Add support for SCRAM-SHA-* #5

singpolyma · 2022-05-18T20:06:53Z

No -PLUS yet, no support for channel binding.
Also supports SCRAM-* for any OpenSSL-supported digest function with just one
line of code in the future.

No -PLUS yet, no support for channel binding. Also supports SCRAM-* for any OpenSSL-supported digest function with just one line of code in the future.

nevans · 2022-05-21T03:20:38Z

Thanks! This is great!

I haven't really given this project the attention I'd hoped. But I'm glad someone else sees the value in it. I'll take a closer look at all three PRs next week!

nevans

FYI, I merged most of your changes with a bunch of updates of my own in some ruby/net-imap branches. I've pushed one PR but I'll send in some others that correspond to all of your other PRs.

Adding in a pure-ruby saslprep has been the main thing holding me back from finishing all this. The other big thing holding me back is that I have some fundamental (but backwards-compatible) API changes I needed to make, so the SASL gem can more generally useful than the embedded net/imap API: standardizing properties across all mechanisms, and property callbacks. I wanted to finish than before I could merge your PRs. Actually, I've used your PRs to help guide the design choices.

Once I'm happy with the net-imap PRs, I'll merge all of the changes back into net-sasl. I'll keep your PRs open until then. :) Thanks!

nevans · 2022-05-22T18:39:38Z

net-sasl.gemspec

@@ -28,5 +28,6 @@ Gem::Specification.new do |spec|
  spec.require_paths = ["lib"]

  spec.add_dependency "digest"
+  spec.add_dependency "idn-ruby"


I'd prefer a pure-ruby saslprep, to avoid dependencies. I'm hoping to replace the SASL implementations in net/imap and net/pop and net/smtp, which are all bundled gems. Adding another dependency might prevent that from happening; especially a C dependency, which isn't ideal for for JRuby and TruffleRuby.

My plan is to allow a swap-able saslprep library, for performance. But that will need benchmarks.

nevans

FWIW, I wrote this up a while back, but never submitted the review! Oops. I recently finished a pure ruby SASLprep (I want some more test-cases before I'm happy with it), which generates Regexps directly from the stringprep RFC's tables.

I think I'll have time to address all of these comments before the end of the week!

nevans · 2022-07-19T00:33:06Z

lib/net/sasl/scram_authenticator.rb

+
+      # responds to the server's challenges
+      def process(challenge)
+        return "n,#{'a=' + @authzid if @authzid},#{initial_message}" if challenge.nil?


We can't rely on nil, because not every protocol supports initial response. E.g. IMAP4rev1 without the SASL-IR extension. Per the SCRAM RFC, it should just be an empty string in that case. But it's still better to just explicitly track the state, IMO. The first time the method is called, challenge will be ignored and the initial response will be sent.

nevans · 2022-07-19T00:42:00Z

lib/net/sasl/scram_authenticator.rb

+
+        bare = "c=biws,r=#{sparams['r']}"
+        salted_password = OpenSSL::KDF.pbkdf2_hmac(
+          IDN::Stringprep.with_profile(@password.encode("utf-8"), "SASLprep"),


I have a pure ruby stdlib implementation of SASLprep in mind for this but it isn't ready yet. nearly done, but not quite. The RFC does give us an official way to cheat: only allow ASCII strings. ;) So this could just guard raise unless password.ascii_only? and I'd replace it with my Net::SASL::SASLprep.saslprep when it's done.

nevans · 2022-07-19T00:44:02Z

lib/net/sasl/scram_authenticator.rb

+    protected
+
+      def initial_message
+        "n=#{@username},r=#{@cnonce}"


we should do some error checking or formatting on these values, here or in #initialize? eg "\0" or "," chars.

nevans · 2022-07-19T00:53:15Z

lib/net/sasl.rb

+    add_authenticator "SCRAM-SHA-224", ScramAuthenticator.for("SHA224")
+    add_authenticator "SCRAM-SHA-256", ScramAuthenticator.for("SHA256")
+    add_authenticator "SCRAM-SHA-384", ScramAuthenticator.for("SHA384")
+    add_authenticator "SCRAM-SHA-512", ScramAuthenticator.for("SHA512")


I like that it only takes one line of code. But, even though plugging in alternate hashes is fairly straightforward and predictable, I'd prefer to only include official SASL mechanisms in the default set. I think that the IETF drafts for 512 (or at least, the ones that I knew about) recently expired, although I haven't researched why.

For any that are left out, we can include documentation so that users can add it themselves, if they really need it sooner.

nevans · 2022-07-19T00:58:54Z

lib/net/sasl/scram_authenticator.rb

+      # This should generally be instantiated via Net::SASL.authenticator.
+      def initialize(username, password, authzid = nil, hash:, **options)
+        super
+        @hash = OpenSSL::Digest.new(hash)


I think I'd prefer to define hash on the subclasses. It loses the nice one liner for new SCRAM-* mechanisms, but it'll be simpler and only like... three lines.

class SCRAMAuthenticator def hash; @hash ||= OpenSSL::Digest.new(self.class::HASH) end end # Defined in RFC5802, etc etc etc class SCRAM256Authenticator < SCRAMAuthenticator HASH = "SHA256" end

The one-liner could be recreated with only a little bit of extra work. But I'm not sure it's worth it, especially because I prefer an explicitly named class constant to which we can add rdoc for each specific mechanism.

