Re: [Openh323gk-developer] gnugk eats memory [PATCH]
H.323 Gatekeeper for VoIP and videconferencing
Brought to you by:
willamowius
Menu
▾
▴
Re: [Openh323gk-developer] gnugk eats memory [PATCH]
|
From: Thomas J. <ad...@po...> - 2002-10-14 07:37:17
|
Hello,
we found a bug in the pwlib code. Thanks to Alex, the problem
source was identified: when the server has netmask not 0 at the end (e.g.
255.255.255.224), the pwlib fails.
I searched the code and find the problem in the
PIPSocket::GetRouteTable function (pwlib/src/ptlib/unix/socket.cxx)
Simply:
long net_addr, dest_addr, net_mask;
is wrong, because these variables shoud have 31bit unsigned number while
IP Addresses and netmasks has 32bit length. Therefore when I had e.g. .224
netmask, it shouldn be placed into this variable and it made a loop.
The right way:
unsigned long net_addr, dest_addr, net_mask;
I tested it and now it works like a charm with any network
netmask.
Regards,
Thomas
patch:
--- pwlib/src/ptlib/unix/socket.cxx.orig 2002-10-12
13:22:50.000000000 +0200
+++ pwlib/src/ptlib/unix/socket.cxx 2002-10-12 13:23:14.000000000
+0200
@@ -1030,7 +1030,7 @@
return TRUE;
char iface[20];
- long net_addr, dest_addr, net_mask;
+ unsigned long net_addr, dest_addr, net_mask;
int flags, refcnt, use, metric;
procfile >> iface >> ::hex >> net_addr >> dest_addr >> flags
>> ::dec >> refcnt >> use >> metric
On Mon, 7 Oct 2002, Thomas Jalsovsky wrote:
>
> > > I have the same problem, that the gk loops into infinity
> > > and eats up all mem in Toolkit::RouteTable::InitTable()
> Yes, it looks like a BUG...
>
> > >
> > > I use: gk-2.0a, openh323-1.9.9, pwlib-1.3.9,
> > > kernel-2.4.19-gentoo1.4rc1, gcc3.2
> > >
> > > I assume that the problem lies in the pwlib.
> > > Note: This only happens if the netmask is
> > > not 'even' (not on 8 bit boundaries)
> > > I mean, eg: 255.255.255.248
> > Great! I changed the netmask a week ago and about that time gnugk won't
> > start. So probably there is something wrong around networking.
> > But, I had the same problem on that NAT firewall when I had
> > netmasks: 255.255.255.0.
> >
> > >
> > > As a workaround, to get it running i return just after
> > > PIPSocket::GetHostAddress(defAddr);
> I'm looking for the code, but unfortunately I'm not a C++ expert :-(
> Did you try to make the return point not directly after the mentioned line
> but above?
>
> > > But i dont know what sideeffects this can cause,
> > > anyhow it runs now in our test environment.
> > I will try it, too.
> I tryed. You are right. When I use something not 'even' netmasks it eats
> the memory and if I have e.g. /24 it works fine.
> Unfortunately I have to use /29, so I need to solve this problem....
>
> Regards,
> Thomas
>
> > Unfortunately developers didn't answer for my question :((
> >
> > Regards,
> > Thomas
> >
> > > Cheers
> > > Alex
> > >
> > >
> > >
> > > On Fri, 2002-09-27 at 16:02, Thomas Jalsovsky wrote:
> > > >
> > > >
> > > > Hello,
> > > > gnugk-2.0a compiled with today's pwlib & openh323 libraries (CVS)
> > > > does not start on RedHat Linux 7.3, kernel-2.4.20-pre7-ac3,
> > > > glibc-2.2.90-26 (Rawhide), gcc-3.2-0.3 (compiled with).
> > > >
> > > > [root@fw-lan OK-debug]# gdb gnugk
> > > > GNU gdb Red Hat Linux (5.2-2)
> > > > Copyright 2002 Free Software Foundation, Inc.
> > > > GDB is free software, covered by the GNU General Public License, and you
> > > > are
> > > > welcome to change it and/or distribute copies of it under certain
> > > > conditions.
> > > > Type "show copying" to see the conditions.
> > > > There is absolutely no warranty for GDB. Type "show warranty" for
> > > > details.
> > > > This GDB was configured as "i386-redhat-linux"...
> > > > (gdb) run
> > > > Starting program: /DEVEL/OK-debug/gnugk
> > > > [New Thread 1024 (LWP 9249)]
> > > > [New Thread 2049 (LWP 9257)]
> > > > [New Thread 1026 (LWP 9258)]
> > > >
> > > > Program received signal SIGINT, Interrupt.
