Next batch of memory miserdom: mmap cached-routers file. This is sure to break somewhere.

svn:r6675

8 files changed, 119 insertions, 43 deletions

diff --git a/src/or/control.c b/src/or/control.c
index 7e06ec3ef9..8e19aea94a 100644
--- a/src/or/control.c
+++ b/src/or/control.c
@@ -1330,14 +1330,14 @@ handle_getinfo_helper(const char *question, char **answer)
     if (ri) {
       const char *body = signed_descriptor_get_body(&ri->cache_info);
       if (body)
-        *answer = tor_strdup(body);
+        *answer = tor_strndup(body, ri->cache_info.signed_descriptor_len);
     }
   } else if (!strcmpstart(question, "desc/name/")) {
     routerinfo_t *ri = router_get_by_nickname(question+strlen("desc/name/"),1);
     if (ri) {
       const char *body = signed_descriptor_get_body(&ri->cache_info);
       if (body)
-        *answer = tor_strdup(body);
+        *answer = tor_strndup(body, ri->cache_info.signed_descriptor_len);
     }
   } else if (!strcmp(question, "desc/all-recent")) {
     routerlist_t *routerlist = router_get_routerlist();
@@ -1347,7 +1347,8 @@ handle_getinfo_helper(const char *question, char **answer)
       {
         const char *body = signed_descriptor_get_body(&ri->cache_info);
         if (body)
-          smartlist_add(sl, tor_strdup(body));
+          smartlist_add(sl,
+                  tor_strndup(body, ri->cache_info.signed_descriptor_len));
       });
     }
     *answer = smartlist_join_strings(sl, "", 0, NULL);
diff --git a/src/or/directory.c b/src/or/directory.c
index 3430e12c22..47af8f1504 100644
--- a/src/or/directory.c
+++ b/src/or/directory.c
@@ -1076,7 +1076,7 @@ connection_dir_client_reached_eof(connection_t *conn)
     if (which || (conn->requested_resource &&
                   !strcmpstart(conn->requested_resource, "all"))) {
       /* as we learn from them, we remove them from 'which' */
-      router_load_routers_from_string(body, 0, which);
+      router_load_routers_from_string(body, SAVED_NOWHERE, which);
       directory_info_has_arrived(time(NULL), 0);
     }
     if (which) { /* mark remaining ones as failed */
diff --git a/src/or/dirserv.c b/src/or/dirserv.c
index 562fad4910..7af40e9cb6 100644
--- a/src/or/dirserv.c
+++ b/src/or/dirserv.c
@@ -481,7 +481,7 @@ dirserv_add_descriptor(const char *desc, const char **msg)
   *msg = NULL;
 
   /* Check: is the descriptor syntactically valid? */
-  ri = router_parse_entry_from_string(desc, NULL);
+  ri = router_parse_entry_from_string(desc, NULL, 1);
   if (!ri) {
     log_warn(LD_DIRSERV, "Couldn't parse uploaded server descriptor");
     *msg = "Rejected: Couldn't parse server descriptor.";
@@ -1808,6 +1808,7 @@ connection_dirserv_add_servers_to_outbuf(connection_t *conn)
 
   while (smartlist_len(conn->fingerprint_stack) &&
          buf_datalen(conn->outbuf) < DIRSERV_BUFFER_MIN) {
+    char *body;
     char *fp = smartlist_pop_last(conn->fingerprint_stack);
     signed_descriptor_t *sd = NULL;
     if (by_fp) {
@@ -1824,13 +1825,14 @@ connection_dirserv_add_servers_to_outbuf(connection_t *conn)
     tor_free(fp);
     if (!sd)
       continue;
+    body = signed_descriptor_get_body(sd);
     if (conn->zlib_state) {
       connection_write_to_buf_zlib(
                         conn, conn->zlib_state,
-                        sd->signed_descriptor_body, sd->signed_descriptor_len,
+                        body, sd->signed_descriptor_len,
                         0);
     } else {
-      connection_write_to_buf(sd->signed_descriptor_body,
+      connection_write_to_buf(body,
                               sd->signed_descriptor_len,
                               conn);
     }
diff --git a/src/or/or.h b/src/or/or.h
index 92742848d1..113fda026f 100644
--- a/src/or/or.h
+++ b/src/or/or.h
@@ -762,6 +762,10 @@ typedef struct cached_dir_t {
   int refcnt; /**< Reference count for this cached_dir_t. */
 } cached_dir_t;
 
