Thanks Robert, and Balasz, I understand the functioon of libnet :).


I beleive it to be the source of my memory problems, and not syslog-ng necessarily.



>The funny part is that this version of libnet seem to expect port
>numbers in host byte order whereas I pass it to libnet in network byte
>order. I'm almost confident that this used to work when I originally did
>the libnet support, judging the libnet changelog again, this was a
>change between 1.0 <-> 1.1
>

>Is your syslog-ng sending messages to the correct port? Can you check
>that with tcpdump for example? Or maybe you are using a big-endian
>machine?


Balasz,  spoofed UDP packets are being  sent properly, as far as I can tell  - the data is getting to the target properly.
tcpdump shows some minor strangeness - the source  address is that of the spoofed syslog host, which is to be expected, and the target host is correct, as is the target port (514/UDP).  What is strange is that all the spopofed packets are all useing UDP/514 as the source.

An example of a tcpdump ron on the UDP spoofer syslogmachine(syslogng1.testdomain.org):

10:10:39.4092332 IP cisco2121.testdomain.org.syslog > syslogng2.testdomain.org.syslog: UDP, length 150

I don't thing the endianness is coming into play here.  Also, I verified that libnet was not installed prior to the 1.2.2 installtion, I am certain that syslog-ng was compiled against 1.2.2.  

The output of libnet-config --defines:
-D_BSD_SOURCE -D__BSD_SOURCE -D__FAVOR_BSD -DHAVE_NET_ETHERNET_H

The output of libnet-config --cflags:


The output of libnet-config --libs:
-lnet

One more thing, the only confiure options I used when compiling syslog-ng was "--enable-spoof-source"

Oh, and I reran valgrind with Robert's exact command line - the output is pretty much the same as I had prior:


# valgrind --tool=memcheck --trace-children=yes --leak-check=yes syslog-ng

==9888== Memcheck, a memory error detector for x86-linux.

==9888== Copyright (C) 2002-2004, and GNU GPL'd, by Julian Seward et al.

==9888== Using valgrind-2.2.0, a program supervision framework for x86-linux.

==9888== Copyright (C) 2000-2004, and GNU GPL'd, by Julian Seward et al.

==9888== For more details, rerun with: -v

==9888==

==9888==

==9888== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 19 from 1)

==9888== malloc/free: in use at exit: 32811 bytes in 508 blocks.

==9888== malloc/free: 734 allocs, 226 frees, 40605 bytes allocated.

==9888== For counts of detected errors, rerun with: -v

==9888== searching for pointers to 508 not-freed blocks.

==9888== checked 1598824 bytes.

==9888==

==9888== LEAK SUMMARY:

==9888==    definitely lost: 0 bytes in 0 blocks.

==9888==    possibly lost:   0 bytes in 0 blocks.

==9888==    still reachable: 32811 bytes in 508 blocks.

==9888==         suppressed: 0 bytes in 0 blocks.

==9888== Reachable blocks (those to which a pointer was found) are not shown.

==9888== To see them, rerun with: --show-reachable=yes

==9890== Invalid read of size 2

==9890==    at 0x805A987: libnet_in_cksum (in /root/syslog-ng-1.6.6/src/syslog-ng)

==9890==  Address 0x1BA790BA is 178 bytes inside a block of size 179 alloc'd

==9890==    at 0x1B902E28: malloc (vg_replace_malloc.c:131)

==9890==    by 0x805912D: libnet_pblock_coalesce (in /root/syslog-ng-1.6.6/src/syslog-ng)

==9890==    by 0x804C063: do_handle_log (destinations.c:103)

==9890==    by 0x804B5EC: do_distribute_log (center.c:149)




Balazs Scheidler <bazsi@balabit.hu>
Sent by: syslog-ng-admin@lists.balabit.hu

03/03/2005 07:27 AM
Please respond to
syslog-ng@lists.balabit.hu

To
syslog-ng@lists.balabit.hu
cc
Subject
Re: [syslog-ng]Syslog-NG 1.6.6 memory leak when sending UDP logs





On Wed, 2005-03-02 at 22:19 +0100, Roberto Nibali wrote:

> > io.c: Preparing fd 6 for writing
> > ==27361== Invalid read of size 2
>
> There seems to be a off-by-one error in a string. This is the result if
> you do something like follows:

This message is not triggered for me, but I'm going to try to use your
exact configuration as well.

