Re: Bug: Expiring a dns resolve that's still pending. Forgot to cull it?

[yancm@xxxxxxxxxxxxxxxx]

Following some post 0.1.0.4-rc updates yesterday from CVS
I received the following crash this AM:

From Warnings.log:
Apr 26 06:57:14.507 [notice] Tor 0.1.0.4-rc opening log file.
Apr 26 09:28:33.515 [warn] purge_expired_resolves():
   Bug: Expiring a dns resolve that's still pending. Forgot to cull it?
Apr 26 09:28:33.516 [warn] purge_expired_resolves():
   Bug: Expiring a dns resolve that's still pending. Forgot to cull it?
Apr 26 09:44:33.235 [warn] connection_dns_process_inbuf():
   Bug: read data (5 bytes) from an idle dns worker.  Please report.
Apr 26 09:44:33.940 [warn] connection_dns_process_inbuf():
   Bug: read data (10 bytes) from an idle dns worker.  Please report.
Apr 26 09:44:34.540 [warn] connection_dns_process_inbuf():
   Bug: read data (20 bytes) from an idle dns worker.  Please report.
Apr 26 09:48:41.403 [warn] purge_expired_resolves():
   Bug: Expiring a dns resolve that's still pending. Forgot to cull it?
Apr 26 09:48:41.403 [warn] purge_expired_resolves():
   Bug: Expiring a dns resolve that's still pending. Forgot to cull it?
Apr 26 09:49:33.145 [err] dns.c:649: connection_dns_process_inbuf:
   Assertion buf_datalen(conn->inbuf) == 5 failed; aborting.

gdb backtrace:
clarity 9 -> gdb /usr/local/bin/tor tor.core
(gdb) bt
#0  0x4821dfeb in kill () from /usr/lib/libc.so.12
#1  0x48292a3f in abort () from /usr/lib/libc.so.12
#2  0x0806c28e in connection_dns_process_inbuf (conn=0x80fa900) at dns.c:668
#3  0x0805bfe0 in connection_handle_read (conn=0x80fa900)
    at connection.c:1031
#4  0x0806eba9 in conn_read_callback (fd=21, event=2,
    _conn=0x80fa900) at main.c:354
#5  0x0808d834 in event_process_active ()
#6  0x0808da2e in event_base_loop ()
#7  0x0808d8cc in event_loop ()
#8  0x0808d855 in event_dispatch ()
#9  0x0806fbd0 in do_main_loop () at main.c:934
#10 0x080705c1 in tor_main (argc=6, argv=0xbfbffba4) at main.c:1573
#11 0x080807ab in main (argc=6, argv=0xbfbffba4) at tor_main.c:19
#12 0x0804bbb2 in ___start ()




> I received the following crash info this AM...
> From Warning.log
> Apr 24 22:08:06.798 [notice] Tor 0.1.0.4-rc opening log file.
> Apr 25 00:01:26.395 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:01:26.395 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:01:26.395 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:01:26.395 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:39:42.072 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:39:42.073 [warn] purge_expired_resolves(): Bug: Expiring a dns
> resolve that's still pending. Forgot to cull it?
> Apr 25 00:51:08.710 [err] dns.c:649: connection_dns_process_inbuf:
> Assertion buf_datalen(conn->inbuf) == 5 failed; aborting.
>
> Gdb backtrace (bt):
> (gdb) bt
> #0  0x4821dfeb in kill () from /usr/lib/libc.so.12
> #1  0x48292a3f in abort () from /usr/lib/libc.so.12
> #2  0x0806c286 in connection_dns_process_inbuf (conn=0x83b0700) at
> dns.c:668
> #3  0x0805bfd4 in connection_handle_read (conn=0x83b0700) at
> connection.c:1031
> #4  0x0806eba1 in conn_read_callback (fd=73, event=2, _conn=0x83b0700)
>     at main.c:354
> #5  0x0808d82c in event_process_active ()
> #6  0x0808da26 in event_base_loop ()
> #7  0x0808d8c4 in event_loop ()
> #8  0x0808d84d in event_dispatch ()
> #9  0x0806fbc8 in do_main_loop () at main.c:934
> #10 0x080705b9 in tor_main (argc=6, argv=0xbfbffba4) at main.c:1573
> #11 0x080807a3 in main (argc=6, argv=0xbfbffba4) at tor_main.c:19
> #12 0x0804bbb2 in ___start ()
> (gdb)
>
>
>
>
>
>> Hi Roger, Nick, et. al.
>>
>> This bug is still with me with the new Tor 0.1.0.4-rc
>> CVS build.
>>
>> I managed to update my libevent to 1.0c, but I still get
>> a bunch of the messages;
>>  Bug: Expiring a dns resolve that's still pending. Forgot to cull it?
>> Then after a while (1-10 hours) the process seems to just
>> lock up (no longer seems to route traffic), without actually crashing.
>>
>> I'm running under NetBSD 2.0_Stable...i386
>>
>> I started a flyspray entry on this bug a few weeks back
>> and never heard back from you...
>>
>> How can I help you track it down?
>>
>> I can always just turn off the PTHREAD_CREATE switch, but this seems
>> to increase resource utilization...I'd like to get this back
>> to normal...
>>
>> Help!
>> gene
>>
>>
>
>