Fix SMB2 compoud response signing

[jborean93]

Fix the signing the logic when responding with an SMB2 compount response. The signature will include the padding of each compound element and include the next offset value before signing the data.

The current logic was adding the signature but before it set the NextCommand entry and it did not include any of the padding added between each compound entry. By delaying the signing until after the compound entries are build we can ensure that the signatures are generated correctly so the client can verify them.

The padding data must be included in the signature as per MS-SMB2 https://learn.microsoft.com/en-us/openspecs/windows_protocols/ms-smb2/a3e9ea1e-53c8-4cff-94bd-d98fb20417c0

If the message is part of a compounded chain, any padding at the end of the message MUST be used in the hash computation

[gabrielg5]

Hey @jborean93, thanks for all these PRs!

Reading the code changes for this one, seems aligned now with the documentation.
I have only one note related to the alignment of the last message in the compounded chain. Shouldn't it be padded to a multiple of 8b as well? Currently it is not being done (neither it wasn't previously)

Will do some tests on this branch before moving forward!

Update: Documentation answered my question... it's advised but not mandatory (3.3.4.1.3 Sending Compounded Responses)

The length of the last response in the compounded responses SHOULD be padded to a multiple of 8 bytes.

[jborean93]

That's a good point, I'll update the code to always pad each compound message including the last one to align with the SHOULD in the docs.

[gabrielg5]

Awesome!
Thank you!!

[jborean93]

Thanks for the reviews on the various PRs.