xref: /spdk/lib/blobfs/blobfs.c (revision 981607175be8fa4658ff77eb478facfaaa64fa1a)
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 fs_load_done(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, fs_load_done, req);
544 		free(deleted_file);
545 		return 0;
546 	}
547 
548 	return 1;
549 }
550 
551 static void
552 fs_load_done(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 	/* The filesystem has been loaded.  Now check if there are any files that
559 	 *  were marked for deletion before last unload.  Do not complete the
560 	 *  fs_load callback until all of them have been deleted on disk.
561 	 */
562 	if (_handle_deleted_files(req) == 0) {
563 		/* We found a file that's been marked for deleting but not actually
564 		 *  deleted yet.  This function will get called again once the delete
565 		 *  operation is completed.
566 		 */
567 		return;
568 	}
569 
570 	args->fn.fs_op_with_handle(args->arg, fs, 0);
571 	free_fs_request(req);
572 
573 }
574 
575 static void
576 iter_cb(void *ctx, struct spdk_blob *blob, int rc)
577 {
578 	struct spdk_fs_request *req = ctx;
579 	struct spdk_fs_cb_args *args = &req->args;
580 	struct spdk_filesystem *fs = args->fs;
581 	uint64_t *length;
582 	const char *name;
583 	uint32_t *is_deleted;
584 	size_t value_len;
585 
586 	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 
640 static void
641 load_cb(void *ctx, struct spdk_blob_store *bs, int bserrno)
642 {
643 	struct spdk_fs_request *req = ctx;
644 	struct spdk_fs_cb_args *args = &req->args;
645 	struct spdk_filesystem *fs = args->fs;
646 	struct spdk_bs_type bstype;
647 	static const struct spdk_bs_type blobfs_type = {"BLOBFS"};
648 	static const struct spdk_bs_type zeros;
649 
650 	if (bserrno != 0) {
651 		args->fn.fs_op_with_handle(args->arg, NULL, bserrno);
652 		free_fs_request(req);
653 		free(fs);
654 		return;
655 	}
656 
657 	bstype = spdk_bs_get_bstype(bs);
658 
659 	if (!memcmp(&bstype, &zeros, sizeof(bstype))) {
660 		SPDK_DEBUGLOG(SPDK_LOG_BLOB, "assigning bstype\n");
661 		spdk_bs_set_bstype(bs, blobfs_type);
662 	} else if (memcmp(&bstype, &blobfs_type, sizeof(bstype))) {
663 		SPDK_DEBUGLOG(SPDK_LOG_BLOB, "not blobfs\n");
664 		SPDK_TRACEDUMP(SPDK_LOG_BLOB, "bstype", &bstype, sizeof(bstype));
665 		args->fn.fs_op_with_handle(args->arg, NULL, bserrno);
666 		free_fs_request(req);
667 		free(fs);
668 		return;
669 	}
670 
671 	common_fs_bs_init(fs, bs);
672 	fs_load_done(req, 0);
673 }
674 
675 void
676 spdk_fs_load(struct spdk_bs_dev *dev, fs_send_request_fn send_request_fn,
677 	     spdk_fs_op_with_handle_complete cb_fn, void *cb_arg)
678 {
679 	struct spdk_filesystem *fs;
680 	struct spdk_fs_cb_args *args;
681 	struct spdk_fs_request *req;
682 	struct spdk_bs_opts	bs_opts;
683 
684 	fs = fs_alloc(dev, send_request_fn);
685 	if (fs == NULL) {
686 		cb_fn(cb_arg, NULL, -ENOMEM);
687 		return;
688 	}
689 
690 	fs_conf_parse();
691 
692 	req = alloc_fs_request(fs->md_target.md_fs_channel);
693 	if (req == NULL) {
694 		spdk_put_io_channel(fs->md_target.md_io_channel);
695 		spdk_io_device_unregister(&fs->md_target, NULL);
696 		spdk_put_io_channel(fs->sync_target.sync_io_channel);
697 		spdk_io_device_unregister(&fs->sync_target, NULL);
698 		spdk_io_device_unregister(&fs->io_target, NULL);
699 		free(fs);
700 		cb_fn(cb_arg, NULL, -ENOMEM);
701 		return;
702 	}
703 
704 	args = &req->args;
705 	args->fn.fs_op_with_handle = cb_fn;
706 	args->arg = cb_arg;
707 	args->fs = fs;
708 	TAILQ_INIT(&args->op.fs_load.deleted_files);
709 	spdk_bs_opts_init(&bs_opts);
710 	bs_opts.iter_cb_fn = iter_cb;
711 	bs_opts.iter_cb_arg = req;
712 	spdk_bs_load(dev, &bs_opts, load_cb, req);
713 }
714 
715 static void
716 unload_cb(void *ctx, int bserrno)
717 {
718 	struct spdk_fs_request *req = ctx;
719 	struct spdk_fs_cb_args *args = &req->args;
720 	struct spdk_filesystem *fs = args->fs;
721 
722 	pthread_mutex_lock(&g_cache_init_lock);
723 	g_fs_count--;
724 	if (g_fs_count == 0) {
725 		__free_cache();
726 	}
727 	pthread_mutex_unlock(&g_cache_init_lock);
728 
729 	args->fn.fs_op(args->arg, bserrno);
730 	free(req);
731 
732 	spdk_io_device_unregister(&fs->io_target, NULL);
733 	spdk_io_device_unregister(&fs->sync_target, NULL);
734 	spdk_io_device_unregister(&fs->md_target, NULL);
735 
736 	free(fs);
737 }
738 
739 void
740 spdk_fs_unload(struct spdk_filesystem *fs, spdk_fs_op_complete cb_fn, void *cb_arg)
741 {
742 	struct spdk_fs_request *req;
743 	struct spdk_fs_cb_args *args;
744 
745 	/*
746 	 * We must free the md_channel before unloading the blobstore, so just
747 	 *  allocate this request from the general heap.
748 	 */
749 	req = calloc(1, sizeof(*req));
750 	if (req == NULL) {
751 		cb_fn(cb_arg, -ENOMEM);
752 		return;
753 	}
754 
755 	args = &req->args;
756 	args->fn.fs_op = cb_fn;
757 	args->arg = cb_arg;
758 	args->fs = fs;
759 
760 	spdk_fs_free_io_channel(fs->md_target.md_io_channel);
761 	spdk_fs_free_io_channel(fs->sync_target.sync_io_channel);
762 	spdk_bs_unload(fs->bs, unload_cb, req);
763 }
764 
765 static struct spdk_file *
766 fs_find_file(struct spdk_filesystem *fs, const char *name)
767 {
768 	struct spdk_file *file;
769 
770 	TAILQ_FOREACH(file, &fs->files, tailq) {
771 		if (!strncmp(name, file->name, SPDK_FILE_NAME_MAX)) {
772 			return file;
773 		}
774 	}
775 
776 	return NULL;
777 }
778 
779 void
780 spdk_fs_file_stat_async(struct spdk_filesystem *fs, const char *name,
781 			spdk_file_stat_op_complete cb_fn, void *cb_arg)
782 {
783 	struct spdk_file_stat stat;
784 	struct spdk_file *f = NULL;
785 
786 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
787 		cb_fn(cb_arg, NULL, -ENAMETOOLONG);
788 		return;
789 	}
790 
791 	f = fs_find_file(fs, name);
792 	if (f != NULL) {
793 		stat.blobid = f->blobid;
794 		stat.size = f->append_pos >= f->length ? f->append_pos : f->length;
795 		cb_fn(cb_arg, &stat, 0);
796 		return;
797 	}
798 
799 	cb_fn(cb_arg, NULL, -ENOENT);
800 }
801 
802 static void
803 __copy_stat(void *arg, struct spdk_file_stat *stat, int fserrno)
804 {
