pf_test: kif == NULL, if_xname on multi-WAN and "Reset all states if WAN IP Address changes"

w0w

Hmm... looks like this thing came back, after I've configured CARP and started to play with PPPoE on secondary firewall and re-plugged main ethernet cable that cames from ISP (PPPoE).

Crash report begins.  Anonymous machine information:

amd64
12.1-STABLE
FreeBSD 12.1-STABLE f1de4082be8(devel-12) pfSense

Crash report details:

No PHP errors found.

..........................................
<118>Starting CRON... done.
<118> Starting package Cron...done.
<118> Starting package System Patches...done.
<118> Starting package Service Watchdog...done.
<118> Starting package nut...done.
<118> Starting package Shellcmd...done.
<118> Starting package Backup...done.
<118> Starting package iperf...done.
<118> Starting /usr/local/etc/rc.d/shutdown.nut.sh...done.
<118>pfSense 2.5.0-DEVELOPMENT amd64 Fri May 22 07:43:46 EDT 2020
<118>Bootup complete
<6>ix0: link state changed to DOWN
<6>ix0: link state changed to UP
<6>ix0: link state changed to DOWN
<6>ix0: link state changed to UP
<6>gif0: link state changed to DOWN
<6>ng0: changing name to 'pppoe0'
<6>gif0: link state changed to DOWN
<6>gif0: link state changed to UP
<6>gif0: link state changed to DOWN
<6>gif0: link state changed to UP
<6>gif0: link state changed to DOWN
pf_test: kif == NULL, if_xname 
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname @��
<6>ng0: changing name to 'pppoe0'
pf_test: kif == NULL, if_xname @��
pf_test: kif == NULL, if_xname 
<6>gif0: link state changed to DOWN
<6>gif0: link state changed to UP


Fatal trap 12: page fault while in kernel mode
cpuid = 2; apic id = 04
fault virtual address	= 0x70
fault code		= supervisor read data, page not present
instruction pointer	= 0x20:0xffffffff80f25037
stack pointer	        = 0x28:0xfffffe0095dd2370
frame pointer	        = 0x28:0xfffffe0095dd23b0
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		= 80273 (unbound)
trap number		= 12
panic: page fault
cpuid = 2
time = 1590480096
KDB: enter: panic

The crash occurs only on primary firewall.
Full log is here

w0w

Got those topic messages on backup firewall also, when experimented with PPPoE (same link)

May 26 19:34:51 	kernel 		
May 26 19:34:51 	root 	54561 	PPPoE put down
May 26 19:34:51 	kernel 		
May 26 19:34:50 	kernel 		
May 26 19:34:50 	kernel 		
May 26 19:34:49 	kernel 		
May 26 19:34:49 	kernel 		
May 26 19:34:48 	kernel 		
May 26 19:34:48 	kernel 		
May 26 19:34:47 	kernel 		
May 26 19:34:47 	kernel 		
May 26 19:34:47 	kernel 		
May 26 19:34:46 	kernel 		
May 26 19:34:46 	kernel 		
May 26 19:34:46 	kernel 		
May 26 19:34:45 	kernel 		
May 26 19:34:45 	kernel 		
May 26 19:34:44 	kernel 		pf_test: kif == NULL, if_xname
May 26 19:34:44 	kernel 		pf_test: kif == NULL, if_xname
May 26 19:34:43 	kernel 		pf_test: kif == NULL, if_xname
May 26 19:34:43 	kernel 		pf_test: kif == NULL, if_xname
May 26 19:34:42 	kernel 		
May 26 19:34:42 	kernel 		
May 26 19:34:42 	kernel 		pf_test: kif == NULL, if_xname

Crashdump

w0w

May be it's https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=230498

w0w

I've played around a bit and the result is that 2.4.5-RELEASE is still not affected, and 2.5 is crashing not every time but mostly, when I issue 'killall mpd5' or even 'rc.linkup stop wan'
The configuration is the same. PPPoE as WAN, DHCP as WAN2, no IPv6, only failover is configured. PPPoE as tier1 and DHCP as tier2.
I don't think that unbound is the real cause of crash... just because I've tried to stop unbound service and repeat the sequence (multiple times disconnected and connected WAN port cable) and got another dump. crash00.txt

Fatal trap 12: page fault while in kernel mode
cpuid = 3; apic id = 06
fault virtual address	= 0x0
fault code		= supervisor read instruction, page not present
instruction pointer	= 0x20:0x0
stack pointer	        = 0x28:0xfffffe0074d614a8
frame pointer	        = 0x28:0xfffffe0074d615f0
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 (swi4: clock (0))
trap number		= 12
panic: page fault
cpuid = 3
time = 1590747377
KDB: enter: panic
����������������

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234296 looks very similar.

w0w

Do we have patch provided in https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234296 enabled on pfSense latest 2.5 version? I mean https://svnweb.freebsd.org/base?view=revision&revision=343787

jimp

If it was marked fixed in 12.0-RELEASE then yes, that would be in 2.5.0. 2.5.0 snapshots are on 12.1-STABLE now, so well past that point.

w0w

@jimp
Can you suggest me something?

jimp

Nothing I'm aware of for that, I'm afraid.

w0w

@jimp
What do you think, сould this fatal trap be due to the radix_mpath option enabled in the kernel?
Any chances to get some snapshot with this option disabled?

jimp

It's possible. There are other problems with RADIX_MPATH as well but I'm not sure if we're going to look into fixing them or back that out.