fs_request(req);
1830 	}
1831 }
1832 
1833 static void
1834 __check_sync_reqs(struct spdk_file *file)
1835 {
1836 	struct spdk_fs_request *sync_req;
1837 
1838 	pthread_spin_lock(&file->lock);
1839 
1840 	TAILQ_FOREACH(sync_req, &file->sync_requests, args.op.sync.tailq) {
1841 		if (sync_req->args.op.sync.offset <= file->length_flushed) {
1842 			break;
1843 		}
1844 	}
1845 
1846 	if (sync_req != NULL && !sync_req->args.op.sync.xattr_in_progress) {
1847 		BLOBFS_TRACE(file, "set xattr length 0x%jx\n", file->length_flushed);
1848 		sync_req->args.op.sync.xattr_in_progress = true;
1849 		spdk_blob_set_xattr(file->blob, "length", &file->length_flushed,
1850 				    sizeof(file->length_flushed));
1851 
1852 		pthread_spin_unlock(&file->lock);
1853 		spdk_blob_sync_md(file->blob, __file_cache_finish_sync_bs_cb, file);
1854 	} else {
1855 		pthread_spin_unlock(&file->lock);
1856 	}
1857 }
1858 
1859 static void
1860 __file_flush_done(void *arg, int bserrno)
1861 {
1862 	struct spdk_fs_cb_args *args = arg;
1863 	struct spdk_file *file = args->file;
1864 	struct cache_buffer *next = args->op.flush.cache_buffer;
1865 
1866 	BLOBFS_TRACE(file, "length=%jx\n", args->op.flush.length);
1867 
1868 	pthread_spin_lock(&file->lock);
1869 	next->in_progress = false;
1870 	next->bytes_flushed += args->op.flush.length;
1871 	file->length_flushed += args->op.flush.length;
1872 	if (file->length_flushed > file->length) {
1873 		file->length = file->length_flushed;
1874 	}
1875 	if (next->bytes_flushed == next->buf_size) {
1876 		BLOBFS_TRACE(file, "write buffer fully flushed 0x%jx\n", file->length_flushed);
1877 		next = spdk_tree_find_buffer(file->tree, file->length_flushed);
1878 	}
1879 
1880 	/*
1881 	 * Assert that there is no cached data that extends past the end of the underlying
1882 	 *  blob.
1883 	 */
1884 	assert(next == NULL || next->offset < __file_get_blob_size(file) ||
1885 	       next->bytes_filled == 0);
1886 
1887 	pthread_spin_unlock(&file->lock);
1888 
1889 	__check_sync_reqs(file);
1890 
1891 	__file_flush(args);
1892 }
1893 
1894 static void
1895 __file_flush(void *_args)
1896 {
1897 	struct spdk_fs_cb_args *args = _args;
1898 	struct spdk_file *file = args->file;
1899 	struct cache_buffer *next;
1900 	uint64_t offset, length, start_page, num_pages;
1901 	uint32_t page_size;
1902 
1903 	pthread_spin_lock(&file->lock);
1904 	next = spdk_tree_find_buffer(file->tree, file->length_flushed);
1905 	if (next == NULL || next->in_progress) {
1906 		/*
1907 		 * There is either no data to flush, or a flush I/O is already in
1908 		 *  progress.  So return immediately - if a flush I/O is in
1909 		 *  progress we will flush more data after that is completed.
