Yes, certain - PD7 has an internal pullup, but PD1 floats, which is never a good thing, even on pins that aren't connected to debug circuitry.
Had 3 units that reliably would reset after 10-20 seconds running. Grounding PD1 fixed them reliably.
Enabling pullup/down didn't actually enable pullup/down until I set bit 26 of the AFIO->PCFR1 register.
The SPI port was routed to a header alongside PD1, and it always reset at some random time within a SPI access, so I suspect it was noise coupling.