More aggressively use memory-mapped data for cloud log items.

Replace a string with a memory-mapped version as soon as possible when the
item is written out.  The intent is that memory-mapped versions rely on the
kernel's memory management, and don't need to be written to swap like a
private copy would, and so should give better overall system memory
management.

diff --git a/bluesky/log.c b/bluesky/log.c
index c45ffc8..99b5ae5 100644 (file)
--- a/bluesky/log.c
+++ b/bluesky/log.c
@@ -204,6 +204,14 @@ static gpointer log_thread(gpointer d)
 
         offset += sizeof(header) + sizeof(footer) + item->data->len;
 
+        /* Replace the log item's string data with a memory-mapped copy of the
+         * data, now that it has been written to the log file.  (Even if it
+         * isn't yet on disk, it should at least be in the page cache and so
+         * available to memory map.) */
+        bluesky_string_unref(item->data);
+        item->data = NULL;
+        bluesky_cloudlog_fetch(item);
+
         log->committed  = g_slist_prepend(log->committed, item);
         g_atomic_int_add(&item->data_lock_count, -1);
         g_mutex_unlock(item->lock);