Add in some support for journal replay.

[bluesky.git] / bluesky / bluesky-private.h

diff --git a/bluesky/bluesky-private.h b/bluesky/bluesky-private.h
index b420169..837c6e5 100644 (file)
--- a/bluesky/bluesky-private.h
+++ b/bluesky/bluesky-private.h
@@ -224,18 +224,20 @@ struct _BlueSkyCloudLogState {
     GString *data;
     BlueSkyCloudPointer location;
     GList *inode_list;
+    GSList *writeback_list;     // Items which are being serialized right now
+    GList *pending_segments;    // Segments which are being uploaded now
 };
 
 gboolean bluesky_cloudlog_equal(gconstpointer a, gconstpointer b);
 guint bluesky_cloudlog_hash(gconstpointer a);
-BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs);
+BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs, const BlueSkyCloudID *id);
 gchar *bluesky_cloudlog_id_to_string(BlueSkyCloudID id);
 BlueSkyCloudID bluesky_cloudlog_id_from_string(const gchar *idstr);
 void bluesky_cloudlog_ref(BlueSkyCloudLog *log);
 void bluesky_cloudlog_unref(BlueSkyCloudLog *log);
+void bluesky_cloudlog_stats_update(BlueSkyCloudLog *log, int type);
 void bluesky_cloudlog_sync(BlueSkyCloudLog *log);
 void bluesky_cloudlog_insert(BlueSkyCloudLog *log);
-void bluesky_cloudlog_write_log(BlueSkyFS *fs);
 void bluesky_cloudlog_fetch(BlueSkyCloudLog *log);
 BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
                                                BlueSkyFS *fs);
@@ -248,34 +250,107 @@ uint32_t crc32c(uint32_t crc, const char *buf, unsigned int length);
 uint32_t crc32c_finalize(uint32_t crc);
 
 struct _BlueSkyLog {
+    BlueSkyFS *fs;
     char *log_directory;
     GAsyncQueue *queue;
     int fd, dirfd;
     int seq_num;
     GSList *committed;
 
+    /* The currently-open log file. */
+    BlueSkyCacheFile *current_log;
+
     /* Cache of log segments which have been memory-mapped. */
     GMutex *mmap_lock;
     GHashTable *mmap_cache;
+
+    /* A count of the disk space consumed (in 1024-byte units) by all files
+     * tracked by mmap_cache (whether mapped or not, actually). */
+    gint disk_used;
+
+    /* The smallest journal sequence number which may still contain data that
+     * must be preserved (since it it not yet in the cloud). */
+    int journal_watermark;
 };
 
-/* Reference-counted blocks of memory, used for passing data in and out of
- * storage backends and in other places.  This may also refer to read-only
- * mmaped data. */
-struct _BlueSkyMmap {
+/* An object for tracking log files which are stored locally--either the
+ * journal for filesystem consistency or log segments which have been fetched
+ * back from cloud storage. */
+struct _BlueSkyCacheFile {
+    GMutex *lock;
+    GCond *cond;
     gint refcount;
+    int type;                   // Only one of CLOUDLOG_{JOURNAL,CLOUD}
+    int log_dir;
     int log_seq;
-    const char *addr;
+    char *filename;             // Local filename, relateive to log directory
+    gint mapcount;              // References to the mmaped data
+    const char *addr;           // May be null if data is not mapped in memory
     size_t len;
+    BlueSkyFS *fs;
     BlueSkyLog *log;
+    gboolean fetching, ready;   // Cloud data: downloading or ready for use
+    int64_t atime;              // Access time, for cache management
 };
 
 BlueSkyLog *bluesky_log_new(const char *log_directory);
 void bluesky_log_item_submit(BlueSkyCloudLog *item, BlueSkyLog *log);
 void bluesky_log_finish_all(GList *log_items);
-BlueSkyRCStr *bluesky_log_map_object(BlueSkyLog *log, int log_seq,
+BlueSkyRCStr *bluesky_log_map_object(BlueSkyFS *fs, int log_dir, int log_seq,
                                      int log_offset, int log_size);
-void bluesky_mmap_unref(BlueSkyMmap *mmap);
+void bluesky_mmap_unref(BlueSkyCacheFile *mmap);
+void bluesky_cachefile_unref(BlueSkyCacheFile *cachefile);
+
+BlueSkyCacheFile *bluesky_cachefile_lookup(BlueSkyFS *fs,
+                                           int clouddir, int log_seq);
+void bluesky_cachefile_gc(BlueSkyFS *fs);
+
+void bluesky_replay(BlueSkyFS *fs);
+
+/* Used to track log segments that are being written to the cloud. */
+typedef struct {
+    BlueSkyRCStr *data;
+    GSList *items;
+    GMutex *lock;
+    GCond *cond;
+    gboolean complete;
+} SerializedRecord;
+
+/***** Inode map management *****/
+
+/* Mapping information for a single inode number.  These are grouped together
+ * into InodeMapRange objects. */
+typedef struct {
+    uint64_t inum;
+
+    /* The ID of the most recent version of the inode. */
+    BlueSkyCloudID id;
+
+    /* The location where that version is written in the cloud. */
+    BlueSkyCloudPointer location;
+
+    /* If the cloud log entry exists in memory, then a pointer to it, otherwise
+     * NULL. */
+    BlueSkyCloudLog *item;
+} InodeMapEntry;
+
+typedef struct {
+    /* Starting and ending inode number values that fall in this section.
+     * Endpoint values are inclusive. */
+    uint64_t start, end;
+
+    /* A sorted list (by inode number) of InodeMapEntry objects. */
+    GSequence *map_entries;
+
+    /* The location where this inode map section is stored in the cloud. */
+    BlueSkyCloudPointer location;
+
+    /* Have there been changes that require writing this section out again? */
+    gboolean dirty;
+} InodeMapRange;
+
+InodeMapEntry *bluesky_inode_map_lookup(GSequence *inode_map, uint64_t inum,
+                                        int action);
 
 #ifdef __cplusplus
 }