[Bug 284457] periodic pf panic on busy jailhost

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=284457

            Bug ID: 284457
           Summary: periodic pf panic on busy jailhost
           Product: Base System
           Version: 13.4-STABLE
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: bugs@FreeBSD.org
          Reporter: thomas@gibfest.dk

Hello,

I recently (late december) upgraded a jailhost from 13-STABLE-e6bb702fc to
13-STABLE-11c7eb30cc and since then it has been regularly panicing.

The latest panic can be seen here:

Jan 29 21:06:17 hash1 kernel: Fatal trap 9: general protection fault while in
kernel mode
Jan 29 21:06:17 hash1 kernel: cpuid = 5; apic id = 0a
Jan 29 21:06:17 hash1 kernel: instruction pointer       =
0x20:0xffffffff810ae8d0
Jan 29 21:06:17 hash1 kernel: stack pointer             =
0x28:0xfffffe0115b6c6e0
Jan 29 21:06:17 hash1 kernel: frame pointer             =
0x28:0xfffffe0115b6c6e0
Jan 29 21:06:17 hash1 kernel: code segment              = base 0x0, limit
0xfffff, type 0x1b
Jan 29 21:06:17 hash1 kernel:                   = DPL 0, pres 1, long 1, def32
0, gran 1
Jan 29 21:06:17 hash1 kernel: processor eflags  = interrupt enabled, resume,
IOPL = 0
Jan 29 21:06:17 hash1 kernel: current process           = 0 (if_io_tqg_5)
Jan 29 21:06:17 hash1 kernel: trap number               = 9
Jan 29 21:06:17 hash1 kernel: panic: general protection fault
Jan 29 21:06:17 hash1 kernel: cpuid = 5
Jan 29 21:06:17 hash1 kernel: time = 1738184402
Jan 29 21:06:17 hash1 kernel: KDB: stack backtrace:
Jan 29 21:06:17 hash1 kernel: #0 0xffffffff80c42835 at kdb_backtrace+0x65
Jan 29 21:06:17 hash1 kernel: #1 0xffffffff80bf6182 at vpanic+0x182
Jan 29 21:06:17 hash1 kernel: #2 0xffffffff80bf5ff3 at panic+0x43
Jan 29 21:06:17 hash1 kernel: #3 0xffffffff810b3169 at trap_fatal+0x389
Jan 29 21:06:17 hash1 kernel: #4 0xffffffff8108a968 at calltrap+0x8
Jan 29 21:06:17 hash1 kernel: #5 0xffffffff82f430e0 at pf_find_state+0xb0
Jan 29 21:06:17 hash1 kernel: #6 0xffffffff82f45118 at pf_test_state_tcp+0x1b8
Jan 29 21:06:17 hash1 kernel: #7 0xffffffff82f441f1 at pf_test+0xee1
Jan 29 21:06:17 hash1 kernel: #8 0xffffffff82f63654 at pf_check_in+0x24
Jan 29 21:06:17 hash1 kernel: #9 0xffffffff80d33a36 at pfil_run_hooks+0xc6
Jan 29 21:06:17 hash1 kernel: #10 0xffffffff80dabdaf at ip_input+0x41f
Jan 29 21:06:17 hash1 kernel: #11 0xffffffff80d30c3f at
netisr_dispatch_src+0x9f
Jan 29 21:06:17 hash1 kernel: #12 0xffffffff80d14ae9 at ether_demux+0x159
Jan 29 21:06:17 hash1 kernel: #13 0xffffffff80d15daf at ether_nh_input+0x33f
Jan 29 21:06:17 hash1 kernel: #14 0xffffffff80d30c3f at
netisr_dispatch_src+0x9f
Jan 29 21:06:17 hash1 kernel: #15 0xffffffff80d14e46 at ether_input+0x56
Jan 29 21:06:17 hash1 kernel: #16 0xffffffff80dc2464 at tcp_lro_flush+0x2d4
Jan 29 21:06:17 hash1 kernel: #17 0xffffffff80dc27b3 at tcp_lro_flush_all+0x1b3
Jan 29 21:06:17 hash1 kernel: Timeout initializing vt_vga

The last panic before this was on Jan 20th so about 10 days before. I don't
have a paste of earlier panics but they also mentioned pf. When it happens
again I will update this PR with another backtrace.

The server is a jailhost with a handful of jails, one of which is a busy
webserver with around 20 million hits per day, for the https://ip.tyk.nu
service. So the firewall sees a lot of sessions.

Let me know if I can do or add anything else to help.

Thanks! :)

-- 
You are receiving this mail because:
You are the assignee for the bug.