Fix various bugs in microdescriptor caching.

1 files changed, 83 insertions, 22 deletions

diff --git a/src/or/microdesc.c b/src/or/microdesc.c
index 0128fbbab..2533564dd 100644
--- a/src/or/microdesc.c
+++ b/src/or/microdesc.c
@@ -40,19 +40,28 @@ HT_GENERATE(microdesc_map, microdesc_t, node,
              _microdesc_hash, _microdesc_eq, 0.6,
              _tor_malloc, _tor_realloc, _tor_free);
 
+/* returns n bytes written */
 static int
-dump_microdescriptor(FILE *f, microdesc_t *md)
+dump_microdescriptor(FILE *f, microdesc_t *md, int *annotation_len_out)
 {
+  int r = 0;
   /* XXXX drops unkown annotations. */
   if (md->last_listed) {
     char buf[ISO_TIME_LEN+1];
+    char annotation[ISO_TIME_LEN+32];
     format_iso_time(buf, md->last_listed);
-    fprintf(f, "@last-listed %s\n", buf);
+    tor_snprintf(annotation, sizeof(annotation), "@last-listed %s\n", buf);
+    fputs(annotation, f);
+    r += strlen(annotation);
+    *annotation_len_out = r;
+  } else {
+    *annotation_len_out = 0;
   }
 
   md->off = (off_t) ftell(f);
   fwrite(md->body, 1, md->bodylen, f);
-  return 0;
+  r += md->bodylen;
+  return r;
 }
 
 static microdesc_cache_t *the_microdesc_cache = NULL;
@@ -72,7 +81,7 @@ get_microdesc_cache(void)
 }
 
 /* There are three sources of microdescriptors:
-   1) Generated us while acting as a directory authority.
+   1) Generated by us while acting as a directory authority.
    2) Loaded from the cache on disk.
    3) Downloaded.
 */
@@ -98,7 +107,8 @@ microdescs_add_to_cache(microdesc_cache_t *cache,
   return added;
 }
 
-/* Returns list of added microdesc_t. Frees any not added. */
+/* Returns list of added microdesc_t. Frees any not added. Updates last_listed.
+ */
 smartlist_t *
 microdescs_add_list_to_cache(microdesc_cache_t *cache,
                              smartlist_t *descriptors, saved_location_t where,
@@ -108,13 +118,17 @@ microdescs_add_list_to_cache(microdesc_cache_t *cache,
   open_file_t *open_file = NULL;
   FILE *f = NULL;
   //  int n_added = 0;
+  size_t size = 0;
 
   if (where == SAVED_NOWHERE && !no_save) {
-    f = start_writing_to_stdio_file(cache->journal_fname, OPEN_FLAGS_APPEND,
+    f = start_writing_to_stdio_file(cache->journal_fname,
+                                    OPEN_FLAGS_APPEND|O_BINARY,
                                     0600, &open_file);
-    if (!f)
-      log_warn(LD_DIR, "Couldn't append to journal in %s",
-               cache->journal_fname);
+    if (!f) {
+      log_warn(LD_DIR, "Couldn't append to journal in %s: %s",
+               cache->journal_fname, strerror(errno));
+      return NULL;
+    }
   }
 
   added = smartlist_create();
@@ -131,8 +145,10 @@ microdescs_add_list_to_cache(microdesc_cache_t *cache,
 
     /* Okay, it's a new one. */
     if (f) {
-      dump_microdescriptor(f, md);
+      int annotation_len;
+      size = dump_microdescriptor(f, md, &annotation_len);
       md->saved_location = SAVED_IN_JOURNAL;
+      cache->journal_len += size;
     } else {
       md->saved_location = where;
     }
@@ -143,7 +159,18 @@ microdescs_add_list_to_cache(microdesc_cache_t *cache,
     smartlist_add(added, md);
   } SMARTLIST_FOREACH_END(md);
 
-  finish_writing_to_file(open_file); /*XXX Check me.*/
+  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);
+    }
+  }
+
   return added;
 }
 
