TCP and NSSEARCH = Core Dump

Thu Nov 28 12:03:49 UTC 2013

Alas, another revelation.

The 'hanging' bug is due to a single server not responding.
But, in TCP mode, you can't control-c out(whilst it's waiting for
timeout). But in the normal, UDP mode, you can.

So, to simplify it..

In TCP mode, you can't control-c if a single host doesn't respond. (DiG
9.10.0a1)Example:
# ./dig +tcp +nssearch tynyturi.com
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 59.63.165.131 in 246 ms.
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 59.63.181.21 in 262 ms.
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 60.174.233.164 in 256 ms.
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 121.10.104.50 in 269 ms.
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 60.214.139.75 in 276 ms.
SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600 from
server 222.163.192.23 in 302 ms.
^C^C^C
^C^C^C^C^C^C^C^C
^C^C^C^C^C^C^C^C^C^C
;; Connection to 61.132.227.22#53(61.132.227.22) for tynyturi.com
failed: timed out.

And, if you control-c during a nssearch&tcp dig, you recieve a segfault.
Example:
# ./dig +tcp +nssearch tynyturi.com
^Csocket.c:4768: REQUIRE(task != ((void *)0)) failed, back trace
#0 0x561799 in ??
#1 0x5618fa in ??
#2 0x596ce9 in ??
#3 0x41419a in ??
#4 0x416865 in ??
#5 0x582444 in ??
#6 0x7f78d2f8ee9a in ??
#7 0x7f78d295f3fd in ??
Aborted (core dumped)

I hope that makes it 100% clear :)

