Merge remote-tracking branch 'origin/maint-0.2.2'

Conflicts:
	src/or/microdesc.c

3 files changed, 35 insertions, 16 deletions

diff --git a/src/or/microdesc.c b/src/or/microdesc.c
index 521ee0709..5d3f1c372 100644
--- a/src/or/microdesc.c
+++ b/src/or/microdesc.c
@@ -214,6 +214,8 @@ microdescs_add_list_to_cache(microdesc_cache_t *cache,
       if (md2->last_listed < md->last_listed)
         md2->last_listed = md->last_listed;
       microdesc_free(md);
+      if (where != SAVED_NOWHERE)
+        cache->bytes_dropped += size;
       continue;
     }
 
@@ -243,14 +245,7 @@ microdescs_add_list_to_cache(microdesc_cache_t *cache,
   if (f)
     finish_writing_to_file(open_file); /*XXX Check me.*/
 
-  {
-    size_t old_content_len =
-      cache->cache_content ? cache->cache_content->size : 0;
-    if (cache->journal_len > 16384 + old_content_len &&
-        cache->journal_len > old_content_len / 2) {
-      microdesc_cache_rebuild(cache);
-    }
-  }
+  microdesc_cache_rebuild(cache, 0/* only as needed */);
 
   {
     networkstatus_t *ns = networkstatus_get_latest_consensus();
@@ -318,6 +313,9 @@ microdesc_cache_reload(microdesc_cache_t *cache)
   }
   log_notice(LD_DIR, "Reloaded microdescriptor cache.  Found %d descriptors.",
              total);
+
+  microdesc_cache_clean(cache, 0, 0);
+
   return 0;
 }
 
@@ -368,11 +366,30 @@ microdesc_cache_clean(microdesc_cache_t *cache, time_t cutoff, int force)
   }
 }
 
+static int
+should_rebuild_md_cache(microdesc_cache_t *cache)
+{
+    const size_t old_len =
+      cache->cache_content ? cache->cache_content->size : 0;
+    const size_t journal_len = cache->journal_len;
+    const size_t dropped = cache->bytes_dropped;
+
+    if (journal_len < 16384)
+      return 0; /* Don't bother, not enough has happened yet. */
+    if (dropped > (journal_len + old_len) / 3)
+      return 1; /* We could save 1/3 or more of the currently used space. */
+    if (journal_len > old_len / 2)
+      return 1; /* We should append to the regular file */
+
+    return 0;
+}
+
 /** Regenerate the main cache file for <b>cache</b>, clear the journal file,
  * and update every microdesc_t in the cache with pointers to its new
- * location. */
+ * location.  If <b>force</b> is true, do this unconditionally.  If
+ * <b>force</b> is false, do it only if we expect to save space on disk. */
 int
-microdesc_cache_rebuild(microdesc_cache_t *cache)
+microdesc_cache_rebuild(microdesc_cache_t *cache, int force)
 {
   open_file_t *open_file;
   FILE *f;
@@ -382,12 +399,14 @@ microdesc_cache_rebuild(microdesc_cache_t *cache)
   off_t off = 0;
   int orig_size, new_size;
 
-  log_info(LD_DIR, "Rebuilding the microdescriptor cache...");
-
   /* Remove dead descriptors */
   microdesc_cache_clean(cache, 0/*cutoff*/, 0/*force*/);
 
-  /* Calculate starting disk usage */
+  if (!force && !should_rebuild_md_cache(cache))
+    return 0;
+
+  log_info(LD_DIR, "Rebuilding the microdescriptor cache...");
+
   orig_size = (int)(cache->cache_content ? cache->cache_content->size : 0);
   orig_size += (int)cache->journal_len;
 
diff --git a/src/or/microdesc.h b/src/or/microdesc.h
index d96d34a77..c967742aa 100644
--- a/src/or/microdesc.h
+++ b/src/or/microdesc.h
@@ -23,7 +23,7 @@ smartlist_t *microdescs_add_list_to_cache(microdesc_cache_t *cache,
                         int no_save);
 
 void microdesc_cache_clean(microdesc_cache_t *cache, time_t cutoff, int force);
-int microdesc_cache_rebuild(microdesc_cache_t *cache);
+int microdesc_cache_rebuild(microdesc_cache_t *cache, int force);
 int microdesc_cache_reload(microdesc_cache_t *cache);
 void microdesc_cache_clear(microdesc_cache_t *cache);
 
diff --git a/src/test/test_microdesc.c b/src/test/test_microdesc.c
index 5b14cdb3e..5a61c5dad 100644
--- a/src/test/test_microdesc.c
+++ b/src/test/test_microdesc.c
@@ -160,7 +160,7 @@ test_md_cache(void *data)
   tt_str_op(smartlist_get(md3->family, 0), ==, "nodeX");
 
   /* Now rebuild the cache! */
-  tt_int_op(microdesc_cache_rebuild(mc), ==, 0);
+  tt_int_op(microdesc_cache_rebuild(mc, 1), ==, 0);
 
   tt_int_op(md1->saved_location, ==, SAVED_IN_CACHE);
   tt_int_op(md2->saved_location, ==, SAVED_IN_CACHE);
@@ -208,7 +208,7 @@ test_md_cache(void *data)
   md3 = NULL; /* it's history now! */
 
   /* rebuild again, make sure it stays gone. */
-  microdesc_cache_rebuild(mc);
+  microdesc_cache_rebuild(mc, 1);
   tt_ptr_op(md1, ==, microdesc_cache_lookup_by_digest256(mc, d1));
   tt_ptr_op(md2, ==, microdesc_cache_lookup_by_digest256(mc, d2));
   tt_ptr_op(NULL, ==, microdesc_cache_lookup_by_digest256(mc, d3));