At low temperature we systematically ran into this issue (Advisory 3.0.24) documented on p. 23 of the silicon errata of the OMAP L137 which we simply fixed for now by asserting the reset of the USB0 PHY for 500ms every minute. We are now looking into a more efficient way to implement the workaround into the Linux driver but it’s not clear what would be the proper parameters mainly:
- Width of the reset pulse. We tested 1ms as coded in the errata on p. 24. That wasn’t enough. 500ms (as commented) did the trick. The technical reference manual says a few clock cycles (p. 1428) but that’s the only specification we could find.
- For how long shall we wait after asserting the reset before checking again for the condition