browser: get session only once - trying to fix #4125

[mstoykov]

What?

Use the session we get a bit earlier in attachIFrameToTarget instead of regetting it there.

Why?

This should fix #4125 as we already check the session isn't nil at the time.

It is still strange it suddenly disappears, and my expectation is that it is likely something to do with k6 stopping and session going away.

Also reuse the session we get in other cases within onAttachedToTarget as we already have it.

Checklist

• I have performed a self-review of my code.
• I have added tests for my changes.
• I have run linter locally (make lint) and all checks pass.
• I have run tests locally (make tests) and all tests pass.
• I have commented on my code, particularly in hard-to-understand areas.

Related PR(s)/Issue(s)

[mstoykov]

@inancgumus you are probably the person who can currently say if this is good idea at all.

Unfortunately I do not have a reproducible example that we can test with.

[inancgumus]

@inancgumus you are probably the person who can currently say if this is good idea at all.

Unfortunately I do not have a reproducible example that we can test with.

This change may solve the nil-pointer issue 🤞 However, the racy conditions will remain. Sessions in Connection are protected by a mutex. We get the session by locking and unlocking the mutex. After getting the session, while onAttachedToTarget is working, in the meantime, Connection may delete the session. Communicating over that Session can result in unexpected errors (because it may be gone from the browser and Connection side). The session management in the browser should definitely be improved.