Time series - stationarity & trend models

Stationarity assumption is a must for regression models.
Linear and log-liner models assume trending.
Trending violates the stationarity assumption.

Can someone help me understand how come linear and log-linear models still work despite the above three statements.