Bugfix: correct cleanup after sending a HelloRequest
authorBodo Möller <bodo@openssl.org>
Fri, 21 Sep 2001 00:04:15 +0000 (00:04 +0000)
committerBodo Möller <bodo@openssl.org>
Fri, 21 Sep 2001 00:04:15 +0000 (00:04 +0000)
CHANGES
ssl/s3_srvr.c

diff --git a/CHANGES b/CHANGES
index 1d2c6cf98a308c52b027b72e91066dfa180b6bed..583b24acf17521cbc8a5ebf2bef024c8893d781f 100644 (file)
--- a/CHANGES
+++ b/CHANGES
      session resumption is possible only in the first handshake.
      [Bodo Moeller]
 
-  *) Fix ssl3_accept (ssl/s3_srvr.c): Do not call ssl_init_wbio_buffer()
-     when just sending a HelloRequest as this could interfere with
-     application data writes (and is totally unnecessary).
-     [Bodo Moeller; bug pointed out by Eric Rescorla <ekr@rtfm.com>]
+  *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
+     should end in 'break', not 'goto end' which circuments various
+     cleanups.
+
+     Also avoid some overhead by not calling ssl_init_wbio_buffer()
+     before just sending a HelloRequest.
+     [Bodo Moeller]
 
   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
      reveal whether illegal block cipher padding was found or a MAC
index 94da180d08be65b0798f6f6efc3cb358001302b2..eb5c27928b9224fc8b2379c7db7d7a5620e21a98 100644 (file)
@@ -273,9 +273,7 @@ int ssl3_accept(SSL *s)
 
                case SSL3_ST_SW_HELLO_REQ_C:
                        s->state=SSL_ST_OK;
-                       ret=1;
-                       goto end;
-                       /* break; */
+                       break;
 
                case SSL3_ST_SR_CLNT_HELLO_A:
                case SSL3_ST_SR_CLNT_HELLO_B: