Troubleshoot Model Coverage for Conditional Transitions in Stateflow Charts

If you generate coverage results for a model that contains a Stateflow^® chart, and the chart contains a junction that has a conditional transition, the model coverage results may not include complete coverage data about all execution paths. Consequently, if you generate a code coverage report, the model and code coverage reports may not match. To collect complete coverage data for the conditional transitions in the chart, you can remove the junctions or guard the conditional transition by using the in operator.

Issue

Consider a Stateflow chart that contains three states and one junction. States A and B transition to a junction, and the transition exiting the junction has a condition, c that transitions to state C.

When you analyze model coverage for this chart, the transition condition c reports decision coverage. The decision coverage for the transition provides no information about whether the transition originates from state A or state B. The report contains model coverage data only about whether decision c is true or false.

When you simulate the same chart in software-in-the-loop (SIL) mode, Embedded Coder^® creates a switch case block to define which state is currently active, and two of if statements inside the case blocks that capture the logic of the transition. In this example, the code generator generates this code for the junction:

case junct_cov_discrep_IN_A:
      if (junct_cov_discrep_U.In1 != 0) 

case junct_cov_discrep_IN_B:
      if (junct_cov_discrep_U.In1 != 0)

Because the if statements are inside the case blocks that define the active state, the code coverage for the generated code contains the full information about the transition. When you analyze code coverage for the generated code, Simulink^® Coverage™ reports a decision for each if statement, with a total of four decision outcomes.

Possible Solutions

To collect complete model coverage data for Stateflow charts that contain junctions with conditional transitions, you can remove the junctions, or guard the conditional transition by using the in operator.

For simple charts, remove the junctions. If removing the junctions is difficult to implement, or makes the chart less readable, use the in operator instead.

Remove Conditional Junctions with Multiple Source States

To ensure that the model coverage captures all condition coverage information, avoid using junctions that can be reached from multiple source states and have a condition on the exit junction. In this example, replace the junction with transitions from state A and B to state C, and add the condition c on both transitions to state C.

The coverage results now provide information about the full execution path because there are two separate conditional transitions.

Guard the Transition Logic by Using the in Operator

Alternatively, use the in (Stateflow) operator to duplicate the logic in the transition condition. In this example, replace the transition condition c with in(A) && c || in(B) && c.

The coverage results now provide information about the full execution path of the transitions before and after the junction.

In this case, the transition also reports modified condition decision coverage (MCDC).