diff options
author | Boaz Harrosh <bharrosh@panasas.com> | 2008-10-28 10:11:41 -0400 |
---|---|---|
committer | Boaz Harrosh <bharrosh@panasas.com> | 2009-03-31 12:44:34 -0400 |
commit | ba9e5e98ca2f808fe92b103a8e6ce5271b10cc89 (patch) | |
tree | 0ec6eb23adfe0f762585e3a0c0f559bd91b74b11 /fs/exofs | |
parent | e6af00f1d1697ca41ab6a55307066ef3466833a9 (diff) |
exofs: super_operations and file_system_type
This patch ties all operation vectors into a file system superblock
and registers the exofs file_system_type at module's load time.
* The file system control block (AKA on-disk superblock) resides in
an object with a special ID (defined in common.h).
Information included in the file system control block is used to
fill the in-memory superblock structure at mount time. This object
is created before the file system is used by mkexofs.c It contains
information such as:
- The file system's magic number
- The next inode number to be allocated
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Diffstat (limited to 'fs/exofs')
-rw-r--r-- | fs/exofs/Kbuild | 2 | ||||
-rw-r--r-- | fs/exofs/exofs.h | 2 | ||||
-rw-r--r-- | fs/exofs/inode.c | 186 | ||||
-rw-r--r-- | fs/exofs/super.c | 531 |
4 files changed, 720 insertions, 1 deletions
diff --git a/fs/exofs/Kbuild b/fs/exofs/Kbuild index 0a23f24db8b2..cc2d22db119c 100644 --- a/fs/exofs/Kbuild +++ b/fs/exofs/Kbuild | |||
@@ -12,5 +12,5 @@ | |||
12 | # Kbuild - Gets included from the Kernels Makefile and build system | 12 | # Kbuild - Gets included from the Kernels Makefile and build system |
13 | # | 13 | # |
14 | 14 | ||
15 | exofs-y := osd.o inode.o file.o symlink.o namei.o dir.o | 15 | exofs-y := osd.o inode.o file.o symlink.o namei.o dir.o super.o |
16 | obj-$(CONFIG_EXOFS_FS) += exofs.o | 16 | obj-$(CONFIG_EXOFS_FS) += exofs.o |
diff --git a/fs/exofs/exofs.h b/fs/exofs/exofs.h index d11982c67e2a..11e11d5c97ba 100644 --- a/fs/exofs/exofs.h +++ b/fs/exofs/exofs.h | |||
@@ -140,6 +140,8 @@ int exofs_write_begin(struct file *file, struct address_space *mapping, | |||
140 | struct page **pagep, void **fsdata); | 140 | struct page **pagep, void **fsdata); |
141 | extern struct inode *exofs_iget(struct super_block *, unsigned long); | 141 | extern struct inode *exofs_iget(struct super_block *, unsigned long); |
142 | struct inode *exofs_new_inode(struct inode *, int); | 142 | struct inode *exofs_new_inode(struct inode *, int); |
143 | extern int exofs_write_inode(struct inode *, int); | ||
144 | extern void exofs_delete_inode(struct inode *); | ||
143 | 145 | ||
144 | /* dir.c: */ | 146 | /* dir.c: */ |
145 | int exofs_add_link(struct dentry *, struct inode *); | 147 | int exofs_add_link(struct dentry *, struct inode *); |
diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 7c26ba371e1b..ba8d9fab4693 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c | |||
@@ -1115,3 +1115,189 @@ struct inode *exofs_new_inode(struct inode *dir, int mode) | |||
1115 | 1115 | ||
1116 | return inode; | 1116 | return inode; |
1117 | } | 1117 | } |
1118 | |||
1119 | /* | ||
1120 | * struct to pass two arguments to update_inode's callback | ||
1121 | */ | ||
1122 | struct updatei_args { | ||
1123 | struct exofs_sb_info *sbi; | ||
1124 | struct exofs_fcb fcb; | ||
1125 | }; | ||
1126 | |||
1127 | /* | ||
1128 | * Callback function from exofs_update_inode(). | ||
1129 | */ | ||
1130 | static void updatei_done(struct osd_request *or, void *p) | ||
1131 | { | ||