1910 		 */
1911 		__free_args(args);
1912 		pthread_spin_unlock(&file->lock);
1913 		return;
1914 	}
1915 
1916 	offset = next->offset + next->bytes_flushed;
1917 	length = next->bytes_filled - next->bytes_flushed;
1918 	if (length == 0) {
1919 		__free_args(args);
1920 		pthread_spin_unlock(&file->lock);
1921 		return;
1922 	}
1923 	args->op.flush.length = length;
1924 	args->op.flush.cache_buffer = next;
1925 
1926 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
1927 
1928 	next->in_progress = true;
1929 	BLOBFS_TRACE(file, "offset=%jx length=%jx page start=%jx num=%jx\n",
1930 		     offset, length, start_page, num_pages);
1931 	pthread_spin_unlock(&file->lock);
1932 	spdk_bs_io_write_blob(file->blob, file->fs->sync_target.sync_fs_channel->bs_channel,
1933 			      next->buf + (start_page * page_size) - next->offset,
1934 			      start_page, num_pages,
1935 			      __file_flush_done, args);
1936 }
1937 
1938 static void
1939 __file_extend_done(void *arg, int bserrno)
1940 {
1941 	struct spdk_fs_cb_args *args = arg;
1942 
1943 	__wake_caller(args);
1944 }
1945 
1946 static void
1947 __file_extend_blob(void *_args)
1948 {
1949 	struct spdk_fs_cb_args *args = _args;
1950 	struct spdk_file *file = args->file;
1951 
1952 	spdk_blob_resize(file->blob, args->op.resize.num_clusters);
1953 
1954 	spdk_blob_sync_md(file->blob, __file_extend_done, args);
1955 }
1956 
1957 static void
1958 __rw_from_file_done(void *arg, int bserrno)
1959 {
1960 	struct spdk_fs_cb_args *args = arg;
1961 
1962 	__wake_caller(args);
1963 	__free_args(args);
1964 }
1965 
1966 static void
1967 __rw_from_file(void *_args)
1968 {
1969 	struct spdk_fs_cb_args *args = _args;
1970 	struct spdk_file *file = args->file;
1971 
1972 	if (args->op.rw.is_read) {
1973 		spdk_file_read_async(file, file->fs->sync_target.sync_io_channel, args->op.rw.user_buf,
1974 				     args->op.rw.offset, args->op.rw.length,
1975 				     __rw_from_file_done, args);
1976 	} else {
1977 		spdk_file_write_async(file, file->fs->sync_target.sync_io_channel, args->op.rw.user_buf,
1978 				      args->op.rw.offset, args->op.rw.length,
1979 				      __rw_from_file_done, args);
1980 	}
1981 }
1982 
1983 static int
1984 __send_rw_from_file(struct spdk_file *file, sem_t *sem, void *payload,
1985 		    uint64_t offset, uint64_t length, bool is_read)
1986 {
1987 	struct spdk_fs_cb_args *args;
1988 
1989 	args = calloc(1, sizeof(*args));
1990 	if (args == NULL) {
1991 		sem_post(sem);
1992 		return -ENOMEM;
1993 	}
1994 
1995 	args->file = file;
1996 	args->sem = sem;
1997 	args->op.rw.user_buf = payload;
1998 	args->op.rw.offset = offset;
1999 	args->op.rw.length = length;
2000 	args->op.rw.is_read = is_read;
2001 	file->fs->send_request(__rw_from_file, args);
2002 	return 0;
2003 }
2004 
2005 int
2006 spdk_file_write(struct spdk_file *file, struct spdk_io_channel *_channel,
2007 		void *payload, uint64_t offset, uint64_t length)
2008 {
2009 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2010 	struct spdk_fs_cb_args *args;
2011 	uint64_t rem_length, copy, blob_size, cluster_sz;
2012 	uint32_t cache_buffers_filled = 0;
2013 	uint8_t *cur_payload;
2014 	struct cache_buffer *last;
2015 
2016 	BLOBFS_TRACE_RW(file, "offset=%jx length=%jx\n", offset, length);
2017 
2018 	if (length == 0) {
2019 		return 0;
2020 	}
2021 
2022 	if (offset != file->append_pos) {
2023 		BLOBFS_TRACE(file, " error offset=%jx append_pos=%jx\n", offset, file->append_pos);
