Re: cvs commit: openssl/ssl s3_enc.c ssl.h ssl_err.c ssl_lib.c

[Richard Levitte - VMS Whacker]

From: [EMAIL PROTECTED]

bodo> bodo        08-Apr-2001 15:47:55
bodo> 
bodo>   Modified:    ssl      s3_enc.c ssl.h ssl_err.c ssl_lib.c
bodo>   Log:
bodo>   Avoid assert() in the library.
bodo>   
bodo>   Revision  Changes    Path
bodo>   1.22      +10 -7     openssl/ssl/s3_enc.c
bodo>   1.105     +1 -0      openssl/ssl/ssl.h
bodo>   1.38      +1 -0      openssl/ssl/ssl_err.c
bodo>   1.93      +3 -1      openssl/ssl/ssl_lib.c
bodo>   
bodo>   Index: s3_enc.c
bodo>   ===================================================================
bodo>   RCS file: /e/openssl/cvs/openssl/ssl/s3_enc.c,v
bodo>   retrieving revision 1.21
bodo>   retrieving revision 1.22
bodo>   diff -u -r1.21 -r1.22
bodo>   --- s3_enc.c    2001/04/08 05:41:42     1.21
bodo>   +++ s3_enc.c    2001/04/08 13:47:50     1.22
bodo>   @@ -57,7 +57,6 @@
bodo>     */
bodo>    
bodo>    #include <stdio.h>
bodo>   -#include <assert.h>
bodo>    #include <openssl/md5.h>
bodo>    #include <openssl/sha.h>
bodo>    #include <openssl/evp.h>
bodo>   @@ -82,7 +81,7 @@
bodo>    static int ssl3_handshake_mac(SSL *s, EVP_MD_CTX *in_ctx,
bodo>           const char *sender, int len, unsigned char *p);
bodo>    
bodo>   -static void ssl3_generate_key_block(SSL *s, unsigned char *km, int num)
bodo>   +static int ssl3_generate_key_block(SSL *s, unsigned char *km, int num)
bodo>           {
bodo>           MD5_CTX m5;
bodo>           SHA_CTX s1;
bodo>   @@ -97,9 +96,13 @@
bodo>           for (i=0; i<num; i+=MD5_DIGEST_LENGTH)
bodo>                   {
bodo>                   k++;
bodo>   -               /* If this assert is triggered, it means buf needs to be
bodo>   -                  resized.  This should never be triggered in a release. */
bodo>   -               assert(k <= sizeof(buf));
bodo>   +               if (k > sizeof buf)
bodo>   +                       {
bodo>   +                       /* bug: 'buf' is too small for this ciphersuite */
bodo>   +                       SSLerr(SSL_F_SSL3_GENERATE_KEY_BLOCK, 
ERR_R_INTERNAL_ERROR);
bodo>   +                       return 0;
bodo>   +                       }
bodo>   +               
bodo>                   for (j=0; j<k; j++)
bodo>                           buf[j]=c;
bodo>                   c++;

Hmm, my thinking with that assert is that it should be caught
*EARLY*.  If we release anything with a buf too small for any ciphers
and digests that we support, it's a library bug and should be caught
early.  The best way I saw for it to be caught early was to generate a
violent death, so we get it in our face.  I'm not sure I understand
the error in that thinking.  Could that assert be triggered in some
other way, or whall we leave it to windows users to discover such a
bug? :-)

-- 
Richard Levitte   \ Spannvägen 38, II \ [EMAIL PROTECTED]
Chairman@Stacken   \ S-168 35  BROMMA  \ T: +46-8-26 52 47
Redakteur@Stacken   \      SWEDEN       \ or +46-709-50 36 10
Procurator Odiosus Ex Infernis                -- [EMAIL PROTECTED]
Member of the OpenSSL development team: http://www.openssl.org/
Software Engineer, Celo Communications: http://www.celocom.com/

Unsolicited commercial email is subject to an archival fee of $400.
See <http://www.stacken.kth.se/~levitte/mail/> for more info.

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]