Help: p-channel MOSFET won't turn off

[Siwastaja]

With drain (aka Vin of regulator) shorted to ground, and still voltage measured at Vout of regulator, proves the current is coming somewhere from the output side. It's just passing through the regulator to its input in reverse direction because there is internal diode structure from out to in inside the regulator.

[rea5245]

OK, thanks to all your input, I've found the problem.

Power was, indeed, leaking back through the 10K pullups on the I2C pins. When I turned on the radio, code deep down in the library configured those pins to be I2C instead of GPIO. When I turned the radio off, those pins remained as I2C. Thus, my setting the GPIO to low outputs had no effect.

Thank you all for your advice!