Kernel panic on FreeBSD 9.0-beta2

[dave jones]

On Tue, Sep 27, 2011 at 9:24 AM, Arnaud Lacombe wrote:
> Hi,
>
> On Mon, Sep 26, 2011 at 8:51 PM, dave jones <s.dave.jones at gmail.com> wrote:
>> On Mon, Sep 26, 2011 at 1:41 PM, Arnaud Lacombe wrote:
>>> Hi,
>>>
>>> On Mon, Sep 26, 2011 at 1:12 AM, Arnaud Lacombe wrote:
>>>> Hi,
>>>>
>>>> On Mon, Sep 26, 2011 at 12:43 AM, dave jones wrote:
>>>>> Hi,
>>>>> I have two production machines running on freebsd 9.0-beta2 and both got
>>>>> kernel panic related to networking. Any idea how to solve it? thanks.
>>>>>
>>>>> http://http://60.248.161.9/p1.jpg
>>>>> http://http://60.248.161.9/p2.jpg
>>>>>
>>>> this host is really slow :-)
>>>>
>>>> To avoid the waiting time, the backtrace is:
>>>>
>>>> in_pcbbind_setup()+0x28f
>>>> in_pcbbind()+0xa9
>>>> udp_bind()
>>>> bind()
>>>> kern_bind()
>>>> syscall_enter()
>>>> syscall()
>>>>
>>>> faulted at VA 0x07. Origin process in named.
>>>>
>>> AFAICT, the crash happens in the following block:
>>>
>>>  /*
>>>   * XXX
>>>   * This entire block sorely needs a rewrite.
>>>   */
>>>        if (t &&
>>>            ((t->inp_flags & INP_TIMEWAIT) == 0) &&
>>>            (so->so_type != SOCK_STREAM ||
>>>             ntohl(t->inp_faddr.s_addr) == INADDR_ANY) &&
>>>            (ntohl(sin->sin_addr.s_addr) != INADDR_ANY ||
>>>             ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
>>>             (t->inp_socket->so_options &
>>>           SO_REUSEPORT) == 0) &&
>>>            (inp->inp_cred->cr_uid !=
>>>             t->inp_cred->cr_uid))
>>>          return (EADDRINUSE);
>>>      }
>>>
>>> more specifically, `t->inp_socket' is NULL. The top comment may not be
>>> relevant, as it's been here for the past 8 years.
>>
>> Hi Arnaud,
>>
>> Ah, thanks for the info. I'm wondering if you have a patch to fix that issue?
>> Guess what? another production machine got the same panic, oh my~
> Can you give us more precision on the condition of the crash ? is it
> immediately at boot-time ? If not, is it when a special network
> condition occurs ? What kind of load is named(9) having on average
> before the panic ? Are you using any fancy configuration or thing out
> of the ordinary ? Are you using named from the base system, or from
> port ?

Hi Arnaud,

Sorry for the confusion. It's not crashed at boot-time and named(9) used to
create a caching-only name server and it worked on FreeBSD <= 8.2 for years.
I've been using named from the base system.
When I switched to 9.0-beta2, it got panic after running three days *sigh*

> Thanks,
>  - Arnaud

BR,
Dave.