+typedef enum {
+   SAVED_NOWHERE=0, SAVED_IN_CACHE, SAVED_IN_JOURNAL
+} saved_location_t;
+
 /** Information need to cache an onion router's descriptor. */
 typedef struct signed_descriptor_t {
   char *signed_descriptor_body;
@@ -769,7 +773,7 @@ typedef struct signed_descriptor_t {
   char signed_descriptor_digest[DIGEST_LEN];
   char identity_digest[DIGEST_LEN];
   time_t published_on;
-  enum { SAVED_NOWHERE=0, SAVED_IN_CACHE, SAVED_IN_JOURNAL } saved_location;
+  saved_location_t saved_location;
   off_t saved_offset;
 } signed_descriptor_t;
 
@@ -933,6 +937,9 @@ typedef struct {
   /** List of signed_descriptor_t for older router descriptors we're
    * caching. */
   smartlist_t *old_routers;
+  /** DOCDOC */
+  const char *mmap_descriptors;
+  size_t mmap_descriptors_len;
 } routerlist_t;
 
 /** Information on router used when extending a circuit.  (We don't need a
@@ -2367,7 +2374,8 @@ int router_add_to_routerlist(routerinfo_t *router, const char **msg,
                              int from_cache, int from_fetch);
 int router_load_single_router(const char *s, uint8_t purpose,
                               const char **msg);
-void router_load_routers_from_string(const char *s, int from_cache,
+void router_load_routers_from_string(const char *s,
+                                     saved_location_t saved_location,
                                      smartlist_t *requested_fingerprints);
 typedef enum {
   NS_FROM_CACHE, NS_FROM_DIR, NS_GENERATED
@@ -2440,7 +2448,7 @@ int router_append_dirobj_signature(char *buf, size_t buf_len,
                                    crypto_pk_env_t *private_key);
 int router_parse_list_from_string(const char **s,
                                   smartlist_t *dest,
-                                  int from_cache);
+                                  saved_location_t saved_location);
 int router_parse_routerlist_from_directory(const char *s,
                                            routerlist_t **dest,
                                            crypto_pk_env_t *pkey,
@@ -2448,7 +2456,8 @@ int router_parse_routerlist_from_directory(const char *s,
                                            int write_to_cache);
 int router_parse_runningrouters(const char *str);
 int router_parse_directory(const char *str);
-routerinfo_t *router_parse_entry_from_string(const char *s, const char *end);
+routerinfo_t *router_parse_entry_from_string(const char *s, const char *end,
+                                             int cache_copy);
 addr_policy_t *router_parse_addr_policy_from_string(const char *s,
                                                     int assume_action);
 version_status_t tor_version_is_obsolete(const char *myversion,
diff --git a/src/or/router.c b/src/or/router.c
index 47224de14b..43a1834dd0 100644
--- a/src/or/router.c
+++ b/src/or/router.c
@@ -777,7 +777,10 @@ router_get_my_descriptor(void)
   const char *body;
   if (!router_get_my_routerinfo())
     return NULL;
+  /* Make sure this is nul-terminated. */
+  tor_assert(desc_routerinfo->cache_info.saved_location == SAVED_NOWHERE);
   body = signed_descriptor_get_body(&desc_routerinfo->cache_info);
+  tor_assert(!body[desc_routerinfo->cache_info.signed_descriptor_len]);
   log_debug(LD_GENERAL,"my desc is '%s'", body);
   return body;
 }
@@ -1187,7 +1190,7 @@ router_dump_router_to_string(char *s, size_t maxlen, routerinfo_t *router,
 