1132 | struct updatei_args *args = p; | ||
1133 | |||
1134 | osd_end_request(or); | ||
1135 | |||
1136 | atomic_dec(&args->sbi->s_curr_pending); | ||
1137 | |||
1138 | kfree(args); | ||
1139 | } | ||
1140 | |||
1141 | /* | ||
1142 | * Write the inode to the OSD. Just fill up the struct, and set the attribute | ||
1143 | * synchronously or asynchronously depending on the do_sync flag. | ||
1144 | */ | ||
1145 | static int exofs_update_inode(struct inode *inode, int do_sync) | ||
1146 | { | ||
1147 | struct exofs_i_info *oi = exofs_i(inode); | ||
1148 | struct super_block *sb = inode->i_sb; | ||
1149 | struct exofs_sb_info *sbi = sb->s_fs_info; | ||
1150 | struct osd_obj_id obj = {sbi->s_pid, inode->i_ino + EXOFS_OBJ_OFF}; | ||
1151 | struct osd_request *or; | ||
1152 | struct osd_attr attr; | ||
1153 | struct exofs_fcb *fcb; | ||
1154 | struct updatei_args *args; | ||
1155 | int ret; | ||
1156 | |||
1157 | args = kzalloc(sizeof(*args), GFP_KERNEL); | ||
1158 | if (!args) | ||
1159 | return -ENOMEM; | ||
1160 | |||
1161 | fcb = &args->fcb; | ||
1162 | |||
1163 | fcb->i_mode = cpu_to_le16(inode->i_mode); | ||
1164 | fcb->i_uid = cpu_to_le32(inode->i_uid); | ||
1165 | fcb->i_gid = cpu_to_le32(inode->i_gid); | ||
1166 | fcb->i_links_count = cpu_to_le16(inode->i_nlink); | ||
1167 | fcb->i_ctime = cpu_to_le32(inode->i_ctime.tv_sec); | ||
1168 | fcb->i_atime = cpu_to_le32(inode->i_atime.tv_sec); | ||
1169 | fcb->i_mtime = cpu_to_le32(inode->i_mtime.tv_sec); | ||
1170 | oi->i_commit_size = i_size_read(inode); | ||
1171 | fcb->i_size = cpu_to_le64(oi->i_commit_size); | ||
1172 | fcb->i_generation = cpu_to_le32(inode->i_generation); | ||
1173 | |||
1174 | if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) { | ||
1175 | if (old_valid_dev(inode->i_rdev)) { | ||
1176 | fcb->i_data[0] = | ||
1177 | cpu_to_le32(old_encode_dev(inode->i_rdev)); | ||
1178 | fcb->i_data[1] = 0; | ||
1179 | } else { | ||
1180 | fcb->i_data[0] = 0; | ||
1181 | fcb->i_data[1] = | ||
1182 | cpu_to_le32(new_encode_dev(inode->i_rdev)); | ||
1183 | fcb->i_data[2] = 0; | ||
1184 | } | ||
1185 | } else | ||
1186 | memcpy(fcb->i_data, oi->i_data, sizeof(fcb->i_data)); | ||
1187 | |||
1188 | or = osd_start_request(sbi->s_dev, GFP_KERNEL); | ||
1189 | if (unlikely(!or)) { | ||
1190 | EXOFS_ERR("exofs_update_inode: osd_start_request failed.\n"); | ||
1191 | ret = -ENOMEM; | ||
1192 | goto free_args; | ||
1193 | } | ||
1194 | |||
1195 | osd_req_set_attributes(or, &obj); | ||
1196 | |||
1197 | attr = g_attr_inode_data; | ||
1198 | attr.val_ptr = fcb; | ||
1199 | osd_req_add_set_attr_list(or, &attr, 1); | ||
1200 | |||
1201 | if (!obj_created(oi)) { | ||
1202 | EXOFS_DBGMSG("!obj_created\n"); | ||
1203 | BUG_ON(!obj_2bcreated(oi)); | ||
1204 | wait_event(oi->i_wq, obj_created(oi)); | ||
1205 | EXOFS_DBGMSG("wait_event done\n"); | ||
1206 | } | ||
1207 | |||
1208 | if (do_sync) { | ||
1209 | ret = exofs_sync_op(or, sbi->s_timeout, oi->i_cred); | ||
1210 | osd_end_request(or); | ||
1211 | goto free_args; | ||
1212 | } else { | ||
1213 | args->sbi = sbi; | ||
1214 | |||
1215 | ret = exofs_async_op(or, updatei_done, args, oi->i_cred); | ||
1216 | if (ret) { | ||
1217 | osd_end_request(or); | ||
1218 | goto free_args; | ||
1219 | } | ||
1220 | atomic_inc(&sbi->s_curr_pending); | ||
1221 | goto out; /* deallocation in updatei_done */ | ||
1222 | } | ||
1223 | |||
1224 | free_args: | ||
1225 | kfree(args); | ||
1226 | out: | ||
