From: Matt Caswell <matt@openssl.org>
Date: Thu, 27 Oct 2016 09:46:25 +0000 (+0100)
Subject: Partial revert of "Fix client verify mode to check SSL_VERIFY_PEER"
X-Git-Tag: OpenSSL_1_1_1-pre1~3169
X-Git-Url: https://git.openssl.org/gitweb/?p=openssl.git;a=commitdiff_plain;h=c8e2f98c97ff3327784843946c2d62761572e5d5

Partial revert of "Fix client verify mode to check SSL_VERIFY_PEER"

This partially reverts commit c636c1c47. It also tweaks the documentation
and comments in this area. On the client side the documented interface for
SSL_CTX_set_verify()/SSL_set_verify() is that setting the flag
SSL_VERIFY_PEER causes verfication of the server certificate to take place.
Previously what was implemented was that if *any* flag was set then
verification would take place. The above commit improved the semantics to
be as per the documented interface.

However, we have had a report of at least one application where an
application was incorrectly using the interface and used *only*
SSL_VERIFY_FAIL_IF_NO_PEER_CERT on the client side. In OpenSSL prior to
the above commit this still caused verification of the server certificate
to take place. After this commit the application silently failed to verify
the server certificate.

Ideally SSL_CTX_set_verify()/SSL_set_verify() could be modified to indicate
if invalid flags were being used. However these are void functions!

The simplest short term solution is to revert to the previous behaviour
which at least means we "fail closed" rather than "fail open".

Thanks to Cory Benfield for reporting this issue.

Reviewed-by: Richard Levitte <levitte@openssl.org>
---

diff --git a/doc/man3/SSL_CTX_set_verify.pod b/doc/man3/SSL_CTX_set_verify.pod
index 96a98acaac..d2d3d03a89 100644
--- a/doc/man3/SSL_CTX_set_verify.pod
+++ b/doc/man3/SSL_CTX_set_verify.pod
@@ -145,6 +145,13 @@ Its return value is identical to B<preverify_ok>, so that any verification
 failure will lead to a termination of the TLS/SSL handshake with an
 alert message, if SSL_VERIFY_PEER is set.
 
+=head1 BUGS
+
+In client mode, it is not checked whether the SSL_VERIFY_PEER flag
+is set, but whether any flags are set. This can lead to
+unexpected behaviour if SSL_VERIFY_PEER and other flags are not used as
+required.
+
 =head1 RETURN VALUES
 
 The SSL*_set_verify*() functions do not provide diagnostic information.
diff --git a/ssl/statem/statem_clnt.c b/ssl/statem/statem_clnt.c
index d8fbf582e6..6a05b9dd24 100644
--- a/ssl/statem/statem_clnt.c
+++ b/ssl/statem/statem_clnt.c
@@ -1227,7 +1227,21 @@ MSG_PROCESS_RETURN tls_process_server_certificate(SSL *s, PACKET *pkt)
     }
 
     i = ssl_verify_cert_chain(s, sk);
-    if ((s->verify_mode & SSL_VERIFY_PEER) && i <= 0) {
+    /*
+     * The documented interface is that SSL_VERIFY_PEER should be set in order
+     * for client side verification of the server certificate to take place.
+     * However, historically the code has only checked that *any* flag is set
+     * to cause server verification to take place. Use of the other flags makes
+     * no sense in client mode. An attempt to clean up the semantics was
+     * reverted because at least one application *only* set
+     * SSL_VERIFY_FAIL_IF_NO_PEER_CERT. Prior to the clean up this still caused
+     * server verification to take place, after the clean up it silently did
+     * nothing. SSL_CTX_set_verify()/SSL_set_verify() cannot validate the flags
+     * sent to them because they are void functions. Therefore, we now use the
+     * (less clean) historic behaviour of performing validation if any flag is
+     * set. The *documented* interface remains the same.
+     */
+    if (s->verify_mode != SSL_VERIFY_NONE && i <= 0) {
         al = ssl_verify_alarm_type(s->verify_result);
         SSLerr(SSL_F_TLS_PROCESS_SERVER_CERTIFICATE,
                SSL_R_CERTIFICATE_VERIFY_FAILED);