Fix some resource leaks in journal replay.

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

diff --git a/bluesky/bluesky-private.h b/bluesky/bluesky-private.h
index c066fd3..3cfec78 100644 (file)
--- a/bluesky/bluesky-private.h
+++ b/bluesky/bluesky-private.h
@@ -33,6 +33,8 @@ extern int bluesky_watermark_high_total;
 /* TODO: Make this go away entirely. */
 BlueSkyFS *bluesky_new_fs(gchar *name);
 
+void bluesky_inode_free_resources(BlueSkyInode *inode);
+
 /* Linked list update functions for LRU lists. */
 void bluesky_list_unlink(GList *head, GList *item);
 GList *bluesky_list_prepend(GList *head, BlueSkyInode *inode);
@@ -45,7 +47,12 @@ BlueSkyInode *bluesky_list_tail(GList *head);
 void bluesky_serialize_superblock(GString *out, BlueSkyFS *fs);
 BlueSkyFS *bluesky_deserialize_superblock(const gchar *buf);
 BlueSkyCloudLog *bluesky_serialize_inode(BlueSkyInode *inode);
-gboolean bluesky_deserialize_inode(BlueSkyInode *inode, const gchar *buf);
+gboolean bluesky_deserialize_inode(BlueSkyInode *inode, BlueSkyCloudLog *item);
+
+void bluesky_serialize_cloudlog(BlueSkyCloudLog *log,
+                                GString *encrypted,
+                                GString *authenticated,
+                                GString *writable);
 
 /* Storage layer.  Requests can be performed asynchronously, so these objects
  * help keep track of operations in progress. */
@@ -192,13 +199,6 @@ struct _BlueSkyCloudLog {
     int location_flags;
     int pending_read, pending_write;
 
-    // If the object is not yet flushed to cloud storage but is written to a
-    // journal file locally, a reference to that journal file so that we can
-    // keep the dirty_refs count updated.  When the object is deleted or
-    // becomes clean, decrement the dirty_refs counter of the journal file and
-    // set this pointer to NULL.
-    BlueSkyCacheFile *dirty_journal;
-
     // A stable identifier for the object (only changes when authenticated data
     // is written out, but stays the same when the in-cloud cleaner relocates
     // the object).
@@ -231,11 +231,13 @@ 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);
@@ -251,6 +253,7 @@ void bluesky_cloudlog_flush(BlueSkyFS *fs);
 /* Logging infrastructure for ensuring operations are persistently recorded to
  * disk. */
 #define BLUESKY_CRC32C_SEED (~(uint32_t)0)
+#define BLUESKY_CRC32C_VALIDATOR ((uint32_t)0xb798b438UL)
 uint32_t crc32c(uint32_t crc, const char *buf, unsigned int length);
 uint32_t crc32c_finalize(uint32_t crc);
 
@@ -272,6 +275,10 @@ struct _BlueSkyLog {
     /* 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;
 };
 
 /* An object for tracking log files which are stored locally--either the
@@ -292,11 +299,6 @@ struct _BlueSkyCacheFile {
     BlueSkyLog *log;
     gboolean fetching, ready;   // Cloud data: downloading or ready for use
     int64_t atime;              // Access time, for cache management
-
-    /* Journal: Count of objects which are not yet committed to cloud storage
-     * but need to be; a non-zero value prevents the journal file from being
-     * deleted. */
-    gint dirty_refs;
 };
 
 BlueSkyLog *bluesky_log_new(const char *log_directory);
@@ -311,6 +313,53 @@ 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
 }
 #endif