Use a common state to begin the handshake.
This simplifies the logic around SSL_clear to reset the state for a new
handshake. The state around here is still a little iffy, but this is a
slight improvement.
The SSL_ST_CONNECT and SSL_ST_ACCEPT states are still kept separate to
avoid problems with the info callback reporting SSL_ST_INIT. Glancing
through info callback consumers, although they're all debugging, they
tend to assume that all intermediate states either have only
SSL_ST_CONNECT set or only SSL_ST_ACCEPT set.
(They also all look identical which makes me think it's copy-and-pasted
from OpenSSL command-line tool or something.)
Change-Id: I55503781e52b51b4ca829256c14de6f5942dae51
Reviewed-on: https://boringssl-review.googlesource.com/10760
Reviewed-by: Adam Langley <agl@google.com>
diff --git a/ssl/handshake_client.c b/ssl/handshake_client.c
index 93562e5..9a5cd02 100644
--- a/ssl/handshake_client.c
+++ b/ssl/handshake_client.c
@@ -196,6 +196,11 @@
state = ssl->state;
switch (ssl->state) {
+ case SSL_ST_INIT:
+ ssl->state = SSL_ST_CONNECT;
+ skip = 1;
+ break;
+
case SSL_ST_CONNECT:
ssl_do_info_callback(ssl, SSL_CB_HANDSHAKE_START, 1);