Schwarzschild coordinates beyond the event horizon

Coordinates are not sacred objects in GR. Any coordinate system is just as good as any other coordinate system. So to ask whether the Schwarzschild coordinates are valid or not is a meaningless question ${}^1$.

However it is reasonable to ask if coordinates have an intuitive meaning for some specified observer. So for example if we take an observer far from the massive object then the Schwarzschild time coordinate is the time as measured by that observer's clock, and the Schwarzschild radial coordinate is the circumference of a circle centred on the object divided by $2\pi$. Both these are intuitively meaningful measurements for our observer.

The problem with the interior of the black hole is that for our external observer anything falling into the black hole takes an infinite time even to reach the event horizon, let alone pass through it, so that makes us sceptical about any physical meaning for the Schwarzschild time inside the event horizon. And indeed if we write the Schwarzschild metric:

$$ ds^2 = -\left(1-\frac{r_s}{r}\right)dt^2 + \frac{dr^2}{1-\frac{r_s}{r}}+r^2d\theta^2 + r^2\sin^2\theta d\phi^2 $$

We find that the sign of the $dt^2$ and $dr^2$ terms changes as we move through the horizon. Since the sign tells us whether a term in the metric is spacelike or timelike this means inside the horizon $t$ behaves like a spatial coordinate and $r$ like a time coordinate.

This doesn't mean anything freaky, like time turning into space and vice versa as the more lurid popular science articles would have you believe, it just means the coordinates don't have the intuitive meaning that we associate with them outside the event horizon. However as I said right at the outset, they remain perfectly good coordinates and we just have to be careful about interpreting them.

${}^1$ the coordinates are singular at the horizon, i.e. at $r=r_s$, so they are not useful exactly at the horizon.

I think the coordinates are valid inside the horizon as well, albeit inside the horizon $r$ is timelike and $t$ is spacelike. They are not valid at the horizon since both $r$ and $t$ have zero coefficients in the metric there. That means you need some other coordinate system there to connect the exterior and interior solutions.