Code Execution Halts

[mahendrickson-maker]

Describe the bug?

Using iOS version 2.1.0

I use Okta to open a browser that sends an email to authenticate. The email contains a link allowing the user to authenticate. Once that authentication is complete the user is redirected back into the app however code execution does not continue where it left off at line 113 of BrowserSignIn.swift. I am still able to interact with the app and cancel out of the web browser but because this particular thread never continued a loss of app functionality has occured.

What is expected to happen?

code execution continues, allowing app to dismiss the browser and continue as normal.

What is the actual behavior?

User is redirected into the app where they left off, nothing happens.

Reproduction Steps?

This happens within a specific application and will not be reproducible to Okta devs. However, I can provide a demo to any devs that wish to see one.

Additional Information?

No response

SDK Version(s)

2.1.0

Build Information

No response

[mahendrickson-maker]

@AlexNachbaur

[mahendrickson-maker]

@oleggnidets-okta @chrisdmills-okta @rajdeepnanua-okta friendly bump up to this :)

[AlexNachbaur]

Hello @mahendrickson-maker I apologize that I haven't responded yet and have kept you waiting. I can certainly look into this. Do you have a backtrace that shows the call tree to where the hang is occurring? And do you have any sample code that demonstrates the workflow you're using? (for example, if you're using async/await or completion blocks, etc).

[SwiftNativeDeveloper]

@AlexNachbaur

I might be seeing something similar to this in the BrowserSignIn target.

Our symptoms are the same, the Okta code freezes up, but we successfully get a token. There isn't much of a back trace to show but what I can share outside of this screenshot is that we have several services, actors, types, that are all wanting to do networking in their own context and asking to refresh the token if necessary.

So for illustration purposes we have 3 unique networking bits of code that are all trying to refresh the token at the same time.

When I pause execution in Xcode and inspect my threads (which I can't share full details on GitHub), I found 5+ threads that are all waiting for the semaphore to complete.

For reference we tried on both 2.1.2 and 2.1.3 with the same issue. If we revert our code to a 1.x version, the issue doesn't exist to the user.

Further debugging on my end, the isRefreshing property on the credential's token seems to be the first point of failure. If we comment out our code to ensure the token is refreshed if necessary our app runs correctly until our token expires.

[AlexNachbaur]

That's very helpful @SwiftNativeDeveloper, thank you. The comment about the Xcode 26 beta regression might be an indicator of what might be going on.

From the looks of things, you have some code that is checking isRefreshing while other portions of the code is performing a refresh or refreshIfNeeded?

[AlexNachbaur]

One more thing @SwiftNativeDeveloper, if you end up encountering this again, is there a chance you can give the full backtrace for this issue? I'm curious what the call tree is within the other threads, to see where the data race is occurring.

You should be able to pause the debugger, and run bt all in the debug console. If it contains sensitive information, you can pass the information through your support contact so I can help diagnose this better.

[SwiftNativeDeveloper]

@AlexNachbaur

It is reproducible. The team decided to downgrade to 1.x release even though it is retiring. Is there a way I can provide more details outside of this public issue where I'd be open to providing a little more detail about my product?

Edit, I'll ask around internally to find out who my support contact is.