14-CURRENT Corrupting SPI Flash on Panic?

From: Daniel Morante <daniel_at_morante.net>
Date: Sun, 27 Feb 2022 08:51:38 UTC
Sometimes (not always) when there is a kernel panic the hardware will no 
longer POST after resetting.  I would have to power down the system AND 
remove all power cables.  Then after reconnecting power its back to a 
bootable state.  I first noticed this happening on 
14-CURRENT-3ede04c78c7.  Snapshots prior to that didn't exhibit this 
behavior. The most recent version I've experienced it on is 
14-CURRENT-45c23c2608e.  This is a ARM64 server (Gigabyte R281-T91).  At 
first I thought it could be bad memory.  I replaced the memory and saw 
no changes.  I sent the server back to the vendor for testing/repair.  
It came back as OK/passing all tests.

Here is the serial over LAN capture of the most recent kernel panic that 
"broke" my server (also attached):
http://venus.morante.net/downloads/unibia/screenshots/freebsd/R281-T91/14-CURRENT-45c23c2608e/panic1.txt

Rebooting the server is no longer possible, the UEFI firmware gets stuck 
here (also attached):
http://venus.morante.net/downloads/unibia/screenshots/freebsd/R281-T91/14-CURRENT-45c23c2608e/stuckboot.txt

This is what a "good" boot should start out like (also attached):
http://venus.morante.net/downloads/unibia/screenshots/freebsd/R281-T91/14-CURRENT-45c23c2608e/goodboot.txt

It seems as if the kernel panic is corrupting the firmware or 
overwriting memory somewhere (sorry I don't fully understand UEFI yet) 
with 1's