Merge branch 'optimistic-client'

The conflicts are with the proposal 171 circuit isolation code, and
they're all trivial: they're just a matter of both branches adding
some unrelated code in the same places.

Conflicts:
	src/or/circuituse.c
	src/or/connection.c

src/or/buffers.c

@@ -557,6 +557,39 @@ buf_free(buf_t *buf)
   tor_free(buf);
 }
 
+/** Return a new copy of <b>in_chunk</b> */
+static chunk_t *
+chunk_copy(const chunk_t *in_chunk)
+{
+  chunk_t *newch = tor_memdup(in_chunk, CHUNK_ALLOC_SIZE(in_chunk->memlen));
+  newch->next = NULL;
+  if (in_chunk->data) {
+    off_t offset = in_chunk->data - in_chunk->mem;
+    newch->data = newch->mem + offset;
+  }
+  return newch;
+}
+
+/** Return a new copy of <b>buf</b> */
+buf_t *
+buf_copy(const buf_t *buf)
+{
+  chunk_t *ch;
+  buf_t *out = buf_new();
+  out->default_chunk_size = buf->default_chunk_size;
+  for (ch = buf->head; ch; ch = ch->next) {
+    chunk_t *newch = chunk_copy(ch);
+    if (out->tail) {
+      out->tail->next = newch;
+      out->tail = newch;
+    } else {
+      out->head = out->tail = newch;
+    }
+  }
+  out->datalen = buf->datalen;
+  return out;
+}
+
 /** Append a new chunk with enough capacity to hold <b>capacity</b> bytes to
  * the tail of <b>buf</b>.  If <b>capped</b>, don't allocate a chunk bigger
  * than MAX_CHUNK_ALLOC. */
@@ -2374,6 +2407,43 @@ write_to_evbuffer_zlib(struct evbuffer *buf, tor_zlib_state_t *state,
 }
 #endif
 
+/** Set *<b>output</b> to contain a copy of the data in *<b>input</b> */
+int
+generic_buffer_set_to_copy(generic_buffer_t **output,
+                           const generic_buffer_t *input)
+{
+#ifdef USE_BUFFEREVENTS
+  struct evbuffer_ptr ptr;
+  size_t remaining = evbuffer_get_length(input);
+  if (*output) {
+    evbuffer_drain(*output, evbuffer_get_length(*output));
+  } else {
+    if (!(*output = evbuffer_new()))
+      return -1;
+  }
+  evbuffer_ptr_set((struct evbuffer*)input, &ptr, 0, EVBUFFER_PTR_SET);
+  while (remaining) {
+    struct evbuffer_iovec v[4];
+    int n_used, i;
+    n_used = evbuffer_peek((struct evbuffer*)input, -1, &ptr, v, 4);
+    if (n_used < 0)
+      return -1;
+    for (i=0;i<n_used;++i) {
+      evbuffer_add(*output, v[i].iov_base, v[i].iov_len);
+      tor_assert(v[i].iov_len <= remaining);
+      remaining -= v[i].iov_len;
+      evbuffer_ptr_set((struct evbuffer*)input,
+                       &ptr, v[i].iov_len, EVBUFFER_PTR_ADD);
+    }
+  }
+#else
+  if (*output)
+    buf_free(*output);
+  *output = buf_copy(input);
+#endif
+  return 0;
+}
+
 /** Log an error and exit if <b>buf</b> is corrupted.
  */
 void

src/or/buffers.h

@@ -16,6 +16,7 @@ buf_t *buf_new(void);
 buf_t *buf_new_with_capacity(size_t size);
 void buf_free(buf_t *buf);
 void buf_clear(buf_t *buf);
+buf_t *buf_copy(const buf_t *buf);
 void buf_shrink(buf_t *buf);
 void buf_shrink_freelists(int free_all);
 void buf_dump_freelist_sizes(int severity);
@@ -67,6 +68,24 @@ int write_to_evbuffer_zlib(struct evbuffer *buf, tor_zlib_state_t *state,
                            int done);
 #endif
 
