Alright... lets see what we've found now..
(I am using 2.1.2, I will try again with 2.1.50... but in the mean time)...
also... should I move this discussion to ntop-dev? Or no? I'll keep it on
ntop for now..
at line 272 in hash.c, we call 'free(list->virtualHostName);'
list->virtualHostName == 0xffffffff; Obviously this is an invalid pointer.
Other interesting things:
list->bytesSent==1844674406941584320;
list->bytesRcvd==the same number;
list->next==0x0;
so it looks like we're at what is supposed to be the end...
The host we're trying to free, is m2.doubleclick.net. All of the other
stats inside of this entry look just fine. So it looks like the
httpVirtualHosts entry is getting filled with bogus stuff? or maybe
0xffffffff is supposed to mean "no virtual hosts" ?
I went down(or "up" in gdb...) further into the backtrace and found
nothing more to suggest why httpVirtualHosts had this strange entry in it.
I will continue to run under gdb and see if other crashes are similar, and
I will grap 2.1.50 and try to smack it around in a similar fashion
tonight. ;)
Burton M. Strauss III said:
> The backtrace you posted pointed at freeHostInfo+0x4a6 -- but the hex
> offset is highly dependent upon your version of ntop and of gcc, so I
> can't really work from the hex offset.
>
> Running under gdb you get the benefit of the translation to the source
> line #.
>
> void freeHostInfo(int theDevice, HostTraffic *host, int actualDeviceId)
> {
> u_int j, i;
> IpGlobalSession *nextElement, *element;
>
> ...
>
> freeHostSessions(host->hostTrafficBucket, actualDeviceId);
>
>
> So now what you do is print the values of the variables and follow the
> back trace up through the stack, i.e. to purgeIdleHosts and then
> scanIdleLoop. From the various variables (esp. myGlobals.device[]), we
> should be able to determine what's special about this entry...
>
>
> -----Burton
>
>
>
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
> Clint Byrum
> Sent: Thursday, September 19, 2002 5:12 PM
> To: [EMAIL PROTECTED]
> Subject: RE: [Ntop] ntop 2.1.2 segfaulting on Debian 3.0
>
>
> Burton M. Strauss III said:
>> It's really not that hard to run under gdb and see what line failed --
>> Instructions are at http://snapshot.ntop.org
>>
>
> I have read them. I have followed them. I'm not sure what I'm supposed
> to be looking for, beyond what the backtrace that ntop already spit out
> has said. That ntop died in hash.c, at line 272, because it tried to
> free a NULL pointer.
> Just point at a spot and I'll dig. ;)
>
>> -----Burton
>>
>>
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
>> Clint Byrum
>> Sent: Tuesday, September 17, 2002 4:57 PM
>> To: [EMAIL PROTECTED]
>> Subject: [Ntop] ntop 2.1.2 segfaulting on Debian 3.0
>>
>>
>> I have a few boxes running ntop 2.1.2 on Debian 3.0 ... They all seem
>> to segfault regularly. I read the FAQ and hopefully I have all of the
>> information any of you might need to debug this problem. I ran at
>> trace level 1 for this. I will up it to level 2 and report the next
>> segfault too, but it uses up quite a lot of disk space to go any
>> higher.
>> I tried reading hash.c ... but its extremely difficult to know where
>> this null pointer might be coming from.
>> :-/
>>
>> Command Line:/usr/sbin/ntop -d -L -u ntop -w 3000 -p
>> /etc/ntop/protocol.list -P /var/lib/ntop -S 1 -a
>> /var/lib/ntop/access.log -i eth0,eth1,ipsec0 -t 1 -O /var/log/ntop/ -E
>> -K
>> I downloaded the source and Dennis's diff to debianize it from a
>> debian mirror. I compiled it on a Debian 3.0 box, gcc 2.95.4 20011002
>> (Debian prerelease).
>> Thanks very much for all of your work Burton. I wish I could help you
>> more with ntop.
>> Log messages:
>> Sep 17 13:05:27 plazamon ntop[28449]: Truncated network size to 1024
>> hosts (real netmask 255.255.255.0)Sep 17 14:11:42 plazamon
>> ntop[28454]: WARNING: free of NULL pointer @
>> hash.c:272Sep 17 14:11:42 plazamon ntop[28454]: *****ntop error:
>> Signal(11)
>> Sep 17 14:11:42 plazamon ntop[28454]: backtrace is:
>> Sep 17 14:11:42 plazamon ntop[28454]: 1.
>> /lib/libpthread.so.0 [0x502f54]Sep 17 14:11:42 plazamon ntop[28454]:
>> 2. /lib/libc.so.6 [0x6206b8]
>> Sep 17 14:11:42 plazamon ntop[28454]: 3.
>> /lib/libc.so.6(__libc_free+0xa7) [0x665aa3]Sep 17 14:11:42 plazamon
>> ntop[28454]: 4.
>> /usr/lib/ntop/libntop-2.1.so.2(ntop_safefree+0x55) [0x2a6709]Sep 17
>> 14:11:42 plazamon ntop[28454]: 5.
>> /usr/lib/ntop/libntop-2.1.so.2(freeHostInfo+0x4a6) [0x2a1b62]Sep 17
>> 14:11:42 plazamon ntop[28454]: 6.
>> /usr/lib/ntop/libntop-2.1.so.2(purgeIdleHosts+0x316) [0x2a2236]Sep 17
>> 14:11:42 plazamon ntop[28454]: 7.
>> /usr/lib/ntop/libntop-2.1.so.2(scanIdleLoop+0x5f) [0x2a85af]Sep 17
>> 14:11:42 plazamon ntop[28454]: 8. /lib/libpthread.so.0
>> [0x5000ba]Sep 17 14:11:42 plazamon ntop[28454]: 9.
>> /lib/libc.so.6(__clone+0x3a) [0x6c6d4a]
>>
>> _______________________________________________
>> Ntop mailing list
>> [EMAIL PROTECTED]
>> http://lists.ntop.org/mailman/listinfo/ntop
>>
>> _______________________________________________
>> Ntop mailing list
>> [EMAIL PROTECTED]
>> http://lists.ntop.org/mailman/listinfo/ntop
>
>
>
> _______________________________________________
> Ntop mailing list
> [EMAIL PROTECTED]
> http://lists.ntop.org/mailman/listinfo/ntop
_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://lists.ntop.org/mailman/listinfo/ntop