1227 | EXOFS_DBGMSG("ret=>%d\n", ret); | ||
1228 | return ret; | ||
1229 | } | ||
1230 | |||
1231 | int exofs_write_inode(struct inode *inode, int wait) | ||
1232 | { | ||
1233 | return exofs_update_inode(inode, wait); | ||
1234 | } | ||
1235 | |||
1236 | /* | ||
1237 | * Callback function from exofs_delete_inode() - don't have much cleaning up to | ||
1238 | * do. | ||
1239 | */ | ||
1240 | static void delete_done(struct osd_request *or, void *p) | ||
1241 | { | ||
1242 | struct exofs_sb_info *sbi; | ||
1243 | osd_end_request(or); | ||
1244 | sbi = p; | ||
1245 | atomic_dec(&sbi->s_curr_pending); | ||
1246 | } | ||
1247 | |||
1248 | /* | ||
1249 | * Called when the refcount of an inode reaches zero. We remove the object | ||
1250 | * from the OSD here. We make sure the object was created before we try and | ||
1251 | * delete it. | ||
1252 | */ | ||
1253 | void exofs_delete_inode(struct inode *inode) | ||
1254 | { | ||
1255 | struct exofs_i_info *oi = exofs_i(inode); | ||
1256 | struct super_block *sb = inode->i_sb; | ||
1257 | struct exofs_sb_info *sbi = sb->s_fs_info; | ||
1258 | struct osd_obj_id obj = {sbi->s_pid, inode->i_ino + EXOFS_OBJ_OFF}; | ||
1259 | struct osd_request *or; | ||
1260 | int ret; | ||
1261 | |||
1262 | truncate_inode_pages(&inode->i_data, 0); | ||
1263 | |||
1264 | if (is_bad_inode(inode)) | ||
1265 | goto no_delete; | ||
1266 | |||
1267 | mark_inode_dirty(inode); | ||
1268 | exofs_update_inode(inode, inode_needs_sync(inode)); | ||
1269 | |||
1270 | inode->i_size = 0; | ||
1271 | if (inode->i_blocks) | ||
1272 | exofs_truncate(inode); | ||
1273 | |||
1274 | clear_inode(inode); | ||
1275 | |||
1276 | or = osd_start_request(sbi->s_dev, GFP_KERNEL); | ||
1277 | if (unlikely(!or)) { | ||
1278 | EXOFS_ERR("exofs_delete_inode: osd_start_request failed\n"); | ||
1279 | return; | ||
1280 | } | ||
1281 | |||
1282 | osd_req_remove_object(or, &obj); | ||
1283 | |||
1284 | /* if we are deleting an obj that hasn't been created yet, wait */ | ||
1285 | if (!obj_created(oi)) { | ||
1286 | BUG_ON(!obj_2bcreated(oi)); | ||
1287 | wait_event(oi->i_wq, obj_created(oi)); | ||
1288 | } | ||
1289 | |||
1290 | ret = exofs_async_op(or, delete_done, sbi, oi->i_cred); | ||
1291 | if (ret) { | ||
1292 | EXOFS_ERR( | ||
1293 | "ERROR: @exofs_delete_inode exofs_async_op failed\n"); | ||
1294 | osd_end_request(or); | ||
1295 | return; | ||
1296 | } | ||
1297 | atomic_inc(&sbi->s_curr_pending); | ||
1298 | |||
1299 | return; | ||
1300 | |||
1301 | no_delete: | ||
1302 | clear_inode(inode); | ||
1303 | } | ||
diff --git a/fs/exofs/super.c b/fs/exofs/super.c new file mode 100644 index 000000000000..b736b117a2a4 --- /dev/null +++ b/fs/exofs/super.c | |||
@@ -0,0 +1,531 @@ | |||
1 | /* | ||
2 | * Copyright (C) 2005, 2006 | ||
3 | * Avishay Traeger (avishay@gmail.com) (avishay@il.ibm.com) | ||
4 | * Copyright (C) 2005, 2006 | ||
5 | * International Business Machines | ||
6 | * Copyright (C) 2008, 2009 | ||
7 | * Boaz Harrosh <bharrosh@panasas.com> | ||
8 | * | ||
9 | * Copyrights for code taken from ext2: | ||
10 | * Copyright (C) 1992, 1993, 1994, 1995 | ||
11 | * Remy Card (card@masi.ibp.fr) | ||
12 | * Laboratoire MASI - Institut Blaise Pascal | ||
13 | * Universite Pierre et Marie Curie (Paris VI) | ||
14 | * from | ||
15 | * linux/fs/minix/inode.c | ||
16 | * Copyright (C) 1991, 1992 Linus Torvalds | ||
17 | * | ||
18 | * This file is part of exofs. | ||
19 | * | ||
