Refactor scrypt hashing and source app session mgmt pt 3/3 #5695

nabla-c0d3 · 2021-01-11T06:06:24Z

Status

Ready.

Description of Changes

This PR removes usage of flask.g and flask.session from the source app's test suite and all of the source app. It's a "cleanup" PR following Refactor scrypt hashing and source app session mgmt pt 2/3 #5694.
Update: based on the discussion in Refactor scrypt hashing and source app session mgmt pt 2/3 #5694, this PR also removes the local cache for source user sessions, in order to fix logout errors/bugs.

nabla-c0d3 · 2021-08-19T20:11:06Z

securedrop/tests/conftest.py

+        )
+        journalist_app.crypto_util.genkeypair(source_user)
+        source = source_user.get_db_record()
+        return {'source_user': source_user,


This makes the source_user easy to access in the tests

Fix tests

nabla-c0d3 · 2021-09-12T05:13:34Z

securedrop/sdconfig.py

-            pass
+        self.SESSION_EXPIRATION_MINUTES: int = getattr(
+            _config, "SESSION_EXPIRATION_MINUTES", 120
+        )


I moved the default value from the source app to here.

nabla-c0d3 · 2021-09-12T05:14:01Z

securedrop/source_app/__init__.py

-            timedelta(minutes=getattr(config,
-                                      'SESSION_EXPIRATION_MINUTES',
-                                      120))
-


This was moved to the SessionManager. I think this was the root cause of one of the bugs discussed by @zenmonkeykstop at the end of #5694 (but not 100% sure).

nabla-c0d3 · 2021-09-12T05:15:10Z

securedrop/source_app/decorators.py

@@ -14,10 +16,10 @@ def login_required(f: Callable) -> Callable:
    @wraps(f)
    def decorated_function(*args: Any, **kwargs: Any) -> Any:
        try:
-            logged_in_source = SessionManager.get_logged_in_user()
+            logged_in_source = SessionManager.get_logged_in_user(db_session=db.session)


In this PR, I have removed the local cache (due to the problems encountered in #5694) so a DB session now has to be passed to the SessionManager in order to fetch the source record (instead of fetching it from the now-removed cache).

nabla-c0d3 · 2021-09-12T05:16:15Z

securedrop/source_app/main.py

@@ -54,22 +55,30 @@ def generate() -> Union[str, werkzeug.Response]:
        codenames = session.get('codenames', {})
        codenames[tab_id] = codename
        session['codenames'] = fit_codenames_into_cookie(codenames)
+        session["codenames_expire"] = datetime.utcnow() + timedelta(
+            minutes=config.SESSION_EXPIRATION_MINUTES
+        )


I am not sure why the codenames from the /generate flow need to expire, but this brings the previous behavior back (so that they do expire).

I don't know if this was the original motivation or not, but one reason I can see to expire codenames is that folks may have left their terminal on the "generate codename" page, then returned to it at a later point, increasing the risk that a third party observed the codename.

nabla-c0d3 · 2021-09-12T05:19:50Z

securedrop/source_app/session_manager.py

@@ -27,80 +28,76 @@ class UserHasBeenDeleted(_InvalidUserSession):


 class SessionManager:
-    _CACHE = SimpleCache()  # A mapping of passphrase -> SourceUser


I have removed the local cache here to address session timeout bugs discussed in #5694.

nabla-c0d3 · 2021-09-12T05:20:46Z

securedrop/source_app/session_manager.py

-        )
+        # Save the session expiration date in the user's session cookie
+        session_duration = timedelta(minutes=config.SESSION_EXPIRATION_MINUTES)
+        session[cls._SESSION_COOKIE_KEY_FOR_EXPIRATION_DATE] = datetime.utcnow() + session_duration


I've put back the session expiration date in the session cookie; in my previous PR it was in the local cache.

nabla-c0d3 · 2021-09-12T05:22:19Z

securedrop/source_user.py

@@ -118,6 +122,8 @@ def __init__(
        self._scrypt_p = scrypt_p
        self._backend = default_backend()

+    # Use @lru_cache to not recompute the same values over and over for the same user
+    @lru_cache


This should help a lot with performance, by not having to recompute hashes on every authenticated request.

kushaldas

Worked well in my staging. I will need a second pair of eyes before merging.

eloquence · 2021-09-22T06:33:49Z

I've done some additional testing with packages built from this branch using the staging env, and so far have not encountered any issues, including by following the steps from #5694 (comment) . I'll do additional testing tomorrow 1) with multiple browser instances, 2) with a forced low session expiry setting.

eloquence · 2021-09-22T19:01:15Z

I've done additional testing of the following:

Being logged in as different users in multiple browser instances and logging out / logging back in again. I've observed no issues there.
Reducing session length expiry to 1 minute and repeatedly bouncing source and journalist sessions, including in multiple tabs. Observed no issues there.
Letting the session timer expire on the codename page and observing that I get kicked back to the index page. Observed no issues there.