805 	struct spdk_fs_request *req = arg;
806 	struct spdk_fs_cb_args *args = &req->args;
807 
808 	args->rc = fserrno;
809 	if (fserrno == 0) {
810 		memcpy(args->arg, stat, sizeof(*stat));
811 	}
812 	sem_post(args->sem);
813 }
814 
815 static void
816 __file_stat(void *arg)
817 {
818 	struct spdk_fs_request *req = arg;
819 	struct spdk_fs_cb_args *args = &req->args;
820 
821 	spdk_fs_file_stat_async(args->fs, args->op.stat.name,
822 				args->fn.stat_op, req);
823 }
824 
825 int
826 spdk_fs_file_stat(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
827 		  const char *name, struct spdk_file_stat *stat)
828 {
829 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
830 	struct spdk_fs_request *req;
831 	int rc;
832 
833 	req = alloc_fs_request(channel);
834 	assert(req != NULL);
835 
836 	req->args.fs = fs;
837 	req->args.op.stat.name = name;
838 	req->args.fn.stat_op = __copy_stat;
839 	req->args.arg = stat;
840 	req->args.sem = &channel->sem;
841 	channel->send_request(__file_stat, req);
842 	sem_wait(&channel->sem);
843 
844 	rc = req->args.rc;
845 	free_fs_request(req);
846 
847 	return rc;
848 }
849 
850 static void
851 fs_create_blob_close_cb(void *ctx, int bserrno)
852 {
853 	struct spdk_fs_request *req = ctx;
854 	struct spdk_fs_cb_args *args = &req->args;
855 
856 	args->fn.file_op(args->arg, bserrno);
857 	free_fs_request(req);
858 }
859 
860 static void
861 fs_create_blob_open_cb(void *ctx, struct spdk_blob *blob, int bserrno)
862 {
863 	struct spdk_fs_request *req = ctx;
864 	struct spdk_fs_cb_args *args = &req->args;
865 	struct spdk_file *f = args->file;
866 	uint64_t length = 0;
867 
868 	spdk_blob_resize(blob, 1);
869 	spdk_blob_set_xattr(blob, "name", f->name, strlen(f->name) + 1);
870 	spdk_blob_set_xattr(blob, "length", &length, sizeof(length));
871 
872 	spdk_blob_close(blob, fs_create_blob_close_cb, args);
873 }
874 
875 static void
876 fs_create_blob_create_cb(void *ctx, spdk_blob_id blobid, int bserrno)
877 {
878 	struct spdk_fs_request *req = ctx;
879 	struct spdk_fs_cb_args *args = &req->args;
880 	struct spdk_file *f = args->file;
881 
882 	f->blobid = blobid;
883 	spdk_bs_open_blob(f->fs->bs, blobid, fs_create_blob_open_cb, req);
884 }
885 
886 void
887 spdk_fs_create_file_async(struct spdk_filesystem *fs, const char *name,
888 			  spdk_file_op_complete cb_fn, void *cb_arg)
889 {
890 	struct spdk_file *file;
891 	struct spdk_fs_request *req;
892 	struct spdk_fs_cb_args *args;
893 
894 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
895 		cb_fn(cb_arg, -ENAMETOOLONG);
896 		return;
897 	}
898 
899 	file = fs_find_file(fs, name);
900 	if (file != NULL) {
901 		cb_fn(cb_arg, -EEXIST);
902 		return;
903 	}
904 
905 	file = file_alloc(fs);
906 	if (file == NULL) {
907 		cb_fn(cb_arg, -ENOMEM);
908 		return;
909 	}
910 
911 	req = alloc_fs_request(fs->md_target.md_fs_channel);
912 	if (req == NULL) {
913 		cb_fn(cb_arg, -ENOMEM);
914 		return;
915 	}
916 
917 	args = &req->args;
918 	args->file = file;
919 	args->fn.file_op = cb_fn;
920 	args->arg = cb_arg;
921 
922 	file->name = strdup(name);
923 	spdk_bs_create_blob(fs->bs, fs_create_blob_create_cb, args);
924 }
925 
926 static void
927 __fs_create_file_done(void *arg, int fserrno)
928 {
929 	struct spdk_fs_request *req = arg;
930 	struct spdk_fs_cb_args *args = &req->args;
931 
932 	args->rc = fserrno;
933 	sem_post(args->sem);
934 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.create.name);
935 }
936 
937 static void
938 __fs_create_file(void *arg)
939 {
940 	struct spdk_fs_request *req = arg;
941 	struct spdk_fs_cb_args *args = &req->args;
942 
943 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.create.name);
944 	spdk_fs_create_file_async(args->fs, args->op.create.name, __fs_create_file_done, req);
945 }
946 
947 int
948 spdk_fs_create_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel, const char *name)
949 {
950 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
951 	struct spdk_fs_request *req;
952 	struct spdk_fs_cb_args *args;
953 	int rc;
954 
955 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
956 
957 	req = alloc_fs_request(channel);
958 	assert(req != NULL);
959 
960 	args = &req->args;
961 	args->fs = fs;
962 	args->op.create.name = name;
963 	args->sem = &channel->sem;
964 	fs->send_request(__fs_create_file, req);
965 	sem_wait(&channel->sem);
966 	rc = args->rc;
967 	free_fs_request(req);
968 
969 	return rc;
970 }
971 
972 static void
973 fs_open_blob_done(void *ctx, struct spdk_blob *blob, int bserrno)
974 {
975 	struct spdk_fs_request *req = ctx;
976 	struct spdk_fs_cb_args *args = &req->args;
977 	struct spdk_file *f = args->file;
978 
979 	f->blob = blob;
980 	while (!TAILQ_EMPTY(&f->open_requests)) {
981 		req = TAILQ_FIRST(&f->open_requests);
982 		args = &req->args;
983 		TAILQ_REMOVE(&f->open_requests, req, args.op.open.tailq);
984 		args->fn.file_op_with_handle(args->arg, f, bserrno);
985 		free_fs_request(req);
986 	}
987 }
988 
989 static void
990 fs_open_blob_create_cb(void *ctx, int bserrno)
991 {
992 	struct spdk_fs_request *req = ctx;
993 	struct spdk_fs_cb_args *args = &req->args;
994 	struct spdk_file *file = args->file;
995 	struct spdk_filesystem *fs = args->fs;
996 
997 	if (file == NULL) {
998 		/*
999 		 * This is from an open with CREATE flag - the file
1000 		 *  is now created so look it up in the file list for this
1001 		 *  filesystem.
1002 		 */
1003 		file = fs_find_file(fs, args->op.open.name);
1004 		assert(file != NULL);
1005 		args->file = file;
1006 	}
1007 
1008 	file->ref_count++;
1009 	TAILQ_INSERT_TAIL(&file->open_requests, req, args.op.open.tailq);
1010 	if (file->ref_count == 1) {
1011 		assert(file->blob == NULL);
1012 		spdk_bs_open_blob(fs->bs, file->blobid, fs_open_blob_done, req);
1013 	} else if (file->blob != NULL) {
1014 		fs_open_blob_done(req, file->blob, 0);
1015 	} else {
1016 		/*
1017 		 * The blob open for this file is in progress due to a previous
1018 		 *  open request.  When that open completes, it will invoke the
1019 		 *  open callback for this request.
