summaryrefslogtreecommitdiff
path: root/fs/orangefs/super.c
blob: b4ab1c1e827840965c54b255182873b3d925953c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
/*
 * (C) 2001 Clemson University and The University of Chicago
 *
 * See COPYING in top-level directory.
 */

#include "protocol.h"
#include "orangefs-kernel.h"
#include "orangefs-bufmap.h"

#include <linux/parser.h>

/* a cache for orangefs-inode objects (i.e. orangefs inode private data) */
static struct kmem_cache *orangefs_inode_cache;

/* list for storing orangefs specific superblocks in use */
LIST_HEAD(orangefs_superblocks);

DEFINE_SPINLOCK(orangefs_superblocks_lock);

enum {
	Opt_intr,
	Opt_acl,
	Opt_local_lock,

	Opt_err
};

static const match_table_t tokens = {
	{ Opt_acl,		"acl" },
	{ Opt_intr,		"intr" },
	{ Opt_local_lock,	"local_lock" },
	{ Opt_err,	NULL }
};

uint64_t orangefs_features;

static int parse_mount_options(struct super_block *sb, char *options,
		int silent)
{
	struct orangefs_sb_info_s *orangefs_sb = ORANGEFS_SB(sb);
	substring_t args[MAX_OPT_ARGS];
	char *p;

	/*
	 * Force any potential flags that might be set from the mount
	 * to zero, ie, initialize to unset.
	 */
	sb->s_flags &= ~MS_POSIXACL;
	orangefs_sb->flags &= ~ORANGEFS_OPT_INTR;
	orangefs_sb->flags &= ~ORANGEFS_OPT_LOCAL_LOCK;

	while ((p = strsep(&options, ",")) != NULL) {
		int token;

		if (!*p)
			continue;

		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_acl:
			sb->s_flags |= MS_POSIXACL;
			break;
		case Opt_intr:
			orangefs_sb->flags |= ORANGEFS_OPT_INTR;
			break;
		case Opt_local_lock:
			orangefs_sb->flags |= ORANGEFS_OPT_LOCAL_LOCK;
			break;
		default:
			goto fail;
		}
	}

	return 0;
fail:
	if (!silent)
		gossip_err("Error: mount option [%s] is not supported.\n", p);
	return -EINVAL;
}

static void orangefs_inode_cache_ctor(void *req)
{
	struct orangefs_inode_s *orangefs_inode = req;

	inode_init_once(&orangefs_inode->vfs_inode);
	init_rwsem(&orangefs_inode->xattr_sem);

	orangefs_inode->vfs_inode.i_version = 1;
}

static struct inode *orangefs_alloc_inode(struct super_block *sb)
{
	struct orangefs_inode_s *orangefs_inode;

	orangefs_inode = kmem_cache_alloc(orangefs_inode_cache, GFP_KERNEL);
	if (orangefs_inode == NULL) {
		gossip_err("Failed to allocate orangefs_inode\n");
		return NULL;
	}

	/*
	 * We want to clear everything except for rw_semaphore and the
	 * vfs_inode.
	 */
	memset(&orangefs_inode->refn.khandle, 0, 16);
	orangefs_inode->refn.fs_id = ORANGEFS_FS_ID_NULL;
	orangefs_inode->last_failed_block_index_read = 0;
	memset(orangefs_inode->link_target, 0, sizeof(orangefs_inode->link_target));
	orangefs_inode->pinode_flags = 0;

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_alloc_inode: allocated %p\n",
		     &orangefs_inode->vfs_inode);
	return &orangefs_inode->vfs_inode;
}

static void orangefs_destroy_inode(struct inode *inode)
{
	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);

	gossip_debug(GOSSIP_SUPER_DEBUG,
			"%s: deallocated %p destroying inode %pU\n",
			__func__, orangefs_inode, get_khandle_from_ino(inode));

	kmem_cache_free(orangefs_inode_cache, orangefs_inode);
}

/*
 * NOTE: information filled in here is typically reflected in the
 * output of the system command 'df'
*/
static int orangefs_statfs(struct dentry *dentry, struct kstatfs *buf)
{
	int ret = -ENOMEM;
	struct orangefs_kernel_op_s *new_op = NULL;
	int flags = 0;
	struct super_block *sb = NULL;

	sb = dentry->d_sb;