+#ifdef USE_BUFFEREVENTS
+#define generic_buffer_new() evbuffer_new()
+#define generic_buffer_len(b) evbuffer_get_length((b))
+#define generic_buffer_add(b,dat,len) evbuffer_add((b),(dat),(len))
+#define generic_buffer_get(b,buf,buflen) evbuffer_remove((b),(buf),(buflen))
+#define generic_buffer_clear(b) evbuffer_drain((b), evbuffer_get_length((b)))
+#define generic_buffer_free(b) evbuffer_free((b))
+#else
+#define generic_buffer_new() buf_new()
+#define generic_buffer_len(b) buf_datalen((b))
+#define generic_buffer_add(b,dat,len) write_to_buf((dat),(len),(b))
+#define generic_buffer_get(b,buf,buflen) fetch_from_buf((buf),(buflen),(b))
+#define generic_buffer_clear(b) buf_clear((b))
+#define generic_buffer_free(b) buf_free((b))
+#endif
+int generic_buffer_set_to_copy(generic_buffer_t **output,
+                               const generic_buffer_t *input);
+
 void assert_buf_ok(buf_t *buf);
 
 #ifdef BUFFERS_PRIVATE

src/or/circuituse.c

@@ -738,6 +738,7 @@ circuit_detach_stream(circuit_t *circ, edge_connection_t *conn)
   tor_assert(conn);
 
   conn->cpath_layer = NULL; /* make sure we don't keep a stale pointer */
+  conn->exit_allows_optimistic_data = 0;
   conn->on_circuit = NULL;
 
   if (CIRCUIT_IS_ORIGIN(circ)) {
@@ -1548,6 +1549,8 @@ static void
 link_apconn_to_circ(edge_connection_t *apconn, origin_circuit_t *circ,
                     crypt_path_t *cpath)
 {
+  const node_t *exitnode;
+
   /* add it into the linked list of streams on this circuit */
   log_debug(LD_APP|LD_CIRC, "attaching new conn to circ. n_circ_id %d.",
             circ->_base.n_circ_id);
@@ -1570,6 +1573,24 @@ link_apconn_to_circ(edge_connection_t *apconn, origin_circuit_t *circ,
 
   circ->isolation_any_streams_attached = 1;
   connection_edge_update_circuit_isolation(apconn, circ, 0);
+
+  /* See if we can use optimistic data on this circuit */
+  if (apconn->cpath_layer->extend_info &&
+      (exitnode = node_get_by_id(
+                     apconn->cpath_layer->extend_info->identity_digest)) &&
+      exitnode->rs) {
+    /* Okay; we know what exit node this is. */
+    if (circ->_base.purpose == CIRCUIT_PURPOSE_C_GENERAL &&
+        exitnode->rs->version_supports_optimistic_data)
+      apconn->exit_allows_optimistic_data = 1;
+    else
+      apconn->exit_allows_optimistic_data = 0;
+    log_info(LD_APP, "Looks like completed circuit to %s %s allow "
+             "optimistic data for connection to %s",
+             safe_str_client(node_describe(exitnode)),
+             apconn->exit_allows_optimistic_data ? "does" : "doesn't",
+             safe_str_client(apconn->socks_request->address));
+  }
 }
 
 /** Return true iff <b>address</b> is matched by one of the entries in

src/or/connection.c

@@ -470,6 +470,12 @@ _connection_free(connection_t *conn)
     tor_free(edge_conn->original_dest_address);
     if (edge_conn->socks_request)
       socks_request_free(edge_conn->socks_request);
+    if (edge_conn->pending_optimistic_data) {
+      generic_buffer_free(edge_conn->pending_optimistic_data);
+    }
+    if (edge_conn->sending_optimistic_data) {
+      generic_buffer_free(edge_conn->sending_optimistic_data);
+    }
     rend_data_free(edge_conn->rend_data);
   }
   if (conn->type == CONN_TYPE_CONTROL) {

src/or/connection_edge.c

@@ -57,6 +57,7 @@ static int connection_exit_connect_dir(edge_connection_t *exitconn);
 static int address_is_in_virtual_range(const char *addr);
 static int consider_plaintext_ports(edge_connection_t *conn, uint16_t port);
 static void clear_trackexithost_mappings(const char *exitname);
+static int connection_ap_supports_optimistic_data(const edge_connection_t *);
 