1020 		 */
1021 	}
1022 }
1023 
1024 void
1025 spdk_fs_open_file_async(struct spdk_filesystem *fs, const char *name, uint32_t flags,
1026 			spdk_file_op_with_handle_complete cb_fn, void *cb_arg)
1027 {
1028 	struct spdk_file *f = NULL;
1029 	struct spdk_fs_request *req;
1030 	struct spdk_fs_cb_args *args;
1031 
1032 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1033 		cb_fn(cb_arg, NULL, -ENAMETOOLONG);
1034 		return;
1035 	}
1036 
1037 	f = fs_find_file(fs, name);
1038 	if (f == NULL && !(flags & SPDK_BLOBFS_OPEN_CREATE)) {
1039 		cb_fn(cb_arg, NULL, -ENOENT);
1040 		return;
1041 	}
1042 
1043 	if (f != NULL && f->is_deleted == true) {
1044 		cb_fn(cb_arg, NULL, -ENOENT);
1045 		return;
1046 	}
1047 
1048 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1049 	if (req == NULL) {
1050 		cb_fn(cb_arg, NULL, -ENOMEM);
1051 		return;
1052 	}
1053 
1054 	args = &req->args;
1055 	args->fn.file_op_with_handle = cb_fn;
1056 	args->arg = cb_arg;
1057 	args->file = f;
1058 	args->fs = fs;
1059 	args->op.open.name = name;
1060 
1061 	if (f == NULL) {
1062 		spdk_fs_create_file_async(fs, name, fs_open_blob_create_cb, req);
1063 	} else {
1064 		fs_open_blob_create_cb(req, 0);
1065 	}
1066 }
1067 
1068 static void
1069 __fs_open_file_done(void *arg, struct spdk_file *file, int bserrno)
1070 {
1071 	struct spdk_fs_request *req = arg;
1072 	struct spdk_fs_cb_args *args = &req->args;
1073 
1074 	args->file = file;
1075 	args->rc = bserrno;
1076 	sem_post(args->sem);
1077 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.open.name);
1078 }
1079 
1080 static void
1081 __fs_open_file(void *arg)
1082 {
1083 	struct spdk_fs_request *req = arg;
1084 	struct spdk_fs_cb_args *args = &req->args;
1085 
1086 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", args->op.open.name);
1087 	spdk_fs_open_file_async(args->fs, args->op.open.name, args->op.open.flags,
1088 				__fs_open_file_done, req);
1089 }
1090 
1091 int
1092 spdk_fs_open_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1093 		  const char *name, uint32_t flags, struct spdk_file **file)
1094 {
1095 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1096 	struct spdk_fs_request *req;
1097 	struct spdk_fs_cb_args *args;
1098 	int rc;
1099 
1100 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
1101 
1102 	req = alloc_fs_request(channel);
1103 	assert(req != NULL);
1104 
1105 	args = &req->args;
1106 	args->fs = fs;
1107 	args->op.open.name = name;
1108 	args->op.open.flags = flags;
1109 	args->sem = &channel->sem;
1110 	fs->send_request(__fs_open_file, req);
1111 	sem_wait(&channel->sem);
1112 	rc = args->rc;
1113 	if (rc == 0) {
1114 		*file = args->file;
1115 	} else {
1116 		*file = NULL;
1117 	}
1118 	free_fs_request(req);
1119 
1120 	return rc;
1121 }
1122 
1123 static void
1124 fs_rename_blob_close_cb(void *ctx, int bserrno)
1125 {
1126 	struct spdk_fs_request *req = ctx;
1127 	struct spdk_fs_cb_args *args = &req->args;
1128 
1129 	args->fn.fs_op(args->arg, bserrno);
1130 	free_fs_request(req);
1131 }
1132 
1133 static void
1134 fs_rename_blob_open_cb(void *ctx, struct spdk_blob *blob, int bserrno)
1135 {
1136 	struct spdk_fs_request *req = ctx;
1137 	struct spdk_fs_cb_args *args = &req->args;
1138 	const char *new_name = args->op.rename.new_name;
1139 
1140 	spdk_blob_set_xattr(blob, "name", new_name, strlen(new_name) + 1);
1141 	spdk_blob_close(blob, fs_rename_blob_close_cb, req);
1142 }
1143 
1144 static void
1145 __spdk_fs_md_rename_file(struct spdk_fs_request *req)
1146 {
1147 	struct spdk_fs_cb_args *args = &req->args;
1148 	struct spdk_file *f;
1149 
1150 	f = fs_find_file(args->fs, args->op.rename.old_name);
1151 	if (f == NULL) {
1152 		args->fn.fs_op(args->arg, -ENOENT);
1153 		free_fs_request(req);
1154 		return;
1155 	}
1156 
1157 	free(f->name);
1158 	f->name = strdup(args->op.rename.new_name);
1159 	args->file = f;
1160 	spdk_bs_open_blob(args->fs->bs, f->blobid, fs_rename_blob_open_cb, req);
1161 }
1162 
1163 static void
1164 fs_rename_delete_done(void *arg, int fserrno)
1165 {
1166 	__spdk_fs_md_rename_file(arg);
1167 }
1168 
1169 void
1170 spdk_fs_rename_file_async(struct spdk_filesystem *fs,
1171 			  const char *old_name, const char *new_name,
1172 			  spdk_file_op_complete cb_fn, void *cb_arg)
1173 {
1174 	struct spdk_file *f;
1175 	struct spdk_fs_request *req;
1176 	struct spdk_fs_cb_args *args;
1177 
1178 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "old=%s new=%s\n", old_name, new_name);
1179 	if (strnlen(new_name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1180 		cb_fn(cb_arg, -ENAMETOOLONG);
1181 		return;
1182 	}
1183 
1184 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1185 	if (req == NULL) {
1186 		cb_fn(cb_arg, -ENOMEM);
1187 		return;
1188 	}
1189 
1190 	args = &req->args;
1191 	args->fn.fs_op = cb_fn;
1192 	args->fs = fs;
1193 	args->arg = cb_arg;
1194 	args->op.rename.old_name = old_name;
1195 	args->op.rename.new_name = new_name;
1196 
1197 	f = fs_find_file(fs, new_name);
1198 	if (f == NULL) {
1199 		__spdk_fs_md_rename_file(req);
1200 		return;
1201 	}
1202 
1203 	/*
1204 	 * The rename overwrites an existing file.  So delete the existing file, then
1205 	 *  do the actual rename.
1206 	 */
1207 	spdk_fs_delete_file_async(fs, new_name, fs_rename_delete_done, req);
1208 }
1209 
1210 static void
1211 __fs_rename_file_done(void *arg, int fserrno)
1212 {
1213 	struct spdk_fs_request *req = arg;
1214 	struct spdk_fs_cb_args *args = &req->args;
1215 
1216 	args->rc = fserrno;
1217 	sem_post(args->sem);
1218 }
1219 
1220 static void
1221 __fs_rename_file(void *arg)
1222 {
1223 	struct spdk_fs_request *req = arg;
1224 	struct spdk_fs_cb_args *args = &req->args;
1225 
1226 	spdk_fs_rename_file_async(args->fs, args->op.rename.old_name, args->op.rename.new_name,
1227 				  __fs_rename_file_done, req);
1228 }
1229 
1230 int
1231 spdk_fs_rename_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1232 		    const char *old_name, const char *new_name)
1233 {
1234 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1235 	struct spdk_fs_request *req;
1236 	struct spdk_fs_cb_args *args;
1237 	int rc;
1238 
1239 	req = alloc_fs_request(channel);
1240 	assert(req != NULL);
1241 
1242 	args = &req->args;
1243 
1244 	args->fs = fs;
1245 	args->op.rename.old_name = old_name;
1246 	args->op.rename.new_name = new_name;
1247 	args->sem = &channel->sem;
1248 	fs->send_request(__fs_rename_file, req);
1249 	sem_wait(&channel->sem);
1250 	rc = args->rc;
1251 	free_fs_request(req);
1252 	return rc;
1253 }
1254 
1255 static void
1256 blob_delete_cb(void *ctx, int bserrno)
1257 {
1258 	struct spdk_fs_request *req = ctx;
1259 	struct spdk_fs_cb_args *args = &req->args;
1260 
1261 	args->fn.file_op(args->arg, bserrno);
1262 	free_fs_request(req);
1263 }
1264 
1265 void
1266 spdk_fs_delete_file_async(struct spdk_filesystem *fs, const char *name,
1267 			  spdk_file_op_complete cb_fn, void *cb_arg)
1268 {
1269 	struct spdk_file *f;
1270 	spdk_blob_id blobid;
1271 	struct spdk_fs_request *req;
1272 	struct spdk_fs_cb_args *args;
1273 
1274 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s\n", name);
