PKCS#12 / PFX support

[jklong]

Thoughts on adding PKCS12/RFC7292 and (eventually) complete or near complete PFX support? It's a monster and it sucks but it's in wide use thanks to Windows' use of the format.

I'm happy to take ownership of the feature and start work but can't guarantee any timelines on it

[tarcieri]

This is definitely something we'd love to have, thanks for volunteering! We also happen to be the owners of the pkcs12 crate.

(Also FYI, we have a brand new "formats" repo, and I've just transferred the issue there)

It's a monster and it sucks

My main recommendation would be to keep the scope of the initial implementation small rather than try to provide a comprehensive implementation of the full specification.

I would suggest supporting a single private key and its associated certificate chain in an initial implementation.

Also I believe someone has already implemented the PKCS#12 KDF and wanted to upstream it to this project, but offhand I don't remember where I discussed that. I can attempt to track that work down.

[jklong]

My main recommendation would be to keep the scope of the initial implementation small rather than try to provide a comprehensive implementation of the full specification.

Absolutely. I was just gauging appetite overall before starting.

Also I believe someone has already implemented the PKCS#12 KDF and wanted to upstream it to this project, but offhand I don't remember where I discussed that. I can attempt to track that work down.

If you do have something that would be great.

[xemwebe]

Also I believe someone has already implemented the PKCS#12 KDF and wanted to upstream it to this project, but offhand I don't remember where I discussed that. I can attempt to track that work down.

I require exactly the PKCS#12 KDF to support communication with some legacy system. Is this already somewhere else implemented or still an open issue?

[tarcieri]

Unfortunately I wasn't able to find where I saw that. Perhaps it can be found with GitHub Code Search

[xemwebe]

I searched GitHub already, but without success. Meanwhile, I implemented the KDF in rust myself. It is not that complicated. If you think this is interesting enough, I could prepare a PR for that. Should it go here or would the KDF repository the better place for that.

[tarcieri]

Could go either way on that. Maybe start here and we can factor it out if it makes sense.

[xemwebe]

The POC for that can be found here: https://github.com/xemwebe/formats/tree/pkcs12_kdf/pkcs12
The code needs to be generalized for accepting arbitrary digest methods, and I probably need some time to understand how to fit this in the existing framework.

The KDF accepts in general passwords (which needs to be converted to unicode) and IVs of arbitrary lengths and returns a key of arbitrary length. This seems to require feature "alloc" to be enabled. Is this acceptable, or would you prefer some variant with predefined arrays of fixed length (returning an error if the size does not fit)?

[carl-wallace]

FYI, I am going to try to move #836 from the pkcs7 crate to the cms crate and generally align it with current. Can give a look at the KDF in context once that is done. It may be next week before I get to that though.

[xemwebe]

The KDF works now with various digest methods and I have prepared a PR #1154.
All test cases have been checked against openssl.
If you have any suggestions or improvements, please let me know.

[xemwebe]

I have posted an update: zeroizing intermediate secrets, removing the unused error type and add test case for very long output keys that check some code regions not covered in the other tests.

[xemwebe]

I believe to have addressed all reviewer's remarks regarding PR #1154 so far and will wait for further reviewer suggestions, if any.

[tarcieri]

We've merged #1154 and published a pkcs12 crate, so I'd say we can close this.