1 /* Blue Sky: File Systems in the Cloud
3 * Copyright (C) 2009 The Regents of the University of California
4 * Written by Michael Vrable <mvrable@cs.ucsd.edu>
14 #include "bluesky-private.h"
16 /* The locations hash table in the file system is used to map objects to their locations. Objects are named using 128- */
21 BlueSkyCloudPointer *cloud_loc;
22 } BlueSkyLocationEntry;
24 BlueSkyCloudID bluesky_cloudlog_new_id()
27 bluesky_crypt_random_bytes((uint8_t *)&id.bytes, sizeof(id));
31 gchar *bluesky_cloudlog_id_to_string(BlueSkyCloudID id)
33 char buf[sizeof(BlueSkyCloudID) * 2 + 1];
36 for (int i = 0; i < sizeof(BlueSkyCloudID); i++) {
37 sprintf(&buf[2*i], "%02x", (uint8_t)(id.bytes[i]));
43 BlueSkyCloudID bluesky_cloudlog_id_from_string(const gchar *idstr)
46 memset(&id, 0, sizeof(id));
47 for (int i = 0; i < 2*sizeof(BlueSkyCloudID); i++) {
50 g_warning("Short cloud id: %s\n", idstr);
54 if (c >= '0' && c <= '9')
56 else if (c >= 'a' && c <= 'f')
59 g_warning("Bad character in cloud id: %s\n", idstr);
60 id.bytes[i / 2] += val << (i % 2 ? 0 : 4);
65 gboolean bluesky_cloudlog_equal(gconstpointer a, gconstpointer b)
67 BlueSkyCloudID *id1 = (BlueSkyCloudID *)a, *id2 = (BlueSkyCloudID *)b;
69 return memcmp(id1, id2, sizeof(BlueSkyCloudID)) == 0;
72 guint bluesky_cloudlog_hash(gconstpointer a)
74 BlueSkyCloudID *id = (BlueSkyCloudID *)a;
76 // Assume that bits in the ID are randomly chosen so that any subset of the
77 // bits can be used as a hash key.
78 return *(guint *)(&id->bytes);
81 /* Formatting of cloud log segments. This handles grouping items together
82 * before writing a batch to the cloud, handling indirection through items like
83 * the inode map, etc. */
85 BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs)
87 BlueSkyCloudLog *log = g_new0(BlueSkyCloudLog, 1);
89 log->lock = g_mutex_new();
90 log->cond = g_cond_new();
92 log->type = LOGTYPE_UNKNOWN;
93 log->id = bluesky_cloudlog_new_id();
94 log->pointers = g_array_new(FALSE, TRUE, sizeof(BlueSkyCloudID));
95 g_atomic_int_set(&log->refcount, 1);
100 void bluesky_cloudlog_ref(BlueSkyCloudLog *log)
105 g_atomic_int_inc(&log->refcount);
108 void bluesky_cloudlog_unref(BlueSkyCloudLog *log)
113 if (g_atomic_int_dec_and_test(&log->refcount)) {
114 g_print("Cloud log refcount dropped to zero.\n");
118 /* Start a write of the object to the local log. */
119 void bluesky_cloudlog_sync(BlueSkyCloudLog *log)
121 bluesky_log_item_submit(log, log->fs->log);
124 /* Add the given entry to the global hash table containing cloud log entries.