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_statfs: called on sb %p (fs_id is %d)\n",
		     sb,
		     (int)(ORANGEFS_SB(sb)->fs_id));

	new_op = op_alloc(ORANGEFS_VFS_OP_STATFS);
	if (!new_op)
		return ret;
	new_op->upcall.req.statfs.fs_id = ORANGEFS_SB(sb)->fs_id;

	if (ORANGEFS_SB(sb)->flags & ORANGEFS_OPT_INTR)
		flags = ORANGEFS_OP_INTERRUPTIBLE;

	ret = service_operation(new_op, "orangefs_statfs", flags);

	if (new_op->downcall.status < 0)
		goto out_op_release;

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "%s: got %ld blocks available | "
		     "%ld blocks total | %ld block size | "
		     "%ld files total | %ld files avail\n",
		     __func__,
		     (long)new_op->downcall.resp.statfs.blocks_avail,
		     (long)new_op->downcall.resp.statfs.blocks_total,
		     (long)new_op->downcall.resp.statfs.block_size,
		     (long)new_op->downcall.resp.statfs.files_total,
		     (long)new_op->downcall.resp.statfs.files_avail);

	buf->f_type = sb->s_magic;
	memcpy(&buf->f_fsid, &ORANGEFS_SB(sb)->fs_id, sizeof(buf->f_fsid));
	buf->f_bsize = new_op->downcall.resp.statfs.block_size;
	buf->f_namelen = ORANGEFS_NAME_MAX;

	buf->f_blocks = (sector_t) new_op->downcall.resp.statfs.blocks_total;
	buf->f_bfree = (sector_t) new_op->downcall.resp.statfs.blocks_avail;
	buf->f_bavail = (sector_t) new_op->downcall.resp.statfs.blocks_avail;
	buf->f_files = (sector_t) new_op->downcall.resp.statfs.files_total;
	buf->f_ffree = (sector_t) new_op->downcall.resp.statfs.files_avail;
	buf->f_frsize = sb->s_blocksize;

out_op_release:
	op_release(new_op);
	gossip_debug(GOSSIP_SUPER_DEBUG, "orangefs_statfs: returning %d\n", ret);
	return ret;
}

/*
 * Remount as initiated by VFS layer.  We just need to reparse the mount
 * options, no need to signal pvfs2-client-core about it.
 */
static int orangefs_remount_fs(struct super_block *sb, int *flags, char *data)
{
	gossip_debug(GOSSIP_SUPER_DEBUG, "orangefs_remount_fs: called\n");
	return parse_mount_options(sb, data, 1);
}

/*
 * Remount as initiated by pvfs2-client-core on restart.  This is used to
 * repopulate mount information left from previous pvfs2-client-core.
 *
 * the idea here is that given a valid superblock, we're
 * re-initializing the user space client with the initial mount
 * information specified when the super block was first initialized.
 * this is very different than the first initialization/creation of a
 * superblock.  we use the special service_priority_operation to make
 * sure that the mount gets ahead of any other pending operation that
 * is waiting for servicing.  this means that the pvfs2-client won't
 * fail to start several times for all other pending operations before
 * the client regains all of the mount information from us.
 * NOTE: this function assumes that the request_mutex is already acquired!
 */
int orangefs_remount(struct orangefs_sb_info_s *orangefs_sb)
{
	struct orangefs_kernel_op_s *new_op;
	int ret = -EINVAL;

	gossip_debug(GOSSIP_SUPER_DEBUG, "orangefs_remount: called\n");

	new_op = op_alloc(ORANGEFS_VFS_OP_FS_MOUNT);
	if (!new_op)
		return -ENOMEM;
	strncpy(new_op->upcall.req.fs_mount.orangefs_config_server,
		orangefs_sb->devname,
		ORANGEFS_MAX_SERVER_ADDR_LEN);

