Disable RLS for a measure

[markpatton]

Is there anyway to disable RLS for a measure. I have active (B) and inactive (A) relationships  in my model. In order to make a caclculation(measure) I have a measure where I deactivate a relationship (B) and activate another (A) using crossfilter and userelationship respectively.

Private Flying - Trips - Owner Other = (calculate(sum('GL - GENERAL LEDGER DETAILS'[Line Amount]), crossfilter('RLS-Plane ID'[registration],'Trip filters'[Plane], none), USERELATIONSHIP('RLS-Plane ID'[registration], 'GL - Customers'[CustomerPlane]), 'GL - Customers'[Customer Type] = "Owner", 'public financials'[135_91] = "135", 'GL - Account'[Account Number] = "11000") - calculate(sum('GL - GENERAL LEDGER DETAILS'[Line Amount]), crossfilter('RLS-Plane ID'[registration],'Trip filters'[Plane], none), filter('GL - GENERAL LEDGER DETAILS', related('GL - Customers'[CustomerPlane]) = related('Trip filters'[Plane])), 'GL - Customers'[Customer Type] = "Owner", 'public financials'[135_91] = "135", 'GL - Account'[Account Number] = "11000")) * -1

This works fine until I view with RLS. When using RLS I get a circular dependency from the GL table to the RLS Plane ID table 

This is my model. I have marked the relationships that create the two paths with A and B.

It appears to me that the path of B is not disabled which I thought would be the case with the crossfilter statement in the measure. 

Any suggestions on how this can be done?

[lbendlin]

Are the security filters applied in both directions between the General Ledger table and the dimensions pointing at it?  Why the bidirectional joins there?

[markpatton]

Thanks for the response. I have bi-directional joins to look up data in both directions. But I still get this error when I using relationships in the single direction. I change the model to test. 

This is an example of my relationship to GL Leger. Since I'm using 1 to many the box for security filter in both directions does not show up.