2024 		return -EINVAL;
2025 	}
2026 
2027 	pthread_spin_lock(&file->lock);
2028 	file->open_for_writing = true;
2029 
2030 	if (file->last == NULL) {
2031 		if (file->append_pos % CACHE_BUFFER_SIZE == 0) {
2032 			cache_append_buffer(file);
2033 		} else {
2034 			int rc;
2035 
2036 			file->append_pos += length;
2037 			pthread_spin_unlock(&file->lock);
2038 			rc = __send_rw_from_file(file, &channel->sem, payload,
2039 						 offset, length, false);
2040 			sem_wait(&channel->sem);
2041 			return rc;
2042 		}
2043 	}
2044 
2045 	blob_size = __file_get_blob_size(file);
2046 
2047 	if ((offset + length) > blob_size) {
2048 		struct spdk_fs_cb_args extend_args = {};
2049 
2050 		cluster_sz = file->fs->bs_opts.cluster_sz;
2051 		extend_args.sem = &channel->sem;
2052 		extend_args.op.resize.num_clusters = __bytes_to_clusters((offset + length), cluster_sz);
2053 		extend_args.file = file;
2054 		BLOBFS_TRACE(file, "start resize to %u clusters\n", extend_args.op.resize.num_clusters);
2055 		pthread_spin_unlock(&file->lock);
2056 		file->fs->send_request(__file_extend_blob, &extend_args);
2057 		sem_wait(&channel->sem);
2058 	}
2059 
2060 	last = file->last;
2061 	rem_length = length;
2062 	cur_payload = payload;
2063 	while (rem_length > 0) {
2064 		copy = last->buf_size - last->bytes_filled;
2065 		if (copy > rem_length) {
2066 			copy = rem_length;
2067 		}
2068 		BLOBFS_TRACE_RW(file, "  fill offset=%jx length=%jx\n", file->append_pos, copy);
2069 		memcpy(&last->buf[last->bytes_filled], cur_payload, copy);
2070 		file->append_pos += copy;
2071 		if (file->length < file->append_pos) {
2072 			file->length = file->append_pos;
2073 		}
2074 		cur_payload += copy;
2075 		last->bytes_filled += copy;
2076 		rem_length -= copy;
2077 		if (last->bytes_filled == last->buf_size) {
2078 			cache_buffers_filled++;
2079 			last = cache_append_buffer(file);
2080 			if (last == NULL) {
2081 				BLOBFS_TRACE(file, "nomem\n");
2082 				pthread_spin_unlock(&file->lock);
2083 				return -ENOMEM;
2084 			}
2085 		}
2086 	}
2087 
2088 	pthread_spin_unlock(&file->lock);
2089 
2090 	if (cache_buffers_filled == 0) {
2091 		return 0;
2092 	}
2093 
2094 	args = calloc(1, sizeof(*args));
2095 	if (args == NULL) {
2096 		return -ENOMEM;
2097 	}
2098 
2099 	args->file = file;
2100 	file->fs->send_request(__file_flush, args);
2101 	return 0;
2102 }
2103 
2104 static void
2105 __readahead_done(void *arg, int bserrno)
2106 {
2107 	struct spdk_fs_cb_args *args = arg;
2108 	struct cache_buffer *cache_buffer = args->op.readahead.cache_buffer;
2109 	struct spdk_file *file = args->file;
2110 
2111 	BLOBFS_TRACE(file, "offset=%jx\n", cache_buffer->offset);
2112 
2113 	pthread_spin_lock(&file->lock);
2114 	cache_buffer->bytes_filled = args->op.readahead.length;
2115 	cache_buffer->bytes_flushed = args->op.readahead.length;
2116 	cache_buffer->in_progress = false;
2117 	pthread_spin_unlock(&file->lock);
2118 
2119 	__free_args(args);
2120 }
2121 
2122 static void
2123 __readahead(void *_args)
2124 {
2125 	struct spdk_fs_cb_args *args = _args;
2126 	struct spdk_file *file = args->file;
2127 	uint64_t offset, length, start_page, num_pages;
2128 	uint32_t page_size;
2129 
2130 	offset = args->op.readahead.offset;
2131 	length = args->op.readahead.length;
2132 	assert(length > 0);
2133 
2134 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
2135 
