During a problem where the system goes into recovery mode and back, the EC gets rebooted multiple times. With our current EC console logging infrastructure, all we see in the feedback report is the time before the problem happened and right after that the "UART initialized after sysjump" from when it's all done and the EC switches back to EC-RW on the next normal boot. The interesting part in between is lost because the AP didn't retrieve it while it was happening and the EC throws it away on reboot/sysjump.

Ideally, it would be very cool if we could preserve that somehow. Shawn tells me this is hard because the console buffer is just a part of the .data section and moves around between image versions. I guess if you moved it to its own special place with a linker script that could work? (I also don't know if EC RAM get's preserved across reboots/sysjumps...?)