2 * Copyright (C) 2005, 2006
3 * Avishay Traeger (avishay@gmail.com)
4 * Copyright (C) 2008, 2009
5 * Boaz Harrosh <bharrosh@panasas.com>
7 * Copyrights for code taken from ext2:
8 * Copyright (C) 1992, 1993, 1994, 1995
9 * Remy Card (card@masi.ibp.fr)
10 * Laboratoire MASI - Institut Blaise Pascal
11 * Universite Pierre et Marie Curie (Paris VI)
13 * linux/fs/minix/inode.c
14 * Copyright (C) 1991, 1992 Linus Torvalds
16 * This file is part of exofs.
18 * exofs is free software; you can redistribute it and/or modify
19 * it under the terms of the GNU General Public License as published by
20 * the Free Software Foundation. Since it is based on ext2, and the only
21 * valid version of GPL for the Linux kernel is version 2, the only valid
22 * version of GPL for exofs is version 2.
24 * exofs is distributed in the hope that it will be useful,
25 * but WITHOUT ANY WARRANTY; without even the implied warranty of
26 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
27 * GNU General Public License for more details.
29 * You should have received a copy of the GNU General Public License
30 * along with exofs; if not, write to the Free Software
31 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
34 #include <linux/string.h>
35 #include <linux/parser.h>
36 #include <linux/vfs.h>
37 #include <linux/random.h>
38 #include <linux/exportfs.h>
42 /******************************************************************************
44 *****************************************************************************/
47 * struct to hold what we get from mount options
49 struct exofs_mountopt {
56 * exofs-specific mount-time options.
58 enum { Opt_pid, Opt_to, Opt_mkfs, Opt_format, Opt_err };
61 * Our mount-time options. These should ideally be 64-bit unsigned, but the
62 * kernel's parsing functions do not currently support that. 32-bit should be
63 * sufficient for most applications now.
65 static match_table_t tokens = {
72 * The main option parsing method. Also makes sure that all of the mandatory
73 * mount options were set.
75 static int parse_options(char *options, struct exofs_mountopt *opts)
78 substring_t args[MAX_OPT_ARGS];
82 EXOFS_DBGMSG("parse_options %s\n", options);
84 memset(opts, 0, sizeof(*opts));
85 opts->timeout = BLK_DEFAULT_SG_TIMEOUT;
87 while ((p = strsep(&options, ",")) != NULL) {
94 token = match_token(p, tokens, args);
97 if (0 == match_strlcpy(str, &args[0], sizeof(str)))
99 opts->pid = simple_strtoull(str, NULL, 0);
100 if (opts->pid < EXOFS_MIN_PID) {
101 EXOFS_ERR("Partition ID must be >= %u",
108 if (match_int(&args[0], &option))
111 EXOFS_ERR("Timout must be > 0");
114 opts->timeout = option * HZ;
120 EXOFS_ERR("Need to specify the following options:\n");
121 EXOFS_ERR(" -o pid=pid_no_to_use\n");
128 /******************************************************************************
130 *****************************************************************************/
133 * Our inode cache. Isn't it pretty?
135 static struct kmem_cache *exofs_inode_cachep;
138 * Allocate an inode in the cache
140 static struct inode *exofs_alloc_inode(struct super_block *sb)
142 struct exofs_i_info *oi;
144 oi = kmem_cache_alloc(exofs_inode_cachep, GFP_KERNEL);
148 oi->vfs_inode.i_version = 1;
149 return &oi->vfs_inode;
153 * Remove an inode from the cache
155 static void exofs_destroy_inode(struct inode *inode)
157 kmem_cache_free(exofs_inode_cachep, exofs_i(inode));
161 * Initialize the inode
163 static void exofs_init_once(void *foo)
165 struct exofs_i_info *oi = foo;
167 inode_init_once(&oi->vfs_inode);
171 * Create and initialize the inode cache
173 static int init_inodecache(void)
175 exofs_inode_cachep = kmem_cache_create("exofs_inode_cache",
176 sizeof(struct exofs_i_info), 0,
177 SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
179 if (exofs_inode_cachep == NULL)
185 * Destroy the inode cache
187 static void destroy_inodecache(void)
189 kmem_cache_destroy(exofs_inode_cachep);
192 /******************************************************************************
193 * SUPERBLOCK FUNCTIONS
194 *****************************************************************************/
195 static const struct super_operations exofs_sops;
196 static const struct export_operations exofs_export_ops;
199 * Write the superblock to the OSD
201 int exofs_sync_fs(struct super_block *sb, int wait)
203 struct exofs_sb_info *sbi;
204 struct exofs_fscb *fscb;
205 struct osd_request *or;
206 struct osd_obj_id obj;
209 fscb = kzalloc(sizeof(struct exofs_fscb), GFP_KERNEL);
211 EXOFS_ERR("exofs_write_super: memory allocation failed.\n");
218 fscb->s_nextid = cpu_to_le64(sbi->s_nextid);
219 fscb->s_numfiles = cpu_to_le32(sbi->s_numfiles);
220 fscb->s_magic = cpu_to_le16(sb->s_magic);
223 or = osd_start_request(sbi->s_dev, GFP_KERNEL);
225 EXOFS_ERR("exofs_write_super: osd_start_request failed.\n");
229 obj.partition = sbi->s_pid;
230 obj.id = EXOFS_SUPER_ID;
231 ret = osd_req_write_kern(or, &obj, 0, fscb, sizeof(*fscb));
233 EXOFS_ERR("exofs_write_super: osd_req_write_kern failed.\n");
237 ret = exofs_sync_op(or, sbi->s_timeout, sbi->s_cred);
239 EXOFS_ERR("exofs_write_super: exofs_sync_op failed.\n");
253 static void exofs_write_super(struct super_block *sb)
255 if (!(sb->s_flags & MS_RDONLY))
256 exofs_sync_fs(sb, 1);
262 * This function is called when the vfs is freeing the superblock. We just
263 * need to free our own part.
