Just a small update on the ABE PLL lockup issue:
I've "installed" R195 (read: I've installed a solder blob there) on the
uevm to make it able to experience warm resets normally. I can now
reproduce the abe pll lockup there, although it seems to happen quite
rarely whereas on the pyra it seems to happen on every reboot.
Interestingly, I've discovered that the ABE PLL is not locking up at warm
reset. A register dump in u-boot immediately prior to a lockup in linux
shows everything was still fine before control was handed over to linux. It
seems linux is somehow managing to get it locked-up while trying to get it
locked, and once this happens the lockup is insensitive to warm reset.
Still lots of unanswered questions, but it's a step