1275 
1276 	if (strnlen(name, SPDK_FILE_NAME_MAX + 1) == SPDK_FILE_NAME_MAX + 1) {
1277 		cb_fn(cb_arg, -ENAMETOOLONG);
1278 		return;
1279 	}
1280 
1281 	f = fs_find_file(fs, name);
1282 	if (f == NULL) {
1283 		cb_fn(cb_arg, -ENOENT);
1284 		return;
1285 	}
1286 
1287 	req = alloc_fs_request(fs->md_target.md_fs_channel);
1288 	if (req == NULL) {
1289 		cb_fn(cb_arg, -ENOMEM);
1290 		return;
1291 	}
1292 
1293 	args = &req->args;
1294 	args->fn.file_op = cb_fn;
1295 	args->arg = cb_arg;
1296 
1297 	if (f->ref_count > 0) {
1298 		/* If the ref > 0, we mark the file as deleted and delete it when we close it. */
1299 		f->is_deleted = true;
1300 		spdk_blob_set_xattr(f->blob, "is_deleted", &f->is_deleted, sizeof(bool));
1301 		spdk_blob_sync_md(f->blob, blob_delete_cb, args);
1302 		return;
1303 	}
1304 
1305 	TAILQ_REMOVE(&fs->files, f, tailq);
1306 
1307 	cache_free_buffers(f);
1308 
1309 	blobid = f->blobid;
1310 
1311 	free(f->name);
1312 	free(f->tree);
1313 	free(f);
1314 
1315 	spdk_bs_delete_blob(fs->bs, blobid, blob_delete_cb, req);
1316 }
1317 
1318 static void
1319 __fs_delete_file_done(void *arg, int fserrno)
1320 {
1321 	struct spdk_fs_request *req = arg;
1322 	struct spdk_fs_cb_args *args = &req->args;
1323 
1324 	args->rc = fserrno;
1325 	sem_post(args->sem);
1326 }
1327 
1328 static void
1329 __fs_delete_file(void *arg)
1330 {
1331 	struct spdk_fs_request *req = arg;
1332 	struct spdk_fs_cb_args *args = &req->args;
1333 
1334 	spdk_fs_delete_file_async(args->fs, args->op.delete.name, __fs_delete_file_done, req);
1335 }
1336 
1337 int
1338 spdk_fs_delete_file(struct spdk_filesystem *fs, struct spdk_io_channel *_channel,
1339 		    const char *name)
1340 {
1341 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1342 	struct spdk_fs_request *req;
1343 	struct spdk_fs_cb_args *args;
1344 	int rc;
1345 
1346 	req = alloc_fs_request(channel);
1347 	assert(req != NULL);
1348 
1349 	args = &req->args;
1350 	args->fs = fs;
1351 	args->op.delete.name = name;
1352 	args->sem = &channel->sem;
1353 	fs->send_request(__fs_delete_file, req);
1354 	sem_wait(&channel->sem);
1355 	rc = args->rc;
1356 	free_fs_request(req);
1357 
1358 	return rc;
1359 }
1360 
1361 spdk_fs_iter
1362 spdk_fs_iter_first(struct spdk_filesystem *fs)
1363 {
1364 	struct spdk_file *f;
1365 
1366 	f = TAILQ_FIRST(&fs->files);
1367 	return f;
1368 }
1369 
1370 spdk_fs_iter
1371 spdk_fs_iter_next(spdk_fs_iter iter)
1372 {
1373 	struct spdk_file *f = iter;
1374 
1375 	if (f == NULL) {
1376 		return NULL;
1377 	}
1378 
1379 	f = TAILQ_NEXT(f, tailq);
1380 	return f;
1381 }
1382 
1383 const char *
1384 spdk_file_get_name(struct spdk_file *file)
1385 {
1386 	return file->name;
1387 }
1388 
1389 uint64_t
1390 spdk_file_get_length(struct spdk_file *file)
1391 {
1392 	assert(file != NULL);
1393 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s length=0x%jx\n", file->name, file->length);
1394 	return file->length;
1395 }
1396 
1397 static void
1398 fs_truncate_complete_cb(void *ctx, int bserrno)
1399 {
1400 	struct spdk_fs_request *req = ctx;
1401 	struct spdk_fs_cb_args *args = &req->args;
1402 
1403 	args->fn.file_op(args->arg, bserrno);
1404 	free_fs_request(req);
1405 }
1406 
1407 static uint64_t
1408 __bytes_to_clusters(uint64_t length, uint64_t cluster_sz)
1409 {
1410 	return (length + cluster_sz - 1) / cluster_sz;
1411 }
1412 
1413 void
1414 spdk_file_truncate_async(struct spdk_file *file, uint64_t length,
1415 			 spdk_file_op_complete cb_fn, void *cb_arg)
1416 {
1417 	struct spdk_filesystem *fs;
1418 	size_t num_clusters;
1419 	struct spdk_fs_request *req;
1420 	struct spdk_fs_cb_args *args;
1421 
1422 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s old=0x%jx new=0x%jx\n", file->name, file->length, length);
1423 	if (length == file->length) {
1424 		cb_fn(cb_arg, 0);
1425 		return;
1426 	}
1427 
1428 	req = alloc_fs_request(file->fs->md_target.md_fs_channel);
1429 	if (req == NULL) {
1430 		cb_fn(cb_arg, -ENOMEM);
1431 		return;
1432 	}
1433 
1434 	args = &req->args;
1435 	args->fn.file_op = cb_fn;
1436 	args->arg = cb_arg;
1437 	args->file = file;
1438 	fs = file->fs;
1439 
1440 	num_clusters = __bytes_to_clusters(length, fs->bs_opts.cluster_sz);
1441 
1442 	spdk_blob_resize(file->blob, num_clusters);
1443 	spdk_blob_set_xattr(file->blob, "length", &length, sizeof(length));
1444 
1445 	file->length = length;
1446 	if (file->append_pos > file->length) {
1447 		file->append_pos = file->length;
1448 	}
1449 
1450 	spdk_blob_sync_md(file->blob, fs_truncate_complete_cb, args);
1451 }
1452 
1453 static void
1454 __truncate(void *arg)
1455 {
1456 	struct spdk_fs_request *req = arg;
1457 	struct spdk_fs_cb_args *args = &req->args;
1458 
1459 	spdk_file_truncate_async(args->file, args->op.truncate.length,
1460 				 args->fn.file_op, args->arg);
1461 }
1462 
1463 void
1464 spdk_file_truncate(struct spdk_file *file, struct spdk_io_channel *_channel,
1465 		   uint64_t length)
1466 {
1467 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1468 	struct spdk_fs_request *req;
1469 	struct spdk_fs_cb_args *args;
1470 
1471 	req = alloc_fs_request(channel);
1472 	assert(req != NULL);
1473 
1474 	args = &req->args;
1475 
1476 	args->file = file;
1477 	args->op.truncate.length = length;
1478 	args->fn.file_op = __sem_post;
1479 	args->arg = &channel->sem;
1480 
1481 	channel->send_request(__truncate, req);
1482 	sem_wait(&channel->sem);
1483 	free_fs_request(req);
1484 }
1485 
1486 static void
1487 __rw_done(void *ctx, int bserrno)
1488 {
1489 	struct spdk_fs_request *req = ctx;
1490 	struct spdk_fs_cb_args *args = &req->args;
1491 
1492 	spdk_dma_free(args->op.rw.pin_buf);
1493 	args->fn.file_op(args->arg, bserrno);
1494 	free_fs_request(req);
1495 }
1496 
1497 static void
1498 __read_done(void *ctx, int bserrno)
1499 {
1500 	struct spdk_fs_request *req = ctx;
1501 	struct spdk_fs_cb_args *args = &req->args;
1502 
1503 	if (args->op.rw.is_read) {
1504 		memcpy(args->op.rw.user_buf,
1505 		       args->op.rw.pin_buf + (args->op.rw.offset & 0xFFF),
1506 		       args->op.rw.length);
1507 		__rw_done(req, 0);
1508 	} else {
1509 		memcpy(args->op.rw.pin_buf + (args->op.rw.offset & 0xFFF),
1510 		       args->op.rw.user_buf,
1511 		       args->op.rw.length);
1512 		spdk_blob_io_write(args->file->blob, args->op.rw.channel,
1513 				   args->op.rw.pin_buf,
1514 				   args->op.rw.start_page, args->op.rw.num_pages,
1515 				   __rw_done, req);
1516 	}
1517 }
1518 
1519 static void
1520 __do_blob_read(void *ctx, int fserrno)
1521 {
1522 	struct spdk_fs_request *req = ctx;
1523 	struct spdk_fs_cb_args *args = &req->args;
1524 
1525 	spdk_blob_io_read(args->file->blob, args->op.rw.channel,
1526 			  args->op.rw.pin_buf,
1527 			  args->op.rw.start_page, args->op.rw.num_pages,
1528 			  __read_done, req);
1529 }
1530 
1531 static void
1532 __get_page_parameters(struct spdk_file *file, uint64_t offset, uint64_t length,
1533 		      uint64_t *start_page, uint32_t *page_size, uint64_t *num_pages)
1534 {
1535 	uint64_t end_page;
1536 
1537 	*page_size = spdk_bs_get_page_size(file->fs->bs);