 /** An AP stream has failed/finished. If it hasn't already sent back
  * a socks reply, send one now (based on endreason). Also set
@@ -154,10 +155,25 @@ connection_edge_process_inbuf(edge_connection_t *conn, int package_partial)
         return -1;
       }
       return 0;
+    case AP_CONN_STATE_CONNECT_WAIT:
+      if (connection_ap_supports_optimistic_data(conn)) {
+        log_info(LD_EDGE,
+                 "data from edge while in '%s' state. Sending it anyway. "
+                 "package_partial=%d, buflen=%ld",
+                 conn_state_to_string(conn->_base.type, conn->_base.state),
+                 package_partial, connection_get_inbuf_len(TO_CONN(conn)));
+        if (connection_edge_package_raw_inbuf(conn, package_partial, NULL)<0) {
+          /* (We already sent an end cell if possible) */
+          connection_mark_for_close(TO_CONN(conn));
+          return -1;
+        }
+        return 0;
+      }
+      /* Fall through if the connection is on a circuit without optimistic
+       * data support. */
     case EXIT_CONN_STATE_CONNECTING:
     case AP_CONN_STATE_RENDDESC_WAIT:
     case AP_CONN_STATE_CIRCUIT_WAIT:
-    case AP_CONN_STATE_CONNECT_WAIT:
     case AP_CONN_STATE_RESOLVE_WAIT:
     case AP_CONN_STATE_CONTROLLER_WAIT:
       log_info(LD_EDGE,
@@ -722,6 +738,12 @@ connection_ap_detach_retriable(edge_connection_t *conn, origin_circuit_t *circ,
 {
   control_event_stream_status(conn, STREAM_EVENT_FAILED_RETRIABLE, reason);
   conn->_base.timestamp_lastread = time(NULL);
+
+  if (conn->pending_optimistic_data) {
+    generic_buffer_set_to_copy(&conn->sending_optimistic_data,
+                               conn->pending_optimistic_data);
+  }
+
   if (!get_options()->LeaveStreamsUnattached || conn->use_begindir) {
     /* If we're attaching streams ourself, or if this connection is
      * a tunneled directory connection, then just attach it. */
@@ -2337,6 +2359,22 @@ get_unique_stream_id_by_circ(origin_circuit_t *circ)
   return test_stream_id;
 }
 
+/** Return true iff <b>conn</b> is linked to a circuit and configured to use
+ * an exit that supports optimistic data. */
+static int
+connection_ap_supports_optimistic_data(const edge_connection_t *conn)
+{
+  tor_assert(conn->_base.type == CONN_TYPE_AP);
+  /* We can only send optimistic data if we're connected to an open
+     general circuit. */
+  if (conn->on_circuit == NULL ||
+      conn->on_circuit->state != CIRCUIT_STATE_OPEN ||
+      conn->on_circuit->purpose != CIRCUIT_PURPOSE_C_GENERAL)
+    return 0;
+
+  return conn->exit_allows_optimistic_data;
+}
+
 /** Write a relay begin cell, using destaddr and destport from ap_conn's
  * socks_request field, and send it down circ.
  *
@@ -2398,6 +2436,18 @@ connection_ap_handshake_send_begin(edge_connection_t *ap_conn)
   log_info(LD_APP,"Address/port sent, ap socket %d, n_circ_id %d",
            ap_conn->_base.s, circ->_base.n_circ_id);
   control_event_stream_status(ap_conn, STREAM_EVENT_SENT_CONNECT, 0);
+
+  /* If there's queued-up data, send it now */
+  if ((connection_get_inbuf_len(TO_CONN(ap_conn)) ||
+       ap_conn->sending_optimistic_data) &&
+      connection_ap_supports_optimistic_data(ap_conn)) {
+    log_info(LD_APP, "Sending up to %ld bytes of queued-up data",
+             connection_get_inbuf_len(TO_CONN(ap_conn)));
+    if (connection_edge_package_raw_inbuf(ap_conn, 1, NULL) < 0) {
+      connection_mark_for_close(TO_CONN(ap_conn));
+    }
+  }
+
   return 0;
 }
 

src/or/or.h

@@ -930,6 +930,11 @@ typedef struct {
 
 typedef struct buf_t buf_t;
 typedef struct socks_request_t socks_request_t;
+#ifdef USE_BUFFEREVENTS
+#define generic_buffer_t struct evbuffer
+#else
+#define generic_buffer_t buf_t
+#endif
 
 /* Values for connection_t.magic: used to make sure that downcasts (casts from
 * connection_t to foo_connection_t) are safe. */
@@ -1263,6 +1268,20 @@ typedef struct edge_connection_t {
    * NATd connection */
   unsigned int is_transparent_ap:1;
 
