[cifs-protocol] FW: [REG:111061756137964] Encryption of the key for "netsh branchcache importkey and exportkey.

Christopher R. Hertel crh at samba.org
Thu Jul 7 13:43:32 MDT 2011

Edgar, et. al.,

My guess was incorrect.

Based upon my testing today, the extracted value is the "arbitrary length
binary string stored on the server".  If I use the extracted value as the
signing key, the result does not match the value that I get from the Windows
server against which I am comparing.

If, however, I use the SHA256 of the extracted value, my results do match
those of Windows.

As you are aware, the first 32 bytes of the decrypted plaintext are the
SHA256 of the extracted value, so both the "arbitrary length binary string
stored on the server" and the Server Secret are provided in the extracted
key blob.

Chris -)-----

Christopher R. Hertel wrote:
> One more note.
> The extracted Server Secret is 32 bytes in length and is a binary string
> (clearly neither Unicode nor OEM Charset).
> My guess, at this point, is that the extracted value (which validates
> correctly) is the Server Secret itself, and *not* the "arbitrary length
> binary string stored on the server" that is used to generate the Server Secret.
> In other words, the extracted value is yet another SHA256 hash.
> I have not yet tested this theory, and the documentation I have found so far
> is not clear on the subject.  Working on it...
>>From [MS-PCCRC, 1.1]:
>   server secret: A SHA-256 hash of an arbitrary length binary string stored
>                  on the server.
>>From the BranchCache help:
>   Usage: exportkey [outputfile=]<File Path> [passphrase]=<Pass Phrase>
>   Parameters:
>       Tag          Value
>       outputfile   - The directory path and name of the file to which the
>                      key should be exported
>       passphrase   - A passphrase required in order to import the key
>   Remarks: This command will export the key which the BranchCache service
>            uses to protect content information. The key can then be
>            imported on another machine by using the importkey command.
> Chris -)-----
> Edgar Olougouna wrote:
>> Chris,
>> Thanks for sharing the good news! The fact that you got the decryption working with two different AES implementations (e.g. Mcrypt and OpenSSL) shows that we nailed down most important details required for successful non-Windows implementation. Hopefully, this has value for testing inter-operability of your BranchCache implementation.
>> I will pass your notes to the product team so it can considered when documenting the algorithm details.
>> As always, feel free to contact us, should you need further assistance on the open specifications.
>> Regards,
>> Edgar

"Implementing CIFS - the Common Internet FileSystem" ISBN: 013047116X
Samba Team -- http://www.samba.org/     -)-----   Christopher R. Hertel
jCIFS Team -- http://jcifs.samba.org/   -)-----   ubiqx development, uninq.
ubiqx Team -- http://www.ubiqx.org/     -)-----   crh at ubiqx.mn.org
OnLineBook -- http://ubiqx.org/cifs/    -)-----   crh at ubiqx.org

More information about the cifs-protocol mailing list