Neustradamus · 2022-10-11T23:15:56Z

@nevans: What is the status of this PR (net-sasl)?

cc: @hsbt, @shugo.

@singpolyma

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

nevans · 2023-09-16T19:04:07Z

I know it's taken me forever to finish this, but ruby/net-imap#172 should be merged shortly. After net-imap 0.4.0 is released (probably this week?), I'll copy all of those updates back here.

I still need to fix net-imap so it ensures the authenticator is #done?. As it is now, if the server returns OK before sending the final message, then we can't satisfy the following requirement from the RFC:

The client then authenticates the server by computing the
ServerSignature and comparing it to the value sent by the server. If
the two are different, the client MUST consider the authentication
exchange to be unsuccessful, and it might have to drop the
connection.

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

@singpolyma

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 New authenticators for any digest algorithms supported can be added by subclassing ScramAuthenticator and adding a DIGEST_NAME constant (and then registering with an Authenticators registry). Co-authored-by: Stephen Paul Weber <[email protected]>

Neustradamus · 2023-09-24T20:11:38Z

@nevans: Good job for your work in net-imap!

But it is not possible to have an official net-sasl?
Some projects or people need net-sasl but not net-imap...

cc: @singpolyma.

nevans · 2023-09-25T14:01:17Z

I'll update this gem with the latest net-imap implementation some time "soonish" (after net-imap v0.4.0 is released)

nevans · 2023-09-27T01:30:03Z

But it is not possible to have an official net-sasl?
Some projects or people need net-sasl but not net-imap...

That's the goal, but it would add a new bundled gem to ruby. I wanted to release the basic implementation in the context of one project and then port it to at least one other before I seriously advance that proposal.

I've actually been using a fork of net-smtp in production for several months and I'm hoping to finish PRs for net-smtp and net-pop this week. If the maintainers of those projects are happy with it, then maybe there's a reasonable chance it might make it into ruby 3.3. 🤞🏻

Add support for SCRAM-SHA-*

270e232

No -PLUS yet, no support for channel binding. Also supports SCRAM-* for any OpenSSL-supported digest function with just one line of code in the future.

nevans reviewed Jul 16, 2022

View reviewed changes

nevans requested changes Aug 1, 2022

View reviewed changes

nevans added a commit to ruby/net-imap that referenced this pull request Sep 7, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

6a8ca1e

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 10, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

a311916

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 10, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

2b27764

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans mentioned this pull request Sep 10, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms ruby/net-imap#172

Merged

nevans added a commit to ruby/net-imap that referenced this pull request Sep 11, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

f807d33

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 12, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

3bd0f5e

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 12, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

0c7ec36

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 12, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

ed2e1b8

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 13, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

8764a3a

Based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans added a commit to ruby/net-imap that referenced this pull request Sep 15, 2023

🔒 Add SASL SCRAM-SHA-* mechanisms

317c619

Loosely based on the implementation by @singpolyma at nevans/net-sasl#5 Co-authored-by: Stephen Paul Weber <[email protected]>

nevans mentioned this pull request Oct 26, 2023

Import latest net-imap (~> 0.4) #6

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add support for SCRAM-SHA-* #5

Add support for SCRAM-SHA-* #5

singpolyma commented May 18, 2022

nevans commented May 21, 2022

nevans left a comment •

edited

Loading

nevans May 22, 2022

nevans left a comment •

edited

Loading

nevans Jul 19, 2022

nevans Jul 19, 2022

nevans Jul 19, 2022

nevans Jul 19, 2022 •

edited

Loading

nevans Jul 19, 2022 •

edited

Loading

Neustradamus commented Oct 11, 2022

nevans commented Sep 16, 2023 •

edited

Loading

Neustradamus commented Sep 24, 2023

nevans commented Sep 25, 2023

nevans commented Sep 27, 2023 •

edited

Loading

Add support for SCRAM-SHA-* #5

Are you sure you want to change the base?

Add support for SCRAM-SHA-* #5

Conversation

singpolyma commented May 18, 2022

nevans commented May 21, 2022

nevans left a comment • edited Loading

Choose a reason for hiding this comment

nevans May 22, 2022

Choose a reason for hiding this comment

nevans left a comment • edited Loading

Choose a reason for hiding this comment

nevans Jul 19, 2022

Choose a reason for hiding this comment

nevans Jul 19, 2022

Choose a reason for hiding this comment

nevans Jul 19, 2022

Choose a reason for hiding this comment

nevans Jul 19, 2022 • edited Loading

Choose a reason for hiding this comment

nevans Jul 19, 2022 • edited Loading

Choose a reason for hiding this comment

Neustradamus commented Oct 11, 2022

nevans commented Sep 16, 2023 • edited Loading

Neustradamus commented Sep 24, 2023

nevans commented Sep 25, 2023

nevans commented Sep 27, 2023 • edited Loading

nevans left a comment •

edited

Loading

nevans left a comment •

edited

Loading

nevans Jul 19, 2022 •

edited

Loading

nevans Jul 19, 2022 •

edited

Loading

nevans commented Sep 16, 2023 •

edited

Loading

nevans commented Sep 27, 2023 •

edited

Loading