A new RT-AX86U image has been uploaded. As per Asus' instructions, this build does not contain the cferom component, which is what was causing the crash on certain routers versions.
Also, to be on the safe side if there are people compiling their own image, I have disabled the RT-AX86U build profile on github. People who still want to build their own RT-AX86U must do the following:
1) edit target.mak, and remove the "-disable" portion of the RT-AX86U model name
2) When flashing, flash the image that does not contain _cferom_ in the filename
This is only for the RT-AX86U (and it's variant the RT-AX86S).
Asus are still testing a permanent fix on their end.
As a side note, the RT-AX86U (and RT-AX68U) must now be built from the src-rt-5.02L.07p2axhnd/ SDK folder. I haven't received confirmation whether this was a permanent change, or if it will eventually be merged back in on top of the existing 5.02p1axhnd.675x SDK.
The Broadcom drivers have been crashing for at least 3 years. This is one on V 385.5.2. Do a google search and you will see all sorts of speculation to the cause. Time to abandon Bradcom based routers?
Jun 8 05:57:41 kernel: CPU: 1 PID: 3057 Comm: dcd Tainted: P O 4.1.52 #2
Jun 8 05:57:41 kernel: Hardware name: Broadcom-v8A (DT)
Jun 8 05:57:41 kernel: task: ffffffc02ccd2b00 ti: ffffffc029ac8000 task.ti: ffffffc029ac8000
Jun 8 05:57:41 kernel: PC is at 0x29c60
Jun 8 05:57:41 kernel: LR is at 0x29ee0
Jun 8 05:57:41 kernel: pc : [<0000000000029c60>] lr : [<0000000000029ee0>] pstate: 20070010
Jun 8 05:57:41 kernel: sp : 00000000ffbd7410
Jun 8 05:57:41 kernel: x12: 00000000000a211c
Jun 8 05:57:41 kernel: x11: 0000000000081bfc x10: 000000000000000b
Jun 8 05:57:41 kernel: x9 : 0000000000000003 x8 : 000000000000000b
Jun 8 05:57:41 kernel: x7 : 00000000f5b004bc x6 : 0000000000000036
Jun 8 05:57:41 kernel: x5 : 0000000000000093 x4 : 00000000f5b0051c
Jun 8 05:57:41 kernel: x3 : 0000000000000000 x2 : 000000000000000a
Jun 8 05:57:41 kernel: x1 : 000000000000000b x0 : 0000000000000000
Is Broadcom actually trying to fix this and made it worse? Possibly Asus is getting better at finding the failing module.