More cache behavior tweaks.

[bluesky.git] / bluesky / cloudlog.c

diff --git a/bluesky/cloudlog.c b/bluesky/cloudlog.c
index 84e63d8..c99d181 100644 (file)
--- a/bluesky/cloudlog.c
+++ b/bluesky/cloudlog.c
@@ -86,6 +86,8 @@ BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs)
 {
     BlueSkyCloudLog *log = g_new0(BlueSkyCloudLog, 1);
 
+    log->lock = g_mutex_new();
+    log->cond = g_cond_new();
     log->fs = fs;
     log->type = LOGTYPE_UNKNOWN;
     log->id = bluesky_cloudlog_new_id();
@@ -97,25 +99,26 @@ BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs)
 
 void bluesky_cloudlog_ref(BlueSkyCloudLog *log)
 {
+    if (log == NULL)
+        return;
+
     g_atomic_int_inc(&log->refcount);
 }
 
 void bluesky_cloudlog_unref(BlueSkyCloudLog *log)
 {
+    if (log == NULL)
+        return;
+
     if (g_atomic_int_dec_and_test(&log->refcount)) {
         g_print("Cloud log refcount dropped to zero.\n");
     }
 }
 
 /* Start a write of the object to the local log. */
-BlueSkyLogItem *bluesky_cloudlog_sync(BlueSkyCloudLog *log)
+void bluesky_cloudlog_sync(BlueSkyCloudLog *log)
 {
-    BlueSkyLogItem *log_item = bluesky_log_item_new();
-    log_item->key = bluesky_cloudlog_id_to_string(log->id);
-    log_item->data = log->data;
-    bluesky_string_ref(log->data);
-    bluesky_log_item_submit(log_item, log->fs->log);
-    return log_item;
+    bluesky_log_item_submit(log, log->fs->log);
 }
 
 /* Add the given entry to the global hash table containing cloud log entries.
@@ -127,13 +130,6 @@ void bluesky_cloudlog_insert(BlueSkyCloudLog *log)
     g_mutex_unlock(log->fs->lock);
 }
 
-/* Serialize objects into a log segment to be written to the cloud. */
-struct log_state {
-    GString *data;
-    BlueSkyCloudPointer location;
-    GList *inode_list;
-};
-
 struct log_header {
     char magic[4];
     uint32_t size;
@@ -152,7 +148,7 @@ struct log_footer {
 };
 
 BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
-                                               struct log_state *state)
+                                               BlueSkyCloudLogState *state)
 {
     if (log->location_flags & CLOUDLOG_CLOUD) {
         return log->location;
@@ -198,7 +194,7 @@ BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
 
 static void find_inodes(gpointer key, gpointer value, gpointer user_data)
 {
-    struct log_state *state = (struct log_state *)user_data;
+    BlueSkyCloudLogState *state = (BlueSkyCloudLogState *)user_data;
     BlueSkyCloudLog *item = (BlueSkyCloudLog *)value;
 
     if (item->type != LOGTYPE_INODE)
@@ -210,36 +206,56 @@ static void find_inodes(gpointer key, gpointer value, gpointer user_data)
 
 void bluesky_cloudlog_write_log(BlueSkyFS *fs)
 {
-    g_print("Starting cloudlog write...\n");
-
-    struct log_state state;
-    state.data = g_string_new("");
-    state.location.directory = 0;
-    state.location.sequence = 0;
-    state.location.offset = 0;
-    state.location.size = 0;
-    state.inode_list = NULL;
+    BlueSkyCloudLogState *state = fs->log_state;
+    if (state->data == NULL)
+        state->data = g_string_new("");
 
     g_mutex_lock(fs->lock);
-    g_hash_table_foreach(fs->locations, find_inodes, &state);
+    g_hash_table_foreach(fs->locations, find_inodes, state);
     g_mutex_unlock(fs->lock);
 
-    while (state.inode_list != NULL) {
-        BlueSkyCloudLog *log = (BlueSkyCloudLog *)state.inode_list->data;
-        bluesky_cloudlog_serialize(log, &state);
-        state.inode_list = g_list_delete_link(state.inode_list,
-                                              state.inode_list);
+    while (state->inode_list != NULL) {
+        BlueSkyCloudLog *log = (BlueSkyCloudLog *)state->inode_list->data;
+        bluesky_cloudlog_serialize(log, state);
+        state->inode_list = g_list_delete_link(state->inode_list,
+                                               state->inode_list);
+    }
+
+    if (state->data->len > 0) {
+        g_print("Serialized %zd bytes of data to cloud\n", state->data->len);
+
+        BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
+        async->op = STORE_OP_PUT;
+        async->key = g_strdup_printf("log-%08d-%08d",
+                                     state->location.directory,
+                                     state->location.sequence);
+        async->data = bluesky_string_new_from_gstring(state->data);
+        bluesky_store_async_submit(async);
+        bluesky_store_async_wait(async);
+        bluesky_store_async_unref(async);
+
+        state->location.sequence++;
+        state->location.offset = 0;
     }
 
-    g_print("Serialized %zd bytes of data\n", state.data->len);
-
-    BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
-    async->op = STORE_OP_PUT;
-    async->key = g_strdup_printf("log-%08d-%08d",
-                                 state.location.directory,
-                                 state.location.sequence);
-    async->data = bluesky_string_new_from_gstring(state.data);
-    bluesky_store_async_submit(async);
-    bluesky_store_async_wait(async);
-    bluesky_store_async_unref(async);
+    state->data = NULL;
+}
+
+/* Ensure that a cloud log item is loaded in memory, and if not read it in.
+ * TODO: Make asynchronous, and make this also fetch from the cloud.  Right now
+ * we only read from the log.  Log item must be locked. */
+void bluesky_cloudlog_fetch(BlueSkyCloudLog *log)
+{
+    if (log->data != NULL)
+        return;
+
+    g_print("Re-mapping log entry %d/%d/%d...\n",
+            log->log_seq, log->log_offset, log->log_size);
+
+    g_assert(log->location_flags & CLOUDLOG_JOURNAL);
+
+    log->data = bluesky_log_map_object(log->fs->log, log->log_seq,
+                                       log->log_offset, log->log_size);
+
+    g_cond_broadcast(log->cond);
 }