Considering this purely from the P25 protocol perspective, all currently active group voice grants and group voice updates are signaled on the control channel and active voice channels. That design ensures that a subscriber radio always knows what is going on and whether they need to temporarily switch tgids to receive a higher priority call or something from their scanlist. The specific messages by which this information is conveyed varies depending whether FDMA or TDMA voice is in progress, but the bulk of the same information is present in either.
OP25 (and presumably also DSD+) use these in-call messages to maintain a complete picture of what is going on across the whole RFSS and can therefore switch around received calls depending on user-defined tgid priority. SDRTrunk uses a different methodology and likes to be configured so that the entire RFSS spectrum is captured through one or more concurrent sdr devices so that there is always a decoder listening to the control channel. OP25 multi_rx can do something very similar, but in practice it doesn't net you much benefit other than filling in the short gaps that would otherwise occur during retuning.