2136 	BLOBFS_TRACE(file, "offset=%jx length=%jx page start=%jx num=%jx\n",
2137 		     offset, length, start_page, num_pages);
2138 	spdk_bs_io_read_blob(file->blob, file->fs->sync_target.sync_fs_channel->bs_channel,
2139 			     args->op.readahead.cache_buffer->buf,
2140 			     start_page, num_pages,
2141 			     __readahead_done, args);
2142 }
2143 
2144 static uint64_t
2145 __next_cache_buffer_offset(uint64_t offset)
2146 {
2147 	return (offset + CACHE_BUFFER_SIZE) & ~(CACHE_TREE_LEVEL_MASK(0));
2148 }
2149 
2150 static void
2151 check_readahead(struct spdk_file *file, uint64_t offset)
2152 {
2153 	struct spdk_fs_cb_args *args;
2154 
2155 	offset = __next_cache_buffer_offset(offset);
2156 	if (spdk_tree_find_buffer(file->tree, offset) != NULL || file->length <= offset) {
2157 		return;
2158 	}
2159 
2160 	args = calloc(1, sizeof(*args));
2161 	if (args == NULL) {
2162 		return;
2163 	}
2164 
2165 	BLOBFS_TRACE(file, "offset=%jx\n", offset);
2166 
2167 	args->file = file;
2168 	args->op.readahead.offset = offset;
2169 	args->op.readahead.cache_buffer = cache_insert_buffer(file, offset);
2170 	args->op.readahead.cache_buffer->in_progress = true;
2171 	if (file->length < (offset + CACHE_BUFFER_SIZE)) {
2172 		args->op.readahead.length = file->length & (CACHE_BUFFER_SIZE - 1);
2173 	} else {
2174 		args->op.readahead.length = CACHE_BUFFER_SIZE;
2175 	}
2176 	file->fs->send_request(__readahead, args);
2177 }
2178 
2179 static int
2180 __file_read(struct spdk_file *file, void *payload, uint64_t offset, uint64_t length, sem_t *sem)
2181 {
2182 	struct cache_buffer *buf;
2183 	int rc;
2184 
2185 	buf = spdk_tree_find_filled_buffer(file->tree, offset);
2186 	if (buf == NULL) {
2187 		pthread_spin_unlock(&file->lock);
2188 		rc = __send_rw_from_file(file, sem, payload, offset, length, true);
2189 		pthread_spin_lock(&file->lock);
2190 		return rc;
2191 	}
2192 
2193 	if ((offset + length) > (buf->offset + buf->bytes_filled)) {
2194 		length = buf->offset + buf->bytes_filled - offset;
2195 	}
2196 	BLOBFS_TRACE(file, "read %p offset=%ju length=%ju\n", payload, offset, length);
2197 	memcpy(payload, &buf->buf[offset - buf->offset], length);
2198 	if ((offset + length) % CACHE_BUFFER_SIZE == 0) {
2199 		pthread_spin_lock(&g_caches_lock);
2200 		spdk_tree_remove_buffer(file->tree, buf);
2201 		if (file->tree->present_mask == 0) {
2202 			TAILQ_REMOVE(&g_caches, file, cache_tailq);
2203 		}
2204 		pthread_spin_unlock(&g_caches_lock);
2205 	}
2206 
2207 	sem_post(sem);
2208 	return 0;
2209 }
2210 
2211 int64_t
2212 spdk_file_read(struct spdk_file *file, struct spdk_io_channel *_channel,
2213 	       void *payload, uint64_t offset, uint64_t length)
2214 {
2215 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2216 	uint64_t final_offset, final_length;
2217 	uint32_t sub_reads = 0;
2218 	int rc = 0;
2219 
2220 	pthread_spin_lock(&file->lock);
2221 
2222 	BLOBFS_TRACE_RW(file, "offset=%ju length=%ju\n", offset, length);
2223 
2224 	file->open_for_writing = false;
2225 
2226 	if (length == 0 || offset >= file->append_pos) {
2227 		pthread_spin_unlock(&file->lock);
2228 		return 0;
2229 	}
2230 
2231 	if (offset + length > file->append_pos) {
2232 		length = file->append_pos - offset;
2233 	}
2234 
2235 	if (offset != file->next_seq_offset) {
2236 		file->seq_byte_count = 0;
2237 	}
2238 	file->seq_byte_count += length;
2239 	file->next_seq_offset = offset + length;
