From 0a5db8fb9777ff872240053593fd86af71360ee8 Mon Sep 17 00:00:00 2001
From: Michael Vrable <mvrable@cs.ucsd.edu>
Date: Sun, 19 Sep 2010 15:22:02 -0700
Subject: [PATCH] Work on reducing memory pinned by the inode map.

---
 bluesky/bluesky-private.h | 12 ++++--------
 bluesky/cache.c           |  2 ++
 bluesky/cloudlog.c        |  2 --
 bluesky/debug.c           | 11 ++++++++---
 bluesky/imap.c            | 20 ++++++++++++++++++--
 5 files changed, 32 insertions(+), 15 deletions(-)

diff --git a/bluesky/bluesky-private.h b/bluesky/bluesky-private.h
index 7ebd728..61238eb 100644
--- a/bluesky/bluesky-private.h
+++ b/bluesky/bluesky-private.h
@@ -385,14 +385,9 @@ typedef struct {
 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. */
+    /* A pointer to the cloud log entry for this inode.  This may or may not
+     * actually have data loaded (it might just contain pointers to the data
+     * location, and in fact this will likely often be the case). */
     BlueSkyCloudLog *item;
 } InodeMapEntry;
 
@@ -414,6 +409,7 @@ typedef struct {
 InodeMapEntry *bluesky_inode_map_lookup(GSequence *inode_map, uint64_t inum,
                                         int action);
 BlueSkyCloudLog *bluesky_inode_map_serialize(BlueSkyFS *fs);
+void bluesky_inode_map_minimize(BlueSkyFS *fs);
 
 gboolean bluesky_checkpoint_load(BlueSkyFS *fs);
 
diff --git a/bluesky/cache.c b/bluesky/cache.c
index 3afceca..d1c5c84 100644
--- a/bluesky/cache.c
+++ b/bluesky/cache.c
@@ -195,6 +195,8 @@ static void flushd_cloud(BlueSkyFS *fs)
             journal_seq_start);
 
     fs->log->journal_watermark = journal_seq_start;
+
+    bluesky_inode_map_minimize(fs);
 }
 
 /* Drop cached data for a given inode, if it is clean.  inode must be locked. */
diff --git a/bluesky/cloudlog.c b/bluesky/cloudlog.c
index e50c269..9243948 100644
--- a/bluesky/cloudlog.c
+++ b/bluesky/cloudlog.c
@@ -383,8 +383,6 @@ BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
         g_mutex_lock(fs->lock);
         InodeMapEntry *entry = bluesky_inode_map_lookup(fs->inode_map,
                                                         log->inum, 1);
-        entry->id = log->id;
-        entry->location = log->location;
         entry->item = log;
         bluesky_cloudlog_ref(entry->item);
         g_mutex_unlock(fs->lock);
diff --git a/bluesky/debug.c b/bluesky/debug.c
index ee6e0c5..7de5095 100644
--- a/bluesky/debug.c
+++ b/bluesky/debug.c
@@ -81,9 +81,14 @@ void inode_map_dump(GSequence *inode_map)
              !g_sequence_iter_is_end(j); j = g_sequence_iter_next(j))
         {
             InodeMapEntry *entry = (InodeMapEntry *)g_sequence_get(j);
-            char *id = bluesky_cloudlog_id_to_string(entry->id);
-            g_print("    Entry %"PRIu64" id=%s\n", entry->inum, id);
-            g_free(id);
+            BlueSkyCloudLog *item = entry->item;
+            if (item != NULL) {
+                char *id = bluesky_cloudlog_id_to_string(item->id);
+                g_print("    Entry %"PRIu64" id=%s\n", entry->inum, id);
+                g_free(id);
+            } else {
+                g_print("    Entry %"PRIu64" not available\n", entry->inum);
+            }
         }
     }
 }
diff --git a/bluesky/imap.c b/bluesky/imap.c
index c496f99..6318256 100644
--- a/bluesky/imap.c
+++ b/bluesky/imap.c
@@ -145,6 +145,7 @@ static void bluesky_inode_map_serialize_section(BlueSkyFS *fs,
         InodeMapEntry *entry = (InodeMapEntry *)g_sequence_get(i);
         uint64_t inum = GUINT64_TO_LE(entry->inum);
         g_string_append_len(buf, (const char *)&inum, sizeof(inum));
+        bluesky_cloudlog_ref(entry->item);
         g_array_append_val(log->links, entry->item);
         i = g_sequence_iter_next(i);
     }
@@ -189,6 +190,23 @@ BlueSkyCloudLog *bluesky_inode_map_serialize(BlueSkyFS *fs)
     }
 }
 
+/* Minimize resources consumed the inode map.  This should only be called once
+ * an updated inode map has been serialized to the cloud, and will replace
+ * cloud log objects with skeletal versions that just reference the data
+ * location in the cloud (rather than pinning all object data in memory). */
+void bluesky_inode_map_minimize(BlueSkyFS *fs)
+{
+    GSequenceIter *i = g_sequence_get_begin_iter(fs->inode_map);
+    while (!g_sequence_iter_is_end(i)) {
+        InodeMapRange *range = (InodeMapRange *)g_sequence_get(i);
+
+        if (range->serialized != NULL)
+            bluesky_cloudlog_erase(range->serialized);
+
+        i = g_sequence_iter_next(i);
+    }
+}
+
 /* Reconstruct the inode map from data stored in the cloud. */
 static void bluesky_inode_map_deserialize(BlueSkyFS *fs, BlueSkyCloudLog *imap)
 {
@@ -214,8 +232,6 @@ static void bluesky_inode_map_deserialize(BlueSkyFS *fs, BlueSkyCloudLog *imap)
             entry->item = g_array_index(section->links,
                                         BlueSkyCloudLog *, j);
             bluesky_cloudlog_ref(entry->item);
-            entry->id = entry->item->id;
-            entry->location = entry->item->location;
             fs->next_inum = MAX(fs->next_inum, entry->inum + 1);
             inum++;
         }
-- 
2.20.1