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "Attempting ORANGEFS Remount via host %s\n",
		     new_op->upcall.req.fs_mount.orangefs_config_server);

	/*
	 * we assume that the calling function has already acquired the
	 * request_mutex to prevent other operations from bypassing
	 * this one
	 */
	ret = service_operation(new_op, "orangefs_remount",
		ORANGEFS_OP_PRIORITY | ORANGEFS_OP_NO_MUTEX);
	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_remount: mount got return value of %d\n",
		     ret);
	if (ret == 0) {
		/*
		 * store the id assigned to this sb -- it's just a
		 * short-lived mapping that the system interface uses
		 * to map this superblock to a particular mount entry
		 */
		orangefs_sb->id = new_op->downcall.resp.fs_mount.id;
		orangefs_sb->mount_pending = 0;
	}

	op_release(new_op);

	if (orangefs_userspace_version >= 20907) {
		new_op = op_alloc(ORANGEFS_VFS_OP_FEATURES);
		if (!new_op)
			return -ENOMEM;
		new_op->upcall.req.features.features = 0;
		ret = service_operation(new_op, "orangefs_features", 0);
		orangefs_features = new_op->downcall.resp.features.features;
		op_release(new_op);
	} else {
		orangefs_features = 0;
	}

	return ret;
}

int fsid_key_table_initialize(void)
{
	return 0;
}

void fsid_key_table_finalize(void)
{
}

/* Called whenever the VFS dirties the inode in response to atime updates */
static void orangefs_dirty_inode(struct inode *inode, int flags)
{
	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_dirty_inode: %pU\n",
		     get_khandle_from_ino(inode));
	SetAtimeFlag(orangefs_inode);
}

static const struct super_operations orangefs_s_ops = {
	.alloc_inode = orangefs_alloc_inode,
	.destroy_inode = orangefs_destroy_inode,
	.dirty_inode = orangefs_dirty_inode,
	.drop_inode = generic_delete_inode,
	.statfs = orangefs_statfs,
	.remount_fs = orangefs_remount_fs,
	.show_options = generic_show_options,
};

static struct dentry *orangefs_fh_to_dentry(struct super_block *sb,
				  struct fid *fid,
				  int fh_len,
				  int fh_type)
{
	struct orangefs_object_kref refn;

	if (fh_len < 5 || fh_type > 2)
		return NULL;

	ORANGEFS_khandle_from(&(refn.khandle), fid->raw, 16);
	refn.fs_id = (u32) fid->raw[4];
	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "fh_to_dentry: handle %pU, fs_id %d\n",
		     &refn.khandle,
		     refn.fs_id);

	return d_obtain_alias(orangefs_iget(sb, &refn));
}

static int orangefs_encode_fh(struct inode *inode,
		    __u32 *fh,
		    int *max_len,
		    struct inode *parent)
{
	int len = parent ? 10 : 5;
	int type = 1;
	struct orangefs_object_kref refn;

	if (*max_len < len) {
		gossip_lerr("fh buffer is too small for encoding\n");
		*max_len = len;
		type = 255;
		goto out;
	}

	refn = ORANGEFS_I(inode)->refn;
	ORANGEFS_khandle_to(&refn.khandle, fh, 16);
	fh[4] = refn.fs_id;