2240 	if (file->seq_byte_count >= CACHE_READAHEAD_THRESHOLD) {
2241 		check_readahead(file, offset);
2242 		check_readahead(file, offset + CACHE_BUFFER_SIZE);
2243 	}
2244 
2245 	final_length = 0;
2246 	final_offset = offset + length;
2247 	while (offset < final_offset) {
2248 		length = NEXT_CACHE_BUFFER_OFFSET(offset) - offset;
2249 		if (length > (final_offset - offset)) {
2250 			length = final_offset - offset;
2251 		}
2252 		rc = __file_read(file, payload, offset, length, &channel->sem);
2253 		if (rc == 0) {
2254 			final_length += length;
2255 		} else {
2256 			break;
2257 		}
2258 		payload += length;
2259 		offset += length;
2260 		sub_reads++;
2261 	}
2262 	pthread_spin_unlock(&file->lock);
2263 	while (sub_reads-- > 0) {
2264 		sem_wait(&channel->sem);
2265 	}
2266 	if (rc == 0) {
2267 		return final_length;
2268 	} else {
2269 		return rc;
2270 	}
2271 }
2272 
2273 static void
2274 _file_sync(struct spdk_file *file, struct spdk_fs_channel *channel,
2275 	   spdk_file_op_complete cb_fn, void *cb_arg)
2276 {
2277 	struct spdk_fs_request *sync_req;
2278 	struct spdk_fs_request *flush_req;
2279 	struct spdk_fs_cb_args *sync_args;
2280 	struct spdk_fs_cb_args *flush_args;
2281 
2282 	BLOBFS_TRACE(file, "offset=%jx\n", file->append_pos);
2283 
2284 	pthread_spin_lock(&file->lock);
2285 	if (file->append_pos <= file->length_flushed || file->last == NULL) {
2286 		BLOBFS_TRACE(file, "done - no data to flush\n");
2287 		pthread_spin_unlock(&file->lock);
2288 		cb_fn(cb_arg, 0);
2289 		return;
2290 	}
2291 
2292 	sync_req = alloc_fs_request(channel);
2293 	assert(sync_req != NULL);
2294 	sync_args = &sync_req->args;
2295 
2296 	flush_req = alloc_fs_request(channel);
2297 	assert(flush_req != NULL);
2298 	flush_args = &flush_req->args;
2299 
2300 	sync_args->file = file;
2301 	sync_args->fn.file_op = cb_fn;
2302 	sync_args->arg = cb_arg;
2303 	sync_args->op.sync.offset = file->append_pos;
2304 	sync_args->op.sync.xattr_in_progress = false;
2305 	TAILQ_INSERT_TAIL(&file->sync_requests, sync_req, args.op.sync.tailq);
2306 	pthread_spin_unlock(&file->lock);
2307 
2308 	flush_args->file = file;
2309 	channel->send_request(__file_flush, flush_args);
2310 }
2311 
2312 int
2313 spdk_file_sync(struct spdk_file *file, struct spdk_io_channel *_channel)
2314 {
2315 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2316 
2317 	_file_sync(file, channel, __sem_post, &channel->sem);
2318 	sem_wait(&channel->sem);
2319 
2320 	return 0;
2321 }
2322 
2323 void
2324 spdk_file_sync_async(struct spdk_file *file, struct spdk_io_channel *_channel,
2325 		     spdk_file_op_complete cb_fn, void *cb_arg)
2326 {
2327 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2328 
2329 	_file_sync(file, channel, cb_fn, cb_arg);
2330 }
2331 
2332 void
2333 spdk_file_set_priority(struct spdk_file *file, uint32_t priority)
2334 {
2335 	BLOBFS_TRACE(file, "priority=%u\n", priority);
2336 	file->priority = priority;
2337 
2338 }
2339 
2340 /*
2341  * Close routines
2342  */
2343 
2344 static void
2345 __file_close_async_done(void *ctx, int bserrno)
2346 {
2347 	struct spdk_fs_request *req = ctx;
2348 	struct spdk_fs_cb_args *args = &req->args;
2349 	struct spdk_file *file = args->file;
2350 
2351 	if (file->is_deleted) {
2352 		spdk_fs_delete_file_async(file->fs, file->name, blob_delete_cb, ctx);
2353 		return;
2354 	}
2355 	args->fn.file_op(args->arg, bserrno);
2356 	free_fs_request(req);
2357 }
2358 