20 | * exofs is free software; you can redistribute it and/or modify | ||
21 | * it under the terms of the GNU General Public License as published by | ||
22 | * the Free Software Foundation. Since it is based on ext2, and the only | ||
23 | * valid version of GPL for the Linux kernel is version 2, the only valid | ||
24 | * version of GPL for exofs is version 2. | ||
25 | * | ||
26 | * exofs is distributed in the hope that it will be useful, | ||
27 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
28 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
29 | * GNU General Public License for more details. | ||
30 | * | ||
31 | * You should have received a copy of the GNU General Public License | ||
32 | * along with exofs; if not, write to the Free Software | ||
33 | * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA | ||
34 | */ | ||
35 | |||
36 | #include <linux/string.h> | ||
37 | #include <linux/parser.h> | ||
38 | #include <linux/vfs.h> | ||
39 | #include <linux/random.h> | ||
40 | |||
41 | #include "exofs.h" | ||
42 | |||
43 | /****************************************************************************** | ||
44 | * MOUNT OPTIONS | ||
45 | *****************************************************************************/ | ||
46 | |||
47 | /* | ||
48 | * struct to hold what we get from mount options | ||
49 | */ | ||
50 | struct exofs_mountopt { | ||
51 | const char *dev_name; | ||
52 | uint64_t pid; | ||
53 | int timeout; | ||
54 | }; | ||
55 | |||
56 | /* | ||
57 | * exofs-specific mount-time options. | ||
58 | */ | ||
59 | enum { Opt_pid, Opt_to, Opt_mkfs, Opt_format, Opt_err }; | ||
60 | |||
61 | /* | ||
62 | * Our mount-time options. These should ideally be 64-bit unsigned, but the | ||
63 | * kernel's parsing functions do not currently support that. 32-bit should be | ||
64 | * sufficient for most applications now. | ||
65 | */ | ||
66 | static match_table_t tokens = { | ||
67 | {Opt_pid, "pid=%u"}, | ||
68 | {Opt_to, "to=%u"}, | ||
69 | {Opt_err, NULL} | ||
70 | }; | ||
71 | |||
72 | /* | ||
73 | * The main option parsing method. Also makes sure that all of the mandatory | ||
74 | * mount options were set. | ||
75 | */ | ||
76 | static int parse_options(char *options, struct exofs_mountopt *opts) | ||
77 | { | ||
78 | char *p; | ||
79 | substring_t args[MAX_OPT_ARGS]; | ||
80 | int option; | ||
81 | bool s_pid = false; | ||
82 | |||
83 | EXOFS_DBGMSG("parse_options %s\n", options); | ||
84 | /* defaults */ | ||
85 | memset(opts, 0, sizeof(*opts)); | ||
86 | opts->timeout = BLK_DEFAULT_SG_TIMEOUT; | ||
87 | |||
88 | while ((p = strsep(&options, ",")) != NULL) { | ||
89 | int token; | ||
90 | char str[32]; | ||
91 | |||
92 | if (!*p) | ||
93 | continue; | ||
94 | |||
95 | token = match_token(p, tokens, args); | ||
96 | switch (token) { | ||
97 | case Opt_pid: | ||
98 | if (0 == match_strlcpy(str, &args[0], sizeof(str))) | ||
99 | return -EINVAL; | ||
100 | opts->pid = simple_strtoull(str, NULL, 0); | ||
101 | if (opts->pid < EXOFS_MIN_PID) { | ||
102 | EXOFS_ERR("Partition ID must be >= %u", | ||
103 | EXOFS_MIN_PID); | ||
104 | return -EINVAL; | ||
105 | } | ||
106 | s_pid = 1; | ||
107 | break; | ||
108 | case Opt_to: | ||
109 | if (match_int(&args[0], &option)) | ||
110 | return -EINVAL; | ||
111 | if (option <= 0) { | ||
112 | EXOFS_ERR("Timout must be > 0"); | ||
113 | return -EINVAL; | ||
114 | } | ||
115 | opts->timeout = option * HZ; | ||
116 | break; | ||
117 | } | ||
