2.5.0 panics when using cbq

dotOne

In PFsense 2.4.5 Class Based Queing worked nicely.
After the 2.5.0 installation CBQ results in a kernel panic and reboot.

I can reproduce it by opening a remote Plex connection.
As soon as I start the movie the unit segfaults and crashes.

db:0:kdb.enter.default>  bt
Tracing pid 12 tid 100039 td 0xfffff800053bf000
kdb_enter() at kdb_enter+0x37/frame 0xfffffe000043e610
vpanic() at vpanic+0x197/frame 0xfffffe000043e660
panic() at panic+0x43/frame 0xfffffe000043e6c0
trap_fatal() at trap_fatal+0x391/frame 0xfffffe000043e720
trap_pfault() at trap_pfault+0x4f/frame 0xfffffe000043e770
trap() at trap+0x286/frame 0xfffffe000043e880
calltrap() at calltrap+0x8/frame 0xfffffe000043e880
--- trap 0xc, rip = 0xffffffff80ec014e, rsp = 0xfffffe000043e950, rbp = 0xfffffe000043e980 ---
ether_8021q_frame() at ether_8021q_frame+0x2e/frame 0xfffffe000043e980
vlan_transmit() at vlan_transmit+0xc8/frame 0xfffffe000043e9f0
vlan_altq_start() at vlan_altq_start+0xb4/frame 0xfffffe000043ea20
cbqrestart() at cbqrestart+0x64/frame 0xfffffe000043ea50
rmc_restart() at rmc_restart+0x6f/frame 0xfffffe000043ea80
softclock_call_cc() at softclock_call_cc+0x141/frame 0xfffffe000043eb30
softclock() at softclock+0x79/frame 0xfffffe000043eb50
ithread_loop() at ithread_loop+0x23c/frame 0xfffffe000043ebb0
fork_exit() at fork_exit+0x7e/frame 0xfffffe000043ebf0
fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe000043ebf0

/Andre

viktor_g

unable to reproduce your issue, CBQ is working fine on my 2.5:

pfctl -sq
queue root_vtnet0 on vtnet0 bandwidth 10Mb priority 0 cbq( wrr root ) {def, q1}
queue  def on vtnet0 bandwidth 4Mb cbq( default ) 
queue  q1 on vtnet0 bandwidth 1Mb

Please provide more information:

firewall rules
network interfaces
CBQ configuration

jimp

Given the code path in that backtrace, is it on a VLAN? Perhaps it only has problems on VLAN interfaces, or something else is contributing in that setup.

dotOne

Correct, QoS is configured on vlan.

Intel igb interfaces
Due to tripple play services igb1 has two vlans:
igb1.4 for video services
igb1.60 for internet connectivity

alt text

On LAN and DMZ there are similar queues but they are bound to the physical interface instead of a vlan.

I use different floating rules to direct ingress traffic into the corresponding queues for ingress and egress traffic.

alt text

jimp

Then for the time being you may want to disable CBQ traffic shaping.

dotOne

Which is what I did.
It looks stable now, at least no reboots since cbq was disabled

jimp

I opened https://redmine.pfsense.org/issues/11470 to track this.

Tetz

I can confirm. I'm running two firewalls in high availability (one hardware and the other VMware) and both are experiencing this issue. I'm running cbq on all internal interfaces and all internal interfaces are VLANs. I will disable cbq and see if it clears up.

db:0:kdb.enter.default>  bt
Tracing pid 12 tid 100095 td 0xfffff800057a8740
kdb_enter() at kdb_enter+0x37/frame 0xfffffe0004259610
vpanic() at vpanic+0x197/frame 0xfffffe0004259660
panic() at panic+0x43/frame 0xfffffe00042596c0
trap_fatal() at trap_fatal+0x391/frame 0xfffffe0004259720
trap_pfault() at trap_pfault+0x4f/frame 0xfffffe0004259770
trap() at trap+0x286/frame 0xfffffe0004259880
calltrap() at calltrap+0x8/frame 0xfffffe0004259880
--- trap 0xc, rip = 0xffffffff80ec014e, rsp = 0xfffffe0004259950, rbp = 0xfffffe0004259980 ---
ether_8021q_frame() at ether_8021q_frame+0x2e/frame 0xfffffe0004259980
vlan_transmit() at vlan_transmit+0xc8/frame 0xfffffe00042599f0
vlan_altq_start() at vlan_altq_start+0xb4/frame 0xfffffe0004259a20
cbqrestart() at cbqrestart+0x64/frame 0xfffffe0004259a50
rmc_restart() at rmc_restart+0x6f/frame 0xfffffe0004259a80
softclock_call_cc() at softclock_call_cc+0x141/frame 0xfffffe0004259b30
softclock() at softclock+0x79/frame 0xfffffe0004259b50
ithread_loop() at ithread_loop+0x23c/frame 0xfffffe0004259bb0
fork_exit() at fork_exit+0x7e/frame 0xfffffe0004259bf0
fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe0004259bf0

SteveITS

If it helps we have one router on 2.5 on a PC, and it has been running with CBQ shaping for over a month. Same thing with my SG-2100 at home on 21.02. Neither have VLANs.

gyterpena

I see same behaviour on 2.5.1 after enabling CQB shaping on vlan interface in lagg group(ix0,ix1).
Disabling shaping stopped crashes.

SteveITS

I know this is an older thread but for posterity it should be fixed in 22.01/2.6.