2359 static void
2360 __file_close_async(struct spdk_file *file, struct spdk_fs_request *req)
2361 {
2362 	struct spdk_blob *blob;
2363 
2364 	pthread_spin_lock(&file->lock);
2365 	if (file->ref_count == 0) {
2366 		pthread_spin_unlock(&file->lock);
2367 		__file_close_async_done(req, -EBADF);
2368 		return;
2369 	}
2370 
2371 	file->ref_count--;
2372 	if (file->ref_count > 0) {
2373 		pthread_spin_unlock(&file->lock);
2374 		__file_close_async_done(req, 0);
2375 		return;
2376 	}
2377 
2378 	pthread_spin_unlock(&file->lock);
2379 
2380 	blob = file->blob;
2381 	file->blob = NULL;
2382 	spdk_blob_close(blob, __file_close_async_done, req);
2383 }
2384 
2385 static void
2386 __file_close_async__sync_done(void *arg, int fserrno)
2387 {
2388 	struct spdk_fs_request *req = arg;
2389 	struct spdk_fs_cb_args *args = &req->args;
2390 
2391 	__file_close_async(args->file, req);
2392 }
2393 
2394 void
2395 spdk_file_close_async(struct spdk_file *file, spdk_file_op_complete cb_fn, void *cb_arg)
2396 {
2397 	struct spdk_fs_request *req;
2398 	struct spdk_fs_cb_args *args;
2399 
2400 	req = alloc_fs_request(file->fs->md_target.md_fs_channel);
2401 	if (req == NULL) {
2402 		cb_fn(cb_arg, -ENOMEM);
2403 		return;
2404 	}
2405 
2406 	args = &req->args;
2407 	args->file = file;
2408 	args->fn.file_op = cb_fn;
2409 	args->arg = cb_arg;
2410 
2411 	spdk_file_sync_async(file, file->fs->md_target.md_io_channel, __file_close_async__sync_done, req);
2412 }
2413 
2414 static void
2415 __file_close_done(void *arg, int fserrno)
2416 {
2417 	struct spdk_fs_cb_args *args = arg;
2418 
2419 	args->rc = fserrno;
2420 	sem_post(args->sem);
2421 }
2422 
2423 static void
2424 __file_close(void *arg)
2425 {
2426 	struct spdk_fs_request *req = arg;
2427 	struct spdk_fs_cb_args *args = &req->args;
2428 	struct spdk_file *file = args->file;
2429 
2430 	__file_close_async(file, req);
2431 }
2432 
2433 int
2434 spdk_file_close(struct spdk_file *file, struct spdk_io_channel *_channel)
2435 {
2436 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2437 	struct spdk_fs_request *req;
2438 	struct spdk_fs_cb_args *args;
2439 
2440 	req = alloc_fs_request(channel);
2441 	assert(req != NULL);
2442 
2443 	args = &req->args;
2444 
2445 	spdk_file_sync(file, _channel);
2446 	BLOBFS_TRACE(file, "name=%s\n", file->name);
2447 	args->file = file;
2448 	args->sem = &channel->sem;
2449 	args->fn.file_op = __file_close_done;
2450 	args->arg = req;
2451 	channel->send_request(__file_close, req);
2452 	sem_wait(&channel->sem);
2453 
2454 	return args->rc;
2455 }
2456 
2457 static void
2458 cache_free_buffers(struct spdk_file *file)
2459 {
2460 	BLOBFS_TRACE(file, "free=%s\n", file->name);
2461 	pthread_spin_lock(&file->lock);
2462 	pthread_spin_lock(&g_caches_lock);
2463 	if (file->tree->present_mask == 0) {
2464 		pthread_spin_unlock(&g_caches_lock);
2465 		pthread_spin_unlock(&file->lock);
2466 		return;
2467 	}
2468 	spdk_tree_free_buffers(file->tree);
2469 
2470 	TAILQ_REMOVE(&g_caches, file, cache_tailq);
2471 	/* If not freed, put it in the end of the queue */
2472 	if (file->tree->present_mask != 0) {
2473 		TAILQ_INSERT_TAIL(&g_caches, file, cache_tailq);
2474 	}
2475 	file->last = NULL;
2476 	pthread_spin_unlock(&g_caches_lock);
2477 	pthread_spin_unlock(&file->lock);
2478 }
2479 
2480 SPDK_LOG_REGISTER_COMPONENT("blobfs", SPDK_LOG_BLOBFS)
2481 SPDK_LOG_REGISTER_COMPONENT("blobfs_rw", SPDK_LOG_BLOBFS_RW)
2482