+  /** For AP connections only: Set if this connection's target exit node
+   * allows optimistic data.  (That is, data sent on this stream before
+   * the exit has sent a CONNECTED cell.)*/
+  unsigned int exit_allows_optimistic_data : 1;
+
+  /** For AP connections only: buffer for data that we have sent
+   * optimistically, which we might need to re-send if we have to
+   * retry this connection. */
+  generic_buffer_t *pending_optimistic_data;
+  /* For AP connections only: buffer for data that we previously sent
+  * optimistically which we are currently re-sending as we retry this
+  * connection. */
+  generic_buffer_t *sending_optimistic_data;
+
   /** If this is a DNSPort connection, this field holds the pending DNS
    * request that we're going to try to answer.  */
   struct evdns_server_request *dns_server_request;
@@ -1706,6 +1725,9 @@ typedef struct routerstatus_t {
   /** True iff this router is a version that, if it caches directory info,
    * we can get microdescriptors from. */
   unsigned int version_supports_microdesc_cache:1;
+  /** True iff this router is a version that allows DATA cells to arrive on
+   * a stream before it has sent a CONNECTED cell. */
+  unsigned int version_supports_optimistic_data:1;
 
   unsigned int has_bandwidth:1; /**< The vote/consensus had bw info */
   unsigned int has_exitsummary:1; /**< The vote/consensus had exit summaries */

src/or/relay.c

@@ -944,6 +944,12 @@ connection_edge_process_relay_cell_not_open(
           break;
       }
     }
+    /* This is definitely a success, so forget about any pending data we
+     * had sent. */
+    if (conn->pending_optimistic_data) {
+      generic_buffer_free(conn->pending_optimistic_data);
+      conn->pending_optimistic_data = NULL;
+    }
 
     /* handle anything that might have queued */
     if (connection_edge_package_raw_inbuf(conn, 1, NULL) < 0) {
@@ -1343,6 +1349,10 @@ connection_edge_package_raw_inbuf(edge_connection_t *conn, int package_partial,
   char payload[CELL_PAYLOAD_SIZE];
   circuit_t *circ;
   unsigned domain = conn->cpath_layer ? LD_APP : LD_EXIT;
+  int sending_from_optimistic = 0;
+  const int sending_optimistically =
+    conn->_base.type == CONN_TYPE_AP &&
+    conn->_base.state != AP_CONN_STATE_OPEN;
 
   tor_assert(conn);
 
@@ -1375,7 +1385,18 @@ connection_edge_package_raw_inbuf(edge_connection_t *conn, int package_partial,
     return 0;
   }
 
-  amount_to_process = connection_get_inbuf_len(TO_CONN(conn));
+  sending_from_optimistic = conn->sending_optimistic_data != NULL;
+
+  if (PREDICT_UNLIKELY(sending_from_optimistic)) {
+    amount_to_process = generic_buffer_len(conn->sending_optimistic_data);
+    if (PREDICT_UNLIKELY(!amount_to_process)) {
+      log_warn(LD_BUG, "sending_optimistic_data was non-NULL but empty");
+      amount_to_process = connection_get_inbuf_len(TO_CONN(conn));
+      sending_from_optimistic = 0;
+    }
+  } else {
+    amount_to_process = connection_get_inbuf_len(TO_CONN(conn));
+  }
 
   if (!amount_to_process)
     return 0;
@@ -1391,11 +1412,30 @@ connection_edge_package_raw_inbuf(edge_connection_t *conn, int package_partial,
   stats_n_data_bytes_packaged += length;
   stats_n_data_cells_packaged += 1;
 
-  connection_fetch_from_buf(payload, length, TO_CONN(conn));
+  if (PREDICT_UNLIKELY(sending_from_optimistic)) {
+    /* XXX023 We could be more efficient here by sometimes packing
+     * previously-sent optimistic data in the same cell with data
+     * from the inbuf. */
+    generic_buffer_get(conn->sending_optimistic_data, payload, length);
+    if (!generic_buffer_len(conn->sending_optimistic_data)) {
+        generic_buffer_free(conn->sending_optimistic_data);
+        conn->sending_optimistic_data = NULL;
+    }
+  } else {
+    connection_fetch_from_buf(payload, length, TO_CONN(conn));
+  }
 
   log_debug(domain,"(%d) Packaging %d bytes (%d waiting).", conn->_base.s,
             (int)length, (int)connection_get_inbuf_len(TO_CONN(conn)));
 