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "Encoding fh: handle %pU, fsid %u\n",
		     &refn.khandle,
		     refn.fs_id);


	if (parent) {
		refn = ORANGEFS_I(parent)->refn;
		ORANGEFS_khandle_to(&refn.khandle, (char *) fh + 20, 16);
		fh[9] = refn.fs_id;

		type = 2;
		gossip_debug(GOSSIP_SUPER_DEBUG,
			     "Encoding parent: handle %pU, fsid %u\n",
			     &refn.khandle,
			     refn.fs_id);
	}
	*max_len = len;

out:
	return type;
}

static const struct export_operations orangefs_export_ops = {
	.encode_fh = orangefs_encode_fh,
	.fh_to_dentry = orangefs_fh_to_dentry,
};

static int orangefs_fill_sb(struct super_block *sb,
		struct orangefs_fs_mount_response *fs_mount,
		void *data, int silent)
{
	int ret = -EINVAL;
	struct inode *root = NULL;
	struct dentry *root_dentry = NULL;
	struct orangefs_object_kref root_object;

	/* alloc and init our private orangefs sb info */
	sb->s_fs_info = kzalloc(sizeof(struct orangefs_sb_info_s), GFP_KERNEL);
	if (!ORANGEFS_SB(sb))
		return -ENOMEM;
	ORANGEFS_SB(sb)->sb = sb;

	ORANGEFS_SB(sb)->root_khandle = fs_mount->root_khandle;
	ORANGEFS_SB(sb)->fs_id = fs_mount->fs_id;
	ORANGEFS_SB(sb)->id = fs_mount->id;

	if (data) {
		ret = parse_mount_options(sb, data, silent);
		if (ret)
			return ret;
	}

	/* Hang the xattr handlers off the superblock */
	sb->s_xattr = orangefs_xattr_handlers;
	sb->s_magic = ORANGEFS_SUPER_MAGIC;
	sb->s_op = &orangefs_s_ops;
	sb->s_d_op = &orangefs_dentry_operations;

	sb->s_blocksize = orangefs_bufmap_size_query();
	sb->s_blocksize_bits = orangefs_bufmap_shift_query();
	sb->s_maxbytes = MAX_LFS_FILESIZE;

	root_object.khandle = ORANGEFS_SB(sb)->root_khandle;
	root_object.fs_id = ORANGEFS_SB(sb)->fs_id;
	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "get inode %pU, fsid %d\n",
		     &root_object.khandle,
		     root_object.fs_id);

	root = orangefs_iget(sb, &root_object);
	if (IS_ERR(root))
		return PTR_ERR(root);

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "Allocated root inode [%p] with mode %x\n",
		     root,
		     root->i_mode);

	/* allocates and places root dentry in dcache */
	root_dentry = d_make_root(root);
	if (!root_dentry)
		return -ENOMEM;

	sb->s_export_op = &orangefs_export_ops;
	sb->s_root = root_dentry;
	return 0;
}

struct dentry *orangefs_mount(struct file_system_type *fst,
			   int flags,
			   const char *devname,
			   void *data)
{
	int ret = -EINVAL;
	struct super_block *sb = ERR_PTR(-EINVAL);
	struct orangefs_kernel_op_s *new_op;
	struct dentry *d = ERR_PTR(-EINVAL);

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_mount: called with devname %s\n",
		     devname);

	if (!devname) {
		gossip_err("ERROR: device name not specified.\n");
		return ERR_PTR(-EINVAL);
	}

	new_op = op_alloc(ORANGEFS_VFS_OP_FS_MOUNT);
	if (!new_op)
		return ERR_PTR(-ENOMEM);

	strncpy(new_op->upcall.req.fs_mount.orangefs_config_server,
		devname,
		ORANGEFS_MAX_SERVER_ADDR_LEN);

