This issue has come up a few time in the forum with no clear resolution, and I think it's time for TI to step up and offer a definitive solution.
The problem is that the recommended 420 KHZ resonator for this part is essentially obsolete - we were able to locate only one vendor, which has marked the resonator as not recommended for new designs.
It should be possible to drive the XTAL1 input of the part with a 420KHZ square wave, though signal levels are not documented.
After some analysis and empirical observation of device performance with the resonator attached, we've come up with the circuit below. But we need TI to either approve this design or recommend changes before committing to PCB design and production.