125 * Takes ownership of the caller's reference. */
126 void bluesky_cloudlog_insert(BlueSkyCloudLog *log)
128 g_mutex_lock(log->fs->lock);
129 g_hash_table_insert(log->fs->locations, &log->id, log);
130 g_mutex_unlock(log->fs->lock);
137 uint32_t pointer_count;
138 } __attribute__((packed));
142 BlueSkyCloudPointer location;
143 } __attribute__((packed));
147 struct logref refs[0];
150 BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
151 BlueSkyCloudLogState *state)
153 if (log->location_flags & CLOUDLOG_CLOUD) {
154 return log->location;
157 g_print("Flushing object %s to cloud...\n",
158 bluesky_cloudlog_id_to_string(log->id));
160 for (int i = 0; i < log->pointers->len; i++) {
161 BlueSkyCloudID id = g_array_index(log->pointers, BlueSkyCloudID, i);
162 g_print(" ...checking reference %s...\n",
163 bluesky_cloudlog_id_to_string(id));
164 g_mutex_lock(log->fs->lock);
165 BlueSkyCloudLog *log2
166 = (BlueSkyCloudLog *)g_hash_table_lookup(log->fs->locations, &id);
168 g_assert(log2 != NULL);
169 bluesky_cloudlog_ref(log2);
170 g_mutex_unlock(log->fs->lock);
171 bluesky_cloudlog_serialize(log2, state);
172 bluesky_cloudlog_unref(log2);
175 g_mutex_lock(log->lock);
176 bluesky_cloudlog_fetch(log);
177 g_assert(log->data != NULL);
179 log->location = state->location;
180 log->location.offset = state->data->len;
182 = sizeof(struct log_header) + sizeof(BlueSkyCloudID) * 0
185 struct log_header header;
186 memcpy(header.magic, "AgI ", 4);
187 header.size = GUINT32_TO_LE(log->location.size);
189 header.pointer_count = GUINT32_TO_LE(0);
191 g_string_append_len(state->data, (const char *)&header, sizeof(header));
192 g_string_append_len(state->data, log->data->data, log->data->len);
194 log->location_flags |= CLOUDLOG_CLOUD;
195 g_mutex_unlock(log->lock);
197 return log->location;
200 static void find_inodes(gpointer key, gpointer value, gpointer user_data)
202 BlueSkyCloudLogState *state = (BlueSkyCloudLogState *)user_data;
203 BlueSkyCloudLog *item = (BlueSkyCloudLog *)value;
205 if (item->type != LOGTYPE_INODE)
208 bluesky_cloudlog_ref(item);
209 state->inode_list = g_list_prepend(state->inode_list, item);
212 void bluesky_cloudlog_write_log(BlueSkyFS *fs)
214 BlueSkyCloudLogState *state = fs->log_state;
215 if (state->data == NULL)
216 state->data = g_string_new("");
218 g_mutex_lock(fs->lock);
219 g_hash_table_foreach(fs->locations, find_inodes, state);
220 g_mutex_unlock(fs->lock);
222 while (state->inode_list != NULL) {
223 BlueSkyCloudLog *log = (BlueSkyCloudLog *)state->inode_list->data;
224 bluesky_cloudlog_serialize(log, state);
225 state->inode_list = g_list_delete_link(state->inode_list,
229 if (state->data->len > 0) {
230 g_print("Serialized %zd bytes of data to cloud\n", state->data->len);
232 BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
233 async->op = STORE_OP_PUT;
234 async->key = g_strdup_printf("log-%08d-%08d",
235 state->location.directory,
236 state->location.sequence);
237 async->data = bluesky_string_new_from_gstring(state->data);
238 bluesky_store_async_submit(async);
239 bluesky_store_async_wait(async);
240 bluesky_store_async_unref(async);
242 state->location.sequence++;
243 state->location.offset = 0;
249 /* Ensure that a cloud log item is loaded in memory, and if not read it in.
250 * TODO: Make asynchronous, and make this also fetch from the cloud. Right now
251 * we only read from the log. Log item must be locked. */
252 void bluesky_cloudlog_fetch(BlueSkyCloudLog *log)
254 if (log->data != NULL)
257 g_print("Re-mapping log entry %d/%d/%d...\n",
258 log->log_seq, log->log_offset, log->log_size);
260 g_assert(log->location_flags & CLOUDLOG_JOURNAL);
262 log->data = bluesky_log_map_object(log->fs->log, log->log_seq,
263 log->log_offset, log->log_size);
265 g_cond_broadcast(log->cond);