> > ==27361==    at 0x805A987: libnet_in_cksum (in /usr/local/sbin/syslog-ng)
> > ==27361==  Address 0x1BA764E2 is 178 bytes inside a block of size 179
> > alloc'd
>
> There seems to be a wrong free, not really a missing one.
>
> > ==27361==    at 0x1B902E28: malloc (vg_replace_malloc.c:131)
> > ==27361==    by 0x805912D: libnet_pblock_coalesce (in
> > /usr/local/sbin/syslog-ng)
> > ==27361==    by 0x804C063: do_handle_log (destinations.c:103)
> > ==27361==    by 0x804B5EC: do_distribute_log (center.c:149)
> > ==27361==    by 0x804B02A: do_add_source_name (sources.c:289)
> > ==27361==    by 0x804AA8C: do_handle_line (sources.c:75)
> > ==27361==    by 0x804ADA5: do_read_line (sources.c:134)
> > ==27361==    by 0x8054AF8: read_callback (in /usr/local/sbin/syslog-ng)
> > ==27361==    by 0x804A079: main_loop (main.c:253)
> > ==27361==    by 0x804A75C: main (main.c:549)
> > io.c: Preparing fd 8 for writing
> > io.c: connecting using fd 11
> > io.c: connecting using fd 11


Again, this one does not show up in my valgrind output. In fact it
reports that no blocks are leaked.

I'm using 1.1.2.1-2 Debian package. The libnet changelog shows some
fixed leaks before 1.1.1, but as I see you also have a newer version.

Isn't it possible that you linked syslog-ng to an older libnet
statically and then upgraded your libnet package?

The funny part is that this version of libnet seem to expect port
numbers in host byte order whereas I pass it to libnet in network byte
order. I'm almost confident that this used to work when I originally did
the libnet support, judging the libnet changelog again, this was a
change between 1.0 <-> 1.1

Is your syslog-ng sending messages to the correct port? Can you check
that with tcpdump for example? Or maybe you are using a big-endian
machine?

This patch fixes the byte order issue, and I'm still hunting the memory
leak with your configuration:

diff -u -r1.25.4.6 afinet.c
--- afinet.c    5 Aug 2004 11:35:12 -0000       1.25.4.6
+++ afinet.c    3 Mar 2005 12:26:47 -0000
@@ -617,11 +617,10 @@
                       ADDRESS2SOCKADDR(msg->saddr, sizeof(src_addr), (struct sockaddr *) &src_addr);
                       ADDRESS2SOCKADDR(self->super.dest_addr, sizeof(dst_addr), (struct sockaddr *) &dst_addr);

-
                       libnet_clear_packet(self->lnet_ctx);

-                       udp = libnet_build_udp(src_addr.sin_port,
-                                              dst_addr.sin_port,
+                       udp = libnet_build_udp(ntohs(src_addr.sin_port),
+                                              ntohs(dst_addr.sin_port),
                                              LIBNET_UDP_H + msg_line->length,
                                              0,
                                              msg_line->data,


>
> > Which doesent say too much.  I'm using libnet 1.1.2.1. The valgrind
> > message only appears once - and does not appear as the memory leak
> > contiues.
>
> Was libnet linked statically against syslog-ng?

Yes, libnet is linked in statically by default.

>
> > I'm no valgrind expert, but I'm guessing it leaks one byte for each UDP
> > packet sent. Not sure why spoofing would cause this inside libnet.  
>
> If you need to create a packet, you'd want to use libnet, unless you've
> got enough spare time to code. Otherwise I don't see why libnet could be
> used within syslog-ng.

syslog-ng uses libnet for creating UDP packets sent via a raw socket.

--
Bazsi


_______________________________________________
syslog-ng maillist  -  syslog-ng@lists.balabit.hu
https://lists.balabit.hu/mailman/listinfo/syslog-ng
Frequently asked questions at http://www.campin.net/syslog-ng/faq.html