	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "Attempting ORANGEFS Mount via host %s\n",
		     new_op->upcall.req.fs_mount.orangefs_config_server);

	ret = service_operation(new_op, "orangefs_mount", 0);
	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "orangefs_mount: mount got return value of %d\n", ret);
	if (ret)
		goto free_op;

	if (new_op->downcall.resp.fs_mount.fs_id == ORANGEFS_FS_ID_NULL) {
		gossip_err("ERROR: Retrieved null fs_id\n");
		ret = -EINVAL;
		goto free_op;
	}

	sb = sget(fst, NULL, set_anon_super, flags, NULL);

	if (IS_ERR(sb)) {
		d = ERR_CAST(sb);
		goto free_op;
	}

	ret = orangefs_fill_sb(sb,
	      &new_op->downcall.resp.fs_mount, data,
	      flags & MS_SILENT ? 1 : 0);

	if (ret) {
		d = ERR_PTR(ret);
		goto free_op;
	}

	/*
	 * on successful mount, store the devname and data
	 * used
	 */
	strncpy(ORANGEFS_SB(sb)->devname,
		devname,
		ORANGEFS_MAX_SERVER_ADDR_LEN);

	/* mount_pending must be cleared */
	ORANGEFS_SB(sb)->mount_pending = 0;

	/*
	 * finally, add this sb to our list of known orangefs
	 * sb's
	 */
	gossip_debug(GOSSIP_SUPER_DEBUG,
		     "Adding SB %p to orangefs superblocks\n",
		     ORANGEFS_SB(sb));
	spin_lock(&orangefs_superblocks_lock);
	list_add_tail(&ORANGEFS_SB(sb)->list, &orangefs_superblocks);
	spin_unlock(&orangefs_superblocks_lock);
	op_release(new_op);

	if (orangefs_userspace_version >= 20907) {
		new_op = op_alloc(ORANGEFS_VFS_OP_FEATURES);
		if (!new_op)
			return ERR_PTR(-ENOMEM);
		new_op->upcall.req.features.features = 0;
		ret = service_operation(new_op, "orangefs_features", 0);
		orangefs_features = new_op->downcall.resp.features.features;
		op_release(new_op);
	} else {
		orangefs_features = 0;
	}

	return dget(sb->s_root);

free_op:
	gossip_err("orangefs_mount: mount request failed with %d\n", ret);
	if (ret == -EINVAL) {
		gossip_err("Ensure that all orangefs-servers have the same FS configuration files\n");
		gossip_err("Look at pvfs2-client-core log file (typically /tmp/pvfs2-client.log) for more details\n");
	}

	op_release(new_op);

	return d;
}

void orangefs_kill_sb(struct super_block *sb)
{
	gossip_debug(GOSSIP_SUPER_DEBUG, "orangefs_kill_sb: called\n");

	/* provided sb cleanup */
	kill_anon_super(sb);

	/*
	 * issue the unmount to userspace to tell it to remove the
	 * dynamic mount info it has for this superblock
	 */
	 orangefs_unmount_sb(sb);

	/* remove the sb from our list of orangefs specific sb's */

	spin_lock(&orangefs_superblocks_lock);
	__list_del_entry(&ORANGEFS_SB(sb)->list);	/* not list_del_init */
	ORANGEFS_SB(sb)->list.prev = NULL;
	spin_unlock(&orangefs_superblocks_lock);

	/*
	 * make sure that ORANGEFS_DEV_REMOUNT_ALL loop that might've seen us
	 * gets completed before we free the dang thing.
	 */
	mutex_lock(&request_mutex);
	mutex_unlock(&request_mutex);

	/* free the orangefs superblock private data */
	kfree(ORANGEFS_SB(sb));
}

int orangefs_inode_cache_initialize(void)
{
	orangefs_inode_cache = kmem_cache_create("orangefs_inode_cache",
					      sizeof(struct orangefs_inode_s),
					      0,
					      ORANGEFS_CACHE_CREATE_FLAGS,
					      orangefs_inode_cache_ctor);

	if (!orangefs_inode_cache) {
		gossip_err("Cannot create orangefs_inode_cache\n");
		return -ENOMEM;
	}
	return 0;
}

int orangefs_inode_cache_finalize(void)
{
	kmem_cache_destroy(orangefs_inode_cache);
	return 0;
}