mirror of
https://gitlab.torproject.org/tpo/core/tor.git
synced 2024-11-28 14:23:30 +01:00
Let the controller attachstream conns that are in connect_wait
or resolve_wait: we detach them first and then act as before. Resolves bug 264. svn:r6112
This commit is contained in:
parent
a0865e2f8e
commit
5f8d26496d
@ -1815,7 +1815,9 @@ handle_control_attachstream(connection_t *conn, uint32_t len,
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (ap_conn->state != AP_CONN_STATE_CONTROLLER_WAIT) {
|
||||
if (ap_conn->state != AP_CONN_STATE_CONTROLLER_WAIT &&
|
||||
ap_conn->state != AP_CONN_STATE_CONNECT_WAIT &&
|
||||
ap_conn->state != AP_CONN_STATE_RESOLVE_WAIT) {
|
||||
if (STATE_IS_V0(conn->state)) {
|
||||
send_control0_error(conn, ERR_NO_STREAM,
|
||||
"Connection is not managed by controller.");
|
||||
@ -1827,6 +1829,17 @@ handle_control_attachstream(connection_t *conn, uint32_t len,
|
||||
return 0;
|
||||
}
|
||||
|
||||
/* Do we need to detach it first? */
|
||||
if (ap_conn->state != AP_CONN_STATE_CONTROLLER_WAIT) {
|
||||
circuit_t *circ = circuit_get_by_edge_conn(conn);
|
||||
connection_edge_end(conn, END_STREAM_REASON_TIMEOUT, conn->cpath_layer);
|
||||
/* Un-mark it as ending, since we're going to reuse it. */
|
||||
conn->has_sent_end = 0;
|
||||
if (circ)
|
||||
circuit_detach_stream(circ,conn);
|
||||
conn->state = AP_CONN_STATE_CONTROLLER_WAIT;
|
||||
}
|
||||
|
||||
if (zero_circ) {
|
||||
connection_ap_handshake_rewrite_and_attach(ap_conn);
|
||||
send_control_done(conn);
|
||||
|
Loading…
Reference in New Issue
Block a user