@@ -152,9 +179,11 @@ microdesc_cache_clear(microdesc_cache_t *cache)
 {
   microdesc_t **entry, **next;
   for (entry = HT_START(microdesc_map, &cache->map); entry; entry = next) {
+    microdesc_t *md = *entry;
     next = HT_NEXT_RMV(microdesc_map, &cache->map, entry);
-    microdesc_free(*entry);
+    microdesc_free(md);
   }
+  HT_CLEAR(microdesc_map, &cache->map);
   if (cache->cache_content) {
     tor_munmap_file(cache->cache_content);
     cache->cache_content = NULL;
@@ -176,8 +205,10 @@ microdesc_cache_reload(microdesc_cache_t *cache)
   if (mm) {
     added = microdescs_add_to_cache(cache, mm->data, mm->data+mm->size,
                                     SAVED_IN_CACHE, 0);
-    total += smartlist_len(added);
-    smartlist_free(added);
+    if (added) {
+      total += smartlist_len(added);
+      smartlist_free(added);
+    }
   }
 
   journal_content = read_file_to_str(cache->journal_fname,
@@ -186,8 +217,10 @@ microdesc_cache_reload(microdesc_cache_t *cache)
     added = microdescs_add_to_cache(cache, journal_content,
                                     journal_content+st.st_size,
                                     SAVED_IN_JOURNAL, 0);
-    total += smartlist_len(added);
-    smartlist_free(added);
+    if (added) {
+      total += smartlist_len(added);
+      smartlist_free(added);
+    }
     tor_free(journal_content);
   }
   log_notice(LD_DIR, "Reloaded microdescriptor cache.  Found %d descriptors.",
@@ -202,8 +235,16 @@ microdesc_cache_rebuild(microdesc_cache_t *cache)
   FILE *f;
   microdesc_t **mdp;
   smartlist_t *wrote;
+  int size;
+  off_t off = 0;
+  int orig_size, new_size;
 
-  f = start_writing_to_stdio_file(cache->cache_fname, OPEN_FLAGS_REPLACE,
+  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;
+
+  f = start_writing_to_stdio_file(cache->cache_fname,
+                                  OPEN_FLAGS_REPLACE|O_BINARY,
                                   0600, &open_file);
   if (!f)
     return -1;
@@ -212,15 +253,17 @@ microdesc_cache_rebuild(microdesc_cache_t *cache)
 
   HT_FOREACH(mdp, microdesc_map, &cache->map) {
     microdesc_t *md = *mdp;
+    int annotation_len;
     if (md->no_save)
       continue;
 
-    dump_microdescriptor(f, md);
+    size = dump_microdescriptor(f, md, &annotation_len);
+    md->off = off + annotation_len;
+    off += size;
     if (md->saved_location != SAVED_IN_CACHE) {
       tor_free(md->body);
       md->saved_location = SAVED_IN_CACHE;
     }
-
     smartlist_add(wrote, md);
   }
 
@@ -229,21 +272,29 @@ microdesc_cache_rebuild(microdesc_cache_t *cache)
   if (cache->cache_content)
     tor_munmap_file(cache->cache_content);
   cache->cache_content = tor_mmap_file(cache->cache_fname);
+
   if (!cache->cache_content && smartlist_len(wrote)) {
     log_err(LD_DIR, "Couldn't map file that we just wrote to %s!",
             cache->cache_fname);
+    smartlist_free(wrote);
     return -1;
   }
   SMARTLIST_FOREACH_BEGIN(wrote, microdesc_t *, md) {
-    if (md->no_save)
-      continue;
     tor_assert(md->saved_location == SAVED_IN_CACHE);
     md->body = (char*)cache->cache_content->data + md->off;
     tor_assert(!memcmp(md->body, "onion-key", 9));
-  } SMARTLIST_FOREACH_END(wrote);
+  } SMARTLIST_FOREACH_END(md);
 
   smartlist_free(wrote);
 
+  write_str_to_file(cache->journal_fname, "", 1);
+  cache->journal_len = 0;
+
+  new_size = (int)cache->cache_content->size;
+  log_info(LD_DIR, "Done rebuilding microdesc cache. "
+           "Saved %d bytes; %d still used.",
+           orig_size-new_size, new_size);
+
   return 0;
 }
 
@@ -265,3 +316,13 @@ microdesc_free(microdesc_t *md)
   tor_free(md);
 }
 
+void
+microdesc_free_all(void)
+{
+  if (the_microdesc_cache) {
+    microdesc_cache_clear(the_microdesc_cache);
+    tor_free(the_microdesc_cache->cache_fname);
+    tor_free(the_microdesc_cache->journal_fname);
+    tor_free(the_microdesc_cache);
+  }
+}