1538 	*start_page = offset / *page_size;
1539 	end_page = (offset + length - 1) / *page_size;
1540 	*num_pages = (end_page - *start_page + 1);
1541 }
1542 
1543 static void
1544 __readwrite(struct spdk_file *file, struct spdk_io_channel *_channel,
1545 	    void *payload, uint64_t offset, uint64_t length,
1546 	    spdk_file_op_complete cb_fn, void *cb_arg, int is_read)
1547 {
1548 	struct spdk_fs_request *req;
1549 	struct spdk_fs_cb_args *args;
1550 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
1551 	uint64_t start_page, num_pages, pin_buf_length;
1552 	uint32_t page_size;
1553 
1554 	if (is_read && offset + length > file->length) {
1555 		cb_fn(cb_arg, -EINVAL);
1556 		return;
1557 	}
1558 
1559 	req = alloc_fs_request(channel);
1560 	if (req == NULL) {
1561 		cb_fn(cb_arg, -ENOMEM);
1562 		return;
1563 	}
1564 
1565 	args = &req->args;
1566 	args->fn.file_op = cb_fn;
1567 	args->arg = cb_arg;
1568 	args->file = file;
1569 	args->op.rw.channel = channel->bs_channel;
1570 	args->op.rw.user_buf = payload;
1571 	args->op.rw.is_read = is_read;
1572 	args->op.rw.offset = offset;
1573 	args->op.rw.length = length;
1574 
1575 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
1576 	pin_buf_length = num_pages * page_size;
1577 	args->op.rw.pin_buf = spdk_dma_malloc(pin_buf_length, 4096, NULL);
1578 
1579 	args->op.rw.start_page = start_page;
1580 	args->op.rw.num_pages = num_pages;
1581 
1582 	if (!is_read && file->length < offset + length) {
1583 		spdk_file_truncate_async(file, offset + length, __do_blob_read, req);
1584 	} else {
1585 		__do_blob_read(req, 0);
1586 	}
1587 }
1588 
1589 void
1590 spdk_file_write_async(struct spdk_file *file, struct spdk_io_channel *channel,
1591 		      void *payload, uint64_t offset, uint64_t length,
1592 		      spdk_file_op_complete cb_fn, void *cb_arg)
1593 {
1594 	__readwrite(file, channel, payload, offset, length, cb_fn, cb_arg, 0);
1595 }
1596 
1597 void
1598 spdk_file_read_async(struct spdk_file *file, struct spdk_io_channel *channel,
1599 		     void *payload, uint64_t offset, uint64_t length,
1600 		     spdk_file_op_complete cb_fn, void *cb_arg)
1601 {
1602 	SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "file=%s offset=%jx length=%jx\n",
1603 		      file->name, offset, length);
1604 	__readwrite(file, channel, payload, offset, length, cb_fn, cb_arg, 1);
1605 }
1606 
1607 struct spdk_io_channel *
1608 spdk_fs_alloc_io_channel(struct spdk_filesystem *fs)
1609 {
1610 	struct spdk_io_channel *io_channel;
1611 	struct spdk_fs_channel *fs_channel;
1612 
1613 	io_channel = spdk_get_io_channel(&fs->io_target);
1614 	fs_channel = spdk_io_channel_get_ctx(io_channel);
1615 	fs_channel->bs_channel = spdk_bs_alloc_io_channel(fs->bs);
1616 	fs_channel->send_request = __send_request_direct;
1617 
1618 	return io_channel;
1619 }
1620 
1621 struct spdk_io_channel *
1622 spdk_fs_alloc_io_channel_sync(struct spdk_filesystem *fs)
1623 {
1624 	struct spdk_io_channel *io_channel;
1625 	struct spdk_fs_channel *fs_channel;
1626 
1627 	io_channel = spdk_get_io_channel(&fs->io_target);
1628 	fs_channel = spdk_io_channel_get_ctx(io_channel);
1629 	fs_channel->send_request = fs->send_request;
1630 	fs_channel->sync = 1;
1631 	pthread_spin_init(&fs_channel->lock, 0);
1632 
1633 	return io_channel;
1634 }
1635 
1636 void
1637 spdk_fs_free_io_channel(struct spdk_io_channel *channel)
1638 {
1639 	spdk_put_io_channel(channel);
1640 }
1641 
1642 void
1643 spdk_fs_set_cache_size(uint64_t size_in_mb)
1644 {
1645 	g_fs_cache_size = size_in_mb * 1024 * 1024;
1646 }
1647 
1648 uint64_t
1649 spdk_fs_get_cache_size(void)
1650 {
1651 	return g_fs_cache_size / (1024 * 1024);
1652 }
1653 
1654 static void __file_flush(void *_args);
1655 
1656 static void *
1657 alloc_cache_memory_buffer(struct spdk_file *context)
1658 {
1659 	struct spdk_file *file;
1660 	void *buf;
1661 
1662 	buf = spdk_mempool_get(g_cache_pool);
1663 	if (buf != NULL) {
1664 		return buf;
1665 	}
1666 
1667 	pthread_spin_lock(&g_caches_lock);
1668 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1669 		if (!file->open_for_writing &&
1670 		    file->priority == SPDK_FILE_PRIORITY_LOW &&
1671 		    file != context) {
1672 			break;
1673 		}
1674 	}
1675 	pthread_spin_unlock(&g_caches_lock);
1676 	if (file != NULL) {
1677 		cache_free_buffers(file);
1678 		buf = spdk_mempool_get(g_cache_pool);
1679 		if (buf != NULL) {
1680 			return buf;
1681 		}
1682 	}
1683 
1684 	pthread_spin_lock(&g_caches_lock);
1685 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1686 		if (!file->open_for_writing && file != context) {
1687 			break;
1688 		}
1689 	}
1690 	pthread_spin_unlock(&g_caches_lock);
1691 	if (file != NULL) {
1692 		cache_free_buffers(file);
1693 		buf = spdk_mempool_get(g_cache_pool);
1694 		if (buf != NULL) {
1695 			return buf;
1696 		}
1697 	}
1698 
1699 	pthread_spin_lock(&g_caches_lock);
1700 	TAILQ_FOREACH(file, &g_caches, cache_tailq) {
1701 		if (file != context) {
1702 			break;
1703 		}
1704 	}
1705 	pthread_spin_unlock(&g_caches_lock);
1706 	if (file != NULL) {
1707 		cache_free_buffers(file);
1708 		buf = spdk_mempool_get(g_cache_pool);
1709 		if (buf != NULL) {
1710 			return buf;
1711 		}
1712 	}
1713 
1714 	return NULL;
1715 }
1716 
1717 static struct cache_buffer *
1718 cache_insert_buffer(struct spdk_file *file, uint64_t offset)
1719 {
1720 	struct cache_buffer *buf;
1721 	int count = 0;
1722 
1723 	buf = calloc(1, sizeof(*buf));
1724 	if (buf == NULL) {
1725 		SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "calloc failed\n");
1726 		return NULL;
1727 	}
1728 
1729 	buf->buf = alloc_cache_memory_buffer(file);
1730 	while (buf->buf == NULL) {
1731 		/*
1732 		 * TODO: alloc_cache_memory_buffer() should eventually free
1733 		 *  some buffers.  Need a more sophisticated check here, instead
1734 		 *  of just bailing if 100 tries does not result in getting a
1735 		 *  free buffer.  This will involve using the sync channel's
1736 		 *  semaphore to block until a buffer becomes available.
1737 		 */
1738 		if (count++ == 100) {
1739 			SPDK_ERRLOG("could not allocate cache buffer\n");
1740 			assert(false);
1741 			free(buf);
1742 			return NULL;
1743 		}
1744 		buf->buf = alloc_cache_memory_buffer(file);
1745 	}
1746 
1747 	buf->buf_size = CACHE_BUFFER_SIZE;
1748 	buf->offset = offset;
1749 
1750 	pthread_spin_lock(&g_caches_lock);
1751 	if (file->tree->present_mask == 0) {
1752 		TAILQ_INSERT_TAIL(&g_caches, file, cache_tailq);
1753 	}
1754 	file->tree = spdk_tree_insert_buffer(file->tree, buf);
1755 	pthread_spin_unlock(&g_caches_lock);
1756 
1757 	return buf;
1758 }
1759 
1760 static struct cache_buffer *
1761 cache_append_buffer(struct spdk_file *file)
1762 {
1763 	struct cache_buffer *last;
1764 
1765 	assert(file->last == NULL || file->last->bytes_filled == file->last->buf_size);
1766 	assert((file->append_pos % CACHE_BUFFER_SIZE) == 0);
1767 
1768 	last = cache_insert_buffer(file, file->append_pos);
1769 	if (last == NULL) {
1770 		SPDK_DEBUGLOG(SPDK_LOG_BLOBFS, "cache_insert_buffer failed\n");
1771 		return NULL;
1772 	}
1773 
1774 	file->last = last;