118 | } | ||
119 | |||
120 | if (!s_pid) { | ||
121 | EXOFS_ERR("Need to specify the following options:\n"); | ||
122 | EXOFS_ERR(" -o pid=pid_no_to_use\n"); | ||
123 | return -EINVAL; | ||
124 | } | ||
125 | |||
126 | return 0; | ||
127 | } | ||
128 | |||
129 | /****************************************************************************** | ||
130 | * INODE CACHE | ||
131 | *****************************************************************************/ | ||
132 | |||
133 | /* | ||
134 | * Our inode cache. Isn't it pretty? | ||
135 | */ | ||
136 | static struct kmem_cache *exofs_inode_cachep; | ||
137 | |||
138 | /* | ||
139 | * Allocate an inode in the cache | ||
140 | */ | ||
141 | static struct inode *exofs_alloc_inode(struct super_block *sb) | ||
142 | { | ||
143 | struct exofs_i_info *oi; | ||
144 | |||
145 | oi = kmem_cache_alloc(exofs_inode_cachep, GFP_KERNEL); | ||
146 | if (!oi) | ||
147 | return NULL; | ||
148 | |||
149 | oi->vfs_inode.i_version = 1; | ||
150 | return &oi->vfs_inode; | ||
151 | } | ||
152 | |||
153 | /* | ||
154 | * Remove an inode from the cache | ||
155 | */ | ||
156 | static void exofs_destroy_inode(struct inode *inode) | ||
157 | { | ||
158 | kmem_cache_free(exofs_inode_cachep, exofs_i(inode)); | ||
159 | } | ||
160 | |||
161 | /* | ||
162 | * Initialize the inode | ||
163 | */ | ||
164 | static void exofs_init_once(void *foo) | ||
165 | { | ||
166 | struct exofs_i_info *oi = foo; | ||
167 | |||
168 | inode_init_once(&oi->vfs_inode); | ||
169 | } | ||
170 | |||
171 | /* | ||
172 | * Create and initialize the inode cache | ||
173 | */ | ||
174 | static int init_inodecache(void) | ||
175 | { | ||
176 | exofs_inode_cachep = kmem_cache_create("exofs_inode_cache", | ||
177 | sizeof(struct exofs_i_info), 0, | ||
178 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, | ||
179 | exofs_init_once); | ||
180 | if (exofs_inode_cachep == NULL) | ||
181 | return -ENOMEM; | ||
182 | return 0; | ||
183 | } | ||
184 | |||
185 | /* | ||
186 | * Destroy the inode cache | ||
187 | */ | ||
188 | static void destroy_inodecache(void) | ||
189 | { | ||
190 | kmem_cache_destroy(exofs_inode_cachep); | ||
191 | } | ||
192 | |||
193 | /****************************************************************************** | ||
194 | * SUPERBLOCK FUNCTIONS | ||
195 | *****************************************************************************/ | ||
196 | static const struct super_operations exofs_sops; | ||
197 | |||
198 | /* | ||
199 | * Write the superblock to the OSD | ||
200 | */ | ||
201 | static void exofs_write_super(struct super_block *sb) | ||
202 | { | ||
203 | struct exofs_sb_info *sbi; | ||
204 | struct exofs_fscb *fscb; | ||
205 | struct osd_request *or; | ||
206 | struct osd_obj_id obj; | ||
207 | int ret; | ||
208 | |||
209 | fscb = kzalloc(sizeof(struct exofs_fscb), GFP_KERNEL); | ||
210 | if (!fscb) { | ||
211 | EXOFS_ERR("exofs_write_super: memory allocation failed.\n"); | ||
212 | return; | ||
213 | } | ||
214 | |||
215 | lock_kernel(); | ||
216 | sbi = sb->s_fs_info; | ||
217 | fscb->s_nextid = cpu_to_le64(sbi->s_nextid); | ||
218 | fscb->s_numfiles = cpu_to_le32(sbi->s_numfiles); | ||
219 | fscb->s_magic = cpu_to_le16(sb->s_magic); | ||
220 | fscb->s_newfs = 0; | ||
221 | |||
222 | or = osd_start_request(sbi->s_dev, GFP_KERNEL); | ||
223 | if (unlikely(!or)) { | ||
224 | EXOFS_ERR("exofs_write_super: osd_start_request failed.\n"); | ||
225 | goto out; | ||
226 | } | ||