 #ifdef DEBUG_ROUTER_DUMP_ROUTER_TO_STRING
   cp = s_tmp = s_dup = tor_strdup(s);
-  ri_tmp = router_parse_entry_from_string(cp, NULL);
+  ri_tmp = router_parse_entry_from_string(cp, NULL, 1);
   if (!ri_tmp) {
     log_err(LD_BUG,
             "We just generated a router descriptor we can't parse: <<%s>>",
diff --git a/src/or/routerlist.c b/src/or/routerlist.c
index 2223e319bd..a1ad9fcd0e 100644
--- a/src/or/routerlist.c
+++ b/src/or/routerlist.c
@@ -194,6 +194,21 @@ router_append_to_journal(signed_descriptor_t *desc)
   return 0;
 }
 
+static int
+_compare_old_routers_by_age(const void **_a, const void **_b)
+{
+  const signed_descriptor_t *r1 = *_a, *r2 = *_b;
+  return r1->published_on - r2->published_on;
+}
+
+static int
+_compare_routers_by_age(const void **_a, const void **_b)
+{
+  const routerinfo_t *r1 = *_a, *r2 = *_b;
+  return r1->cache_info.published_on - r2->cache_info.published_on;
+}
+
+
 /** If the journal is too long, or if <b>force</b> is true, then atomically
  * replace the router store with the routers currently in our routerlist, and
  * clear the journal.  Return 0 on success, -1 on failure.
@@ -223,8 +238,16 @@ router_rebuild_store(int force)
   chunk_list = smartlist_create();
 
   for (i = 0; i < 2; ++i) {
-    smartlist_t *lst = (i == 0) ? routerlist->old_routers :
-                                  routerlist->routers;
+    smartlist_t *lst = smartlist_create();
+    /* We sort the routers by age to enhance locality on disk. */
+    if (i==0) {
+      smartlist_add_all(lst, routerlist->old_routers);
+      smartlist_sort(lst, _compare_old_routers_by_age);
+    } else {
+      smartlist_add_all(lst, routerlist->routers);
+      smartlist_sort(lst, _compare_routers_by_age);
+    }
+    /* Now, add the appropriate members to chunk_list */
     SMARTLIST_FOREACH(lst, void *, ptr,
     {
       signed_descriptor_t *sd = (i==0) ?
@@ -233,6 +256,7 @@ router_rebuild_store(int force)
       const char *body = signed_descriptor_get_body(sd);
       if (!body) {
         log_warn(LD_BUG, "Bug! No descriptor available for router.");
+        smartlist_free(lst);
         goto done;
       }
       c = tor_malloc(sizeof(sized_chunk_t));
@@ -240,11 +264,22 @@ router_rebuild_store(int force)
       c->len = sd->signed_descriptor_len;
       smartlist_add(chunk_list, c);
     });
+    smartlist_free(lst);
   }
   if (write_chunks_to_file(fname, chunk_list, 0)<0) {
     log_warn(LD_FS, "Error writing router store to disk.");
     goto done;
   }
+  /* Our mmap is now invalid. */
+  if (routerlist->mmap_descriptors) {
+    tor_munmap_file(routerlist->mmap_descriptors,
+                    routerlist->mmap_descriptors_len);
+    routerlist->mmap_descriptors =
+      tor_mmap_file(fname, &routerlist->mmap_descriptors_len);
+    if (! routerlist->mmap_descriptors)
+      log_warn(LD_FS, "Unable to mmap new descriptor file at '%s'.",fname);
+  }
+
   for (i = 0; i < 2; ++i) {
     smartlist_t *lst = (i == 0) ? routerlist->old_routers :
                                   routerlist->routers;
@@ -256,6 +291,8 @@ router_rebuild_store(int force)
 
       sd->saved_location = SAVED_IN_CACHE;
       sd->saved_offset = offset;
+      if (routerlist->mmap_descriptors)
+        sd->signed_descriptor_body = NULL;
       offset += sd->signed_descriptor_len;
     });
   }
@@ -286,31 +323,32 @@ router_reload_router_list(void)
 {
   or_options_t *options = get_options();
   size_t fname_len = strlen(options->DataDirectory)+32;
-  char *fname = tor_malloc(fname_len);
+  char *fname = tor_malloc(fname_len), *contents;
   struct stat st;
-  int j;
 
   if (!routerlist)
     router_get_routerlist(); /* mallocs and inits it in place */
 
   router_journal_len = router_store_len = 0;
 