1775 
1776 	return last;
1777 }
1778 
1779 static void
1780 __wake_caller(struct spdk_fs_cb_args *args)
1781 {
1782 	sem_post(args->sem);
1783 }
1784 
1785 static void __check_sync_reqs(struct spdk_file *file);
1786 
1787 static void
1788 __file_cache_finish_sync(struct spdk_file *file)
1789 {
1790 	struct spdk_fs_request *sync_req;
1791 	struct spdk_fs_cb_args *sync_args;
1792 
1793 	pthread_spin_lock(&file->lock);
1794 	sync_req = TAILQ_FIRST(&file->sync_requests);
1795 	sync_args = &sync_req->args;
1796 	assert(sync_args->op.sync.offset <= file->length_flushed);
1797 	BLOBFS_TRACE(file, "sync done offset=%jx\n", sync_args->op.sync.offset);
1798 	TAILQ_REMOVE(&file->sync_requests, sync_req, args.op.sync.tailq);
1799 	pthread_spin_unlock(&file->lock);
1800 
1801 	sync_args->fn.file_op(sync_args->arg, 0);
1802 	__check_sync_reqs(file);
1803 
1804 	pthread_spin_lock(&file->lock);
1805 	free_fs_request(sync_req);
1806 	pthread_spin_unlock(&file->lock);
1807 }
1808 
1809 static void
1810 __file_cache_finish_sync_bs_cb(void *ctx, int bserrno)
1811 {
1812 	struct spdk_file *file = ctx;
1813 
1814 	__file_cache_finish_sync(file);
1815 }
1816 
1817 static void
1818 __free_args(struct spdk_fs_cb_args *args)
1819 {
1820 	struct spdk_fs_request *req;
1821 
1822 	if (!args->from_request) {
1823 		free(args);
1824 	} else {
1825 		/* Depends on args being at the start of the spdk_fs_request structure. */
1826 		req = (struct spdk_fs_request *)args;
1827 		free_fs_request(req);
1828 	}
1829 }
1830 
1831 static void
1832 __check_sync_reqs(struct spdk_file *file)
1833 {
1834 	struct spdk_fs_request *sync_req;
1835 
1836 	pthread_spin_lock(&file->lock);
1837 
1838 	TAILQ_FOREACH(sync_req, &file->sync_requests, args.op.sync.tailq) {
1839 		if (sync_req->args.op.sync.offset <= file->length_flushed) {
1840 			break;
1841 		}
1842 	}
1843 
1844 	if (sync_req != NULL && !sync_req->args.op.sync.xattr_in_progress) {
1845 		BLOBFS_TRACE(file, "set xattr length 0x%jx\n", file->length_flushed);
1846 		sync_req->args.op.sync.xattr_in_progress = true;
1847 		spdk_blob_set_xattr(file->blob, "length", &file->length_flushed,
1848 				    sizeof(file->length_flushed));
1849 
1850 		pthread_spin_unlock(&file->lock);
1851 		spdk_blob_sync_md(file->blob, __file_cache_finish_sync_bs_cb, file);
1852 	} else {
1853 		pthread_spin_unlock(&file->lock);
1854 	}
1855 }
1856 
1857 static void
1858 __file_flush_done(void *arg, int bserrno)
1859 {
1860 	struct spdk_fs_cb_args *args = arg;
1861 	struct spdk_file *file = args->file;
1862 	struct cache_buffer *next = args->op.flush.cache_buffer;
1863 
1864 	BLOBFS_TRACE(file, "length=%jx\n", args->op.flush.length);
1865 
1866 	pthread_spin_lock(&file->lock);
1867 	next->in_progress = false;
1868 	next->bytes_flushed += args->op.flush.length;
1869 	file->length_flushed += args->op.flush.length;
1870 	if (file->length_flushed > file->length) {
1871 		file->length = file->length_flushed;
1872 	}
1873 	if (next->bytes_flushed == next->buf_size) {
1874 		BLOBFS_TRACE(file, "write buffer fully flushed 0x%jx\n", file->length_flushed);
1875 		next = spdk_tree_find_buffer(file->tree, file->length_flushed);
1876 	}
1877 
1878 	/*
1879 	 * Assert that there is no cached data that extends past the end of the underlying
1880 	 *  blob.
1881 	 */
1882 	assert(next == NULL || next->offset < __file_get_blob_size(file) ||
1883 	       next->bytes_filled == 0);
1884 
1885 	pthread_spin_unlock(&file->lock);
1886 
1887 	__check_sync_reqs(file);
1888 
1889 	__file_flush(args);
1890 }
1891 
1892 static void
1893 __file_flush(void *_args)
1894 {
1895 	struct spdk_fs_cb_args *args = _args;
1896 	struct spdk_file *file = args->file;
1897 	struct cache_buffer *next;
1898 	uint64_t offset, length, start_page, num_pages;
1899 	uint32_t page_size;
1900 
1901 	pthread_spin_lock(&file->lock);
1902 	next = spdk_tree_find_buffer(file->tree, file->length_flushed);
1903 	if (next == NULL || next->in_progress) {
1904 		/*
1905 		 * There is either no data to flush, or a flush I/O is already in
1906 		 *  progress.  So return immediately - if a flush I/O is in
1907 		 *  progress we will flush more data after that is completed.
1908 		 */
1909 		__free_args(args);
1910 		pthread_spin_unlock(&file->lock);
1911 		return;
1912 	}
1913 
1914 	offset = next->offset + next->bytes_flushed;
1915 	length = next->bytes_filled - next->bytes_flushed;
1916 	if (length == 0) {
1917 		__free_args(args);
1918 		pthread_spin_unlock(&file->lock);
1919 		return;
1920 	}
1921 	args->op.flush.length = length;
1922 	args->op.flush.cache_buffer = next;
1923 
1924 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
1925 
1926 	next->in_progress = true;
1927 	BLOBFS_TRACE(file, "offset=%jx length=%jx page start=%jx num=%jx\n",
1928 		     offset, length, start_page, num_pages);
1929 	pthread_spin_unlock(&file->lock);
1930 	spdk_blob_io_write(file->blob, file->fs->sync_target.sync_fs_channel->bs_channel,
1931 			   next->buf + (start_page * page_size) - next->offset,
1932 			   start_page, num_pages, __file_flush_done, args);
1933 }
1934 
1935 static void
1936 __file_extend_done(void *arg, int bserrno)
1937 {
1938 	struct spdk_fs_cb_args *args = arg;
1939 
1940 	__wake_caller(args);
1941 }
1942 
1943 static void
1944 __file_extend_blob(void *_args)
1945 {
1946 	struct spdk_fs_cb_args *args = _args;
1947 	struct spdk_file *file = args->file;
1948 
1949 	spdk_blob_resize(file->blob, args->op.resize.num_clusters);
1950 
1951 	spdk_blob_sync_md(file->blob, __file_extend_done, args);
1952 }
1953 
1954 static void
1955 __rw_from_file_done(void *arg, int bserrno)
1956 {
1957 	struct spdk_fs_cb_args *args = arg;
1958 
1959 	__wake_caller(args);
1960 	__free_args(args);
1961 }
1962 
1963 static void
1964 __rw_from_file(void *_args)
1965 {
1966 	struct spdk_fs_cb_args *args = _args;
1967 	struct spdk_file *file = args->file;
1968 
1969 	if (args->op.rw.is_read) {
1970 		spdk_file_read_async(file, file->fs->sync_target.sync_io_channel, args->op.rw.user_buf,
1971 				     args->op.rw.offset, args->op.rw.length,
1972 				     __rw_from_file_done, args);
1973 	} else {
1974 		spdk_file_write_async(file, file->fs->sync_target.sync_io_channel, args->op.rw.user_buf,
1975 				      args->op.rw.offset, args->op.rw.length,
1976 				      __rw_from_file_done, args);
1977 	}
1978 }
1979 
1980 static int
1981 __send_rw_from_file(struct spdk_file *file, sem_t *sem, void *payload,
1982 		    uint64_t offset, uint64_t length, bool is_read)
1983 {
1984 	struct spdk_fs_cb_args *args;
1985 
1986 	args = calloc(1, sizeof(*args));
1987 	if (args == NULL) {
1988 		sem_post(sem);
1989 		return -ENOMEM;
1990 	}
1991 
1992 	args->file = file;
1993 	args->sem = sem;
1994 	args->op.rw.user_buf = payload;
1995 	args->op.rw.offset = offset;
1996 	args->op.rw.length = length;
1997 	args->op.rw.is_read = is_read;
1998 	file->fs->send_request(__rw_from_file, args);
1999 	return 0;
2000 }
2001 
2002 int
2003 spdk_file_write(struct spdk_file *file, struct spdk_io_channel *_channel,
2004 		void *payload, uint64_t offset, uint64_t length)
2005 {