227 | |||
228 | obj.partition = sbi->s_pid; | ||
229 | obj.id = EXOFS_SUPER_ID; | ||
230 | ret = osd_req_write_kern(or, &obj, 0, fscb, sizeof(*fscb)); | ||
231 | if (unlikely(ret)) { | ||
232 | EXOFS_ERR("exofs_write_super: osd_req_write_kern failed.\n"); | ||
233 | goto out; | ||
234 | } | ||
235 | |||
236 | ret = exofs_sync_op(or, sbi->s_timeout, sbi->s_cred); | ||
237 | if (unlikely(ret)) { | ||
238 | EXOFS_ERR("exofs_write_super: exofs_sync_op failed.\n"); | ||
239 | goto out; | ||
240 | } | ||
241 | sb->s_dirt = 0; | ||
242 | |||
243 | out: | ||
244 | if (or) | ||
245 | osd_end_request(or); | ||
246 | unlock_kernel(); | ||
247 | kfree(fscb); | ||
248 | } | ||
249 | |||
250 | /* | ||
251 | * This function is called when the vfs is freeing the superblock. We just | ||
252 | * need to free our own part. | ||
253 | */ | ||
254 | static void exofs_put_super(struct super_block *sb) | ||
255 | { | ||
256 | int num_pend; | ||
257 | struct exofs_sb_info *sbi = sb->s_fs_info; | ||
258 | |||
259 | /* make sure there are no pending commands */ | ||
260 | for (num_pend = atomic_read(&sbi->s_curr_pending); num_pend > 0; | ||
261 | num_pend = atomic_read(&sbi->s_curr_pending)) { | ||
262 | wait_queue_head_t wq; | ||
263 | init_waitqueue_head(&wq); | ||
264 | wait_event_timeout(wq, | ||
265 | (atomic_read(&sbi->s_curr_pending) == 0), | ||
266 | msecs_to_jiffies(100)); | ||
267 | } | ||
268 | |||
269 | osduld_put_device(sbi->s_dev); | ||
270 | kfree(sb->s_fs_info); | ||
271 | sb->s_fs_info = NULL; | ||
272 | } | ||
273 | |||
274 | /* | ||
275 | * Read the superblock from the OSD and fill in the fields | ||
276 | */ | ||
277 | static int exofs_fill_super(struct super_block *sb, void *data, int silent) | ||
278 | { | ||
279 | struct inode *root; | ||
280 | struct exofs_mountopt *opts = data; | ||
281 | struct exofs_sb_info *sbi; /*extended info */ | ||
282 | struct exofs_fscb fscb; /*on-disk superblock info */ | ||
283 | struct osd_request *or = NULL; | ||
284 | struct osd_obj_id obj; | ||
285 | int ret; | ||
286 | |||
287 | sbi = kzalloc(sizeof(*sbi), GFP_KERNEL); | ||
288 | if (!sbi) | ||
289 | return -ENOMEM; | ||
290 | sb->s_fs_info = sbi; | ||
291 | |||
292 | /* use mount options to fill superblock */ | ||
293 | sbi->s_dev = osduld_path_lookup(opts->dev_name); | ||
294 | if (IS_ERR(sbi->s_dev)) { | ||
295 | ret = PTR_ERR(sbi->s_dev); | ||
296 | sbi->s_dev = NULL; | ||
297 | goto free_sbi; | ||
298 | } | ||
299 | |||
300 | sbi->s_pid = opts->pid; | ||
301 | sbi->s_timeout = opts->timeout; | ||
302 | |||
303 | /* fill in some other data by hand */ | ||
304 | memset(sb->s_id, 0, sizeof(sb->s_id)); | ||
305 | strcpy(sb->s_id, "exofs"); | ||
306 | sb->s_blocksize = EXOFS_BLKSIZE; | ||
307 | sb->s_blocksize_bits = EXOFS_BLKSHIFT; | ||
308 | sb->s_maxbytes = MAX_LFS_FILESIZE; | ||
309 | atomic_set(&sbi->s_curr_pending, 0); | ||
310 | sb->s_bdev = NULL; | ||
311 | sb->s_dev = 0; | ||
312 | |||
313 | /* read data from on-disk superblock object */ | ||
314 | obj.partition = sbi->s_pid; | ||
315 | obj.id = EXOFS_SUPER_ID; | ||
316 | exofs_make_credential(sbi->s_cred, &obj); | ||
317 | |||
318 | or = osd_start_request(sbi->s_dev, GFP_KERNEL); | ||
319 | if (unlikely(!or)) { | ||
320 | if (!silent) | ||
321 | EXOFS_ERR( | ||
322 | "exofs_fill_super: osd_start_request failed.\n"); | ||
323 | ret = -ENOMEM; | ||
324 | goto free_sbi; | ||
325 | } | ||
326 | ret = osd_req_read_kern(or, &obj, 0, &fscb, sizeof(fscb)); | ||
