Imagine a small capacitor between the gate and source of the P-MOSFET. It is the charge on this capacitor (which is also a measure of its voltage) which determines the operation of the transistor.
You know that with a discharge path a charged capacitor can retain its charge for a long time. The charge can also leak away through the capacitor itself. To see this gate capacitance in action, set up this circuit:
When you press the switch the LED will light up. When you release the switch the LED will still remain lit. In fact, it will remain lit for a long time -- minutes, perhaps even hours. The reason is that pressing the switch charges up the gate capacitor and that charge remains on the gate even after the switch is released because these is no discharge path. There is some gate-source leakage, so the cap will eventually become discharged, but it could take a long time.
A floating gate can also be influenced by external electric fields. A good demonstration of this is any of the various contactless mains voltage detector circuits which use a CMOS logic chip such as:
(Source: https://www.eleccircuit.com/non-contact-ac-detector-voltage-tester-using-cd4060/ )
Here the clock input of the CD4060 is left floating to pick up electric fields from the environment. The oscillating electric fields produced by mains power cord or outlet are enough to create a clock signal to the 4060 counter and thus toggle the LED. This is why @RohatKılıç says that a MOSFET can turn on accidently.
The above phenomena are why you frequently see a pull-up or pull-down resistor on the gates of MOSFETs.
So, in your power supply case perhaps the gate charge on the PMOSFET was such that the transistor operated normally but over time local electric fields changed its voltage and thus its operating point.