+  if (sending_optimistically && !sending_from_optimistic) {
+    /* This is new optimistic data; remember it in case we need to detach and
+       retry */
+    if (!conn->pending_optimistic_data)
+      conn->pending_optimistic_data = generic_buffer_new();
+    generic_buffer_add(conn->pending_optimistic_data, payload, length);
+  }
+
   if (connection_edge_send_command(conn, RELAY_COMMAND_DATA,
                                    payload, length) < 0 )
     /* circuit got marked for close, don't continue, don't need to mark conn */

src/or/routerparse.c

@@ -2092,6 +2092,7 @@ routerstatus_parse_entry_from_string(memarea_t *area,
       rs->version_supports_extrainfo_upload = 1;
       rs->version_supports_conditional_consensus = 1;
       rs->version_supports_microdesc_cache = 1;
+      rs->version_supports_optimistic_data = 1;
     } else {
       rs->version_supports_begindir =
         tor_version_as_new_as(tok->args[0], "0.2.0.1-alpha");
@@ -2109,6 +2110,8 @@ routerstatus_parse_entry_from_string(memarea_t *area,
        */
       rs->version_supports_microdesc_cache =
         tor_version_as_new_as(tok->args[0], "0.2.3.0-alpha");
+      rs->version_supports_optimistic_data =
+        tor_version_as_new_as(tok->args[0], "0.2.3.1-alpha");
     }
     if (vote_rs) {
       vote_rs->version = tor_strdup(tok->args[0]);

src/test/test.c

@@ -569,6 +569,71 @@ test_socks_5_auth_before_negotiation(void *ptr)
   ;
 }
 
+static void
+test_buffer_copy(void *arg)
+{
+  generic_buffer_t *buf=NULL, *buf2=NULL;
+  const char *s;
+  size_t len;
+  char b[256];
+  int i;
+  (void)arg;
+
+  buf = generic_buffer_new();
+  tt_assert(buf);
+
+  /* Copy an empty buffer. */
+  tt_int_op(0, ==, generic_buffer_set_to_copy(&buf2, buf));
+  tt_assert(buf2);
+  tt_int_op(0, ==, generic_buffer_len(buf2));
+
+  /* Now try with a short buffer. */
+  s = "And now comes an act of enormous enormance!";
+  len = strlen(s);
+  generic_buffer_add(buf, s, len);
+  tt_int_op(len, ==, generic_buffer_len(buf));
+  /* Add junk to buf2 so we can test replacing.*/
+  generic_buffer_add(buf2, "BLARG", 5);
+  tt_int_op(0, ==, generic_buffer_set_to_copy(&buf2, buf));
+  tt_int_op(len, ==, generic_buffer_len(buf2));
+  generic_buffer_get(buf2, b, len);
+  test_mem_op(b, ==, s, len);
+  /* Now free buf2 and retry so we can test allocating */
+  generic_buffer_free(buf2);
+  buf2 = NULL;
+  tt_int_op(0, ==, generic_buffer_set_to_copy(&buf2, buf));
+  tt_int_op(len, ==, generic_buffer_len(buf2));
+  generic_buffer_get(buf2, b, len);
+  test_mem_op(b, ==, s, len);
+  /* Clear buf for next test */
+  generic_buffer_get(buf, b, len);
+  tt_int_op(generic_buffer_len(buf),==,0);
+
+  /* Okay, now let's try a bigger buffer. */
+  s = "Quis autem vel eum iure reprehenderit qui in ea voluptate velit "
+    "esse quam nihil molestiae consequatur, vel illum qui dolorem eum "
+    "fugiat quo voluptas nulla pariatur?";
+  len = strlen(s);
+  for (i = 0; i < 256; ++i) {
+    b[0]=i;
+    generic_buffer_add(buf, b, 1);
+    generic_buffer_add(buf, s, len);
+  }
+  tt_int_op(0, ==, generic_buffer_set_to_copy(&buf2, buf));
+  tt_int_op(generic_buffer_len(buf2), ==, generic_buffer_len(buf));
+  for (i = 0; i < 256; ++i) {
+    generic_buffer_get(buf2, b, len+1);
+    tt_int_op((unsigned char)b[0],==,i);
+    test_mem_op(b+1, ==, s, len);
+  }
+
+ done:
+  if (buf)
+    generic_buffer_free(buf);
+  if (buf2)
+    generic_buffer_free(buf2);
+}
+
 /** Run unit tests for buffers.c */
 static void
 test_buffers(void)
@@ -1612,6 +1677,7 @@ const struct testcase_setup_t legacy_setup = {
 
 static struct testcase_t test_array[] = {
   ENT(buffers),
+  { "buffer_copy", test_buffer_copy, 0, NULL, NULL },
   ENT(onion_handshake),
   ENT(circuit_timeout),
   ENT(policies),