327 | if (unlikely(ret)) { | ||
328 | if (!silent) | ||
329 | EXOFS_ERR( | ||
330 | "exofs_fill_super: osd_req_read_kern failed.\n"); | ||
331 | ret = -ENOMEM; | ||
332 | goto free_sbi; | ||
333 | } | ||
334 | |||
335 | ret = exofs_sync_op(or, sbi->s_timeout, sbi->s_cred); | ||
336 | if (unlikely(ret)) { | ||
337 | if (!silent) | ||
338 | EXOFS_ERR("exofs_fill_super: exofs_sync_op failed.\n"); | ||
339 | ret = -EIO; | ||
340 | goto free_sbi; | ||
341 | } | ||
342 | |||
343 | sb->s_magic = le16_to_cpu(fscb.s_magic); | ||
344 | sbi->s_nextid = le64_to_cpu(fscb.s_nextid); | ||
345 | sbi->s_numfiles = le32_to_cpu(fscb.s_numfiles); | ||
346 | |||
347 | /* make sure what we read from the object store is correct */ | ||
348 | if (sb->s_magic != EXOFS_SUPER_MAGIC) { | ||
349 | if (!silent) | ||
350 | EXOFS_ERR("ERROR: Bad magic value\n"); | ||
351 | ret = -EINVAL; | ||
352 | goto free_sbi; | ||
353 | } | ||
354 | |||
355 | /* start generation numbers from a random point */ | ||
356 | get_random_bytes(&sbi->s_next_generation, sizeof(u32)); | ||
357 | spin_lock_init(&sbi->s_next_gen_lock); | ||
358 | |||
359 | /* set up operation vectors */ | ||
360 | sb->s_op = &exofs_sops; | ||
361 | root = exofs_iget(sb, EXOFS_ROOT_ID - EXOFS_OBJ_OFF); | ||
362 | if (IS_ERR(root)) { | ||
363 | EXOFS_ERR("ERROR: exofs_iget failed\n"); | ||
364 | ret = PTR_ERR(root); | ||
365 | goto free_sbi; | ||
366 | } | ||
367 | sb->s_root = d_alloc_root(root); | ||
368 | if (!sb->s_root) { | ||
369 | iput(root); | ||
370 | EXOFS_ERR("ERROR: get root inode failed\n"); | ||
371 | ret = -ENOMEM; | ||
372 | goto free_sbi; | ||
373 | } | ||
374 | |||
375 | if (!S_ISDIR(root->i_mode)) { | ||
376 | dput(sb->s_root); | ||
377 | sb->s_root = NULL; | ||
378 | EXOFS_ERR("ERROR: corrupt root inode (mode = %hd)\n", | ||
379 | root->i_mode); | ||
380 | ret = -EINVAL; | ||
381 | goto free_sbi; | ||
382 | } | ||
383 | |||
384 | ret = 0; | ||
385 | out: | ||
386 | if (or) | ||
387 | osd_end_request(or); | ||
388 | return ret; | ||
389 | |||
390 | free_sbi: | ||
391 | osduld_put_device(sbi->s_dev); /* NULL safe */ | ||
392 | kfree(sbi); | ||
393 | goto out; | ||
394 | } | ||
395 | |||
396 | /* | ||
397 | * Set up the superblock (calls exofs_fill_super eventually) | ||
398 | */ | ||
399 | static int exofs_get_sb(struct file_system_type *type, | ||
400 | int flags, const char *dev_name, | ||
401 | void *data, struct vfsmount *mnt) | ||
402 | { | ||
403 | struct exofs_mountopt opts; | ||
404 | int ret; | ||
405 | |||
406 | ret = parse_options(data, &opts); | ||
407 | if (ret) | ||
408 | return ret; | ||
409 | |||
410 | opts.dev_name = dev_name; | ||
411 | return get_sb_nodev(type, flags, &opts, exofs_fill_super, mnt); | ||
412 | } | ||
413 | |||
414 | /* | ||
415 | * Return information about the file system state in the buffer. This is used | ||
416 | * by the 'df' command, for example. | ||
417 | */ | ||
418 | static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf) | ||
419 | { | ||
420 | struct super_block *sb = dentry->d_sb; | ||
421 | struct exofs_sb_info *sbi = sb->s_fs_info; | ||
422 | struct osd_obj_id obj = {sbi->s_pid, 0}; | ||
423 | struct osd_attr attrs[] = { | ||
424 | ATTR_DEF(OSD_APAGE_PARTITION_QUOTAS, | ||
425 | OSD_ATTR_PQ_CAPACITY_QUOTA, sizeof(__be64)), | ||
426 | ATTR_DEF(OSD_APAGE_PARTITION_INFORMATION, | ||
427 | OSD_ATTR_PI_USED_CAPACITY, sizeof(__be64)), | ||
428 | }; | ||
