Created attachment 199989 [details]
Screenshot of panic.
I'm testing 12.0-RC2 on AWS, and getting double-fault panics. We've observed it at least 3 times, with a stop/start between each, and on two hosts (the only hosts we have on 12). A stop/start should move it to new host hardware, eliminating the plausibility of bad hardware. The last release that has been stable for us is 11.1. (11.2 did not work on AWS due to the crash bug in ena.) This happens on an r4.large instance (r4.large is large enough to use the ena NIC driver rather than xn, fwiw).
Screenshot from AWS attached.
Created attachment 200351 [details]
This problem persists in 12.0-RELEASE.
Often upon rebooting after a panic, it will panic again when rc has barely begun. See new screen shot.