265 static void exofs_put_super(struct super_block *sb)
268 struct exofs_sb_info *sbi = sb->s_fs_info;
273 exofs_write_super(sb);
275 /* make sure there are no pending commands */
276 for (num_pend = atomic_read(&sbi->s_curr_pending); num_pend > 0;
277 num_pend = atomic_read(&sbi->s_curr_pending)) {
278 wait_queue_head_t wq;
279 init_waitqueue_head(&wq);
280 wait_event_timeout(wq,
281 (atomic_read(&sbi->s_curr_pending) == 0),
282 msecs_to_jiffies(100));
285 osduld_put_device(sbi->s_dev);
286 kfree(sb->s_fs_info);
287 sb->s_fs_info = NULL;
293 * Read the superblock from the OSD and fill in the fields
295 static int exofs_fill_super(struct super_block *sb, void *data, int silent)
298 struct exofs_mountopt *opts = data;
299 struct exofs_sb_info *sbi; /*extended info */
300 struct exofs_fscb fscb; /*on-disk superblock info */
301 struct osd_request *or = NULL;
302 struct osd_obj_id obj;
305 sbi = kzalloc(sizeof(*sbi), GFP_KERNEL);
310 /* use mount options to fill superblock */
311 sbi->s_dev = osduld_path_lookup(opts->dev_name);
312 if (IS_ERR(sbi->s_dev)) {
313 ret = PTR_ERR(sbi->s_dev);
318 sbi->s_pid = opts->pid;
319 sbi->s_timeout = opts->timeout;
321 /* fill in some other data by hand */
322 memset(sb->s_id, 0, sizeof(sb->s_id));
323 strcpy(sb->s_id, "exofs");
324 sb->s_blocksize = EXOFS_BLKSIZE;
325 sb->s_blocksize_bits = EXOFS_BLKSHIFT;
326 sb->s_maxbytes = MAX_LFS_FILESIZE;
327 atomic_set(&sbi->s_curr_pending, 0);
331 /* read data from on-disk superblock object */
332 obj.partition = sbi->s_pid;
333 obj.id = EXOFS_SUPER_ID;
334 exofs_make_credential(sbi->s_cred, &obj);
336 or = osd_start_request(sbi->s_dev, GFP_KERNEL);
340 "exofs_fill_super: osd_start_request failed.\n");
344 ret = osd_req_read_kern(or, &obj, 0, &fscb, sizeof(fscb));
348 "exofs_fill_super: osd_req_read_kern failed.\n");
353 ret = exofs_sync_op(or, sbi->s_timeout, sbi->s_cred);
356 EXOFS_ERR("exofs_fill_super: exofs_sync_op failed.\n");
361 sb->s_magic = le16_to_cpu(fscb.s_magic);
362 sbi->s_nextid = le64_to_cpu(fscb.s_nextid);
363 sbi->s_numfiles = le32_to_cpu(fscb.s_numfiles);
365 /* make sure what we read from the object store is correct */
366 if (sb->s_magic != EXOFS_SUPER_MAGIC) {
368 EXOFS_ERR("ERROR: Bad magic value\n");
373 /* start generation numbers from a random point */
374 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
375 spin_lock_init(&sbi->s_next_gen_lock);
377 /* set up operation vectors */
378 sb->s_op = &exofs_sops;
379 sb->s_export_op = &exofs_export_ops;
380 root = exofs_iget(sb, EXOFS_ROOT_ID - EXOFS_OBJ_OFF);
382 EXOFS_ERR("ERROR: exofs_iget failed\n");
386 sb->s_root = d_alloc_root(root);
389 EXOFS_ERR("ERROR: get root inode failed\n");
394 if (!S_ISDIR(root->i_mode)) {
397 EXOFS_ERR("ERROR: corrupt root inode (mode = %hd)\n",
410 osduld_put_device(sbi->s_dev); /* NULL safe */
416 * Set up the superblock (calls exofs_fill_super eventually)
418 static int exofs_get_sb(struct file_system_type *type,
419 int flags, const char *dev_name,
420 void *data, struct vfsmount *mnt)
422 struct exofs_mountopt opts;
425 ret = parse_options(data, &opts);
429 opts.dev_name = dev_name;
430 return get_sb_nodev(type, flags, &opts, exofs_fill_super, mnt);
434 * Return information about the file system state in the buffer. This is used
435 * by the 'df' command, for example.