2006 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2007 	struct spdk_fs_cb_args *args;
2008 	uint64_t rem_length, copy, blob_size, cluster_sz;
2009 	uint32_t cache_buffers_filled = 0;
2010 	uint8_t *cur_payload;
2011 	struct cache_buffer *last;
2012 
2013 	BLOBFS_TRACE_RW(file, "offset=%jx length=%jx\n", offset, length);
2014 
2015 	if (length == 0) {
2016 		return 0;
2017 	}
2018 
2019 	if (offset != file->append_pos) {
2020 		BLOBFS_TRACE(file, " error offset=%jx append_pos=%jx\n", offset, file->append_pos);
2021 		return -EINVAL;
2022 	}
2023 
2024 	pthread_spin_lock(&file->lock);
2025 	file->open_for_writing = true;
2026 
2027 	if (file->last == NULL) {
2028 		if (file->append_pos % CACHE_BUFFER_SIZE == 0) {
2029 			cache_append_buffer(file);
2030 		} else {
2031 			int rc;
2032 
2033 			file->append_pos += length;
2034 			pthread_spin_unlock(&file->lock);
2035 			rc = __send_rw_from_file(file, &channel->sem, payload,
2036 						 offset, length, false);
2037 			sem_wait(&channel->sem);
2038 			return rc;
2039 		}
2040 	}
2041 
2042 	blob_size = __file_get_blob_size(file);
2043 
2044 	if ((offset + length) > blob_size) {
2045 		struct spdk_fs_cb_args extend_args = {};
2046 
2047 		cluster_sz = file->fs->bs_opts.cluster_sz;
2048 		extend_args.sem = &channel->sem;
2049 		extend_args.op.resize.num_clusters = __bytes_to_clusters((offset + length), cluster_sz);
2050 		extend_args.file = file;
2051 		BLOBFS_TRACE(file, "start resize to %u clusters\n", extend_args.op.resize.num_clusters);
2052 		pthread_spin_unlock(&file->lock);
2053 		file->fs->send_request(__file_extend_blob, &extend_args);
2054 		sem_wait(&channel->sem);
2055 	}
2056 
2057 	last = file->last;
2058 	rem_length = length;
2059 	cur_payload = payload;
2060 	while (rem_length > 0) {
2061 		copy = last->buf_size - last->bytes_filled;
2062 		if (copy > rem_length) {
2063 			copy = rem_length;
2064 		}
2065 		BLOBFS_TRACE_RW(file, "  fill offset=%jx length=%jx\n", file->append_pos, copy);
2066 		memcpy(&last->buf[last->bytes_filled], cur_payload, copy);
2067 		file->append_pos += copy;
2068 		if (file->length < file->append_pos) {
2069 			file->length = file->append_pos;
2070 		}
2071 		cur_payload += copy;
2072 		last->bytes_filled += copy;
2073 		rem_length -= copy;
2074 		if (last->bytes_filled == last->buf_size) {
2075 			cache_buffers_filled++;
2076 			last = cache_append_buffer(file);
2077 			if (last == NULL) {
2078 				BLOBFS_TRACE(file, "nomem\n");
2079 				pthread_spin_unlock(&file->lock);
2080 				return -ENOMEM;
2081 			}
2082 		}
2083 	}
2084 
2085 	pthread_spin_unlock(&file->lock);
2086 
2087 	if (cache_buffers_filled == 0) {
2088 		return 0;
2089 	}
2090 
2091 	args = calloc(1, sizeof(*args));
2092 	if (args == NULL) {
2093 		return -ENOMEM;
2094 	}
2095 
2096 	args->file = file;
2097 	file->fs->send_request(__file_flush, args);
2098 	return 0;
2099 }
2100 
2101 static void
2102 __readahead_done(void *arg, int bserrno)
2103 {
2104 	struct spdk_fs_cb_args *args = arg;
2105 	struct cache_buffer *cache_buffer = args->op.readahead.cache_buffer;
2106 	struct spdk_file *file = args->file;
2107 
2108 	BLOBFS_TRACE(file, "offset=%jx\n", cache_buffer->offset);
2109 
2110 	pthread_spin_lock(&file->lock);
2111 	cache_buffer->bytes_filled = args->op.readahead.length;
2112 	cache_buffer->bytes_flushed = args->op.readahead.length;
2113 	cache_buffer->in_progress = false;
2114 	pthread_spin_unlock(&file->lock);
2115 
2116 	__free_args(args);
2117 }
2118 
2119 static void
2120 __readahead(void *_args)
2121 {
2122 	struct spdk_fs_cb_args *args = _args;
2123 	struct spdk_file *file = args->file;
2124 	uint64_t offset, length, start_page, num_pages;
2125 	uint32_t page_size;
2126 
2127 	offset = args->op.readahead.offset;
2128 	length = args->op.readahead.length;
2129 	assert(length > 0);
2130 
2131 	__get_page_parameters(file, offset, length, &start_page, &page_size, &num_pages);
2132 
2133 	BLOBFS_TRACE(file, "offset=%jx length=%jx page start=%jx num=%jx\n",
2134 		     offset, length, start_page, num_pages);
2135 	spdk_blob_io_read(file->blob, file->fs->sync_target.sync_fs_channel->bs_channel,
2136 			  args->op.readahead.cache_buffer->buf,
2137 			  start_page, num_pages, __readahead_done, args);
2138 }
2139 
2140 static uint64_t
2141 __next_cache_buffer_offset(uint64_t offset)
2142 {
2143 	return (offset + CACHE_BUFFER_SIZE) & ~(CACHE_TREE_LEVEL_MASK(0));
2144 }
2145 
2146 static void
2147 check_readahead(struct spdk_file *file, uint64_t offset)
2148 {
2149 	struct spdk_fs_cb_args *args;
2150 
2151 	offset = __next_cache_buffer_offset(offset);
2152 	if (spdk_tree_find_buffer(file->tree, offset) != NULL || file->length <= offset) {
2153 		return;
2154 	}
2155 
2156 	args = calloc(1, sizeof(*args));
2157 	if (args == NULL) {
2158 		return;
2159 	}
2160 
2161 	BLOBFS_TRACE(file, "offset=%jx\n", offset);
2162 
2163 	args->file = file;
2164 	args->op.readahead.offset = offset;
2165 	args->op.readahead.cache_buffer = cache_insert_buffer(file, offset);
2166 	args->op.readahead.cache_buffer->in_progress = true;
2167 	if (file->length < (offset + CACHE_BUFFER_SIZE)) {
2168 		args->op.readahead.length = file->length & (CACHE_BUFFER_SIZE - 1);
2169 	} else {
2170 		args->op.readahead.length = CACHE_BUFFER_SIZE;
2171 	}
2172 	file->fs->send_request(__readahead, args);
2173 }
2174 
2175 static int
2176 __file_read(struct spdk_file *file, void *payload, uint64_t offset, uint64_t length, sem_t *sem)
2177 {
2178 	struct cache_buffer *buf;
2179 	int rc;
2180 
2181 	buf = spdk_tree_find_filled_buffer(file->tree, offset);
2182 	if (buf == NULL) {
2183 		pthread_spin_unlock(&file->lock);
2184 		rc = __send_rw_from_file(file, sem, payload, offset, length, true);
2185 		pthread_spin_lock(&file->lock);
2186 		return rc;
2187 	}
2188 
2189 	if ((offset + length) > (buf->offset + buf->bytes_filled)) {
2190 		length = buf->offset + buf->bytes_filled - offset;
2191 	}
2192 	BLOBFS_TRACE(file, "read %p offset=%ju length=%ju\n", payload, offset, length);
2193 	memcpy(payload, &buf->buf[offset - buf->offset], length);
2194 	if ((offset + length) % CACHE_BUFFER_SIZE == 0) {
2195 		pthread_spin_lock(&g_caches_lock);
2196 		spdk_tree_remove_buffer(file->tree, buf);
2197 		if (file->tree->present_mask == 0) {
2198 			TAILQ_REMOVE(&g_caches, file, cache_tailq);
2199 		}
2200 		pthread_spin_unlock(&g_caches_lock);
2201 	}
2202 
2203 	sem_post(sem);
2204 	return 0;
2205 }
2206 
2207 int64_t
2208 spdk_file_read(struct spdk_file *file, struct spdk_io_channel *_channel,
2209 	       void *payload, uint64_t offset, uint64_t length)
2210 {
2211 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2212 	uint64_t final_offset, final_length;
2213 	uint32_t sub_reads = 0;
2214 	int rc = 0;
2215 
2216 	pthread_spin_lock(&file->lock);
2217 
2218 	BLOBFS_TRACE_RW(file, "offset=%ju length=%ju\n", offset, length);
2219 
2220 	file->open_for_writing = false;
2221 
2222 	if (length == 0 || offset >= file->append_pos) {
2223 		pthread_spin_unlock(&file->lock);
2224 		return 0;
2225 	}
2226 
2227 	if (offset + length > file->append_pos) {
2228 		length = file->append_pos - offset;
