node_crypto: Add support for AES-GCM modes

[KiNgMaR]

This adds support for the authenticated encryption mode GCM to node-crypto.

Before: using aes-128-gcm, aes-192-gcm aes-256-gcm with createCipheriv would work, however decryption with createDecipheriv would always throw in final.

With patch: aforementioned modes are now supported. Users have to use buf = getAuthTag() after Final, transmit the tag with the ciphertext, then use setAuthTag(buf) when decrypting. If the data has been tampered with or if no tag is provided during decryption, final will throw.

The exception thrown from final is currently generic. I can make it more verbose/clear if wanted.

Limitations: it's not possible to add any AAD (additional authenticated data) into the mix (or verify it). This could be achieved by adding another method to Cipheriv, e.g. setAdditionalAuthenticatedData(buf).

Other AEAD modes: The AES-CCM mode cannot be used with the current API because it does not support streaming (the total length of the plain / cipher text has to be known when the first update call is made). Unless limiting users to 1 call of update was ok, of course.
Composite ciphers such as AES-128-CBC-HMAC-SHA1 use a different OpenSSL API, which is rather poorly documented (and only used in OpenSSL's TLS implementation as far as I can tell), so supporting these is not currently in scope of this pull request either.

Thanks for considering.

[Nodejs-Jenkins]

Thank you for contributing this pull request! Here are a few pointers to make sure your submission will be considered for inclusion.

Commit KiNgMaR/node-1@bd44373656a78399f9461b0ac5624bb9a1bc1835 has the following error(s):

• First line of commit message must be no longer than 50 characters
• Commit message must indicate the subsystem this commit changes

You can fix all these things without opening another issue.

Please see CONTRIBUTING.md for more information

[rlidwka]

Unless limiting users to 1 call of update was ok, of course.

It's ok if it's documented and second update() call throws with a good error message. It's odd though that seemingly stream cipher don't support streaming.

can't say anything about the rest of PR due to lack of knowledge

[bnoordhuis]

Allocate the buffer first and copy the tag data directly into it? Saves an allocation.

[trevnorris]

Also note that Buffer uses malloc, so to make valgrind happy you'll want to free().

[trevnorris]

Actually, wait. Use:

Local<Object> buf = Buffer::Use(env, out, out_len);
args.GetReturnValue().Set(buf);

Unless there's some reason you can't hand over control of the memory to Buffer.

[bnoordhuis]

Calling @indutny.

[trevnorris]

Just had a lint rule put in for this. split it to two lines.

[KiNgMaR]

Hey everyone, I changed the things you commented on - except for the int -> unsigned int thing. This way, it's consistent with Update and Final.

Will add documentation update in the next few days.

[RomanMinkin]

👍 for this issue.

[esco]

👍

[trevnorris]

No reason to return here. No logic following the if(). Also, remove the brackets. It's a one liner and you're not using them above.

[wamoyo]

👍 for this issue.

[KiNgMaR]

Rebased, updated, tested -- let me know if there's something else.

[indutny]

final method.

[indutny]

Small style nits, otherwise LGTM.

[KiNgMaR]

Sorry about the delay, nits fixed up :-)

[indutny]

Thank you, landed in e0d31ea