-  for (j = 0; j < 2; ++j) {
-    char *contents;
-    tor_snprintf(fname, fname_len,
-                 (j==0)?"%s/cached-routers":"%s/cached-routers.new",
-                 options->DataDirectory);
-    contents = read_file_to_str(fname, 0);
-    if (contents) {
-      stat(fname, &st);
-      if (j==0)
-        router_store_len = st.st_size;
-      else
-        router_journal_len = st.st_size;
-      router_load_routers_from_string(contents, 1, NULL);
-      tor_free(contents);
-    }
+  tor_snprintf(fname, fname_len, "%s/cached-routers", options->DataDirectory);
+  routerlist->mmap_descriptors =
+    tor_mmap_file(fname, &routerlist->mmap_descriptors_len);
+  if (routerlist->mmap_descriptors) {
+    router_store_len = routerlist->mmap_descriptors_len;
+    router_load_routers_from_string(routerlist->mmap_descriptors,
+                                    SAVED_IN_CACHE, NULL);
   }
+
+  tor_snprintf(fname, fname_len, "%s/cached-routers.new",
+               options->DataDirectory);
+  contents = read_file_to_str(fname, 0);
+  if (contents) {
+    stat(fname, &st);
+    router_load_routers_from_string(contents,
+                                    SAVED_IN_JOURNAL, NULL);
+  }
+
   tor_free(fname);
 
   if (router_journal_len) {
@@ -1103,10 +1141,26 @@ router_get_by_descriptor_digest(const char *digest)
   return digestmap_get(routerlist->desc_digest_map, digest);
 }
 
+/* DOCDOC Not always nul-terminated. */
 const char *
 signed_descriptor_get_body(signed_descriptor_t *desc)
 {
-  return desc->signed_descriptor_body;
+  const char *r;
+  size_t len = desc->signed_descriptor_len;
+  tor_assert(len > 32);
+  if (desc->saved_location == SAVED_IN_CACHE && routerlist &&
+      routerlist->mmap_descriptors) {
+    tor_assert(desc->saved_offset + len <= routerlist->mmap_descriptors_len);
+    r = routerlist->mmap_descriptors + desc->saved_offset;
+  } else {
+    r = desc->signed_descriptor_body;
+  }
+  tor_assert(r);
+  tor_assert(!memcmp("router ", r, 7));
+  tor_assert(!memcmp("\n-----END SIGNATURE-----\n",
+                     r + len - 25, 25));
+
+  return r;
 }
 
 /** Return the current list of all known routers. */
@@ -1887,7 +1941,7 @@ router_load_single_router(const char *s, uint8_t purpose, const char **msg)
   tor_assert(msg);
   *msg = NULL;
 
-  if (!(ri = router_parse_entry_from_string(s, NULL))) {
+  if (!(ri = router_parse_entry_from_string(s, NULL, 1))) {
     log_warn(LD_DIR, "Error parsing router descriptor; dropping.");
     *msg = "Couldn't parse router descriptor.";
     return -1;
@@ -1927,16 +1981,18 @@ router_load_single_router(const char *s, uint8_t purpose, const char **msg)
  * uppercased identity fingerprints.  Do not update any router whose
  * fingerprint is not on the list; after updating a router, remove its
  * fingerprint from the list.
+ * DOCDOC saved_location
  */
 void
-router_load_routers_from_string(const char *s, int from_cache,
+router_load_routers_from_string(const char *s, saved_location_t saved_location,
                                 smartlist_t *requested_fingerprints)
 {
   smartlist_t *routers = smartlist_create(), *changed = smartlist_create();
   char fp[HEX_DIGEST_LEN+1];
   const char *msg;
+  int from_cache = (saved_location != SAVED_NOWHERE);
 