2229 	}
2230 
2231 	if (offset != file->next_seq_offset) {
2232 		file->seq_byte_count = 0;
2233 	}
2234 	file->seq_byte_count += length;
2235 	file->next_seq_offset = offset + length;
2236 	if (file->seq_byte_count >= CACHE_READAHEAD_THRESHOLD) {
2237 		check_readahead(file, offset);
2238 		check_readahead(file, offset + CACHE_BUFFER_SIZE);
2239 	}
2240 
2241 	final_length = 0;
2242 	final_offset = offset + length;
2243 	while (offset < final_offset) {
2244 		length = NEXT_CACHE_BUFFER_OFFSET(offset) - offset;
2245 		if (length > (final_offset - offset)) {
2246 			length = final_offset - offset;
2247 		}
2248 		rc = __file_read(file, payload, offset, length, &channel->sem);
2249 		if (rc == 0) {
2250 			final_length += length;
2251 		} else {
2252 			break;
2253 		}
2254 		payload += length;
2255 		offset += length;
2256 		sub_reads++;
2257 	}
2258 	pthread_spin_unlock(&file->lock);
2259 	while (sub_reads-- > 0) {
2260 		sem_wait(&channel->sem);
2261 	}
2262 	if (rc == 0) {
2263 		return final_length;
2264 	} else {
2265 		return rc;
2266 	}
2267 }
2268 
2269 static void
2270 _file_sync(struct spdk_file *file, struct spdk_fs_channel *channel,
2271 	   spdk_file_op_complete cb_fn, void *cb_arg)
2272 {
2273 	struct spdk_fs_request *sync_req;
2274 	struct spdk_fs_request *flush_req;
2275 	struct spdk_fs_cb_args *sync_args;
2276 	struct spdk_fs_cb_args *flush_args;
2277 
2278 	BLOBFS_TRACE(file, "offset=%jx\n", file->append_pos);
2279 
2280 	pthread_spin_lock(&file->lock);
2281 	if (file->append_pos <= file->length_flushed || file->last == NULL) {
2282 		BLOBFS_TRACE(file, "done - no data to flush\n");
2283 		pthread_spin_unlock(&file->lock);
2284 		cb_fn(cb_arg, 0);
2285 		return;
2286 	}
2287 
2288 	sync_req = alloc_fs_request(channel);
2289 	assert(sync_req != NULL);
2290 	sync_args = &sync_req->args;
2291 
2292 	flush_req = alloc_fs_request(channel);
2293 	assert(flush_req != NULL);
2294 	flush_args = &flush_req->args;
2295 
2296 	sync_args->file = file;
2297 	sync_args->fn.file_op = cb_fn;
2298 	sync_args->arg = cb_arg;
2299 	sync_args->op.sync.offset = file->append_pos;
2300 	sync_args->op.sync.xattr_in_progress = false;
2301 	TAILQ_INSERT_TAIL(&file->sync_requests, sync_req, args.op.sync.tailq);
2302 	pthread_spin_unlock(&file->lock);
2303 
2304 	flush_args->file = file;
2305 	channel->send_request(__file_flush, flush_args);
2306 }
2307 
2308 int
2309 spdk_file_sync(struct spdk_file *file, struct spdk_io_channel *_channel)
2310 {
2311 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2312 
2313 	_file_sync(file, channel, __sem_post, &channel->sem);
2314 	sem_wait(&channel->sem);
2315 
2316 	return 0;
2317 }
2318 
2319 void
2320 spdk_file_sync_async(struct spdk_file *file, struct spdk_io_channel *_channel,
2321 		     spdk_file_op_complete cb_fn, void *cb_arg)
2322 {
2323 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2324 
2325 	_file_sync(file, channel, cb_fn, cb_arg);
2326 }
2327 
2328 void
2329 spdk_file_set_priority(struct spdk_file *file, uint32_t priority)
2330 {
2331 	BLOBFS_TRACE(file, "priority=%u\n", priority);
2332 	file->priority = priority;
2333 
2334 }
2335 
2336 /*
2337  * Close routines
2338  */
2339 
2340 static void
2341 __file_close_async_done(void *ctx, int bserrno)
2342 {
2343 	struct spdk_fs_request *req = ctx;
2344 	struct spdk_fs_cb_args *args = &req->args;
2345 	struct spdk_file *file = args->file;
2346 
2347 	if (file->is_deleted) {
2348 		spdk_fs_delete_file_async(file->fs, file->name, blob_delete_cb, ctx);
2349 		return;
2350 	}
2351 
2352 	args->fn.file_op(args->arg, bserrno);
2353 	free_fs_request(req);
2354 }
2355 
2356 static void
2357 __file_close_async(struct spdk_file *file, struct spdk_fs_request *req)
2358 {
2359 	struct spdk_blob *blob;
2360 
2361 	pthread_spin_lock(&file->lock);
2362 	if (file->ref_count == 0) {
2363 		pthread_spin_unlock(&file->lock);
2364 		__file_close_async_done(req, -EBADF);
2365 		return;
2366 	}
2367 
2368 	file->ref_count--;
2369 	if (file->ref_count > 0) {
2370 		pthread_spin_unlock(&file->lock);
2371 		req->args.fn.file_op(req->args.arg, 0);
2372 		free_fs_request(req);
2373 		return;
2374 	}
2375 
2376 	pthread_spin_unlock(&file->lock);
2377 
2378 	blob = file->blob;
2379 	file->blob = NULL;
2380 	spdk_blob_close(blob, __file_close_async_done, req);
2381 }
2382 
2383 static void
2384 __file_close_async__sync_done(void *arg, int fserrno)
2385 {
2386 	struct spdk_fs_request *req = arg;
2387 	struct spdk_fs_cb_args *args = &req->args;
2388 
2389 	__file_close_async(args->file, req);
2390 }
2391 
2392 void
2393 spdk_file_close_async(struct spdk_file *file, spdk_file_op_complete cb_fn, void *cb_arg)
2394 {
2395 	struct spdk_fs_request *req;
2396 	struct spdk_fs_cb_args *args;
2397 
2398 	req = alloc_fs_request(file->fs->md_target.md_fs_channel);
2399 	if (req == NULL) {
2400 		cb_fn(cb_arg, -ENOMEM);
2401 		return;
2402 	}
2403 
2404 	args = &req->args;
2405 	args->file = file;
2406 	args->fn.file_op = cb_fn;
2407 	args->arg = cb_arg;
2408 
2409 	spdk_file_sync_async(file, file->fs->md_target.md_io_channel, __file_close_async__sync_done, req);
2410 }
2411 
2412 static void
2413 __file_close_done(void *arg, int fserrno)
2414 {
2415 	struct spdk_fs_cb_args *args = arg;
2416 
2417 	args->rc = fserrno;
2418 	sem_post(args->sem);
2419 }
2420 
2421 static void
2422 __file_close(void *arg)
2423 {
2424 	struct spdk_fs_request *req = arg;
2425 	struct spdk_fs_cb_args *args = &req->args;
2426 	struct spdk_file *file = args->file;
2427 
2428 	__file_close_async(file, req);
2429 }
2430 
2431 int
2432 spdk_file_close(struct spdk_file *file, struct spdk_io_channel *_channel)
2433 {
2434 	struct spdk_fs_channel *channel = spdk_io_channel_get_ctx(_channel);
2435 	struct spdk_fs_request *req;
2436 	struct spdk_fs_cb_args *args;
2437 
2438 	req = alloc_fs_request(channel);
2439 	assert(req != NULL);
2440 
2441 	args = &req->args;
2442 
2443 	spdk_file_sync(file, _channel);
2444 	BLOBFS_TRACE(file, "name=%s\n", file->name);
2445 	args->file = file;
2446 	args->sem = &channel->sem;
2447 	args->fn.file_op = __file_close_done;
2448 	args->arg = req;
2449 	channel->send_request(__file_close, req);
2450 	sem_wait(&channel->sem);
2451 
2452 	return args->rc;
2453 }
2454 
2455 static void
2456 cache_free_buffers(struct spdk_file *file)
2457 {
2458 	BLOBFS_TRACE(file, "free=%s\n", file->name);
2459 	pthread_spin_lock(&file->lock);
2460 	pthread_spin_lock(&g_caches_lock);
2461 	if (file->tree->present_mask == 0) {
2462 		pthread_spin_unlock(&g_caches_lock);
2463 		pthread_spin_unlock(&file->lock);
2464 		return;
2465 	}
2466 	spdk_tree_free_buffers(file->tree);
2467 
2468 	TAILQ_REMOVE(&g_caches, file, cache_tailq);
2469 	/* If not freed, put it in the end of the queue */
2470 	if (file->tree->present_mask != 0) {
2471 		TAILQ_INSERT_TAIL(&g_caches, file, cache_tailq);
2472 	}
2473 	file->last = NULL;
2474 	pthread_spin_unlock(&g_caches_lock);
2475 	pthread_spin_unlock(&file->lock);
2476 }
2477 
2478 SPDK_LOG_REGISTER_COMPONENT("blobfs", SPDK_LOG_BLOBFS)
2479 SPDK_LOG_REGISTER_COMPONENT("blobfs_rw", SPDK_LOG_BLOBFS_RW)
2480