RE: *****SPAM(4.2)***** [openssl.org #2086] Resolved: problem with bufferoverflowu.lib on x64 VS2008
Appears to be fixed in 1.0.0 tree; but 0.9.8m-beta still has part of the problem because it didn't get the remove duplicate code part of e.g. http://cvs.openssl.org/chngview?cn=18895 . -- - Ariel Salomon / Security Lead, Senior Software Engineer Real-Time Innovations (RTI) / www.rti.com 408 990-7439 / ar...@rti.com RTI - The Real-Time Middleware Experts -Original Message- From: Andy Polyakov via RT [mailto:r...@openssl.org] Sent: Thursday, January 07, 2010 3:07 AM To: Ariel Salomon Subject: *SPAM(4.2)* [openssl.org #2086] Resolved: problem with bufferoverflowu.lib on x64 VS2008 According to our records, your request has been resolved. If you have any further questions or concerns, please respond to this message. __ OpenSSL Project http://www.openssl.org Development Mailing List openssl-dev@openssl.org Automated List Manager majord...@openssl.org
Re: [openssl.org #1709] DTLS BUG: retransmition of handshake messages does not work
The DTLS code makes some assumptions that it is using a UDP socket BIO to detect the timeout condition for resend. When using another BIO type (e.g. BIO pair) on read, this does not work properly. - Ariel Pavel via RT wrote: Hello, This problem was described by Martin Vladic, but i cant find it in RT. Here is description: Let's suppose that handshake between client and server comes to the point where client sends this message flight to the server: Certificate ClientKeyExchange CertificateVerify ChangeCipherSpec Finished [this message is protected] So, client comes to the stage when all subsequent messages shall be protected. In above message flight only last message (Finished) is protected. First four messages are unprotected. That's all OK. To continue, client needs following response from the server: ChangeCipherSpec Finished [this message is encrypted] What happens if such message doesn't arrive? Retransmission timer expires and client must send last flight again. But, OpenSSL DTLS implementation doesn't handle this situation very well. It sends the last flight of messages, but all messages are protected because implementation thinks that CipherSpec and keys are negotiated. I think that only last message must be protected, and first four must not (like it was in first transmission of the same flight). Also, when client retransmits his last flight (5 messages), message retransmit: message 4 non-existant is printed to stderr. Even if client resends correct last flight (encrypting only Finished message), server will not retransmit his last flight (2 messages). Pavel __ OpenSSL Project http://www.openssl.org Development Mailing List openssl-dev@openssl.org Automated List Manager [EMAIL PROTECTED] -- - Ariel Salomon / Senior Software Engineer Real-Time Innovations (RTI) / www.rti.com 408 990-7439 / [EMAIL PROTECTED] RTI - The Real-Time Middleware Experts __ OpenSSL Project http://www.openssl.org Development Mailing List openssl-dev@openssl.org Automated List Manager [EMAIL PROTECTED]
[openssl.org #1245] fragmentation problems
Does anyone know the status of this patch? The problem still exists in
the openssl-0.9.8 stable tree (and I believe in the head/0.9.9 tree as
well).
The patch still applies; attached is an updated version.
- Ariel
--
- Ariel Salomon / Senior Software Engineer
Real-Time Innovations (RTI) / www.rti.com
408 200-4739 / [EMAIL PROTECTED]
RTI - The Real-Time Middleware Experts
Index: ssl/d1_both.c
===
RCS file: /home/local/cvsroot/usermodules/ariel/openssl/ssl/d1_both.c,v
retrieving revision 1.4.2.4
diff -u -r1.4.2.4 d1_both.c
--- ssl/d1_both.c 30 Sep 2007 21:20:59 - 1.4.2.4
+++ ssl/d1_both.c 4 Oct 2007 00:54:17 -
@@ -725,11 +725,12 @@
/* case (3): received a immediately useful fragment. Determine the
* possible overlap and copy the fragment.
*/
-overlap = (s-init_num - DTLS1_HM_HEADER_LENGTH) - frag_off;
/* retain the header for the first fragment */
if ( s-init_num DTLS1_HM_HEADER_LENGTH)
{
+s-init_num -= DTLS1_HM_HEADER_LENGTH;
+overlap = (s-init_num - DTLS1_HM_HEADER_LENGTH) - frag_off;
memmove((s-init_buf-data[s-init_num]),
(s-init_buf-data[s-init_num + DTLS1_HM_HEADER_LENGTH +
overlap]),
frag_len - overlap);
@@ -737,7 +738,10 @@
s-init_num += frag_len - overlap;
}
else
+{
+overlap = 0;
s-init_num += frag_len;
+}
dtls1_process_handshake_fragment(s, frag_len - overlap);