-  router_parse_list_from_string(&s, routers, from_cache);
+  router_parse_list_from_string(&s, routers, saved_location);
 
   routers_update_status_from_networkstatus(routers, !from_cache);
 
diff --git a/src/or/routerparse.c b/src/or/routerparse.c
index ab1a2ca3bf..608d156c40 100644
--- a/src/or/routerparse.c
+++ b/src/or/routerparse.c
@@ -635,10 +635,11 @@ check_directory_signature(const char *digest,
  * are marked running and valid.  Advances *s to a point immediately
  * following the last router entry.  Ignore any trailing router entries that
  * are not complete. Returns 0 on success and -1 on failure.
+ * DOCDOC saved_location
  */
 int
 router_parse_list_from_string(const char **s, smartlist_t *dest,
-                              int from_cache)
+                              saved_location_t saved_location)
 {
   routerinfo_t *router;
   const char *end, *cp, *start;
@@ -678,15 +679,16 @@ router_parse_list_from_string(const char **s, smartlist_t *dest,
       continue;
     }
 
-    router = router_parse_entry_from_string(*s, end);
+    router = router_parse_entry_from_string(*s, end,
+                                            saved_location != SAVED_IN_CACHE);
 
     *s = end;
     if (!router) {
       log_warn(LD_DIR, "Error reading router; skipping");
       continue;
     }
-    if (from_cache) {
-      router->cache_info.saved_location = SAVED_IN_CACHE;
+    if (saved_location != SAVED_NOWHERE) {
+      router->cache_info.saved_location = saved_location;
       router->cache_info.saved_offset = *s - start;
     }
     smartlist_add(dest, router);
@@ -698,9 +700,11 @@ router_parse_list_from_string(const char **s, smartlist_t *dest,
 /** Helper function: reads a single router entry from *<b>s</b> ...
  * *<b>end</b>.  Mallocs a new router and returns it if all goes well, else
  * returns NULL.
+ * DOCDOC cache_copy
  */
 routerinfo_t *
-router_parse_entry_from_string(const char *s, const char *end)
+router_parse_entry_from_string(const char *s, const char *end,
+                               int cache_copy)
 {
   routerinfo_t *router = NULL;
   char signed_digest[128];
@@ -747,7 +751,8 @@ router_parse_entry_from_string(const char *s, const char *end)
   }
 
   router = tor_malloc_zero(sizeof(routerinfo_t));
-  router->cache_info.signed_descriptor_body = tor_strndup(s, end-s);
+  if (cache_copy)
+    router->cache_info.signed_descriptor_body = tor_strndup(s, end-s);
   router->cache_info.signed_descriptor_len = end-s;
   memcpy(router->cache_info.signed_descriptor_digest, digest, DIGEST_LEN);
 
diff --git a/src/or/test.c b/src/or/test.c
index 14c9cbb62f..732456115d 100644
--- a/src/or/test.c
+++ b/src/or/test.c
@@ -1297,7 +1297,7 @@ test_dir_format(void)
 
   test_assert(router_dump_router_to_string(buf, 2048, &r1, pk2)>0);
   cp = buf;
-  rp1 = router_parse_entry_from_string((const char*)cp,NULL);
+  rp1 = router_parse_entry_from_string((const char*)cp,NULL,1);
   test_assert(rp1);
   test_streq(rp1->address, r1.address);
   test_eq(rp1->or_port, r1.or_port);
@@ -1320,7 +1320,7 @@ test_dir_format(void)
   test_streq(buf, buf2);
 
   cp = buf;
-  rp2 = router_parse_entry_from_string(&cp);
+  rp2 = router_parse_entry_from_string(&cp,1);
   test_assert(rp2);
   test_streq(rp2->address, r2.address);
   test_eq(rp2->or_port, r2.or_port);