Maybe this will help.
The MOTOR_EMF magnitude depends on the speed of the motor. The polarity of the EMF depends on the direction of rotation. We will say forward rotation causes positive EMF. The EMF ONLY depends on the motor rotation direction and speed.
The EMF is NOT the same as the V_EFFECTIVE. V_EFFECTIVE is controlled by the H-bridge and may also be either positive or negative independent of the EMF. MOTOR_R and MOTOR_L are motor parameters. The polarity of V_EFFECTIVE is determined by which elements of the H-bridge are active. The magnitude of V_EFFECTIVE is basically D * V_LINK where D is duty cycle and V_LINK is the DC-link voltage.
Motor torque can also be either positive or negative. The steady state torque is proportional to the current. Let's say positive current gives forward torque. In normal forward motoring operation, V_EFFECTIVE will be positive and higher than the EMF. Therefore current will flow into the motor and torque applied by the motor will be forward. A mechanical load of some sort must be connected to the motor to resist the forward rotation and absorb the power.
Normal forward regen would occur when the motor is spinning forward, and an external force is trying to make the motor spin even faster. In this case EMF and V_EFFECTIVE are the same polarity, but V_EFFECTIVE is slightly lower than the EMF, and so current flows out of the motor (the motor is a generator).
I guess what you are considering is what will happen if you command a positive torque, but an external force causes the motor to spin in reverse direction. This is not a problem conceptually, but it is a case of regen, because current will be flowing out of the motor. If your H-bridge DC-link is not capable of dissipating power, then the voltage will rise and could ultimately destroy the bridge.
You can add some kind of braking resistor to the DC-link, or you can detect when the voltage is too high and disable torque.
simulate this circuit – Schematic created using CircuitLab