Hi.  Thanks for the quick fix to RT #1664.  I've run into two more
bugs in the CMS code.  I can provide isolated test cases if necessary,
but the code that's currently producing these is buried deep in the
guts of a Python extension module, so I'll start by just describing
the problems in the hope that this will all be obvious to Steve.

1) CMS_final() and CMS_verify() both appear to be ignoring an
   EVP_R_COMMAND_NOT_SUPPORTED error.  That is: both functions are
   giving me what I think are meant to be successful returns (non-zero
   return value for CMS_final(), return value greater than zero for
   CMS_verify()), but ERR_get_error() finds an unhandled error
   EVP_R_COMMAND_NOT_SUPPORTED that wasn't there before these calls.

   As far as I can tell, this error is harmless, other than confusing
   subsequent code.  At least, I haven't seen any obvious bad side
   effects of ignoring the error, but of course I might have missed
   something.

2) CMS_add0_crl() dumps core.  Due to the environment in which this is
   occuring I don't have a stack trace available, let me know if you
   need me to generate a test case that produces one.  I don't see any
   code that exercises this particular function, so I don't know if
   it's ever been tested.  If, in the process of fixing this, someone
   were to add CMS_add1_crl(), I wouldn't complain. :)

Please let me know if you need more details or have trouble
reproducing these problems.  I'll be at the RIPE 56 meeting next week
but will try to answer as promptly as circumstances allow.

Thanks for all the good work on this.

--Rob

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

Reply via email to