437 static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf)
439 struct super_block *sb = dentry->d_sb;
440 struct exofs_sb_info *sbi = sb->s_fs_info;
441 struct osd_obj_id obj = {sbi->s_pid, 0};
442 struct osd_attr attrs[] = {
443 ATTR_DEF(OSD_APAGE_PARTITION_QUOTAS,
444 OSD_ATTR_PQ_CAPACITY_QUOTA, sizeof(__be64)),
445 ATTR_DEF(OSD_APAGE_PARTITION_INFORMATION,
446 OSD_ATTR_PI_USED_CAPACITY, sizeof(__be64)),
448 uint64_t capacity = ULLONG_MAX;
449 uint64_t used = ULLONG_MAX;
450 struct osd_request *or;
451 uint8_t cred_a[OSD_CAP_LEN];
454 /* get used/capacity attributes */
455 exofs_make_credential(cred_a, &obj);
457 or = osd_start_request(sbi->s_dev, GFP_KERNEL);
459 EXOFS_DBGMSG("exofs_statfs: osd_start_request failed.\n");
463 osd_req_get_attributes(or, &obj);
464 osd_req_add_get_attr_list(or, attrs, ARRAY_SIZE(attrs));
465 ret = exofs_sync_op(or, sbi->s_timeout, cred_a);
469 ret = extract_attr_from_req(or, &attrs[0]);
471 capacity = get_unaligned_be64(attrs[0].val_ptr);
473 EXOFS_DBGMSG("exofs_statfs: get capacity failed.\n");
475 ret = extract_attr_from_req(or, &attrs[1]);
477 used = get_unaligned_be64(attrs[1].val_ptr);
479 EXOFS_DBGMSG("exofs_statfs: get used-space failed.\n");
481 /* fill in the stats buffer */
482 buf->f_type = EXOFS_SUPER_MAGIC;
483 buf->f_bsize = EXOFS_BLKSIZE;
484 buf->f_blocks = (capacity >> EXOFS_BLKSHIFT);
485 buf->f_bfree = ((capacity - used) >> EXOFS_BLKSHIFT);
486 buf->f_bavail = buf->f_bfree;
487 buf->f_files = sbi->s_numfiles;
488 buf->f_ffree = EXOFS_MAX_ID - sbi->s_numfiles;
489 buf->f_namelen = EXOFS_NAME_LEN;
496 static const struct super_operations exofs_sops = {
497 .alloc_inode = exofs_alloc_inode,
498 .destroy_inode = exofs_destroy_inode,
499 .write_inode = exofs_write_inode,
500 .delete_inode = exofs_delete_inode,
501 .put_super = exofs_put_super,
502 .write_super = exofs_write_super,
503 .sync_fs = exofs_sync_fs,
504 .statfs = exofs_statfs,
507 /******************************************************************************
509 *****************************************************************************/
511 struct dentry *exofs_get_parent(struct dentry *child)
513 unsigned long ino = exofs_parent_ino(child);
518 return d_obtain_alias(exofs_iget(child->d_inode->i_sb, ino));
521 static struct inode *exofs_nfs_get_inode(struct super_block *sb,
522 u64 ino, u32 generation)
526 inode = exofs_iget(sb, ino);
528 return ERR_CAST(inode);
529 if (generation && inode->i_generation != generation) {
530 /* we didn't find the right inode.. */
532 return ERR_PTR(-ESTALE);
537 static struct dentry *exofs_fh_to_dentry(struct super_block *sb,
538 struct fid *fid, int fh_len, int fh_type)
540 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
541 exofs_nfs_get_inode);
544 static struct dentry *exofs_fh_to_parent(struct super_block *sb,
545 struct fid *fid, int fh_len, int fh_type)
547 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
548 exofs_nfs_get_inode);
551 static const struct export_operations exofs_export_ops = {
552 .fh_to_dentry = exofs_fh_to_dentry,
553 .fh_to_parent = exofs_fh_to_parent,
554 .get_parent = exofs_get_parent,
557 /******************************************************************************
559 *****************************************************************************/
562 * struct that describes this file system
564 static struct file_system_type exofs_type = {
565 .owner = THIS_MODULE,
567 .get_sb = exofs_get_sb,
568 .kill_sb = generic_shutdown_super,
571 static int __init init_exofs(void)
575 err = init_inodecache();
579 err = register_filesystem(&exofs_type);
585 destroy_inodecache();
590 static void __exit exit_exofs(void)
592 unregister_filesystem(&exofs_type);
593 destroy_inodecache();
596 MODULE_AUTHOR("Avishay Traeger <avishay@gmail.com>");
597 MODULE_DESCRIPTION("exofs");
598 MODULE_LICENSE("GPL");
600 module_init(init_exofs)
601 module_exit(exit_exofs)