429 | uint64_t capacity = ULLONG_MAX; | ||
430 | uint64_t used = ULLONG_MAX; | ||
431 | struct osd_request *or; | ||
432 | uint8_t cred_a[OSD_CAP_LEN]; | ||
433 | int ret; | ||
434 | |||
435 | /* get used/capacity attributes */ | ||
436 | exofs_make_credential(cred_a, &obj); | ||
437 | |||
438 | or = osd_start_request(sbi->s_dev, GFP_KERNEL); | ||
439 | if (unlikely(!or)) { | ||
440 | EXOFS_DBGMSG("exofs_statfs: osd_start_request failed.\n"); | ||
441 | return -ENOMEM; | ||
442 | } | ||
443 | |||
444 | osd_req_get_attributes(or, &obj); | ||
445 | osd_req_add_get_attr_list(or, attrs, ARRAY_SIZE(attrs)); | ||
446 | ret = exofs_sync_op(or, sbi->s_timeout, cred_a); | ||
447 | if (unlikely(ret)) | ||
448 | goto out; | ||
449 | |||
450 | ret = extract_attr_from_req(or, &attrs[0]); | ||
451 | if (likely(!ret)) | ||
452 | capacity = get_unaligned_be64(attrs[0].val_ptr); | ||
453 | else | ||
454 | EXOFS_DBGMSG("exofs_statfs: get capacity failed.\n"); | ||
455 | |||
456 | ret = extract_attr_from_req(or, &attrs[1]); | ||
457 | if (likely(!ret)) | ||
458 | used = get_unaligned_be64(attrs[1].val_ptr); | ||
459 | else | ||
460 | EXOFS_DBGMSG("exofs_statfs: get used-space failed.\n"); | ||
461 | |||
462 | /* fill in the stats buffer */ | ||
463 | buf->f_type = EXOFS_SUPER_MAGIC; | ||
464 | buf->f_bsize = EXOFS_BLKSIZE; | ||
465 | buf->f_blocks = (capacity >> EXOFS_BLKSHIFT); | ||
466 | buf->f_bfree = ((capacity - used) >> EXOFS_BLKSHIFT); | ||
467 | buf->f_bavail = buf->f_bfree; | ||
468 | buf->f_files = sbi->s_numfiles; | ||
469 | buf->f_ffree = EXOFS_MAX_ID - sbi->s_numfiles; | ||
470 | buf->f_namelen = EXOFS_NAME_LEN; | ||
471 | |||
472 | out: | ||
473 | osd_end_request(or); | ||
474 | return ret; | ||
475 | } | ||
476 | |||
477 | static const struct super_operations exofs_sops = { | ||
478 | .alloc_inode = exofs_alloc_inode, | ||
479 | .destroy_inode = exofs_destroy_inode, | ||
480 | .write_inode = exofs_write_inode, | ||
481 | .delete_inode = exofs_delete_inode, | ||
482 | .put_super = exofs_put_super, | ||
483 | .write_super = exofs_write_super, | ||
484 | .statfs = exofs_statfs, | ||
485 | }; | ||
486 | |||
487 | /****************************************************************************** | ||
488 | * INSMOD/RMMOD | ||
489 | *****************************************************************************/ | ||
490 | |||
491 | /* | ||
492 | * struct that describes this file system | ||
493 | */ | ||
494 | static struct file_system_type exofs_type = { | ||
495 | .owner = THIS_MODULE, | ||
496 | .name = "exofs", | ||
497 | .get_sb = exofs_get_sb, | ||
498 | .kill_sb = generic_shutdown_super, | ||
499 | }; | ||
500 | |||
501 | static int __init init_exofs(void) | ||
502 | { | ||
503 | int err; | ||
504 | |||
505 | err = init_inodecache(); | ||
506 | if (err) | ||
507 | goto out; | ||
508 | |||
509 | err = register_filesystem(&exofs_type); | ||
510 | if (err) | ||
511 | goto out_d; | ||
512 | |||
513 | return 0; | ||
514 | out_d: | ||
515 | destroy_inodecache(); | ||
516 | out: | ||
517 | return err; | ||
518 | } | ||
519 | |||
520 | static void __exit exit_exofs(void) | ||
521 | { | ||
522 | unregister_filesystem(&exofs_type); | ||
523 | destroy_inodecache(); | ||
524 | } | ||
525 | |||
526 | MODULE_AUTHOR("Avishay Traeger <avishay@gmail.com>"); | ||
527 | MODULE_DESCRIPTION("exofs"); | ||
528 | MODULE_LICENSE("GPL"); | ||
529 | |||
530 | module_init(init_exofs) | ||
531 | module_exit(exit_exofs) | ||