RE: Time-stamp issue

[Michael Zolotarev <[e-mail]>]

Prashant,
 
A signature may also contain the CMS signingTime attribute, which is defined
as an authenticated attribute. if a signer adds the SigningTime attribute to
the signature, and then attaches a timestamp, the time in the timestamp must
be close enough to the time specified by the attribute. All assuming that a
timestamp is obtained soon after signing, of course.
 
Therefore, replacing a 'correct' timestamp with the one generated later in
the future, will be easily detectable.
 
But in general, I dont' think that the threat is there at all:
 
A timestamp can be attached to the signature by either the signer or by a
verifier. Why would the signer want to twick his own signature? I dont'
think he would. Now, the verifier: the verifier attaches a timestamp as an
evidence of his own verification process - "I have obtained and verified
that signature at that time, and the signature was valid". What would be a
reason for the verifier to play with his own evidence? I don't see any
practical reason for it.
 
Regards
Michael

-----Original Message-----
From: Prashant Dambe [mailto:[e-mail]]
Sent: Thursday, 15 March 2001 20:26
To: [e-mail]
Subject: Time-stamp issue


As specified in the draft-ietf-pkix-time-stamp-13.txt.
APPENDIX A - Signature Timestamp attribute using CMS
One of the major use of time stamping is to time stamp a digital signature
to prove that the digital signature was created before 
a given time. Should the corresponding public key certificate be revoked
this allows to know whether the signature was created before or after ther
evocation date.
A sensible place to store a time stamp is in a [CMS] structure as an
unsigned attribute.
 
But what happens in the following scenario.
As timestamp token is placed as unsigned attribute, one of the possible
attack is that 
if Time-stamp token it self is replaced with the Time-stamp token of the
same signature value
inside CMS  i.e If the same signature is time-stamped after some later time
and Time-stamp in the 
original CMS is replaced,it not possibled to detect that orignal time-stamp
has been replaced.
So putting time-stamp as unsigned attribute not works fine in all cases.
 
Thanks
Prashant Dambe.


This footnote confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.