This message continues the thread about SSL client failure when using
DES-CBC3-SHA and client authentication...
On Wed, Dec 06, 2000 at 02:36:38PM -0800, Jeffrey Ricks wrote:
> Here are the logs I mentioned. Thanks for offering to look at them.
> Let me know if you need any more info.
Hi Jeff,
I have looked into the logs you sent and see, that there is some new
fact involved that I didn't know before. I actually don't remember
that you mentioned it to the list and it would have changed the discussion
significantly.
First, let me examine the s_client output from the failing connection:
Here, first an SSL connection is successfully established (without a
client certificate requested):
...
---
No client certificate CA names sent
---
SSL handshake has read 1527 bytes and written 240 bytes
---
New, TLSv1/SSLv3, Cipher is DES-CBC3-SHA
Server public key is 1024 bit
SSL-Session:
Protocol : TLSv1
Cipher : DES-CBC3-SHA
...
Then, application data is succcessfully exchanged and only later the server
invokes a renegotiation with the request of a client certificate:
...
SSL_connect:SSL renegotiate ciphers
SSL_connect:SSLv3 write client hello A
SSL_connect:SSLv3 read server hello A
depth=1 /C=US/ST=MyState/L=MyCity/O=MyCompany/OU=Client Certificate Authority/CN
=Test Root (1024RSA) dev
verify return:1
depth=0 /C=US/ST=MyState/L=MyCity/O=MyCompany/OU=Client Certificate Authority/CN
=voodoo
verify return:1
SSL_connect:SSLv3 read server certificate A
SSL3 alert write:fatal:illegal parameter
SSL_connect:error in SSLv3 read server key exchange A
7305:error:1408E098:SSL routines:SSL3_GET_MESSAGE:excessive message size:s3_both
.c:302:
Here the failure occurs. The behaviour shown does match what can be seen from
the ssldump output. After 4.5 seconds the initial handshake is finished and
your application is sending data. After some data has been sent, a
renegotiation is started and excessive data (note the size of the last message)
is being sent from the server:
...
1 6 0.2077 (0.0000) C>SV3.1(1) ChangeCipherSpec
1 7 0.2077 (0.0000) C>SV3.1(40) Handshake
1 8 0.2217 (0.0140) S>CV3.1(1) ChangeCipherSpec
1 9 0.2217 (0.0000) S>CV3.1(40) Handshake
1 10 4.5120 (4.2902) C>SV3.1(80) application_data
1 11 4.5250 (0.0129) C>SV3.1(40) application_data
1 12 4.5250 (0.0000) C>SV3.1(40) application_data
1 13 4.5250 (0.0000) C>SV3.1(24) application_data
1 14 4.5250 (0.0000) C>SV3.1(40) application_data
1 15 4.5250 (0.0000) C>SV3.1(56) application_data
1 16 4.5250 (0.0000) C>SV3.1(48) application_data
1 17 4.5250 (0.0000) C>SV3.1(40) application_data
1 18 4.5257 (0.0007) S>CV3.1(32) Handshake
1 19 4.5260 (0.0003) C>SV3.1(72) Handshake
1 20 4.5268 (0.0007) S>CV3.1(96) Handshake
1 21 4.5298 (0.0030) S>CV3.1(1408) Handshake
1 22 4.5300 (0.0001) S>CV3.1(9920) Handshake
According to the s_client output, the error is caused in
"SSLv3 read server key exchange A", that would mean in ssl3_get_message()
called at line ssl/s3_clnt.c:851. Here a maximum message size is defined:
n=ssl3_get_message(s,
SSL3_ST_CR_KEY_EXCH_A,
SSL3_ST_CR_KEY_EXCH_B,
-1,
1024*8, /* ?? */
&ok);
As 9920>8192, the error condition is triggered.
So by now we know what we are looking for in the next step: is the message
size of 9920 reasonable? In this case the 1024*8 is unreasonable and a
client error. Or is the server in error in sending this large message size?
I must think and read about it, so I hand this analysis to the list,
maybe somebody else also can also add his $0.02.
Best regards,
Lutz
--
Lutz Jaenicke [EMAIL PROTECTED]
BTU Cottbus http://www.aet.TU-Cottbus.DE/personen/jaenicke/
Lehrstuhl Allgemeine Elektrotechnik Tel. +49 355 69-4129
Universitaetsplatz 3-4, D-03044 Cottbus Fax. +49 355 69-4153
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]