*Joshua Rogers* - gpg pubkey
<http://www.internot.info/docs/gpg_pubkey.asc.gpg>
"The system's for sale"
On 28/11/13 22:47, Joshua Rogers wrote:
> After testing a little bit more, I've found that it is partially fixed
> in bind-9.10.0a1.
>
> I was using bind 9.8.1-P1 in the original snippets. - So that's
> outdated I guess.
>
> So I hope you can follow these 'logs'... This explains it more indepth
> in the latest version of bind[9.10.0a1]
>
> 22:38:16 (root at SWAN) /var/tmp/bind-9.10.0a1/bin/dig # dig -v
> DiG 9.8.1-P1
> 22:38:17 (root at SWAN) /var/tmp/bind-9.10.0a1/bin/dig # ./dig -v
> DiG 9.10.0a1
> 22:38:18 (root at SWAN) /var/tmp/bind-9.10.0a1/bin/dig # dig +tcp
> +nssearch  tynyturi.com
> socket.c:2535: REQUIRE(socketp != ((void *)0) && *socketp == ((void
> *)0)) failed, back trace
> #0 0x7fb662acf563 in ??
> #1 0x7fb662acf4ca in ??
> #2 0x7fb662afbef8 in ??
> #3 0x7fb663912d19 in ??
> #4 0x7fb6639130be in ??
> #5 0x7fb662aee6c8 in ??
> #6 0x7fb6628ade9a in ??
> #7 0x7fb6625da3fd in ??
> Aborted (core dumped)
> 22:39:01 (root at SWAN) /var/tmp/bind-9.10.0a1/bin/dig # ./dig +tcp
> +nssearch tynyturi.com
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 60.174.233.164 in 256 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 121.10.104.50 in 263 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 60.214.139.75 in 268 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.181.21 in 288 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 222.163.192.23 in 330 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.165.131 in 450 ms.
>
> From here, it just freezes, and you have to control z to get out of
> it. You can't control C.
>
> So, it seems that +tcp and +nssearch combined is [partially] fixed, I
> guess.
> But it stills hangs nonetheless.
>
>
> Next.. If we just run
> "./dig  +nssearch tynyturi.com" a few times(bind 9.10.0a1), we get..
>  # ./dig  +nssearch tynyturi.com
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.165.131 in 246 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 60.214.139.75 in 273 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.181.21 in 295 ms.
> ;; Truncated, retrying in TCP mode.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.165.131 in 256 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 59.63.181.21 in 266 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 60.214.139.75 in 286 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 121.10.104.50 in 431 ms.
> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
> from server 60.174.233.164 in 464 ms.
>
>
> Don't ask me why it transfers to TCP mode. I don't know why.
>
>
> So, it seems that when in TCP mode, you can't use nssearch without it
> either 1) crashing[in older versions], or 2) hanging forever[in newer
> versions].
>
>
> And one more bug..
> When using TCP mode, if you control-c before it's finished[but it has
> atleast started], it will core dump.
> Examples:
>  # ./dig  +nssearch +tcp tynyturi.com
> ^Csocket.c:4768: REQUIRE(task != ((void *)0)) failed, back trace
> #0 0x561799 in ??
> #1 0x5618fa in ??
> #2 0x596ce9 in ??
> #3 0x41419a in ??
> #4 0x416865 in ??
> #5 0x582444 in ??
> #6 0x7f7e1dec4e9a in ??
> #7 0x7f7e1d8953fd in ??
> Aborted (core dumped)
>
>
>
>
> I hope that all makes sense..
> If it doesn't, please let me know.
>
>
>
> Thanks
>
>
>
>
>
>
> *Joshua Rogers* - gpg pubkey
> <http://www.internot.info/docs/gpg_pubkey.asc.gpg>
> "The system's for sale"
> On 28/11/13 22:21, Joshua Rogers wrote:
>> Hi everyone.
>>
>> I'm unsure of the email I should be writing to, but I chose this one;
>>
>>
>> I've just found a segfault/core dump bug in dig.
>>
>> Running a dig command with +nssearch and +tcp will cause dig to
>> coredump..
>>
>> Example:
>> --snip--
>>
>> $ dig +time=3 +nssearch +tcp google.com
>> socket.c:2535: REQUIRE(socketp != ((void *)0) && *socketp == ((void
>> *)0)) failed, back trace
>> #0 0xb3f77b in ??
>> #1 0xb3f6c4 in ??
>> #2 0xb72062 in ??
>> #3 0xd1d3ef in ??
>> #4 0xd1d7c3 in ??
>> #5 0xb629ac in ??
>> #6 0x972d4c in ??
>> #7 0x55bbae in ??
>> Aborted (core dumped)
>>
>> --/snip--
>>
>>
>> It was discovered whilst running "dig +time=3 +nssearch
>> tynyturi.com", which [sometimes] dumps.
>> --
>> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
>> from server 121.10.104.50 in 258 ms.
>> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
>> from server 59.63.165.131 in 260 ms.
>> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
>> from server 59.63.181.21 in 261 ms.
>> SOA v1s1.xundns.com. nsadmin.xundns.com. 50845 3600 600 604800 600
>> from server 60.174.233.164 in 271 ms.
>> ;; Truncated, retrying in TCP mode.
>> socket.c:2535: REQUIRE(socketp != ((void *)0) && *socketp == ((void
>> *)0)) failed, back trace
>> #0 0x29877b in ??
>> #1 0x2986c4 in ??
>> #2 0x2cb062 in ??
>> #3 0xb8a3ef in ??
>> #4 0xb8a7c3 in ??
>> #5 0x2bb9ac in ??
>> #6 0x828d4c in ??
>> #7 0x92cbae in ??
>> Aborted (core dumped)
>>
>> --
>>
>>
>> And finally, gdb output:
>>
>>
>> --snip--
>> Starting program: /usr/bin/dig +time=3 +nssearch +tcp tynyturi.com
>> [Thread debugging using libthread_db enabled]
>> Using host libthread_db library "/lib/i386-linux-gnu/libthread_db.so.1".
>> [New Thread 0xb7f93b40 (LWP 10758)]
>> [New Thread 0xb7792b40 (LWP 10759)]
>> [New Thread 0xb6f91b40 (LWP 10760)]
>> socket.c:2535: REQUIRE(socketp != ((void *)0) && *socketp == ((void
>> *)0)) failed, back trace
>> #0 0x31977b in ??
>> #1 0x3196c4 in ??
>> #2 0x34c062 in ??
>> #3 0x8000d3ef in ??
>> #4 0x8000d7c3 in ??
>> #5 0x33c9ac in ??
>> #6 0x371d4c in ??
>> #7 0x475bae in ??
>>
>> Program received signal SIGABRT, Aborted.
>> [Switching to Thread 0xb7f93b40 (LWP 10758)]
>> 0x00132416 in __kernel_vsyscall ()
>> (gdb) backtrace
>> #0  0x00132416 in __kernel_vsyscall ()
>> #1  0x003b41df in raise () from /lib/i386-linux-gnu/libc.so.6
>> #2  0x003b7825 in abort () from /lib/i386-linux-gnu/libc.so.6
>> #3  0x003196c9 in isc_assertion_failed () from /usr/lib/libisc.so.83
>> #4  0x0034c062 in isc__socket_create () from /usr/lib/libisc.so.83
>> #5  0x8000d3ef in ?? ()
>> #6  0x8000d7c3 in ?? ()
>> #7  0x0033c9ac in ?? () from /usr/lib/libisc.so.83
>> #8  0x00371d4c in start_thread () from
>> /lib/i386-linux-gnu/libpthread.so.0
>> #9  0x00475bae in clone () from /lib/i386-linux-gnu/libc.so.6
>> --/snip--
>>
>>
>> Hopefully this information is useful :)
>>
>>
>>
>> Thanks,
>> -- 
>> *Joshua Rogers* - gpg pubkey
>> <http://www.internot.info/docs/gpg_pubkey.asc.gpg>
>> "The system's for sale"
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/mailman/private/bind-workers/attachments/20131128/902c181d/attachment.html>