Let me know if you'd like me to test anything else in particular, but so far, the session behavior looks pretty solid in my testing on staging. The behavior also matches what I see in my current 2.0.2 prod environment.

zenmonkeykstop · 2021-09-24T16:54:45Z

Also took a final spin through, via the prod upgrade scenario. Upgrade worked fine, testing with multiple sources worked fine, also loaded a bunch of test data, no issues that I could see.

conorsch

Also took this for a spin, and am pleased to say I can't reproduce any of the session-related problems I was experiencing on develop. Also made sure to follow the clear STR in #5694 (comment) , and everything worked well.

Thanks, @zenmonkeykstop, for all your careful review of these changes. And thanks to @nabla-c0d3 as well, for your collaboration throughout review!

nabla-c0d3 · 2021-09-25T16:40:32Z

Thanks everyone for taking the time to review and test these changes in depth 😀

nabla-c0d3 requested review from kushaldas and redshiftzero as code owners January 11, 2021 06:06

nabla-c0d3 marked this pull request as draft January 11, 2021 06:06

nabla-c0d3 mentioned this pull request Jan 12, 2021

Refactor scrypt hashing and source app session mgmt pt 1/3 #5692

Merged

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from d969087 to c209b4d Compare January 12, 2021 05:43

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch 2 times, most recently from fee40dd to 2de49c4 Compare August 17, 2021 20:48

eloquence added this to the 2.1.0 milestone Aug 18, 2021

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from 2de49c4 to 62fb13e Compare August 18, 2021 20:38

nabla-c0d3 commented Aug 19, 2021

View reviewed changes

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from 62fb13e to 9363869 Compare August 25, 2021 20:18

nabla-c0d3 marked this pull request as ready for review August 25, 2021 20:20

nabla-c0d3 requested a review from a team as a code owner August 25, 2021 20:20

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from 9363869 to 8e10517 Compare September 1, 2021 02:42

kushaldas self-assigned this Sep 7, 2021

Remove usage of flask.g in the source app and fix tests

5bba230

Fix tests

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from 8e10517 to 5bba230 Compare September 8, 2021 02:37

kushaldas mentioned this pull request Sep 9, 2021

Refactor scrypt hashing and source app session mgmt pt 2/3 #5694

Merged

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch 2 times, most recently from e94fa7a to c7bd749 Compare September 11, 2021 19:55

Revert to not caching the source user's details

362c5c6

nabla-c0d3 force-pushed the refactor-scrypt-hashing-pt-3 branch from c7bd749 to 362c5c6 Compare September 12, 2021 05:07

nabla-c0d3 commented Sep 12, 2021

View reviewed changes

kushaldas approved these changes Sep 15, 2021

View reviewed changes

conorsch self-requested a review September 24, 2021 18:52

conorsch approved these changes Sep 24, 2021

View reviewed changes

zenmonkeykstop merged commit 8b17904 into freedomofpress:develop Sep 24, 2021

nabla-c0d3 deleted the refactor-scrypt-hashing-pt-3 branch September 25, 2021 04:43

This was referenced Sep 25, 2021

Session expiry banner on the source index page #5716

Closed

Logged-out user is shown "You were logged out due to inactivity" and redirected to index page #5741

Closed

cfm mentioned this pull request Oct 8, 2021

Release SecureDrop 2.1.0 #6103

Closed

26 tasks

cfm mentioned this pull request Nov 14, 2021

Rebase securedrop@signal-proto-focal against latest develop freedomofpress/securedrop-signal-poc#27

Closed

11 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Refactor scrypt hashing and source app session mgmt pt 3/3 #5695

Refactor scrypt hashing and source app session mgmt pt 3/3 #5695

nabla-c0d3 commented Jan 11, 2021 •

edited

Loading

nabla-c0d3 Aug 19, 2021

nabla-c0d3 Sep 12, 2021

nabla-c0d3 Sep 12, 2021 •

edited

Loading

nabla-c0d3 Sep 12, 2021

nabla-c0d3 Sep 12, 2021

eloquence Sep 22, 2021

nabla-c0d3 Sep 12, 2021

nabla-c0d3 Sep 12, 2021

nabla-c0d3 Sep 12, 2021

kushaldas left a comment

eloquence commented Sep 22, 2021

eloquence commented Sep 22, 2021

zenmonkeykstop commented Sep 24, 2021

conorsch left a comment

nabla-c0d3 commented Sep 25, 2021

		@@ -27,80 +28,76 @@ class UserHasBeenDeleted(_InvalidUserSession):


		class SessionManager:
		_CACHE = SimpleCache() # A mapping of passphrase -> SourceUser

Refactor scrypt hashing and source app session mgmt pt 3/3 #5695

Refactor scrypt hashing and source app session mgmt pt 3/3 #5695

Conversation

nabla-c0d3 commented Jan 11, 2021 • edited Loading

Status

Description of Changes

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nabla-c0d3 Sep 12, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kushaldas left a comment

Choose a reason for hiding this comment

eloquence commented Sep 22, 2021

eloquence commented Sep 22, 2021

zenmonkeykstop commented Sep 24, 2021

conorsch left a comment

Choose a reason for hiding this comment

nabla-c0d3 commented Sep 25, 2021

nabla-c0d3 commented Jan 11, 2021 •

edited

Loading

nabla-c0d3 Sep 12, 2021 •

edited

Loading