wireshark decryption

[Stefan Metzmacher]

Hi Aurélien ,

> Stefan Metzmacher <metze at samba.org> writes:
>> I like the concept, but you need to handle keys of variable size.
> 
> I will look into it, thanks.
> 
>>> The session key is computed by the client, it's different than the
>>> "session key" generated in SMB1 ([2]) which is sent on the wire and not
>>> safe to use for anything cryptographic. I suspect the dissector was
>>> already trying to use this one but it's most likely wrong.
>>
>> No, it tries to use the session key based on the configured
>> NTLMSSP password or the kerberos key provided by a keytab.
> 
> Previously the session key used to derive the keys was taken from the
> local keytab (which I think is safe) but also directly from the wire in
> dissect_smb2_session_setup_request (ntlmssph->session_key) which is the
> part I have a problem with. If the session key is sent on the wire in
> clear then the encryption is no good.
>
> If we assume this is not flawed by design in the protocol, this session
> key serves a different purpose and the previous code is probably not
> right. Am I missing something?

ntlmssph->session_key is not the 32bit "session key" (which seems to be
a key into a session array instead of being a crypto key) used in SMB1,
it's the session key that resulted out of the ntlmssp exchange.

metze

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20170529/cb70573e/signature.sig>