IPv6 UDP panic on CURRENT

[Mark Felder]

I was having nightly crashes for a while but my watchdog was resetting the server. Finally disabled so I could see the screen. It was happening when my acme (Let's Encrypt) script was running to generate new certificates and uses DNS validation which hits a nameserver hosted on the same box and accessible over IPv6.

Fatal trap 12: page fault while in kernel mode
cpuid = 11; apic id = 14
fault virtual address	= 0xcd
fault code		= supervisor read data, page not present
instruction pointer	= 0x20:0xffffffff80e1dc46
stack pointer	        = 0x28:0xfffffe00005ab760
frame pointer	        = 0x28:0xfffffe00005ab800
code segment		= base 0x0, limit 0xfffff, type 0x1b
			= DPL 0, pres 1, long 1, def32 0, gran 1
processor eflags	= interrupt enabled, resume, IOPL = 0
current process		= 12 (swi1: netisr 0)
[ thread pid 12 tid 100086 ]
Stopped at      ip6_savecontrol_v4+0x26:        testb   $0x4,ll+0xac(%rax)
db> bt 
Tracing pid 12 tid 100086 td 0xfffff8000381c580
ip6_savecontrol_v4() at ip6_savecontrol_v4+0x26/frame 0xfffffe00005ab800
ip6_savecontrol() at ip6_savecontrol+0x2c/frame 0xfffffe00005ab850
udp6_append() at udp6_append+0xe7/frame 0xfffffe00005ab8a0
udp6_input() at udp6_input+0x4f8/frame 0xfffffe00005ab9c0
ip6_input() at ip6_input+0xdd8/frame 0xfffffe00005abab0
swi_net() at swi_net+0x17b/frame 0xfffffe00005abb20
intr_event_execute_handlers() at intr_event_execute_handlers+0xe9/frame 0xfffffe00005abb60
ithread_loop() at ithread_loop+0xe7/frame 0xfffffe00005abbb0
fork_exit() at fork_exit+0x83/frame 0xfffffe00005abbf0
fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe00005abbf0
--- trap 0, rip = 0, rsp = 0, rbp = 0 ---

FreeBSD colo.feld.me 12.0-CURRENT FreeBSD 12.0-CURRENT #3 r335699M: Wed Jun 27 09:09:28 UTC 2018     root at colo.feld.me:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-MLX4  amd64



If there is more information I can collect next time please let me know.



-- 
  Mark Felder
  ports-secteam & portmgr member
  feld at FreeBSD.org