Chaotic and eternal inflation are not necessarily exclusive terms -- they describe different aspects of the inflationary peroid. Saying that inflation is "chaotic" is to say something about the initial conditions of the inflaton field. In Linde's original chaotic inflation model based on the [itex]m^2\phi^2[/itex] potential, the field takes on a wide range of values throughout the universe. This is to be contrasted with 'new', or 'hilltop' inflation in which the field begins inflation fairly localized at the local maximum of the potential. Linde described this property as "chaotic".
Meanwhile, the term "eternal" describes the global properties of the inflationary era. Of course inflation was not eternal in our observable universe, because we stopped inflating. But it's entirely possible that inflation is ongoing somewhere else in the universe -- outside our Hubble sphere. Linde's chaotic model is in fact eternal, in the sense that one finds that there are regions of the universe that are always still inflating (depending on the field vev, [itex]\langle \phi \rangle[/itex], quantum fluctuations are large enough to work against the classical slow roll motion towards [itex]V=0[/itex], so that certain regions of the universe always have [itex]\langle \phi \rangle > a\, few\, M_{Pl}[/itex] so that inflation keeps going.)
But there are also regions of the universe where the [itex]m^2\phi^2[/itex] finds its way down close enough to [itex]V=0[/itex] to end inflation. At that time, the field oscillates about the minimum of V, giving the time dependence [itex]\phi(t) \propto \sin(mt)[/itex] quoted by the OP.
So chaotic inflation is eternal, but that's not inconsistent with the fact that we live in a region of the universe in which inflation ended.