Horton

Re: [PoC] Federated Authn/z with OAUTHBEARER

Jacob Champion <jacob.champion@enterprisedb.com>

View thread
From: Jacob Champion <jacob.champion@enterprisedb.com>
To: PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Cc: Daniel Gustafsson <daniel@yesql.se>, Peter Eisentraut <peter@eisentraut.org>, Antonin Houska <ah@cybertec.at>, Kashif Zeeshan <kashi.zeeshan@gmail.com>
Date: 2025-01-17T19:02:15Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. meson: Fix install-quiet after clean

  2. oauth: Run Autoconf tests with correct compiler flags

  3. Link libpq with libdl if the platform needs that.

  4. Doc: correct spelling of meson switch.

  5. oauth: Correct SSL dependency for libpq-oauth.a

  6. oauth: Fix Autoconf build on macOS

  7. oauth: Move the builtin flow into a separate module

  8. Remove a stray "pgrminclude" annotation

  9. oauth: Simplify copy of PGoauthBearerRequest

  10. oauth: Improve validator docs on interruptibility

  11. oauth: Disallow synchronous DNS in libcurl

  12. oauth: Fix postcondition for set_timer on macOS

  13. oauth: Use IPv4-only issuer in oauth_validator tests

  14. Work around OAuth/EVFILT_TIMER quirk on NetBSD.

  15. oauth: Fix incorrect const markers in struct

  16. Add missing entry to oauth_validator test .gitignore

  17. cirrus: Temporarily fix libcurl link error

  18. Add support for OAUTHBEARER SASL mechanism

  19. libpq: Handle asynchronous actions during SASL

  20. require_auth: prepare for multiple SASL mechanisms

  21. Move PG_MAX_AUTH_TOKEN_LENGTH to libpq/auth.h

  22. Make SASL max message length configurable

  23. jsonapi: fully initialize dummy lexer

  24. common/jsonapi: support libpq as a client

  25. Remove fe_memutils from libpgcommon_shlib

  26. Revert ECPG's use of pnstrdup()

  27. Explicitly require password for SCRAM exchange

  28. Refactor SASL exchange to return tri-state status

Attachments

On Mon, Jan 13, 2025 at 5:00 PM Jacob Champion
<jacob.champion@enterprisedb.com> wrote:
> So I think what I'm going to need to do is modify v41-0003 to allow
> the mechanism to politely hang up the connection while the flow is in
> progress.

This is done in v42. The rough conversation now looks like this:

    Issuer    User    libpq   Backend
      |        |        |
      |        x -----> x -----> o      [1] Startup Packet
      |        |        |        |
      |        |        x <----- x      [2] OAUTHBEARER Request
      |        |        |        |
      |        |        x -----> x      [3] Parameter Discovery
      |        |        |        |
      |        |        x <----- x      [4] Parameters Stored
      |        |        |        |
      |        |        x -----> x      [5] Finish OAUTHBEARER
      x <----> x <----> x        |      [6] OAuth Handshake
      x <----> x <----> x <----- o      [7] Server Hangs Up
      x <----> x <----> x
      |        |        |
      |        |        x -----> o      [8] New Startup Packet
      |        |        |        |
      |        |        x <----- x      [9] OAUTHBEARER Request
      |        |        |        |
      o        |        x -----> x      [10] Send Token
               |        |        |
               | <----- x <----- x      [11] Connection Established
               |        |        |
               x <----> x <----> x
               x <----> x <----> x      [11] Use the DB
               .        .        .
               .        .        .
               .        .        .

The key change is that the client sends the final OAUTHBEARER response
_before_ beginning the OAuth flow, allowing the server to concurrently
close its side of the discovery connection (steps 5-7). This requires
only a single change to the proposed SASL_ASYNC feature in v41-0003:
when a mechanism returns SASL_ASYNC during exchange(), it may also
specify an output response packet to send before switching over to
async auth. (This is similar to how mechanisms may send a response
packet when returning SASL_FAILED.)

That change simplifies the description of the flow a bit, too, and
I've updated the documentation. If an OAuth flow needs to run, two
connections to the server will be made. The only way to skip the
discovery connection now is if a (custom) hook has a token cached.

> This further decouples the lifetimes of the mechanism and
> the async auth -- the async state now has to live outside of the SASL
> exchange --

The only part of the state that I had to move was the token itself,
which now lives in conn->oauth_token. This is cleaned up with a new
pqClear- function, so that it can live across connection attempts and
be proactively cleared from memory after a successful connection.

> but I think it's probably more architecturally sound.

As evidence, this change flushed out a few bugs and provided the basis
to fix every TODO in fe-auth-oauth.c, so I'm pretty happy with it:
- an AuthenticationSASLFinal message is not allowed, as OAUTHBEARER
does not specify any additional server data (this bug goes all the way
back to v1)
- a server is not allowed to switch discovery URLs on a client between
connection attempts
- a server is not allowed to override a previously determined oauth_scope
- the connection is retried only if a conn->oauth_token was not
initially set (this simplifies conn->oauth_want_retry considerably)
- require_auth=oauth will complain if a discovery connection lets the client in

The existing Perl tests were not affected by this refactoring, other
than a latent test bug that got caught with the fallout. The advisory
Python tests (which pin behavior on the wire) needed more changes.
I've also added some tests to 002_client.pl which have the custom hook
misbehave in various ways and pin the expected error messages.

v41-0005 has probably outlived its usefulness by now, and I've folded
those changes into v42-0004.

Thanks!
--Jacob