> > > > [Switching to Thread 1026 (LWP 9258)]
> > > > 0x402cf281 in nanosleep () from /lib/libc.so.6
> > > > (gdb) Quit
> > > > (gdb) bt
> > > > #0 0x402cf281 in nanosleep () from /lib/libc.so.6
> > > > #1 0x40024e8d in __pthread_timedsuspend_new () from /lib/libpthread.so.0
> > > > #2 0x400223e8 in pthread_cond_timedwait_relative () from
> > > > /lib/libpthread.so.0
> > > > #3 0x40022108 in pthread_cond_timedwait () from /lib/libpthread.so.0
> > > > #4 0x0838995f in PSyncPoint::Wait(PTimeInterval const&) (this=0x85196b0,
> > > > waitTime=@0xbf7ffa3c) at tlibthrd.cxx:1342
> > > > #5 0x08361c4a in PXConfigWriteThread::Main() (this=0x851ae20) at
> > > > config.cxx:218
> > > > #6 0x08388425 in PThread::PX_ThreadStart(void*) (arg=0x851ae20) at
> > > > tlibthrd.cxx:873
> > > > #7 0x4002371a in pthread_start_thread () from /lib/libpthread.so.0
> > > > #8 0x40023789 in pthread_start_thread_event () from /lib/libpthread.so.0
> > > > (gdb) Quit
> > > > (gdb)
> > > >
> > > > or:
> > > > strace gnugk
> > > > ...
> > > > mprotect(0x4034e000, 1252, PROT_NONE) = 0
> > > > old_mmap(0x4034e000, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED,
> > > > 11, 0x8000) = 0x4034e000
> > > > close(11) = 0
> > > > munmap(0x40012000, 26251) = 0
> > > > open("/etc/host.conf", O_RDONLY) = 11
> > > > fstat64(11, {st_mode=S_IFREG|0644, st_size=29, ...}) = 0
> > > > mmap2(NULL, 29, PROT_READ, MAP_SHARED, 11, 0) = 0x40012000
> > > > _llseek(11, 29, [29], SEEK_SET) = 0
> > > > fstat64(11, {st_mode=S_IFREG|0644, st_size=29, ...}) = 0
> > > > munmap(0x40012000, 29) = 0
> > > > close(11) = 0
> > > > open("/etc/hosts", O_RDONLY) = 11
> > > > fcntl64(11, F_GETFD) = 0
> > > > fcntl64(11, F_SETFD, FD_CLOEXEC) = 0
> > > > fstat64(11, {st_mode=S_IFREG|0644, st_size=1161, ...}) = 0
> > > > mmap2(NULL, 1161, PROT_READ, MAP_SHARED, 11, 0) = 0x40012000
> > > > _llseek(11, 1161, [1161], SEEK_SET) = 0
> > > > munmap(0x40012000, 1161) = 0
> > > > close(11) = 0
> > > > gettimeofday({1033135232, 629006}, NULL) = 0
> > > > socket(PF_INET, SOCK_DGRAM, IPPROTO_IP) = 11
> > > > ioctl(11, 0x5421, [1]) = 0
> > > > fcntl64(11, F_SETFD, FD_CLOEXEC) = 0
> > > > brk(0x8525000) = 0x8525000
> > > > ioctl(11, 0x8912, 0xbffff3a8) = 0
> > > > ioctl(11, 0x8913, 0xbffff380) = 0
> > > > ioctl(11, 0x8927, 0xbffff380) = 0
> > > > brk(0x8526000) = 0x8526000
> > > > ioctl(11, 0x8915, 0xbffff380) = 0
> > > > ioctl(11, 0x891b, 0xbffff380) = 0
> > > > ioctl(11, 0x8913, 0xbffff380) = 0
> > > > ioctl(11, 0x8927, 0xbffff380) = 0
> > > > ioctl(11, 0x8915, 0xbffff380) = 0
> > > > ioctl(11, 0x891b, 0xbffff380) = 0
> > > > ioctl(11, 0x8913, 0xbffff380) = 0
> > > > ioctl(11, 0x8927, 0xbffff380) = 0
> > > > ioctl(11, 0x8915, 0xbffff380) = 0
> > > > ioctl(11, 0x891b, 0xbffff380) = 0
> > > > shutdown(11, 2 /* send and receive */) = -1 ENOTCONN (Transport endpoint
> > > > is not connected)
> > > > close(11) = 0
> > > > open("/proc/net/route", O_RDONLY) = 11
> > > > fcntl64(11, F_SETFD, FD_CLOEXEC) = 0
> > > > read(11, "Iface\tDestination\tGateway \tFlags"..., 1024) = 640
> > > > brk(0x8527000) = 0x8527000
> > > > brk(0x8528000) = 0x8528000
> > > > brk(0x8529000) = 0x8529000
> > > > brk(0x852a000) = 0x852a000
> > > > brk(0x852b000) = 0x852b000
> > > > brk(0x852c000) = 0x852c000
> > > > brk(0x852d000) = 0x852d000
> > > > brk(0x852e000) = 0x852e000
> > > > brk(0x852f000) = 0x852f000
> > > > brk(0x8530000) = 0x8530000
> > > > ...
> > > > only brk(x) = x lines while I don't kill the process.
> > > >
> > > > Any idea? I upgraded kernel, glibc, but it didn't help. The same
> > > > compiled version runs well on another server with the same config (distro,
> > > > glibc, gcc, kernel).
> > > >
> > > > Thanks,
> > > > Thomas
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > -------------------------------------------------------
> > > > This sf.net email is sponsored by:ThinkGeek
> > > > Welcome to geek heaven.
> > > > http://thinkgeek.com/sf
> > > > _______________________________________________
> > > > Openh323gk-developer mailing list
> > > > Ope...@li...
> > > > https://lists.sourceforge.net/lists/listinfo/openh323gk-developer
> > > >
> > > --
> > > Alexander Schupp
> > > R&D
> > >
> > > ISI AG
> > > Landstrasse 1
> > > FL-9490 Vaduz
> > >
> > > Phone +423 232 97 00
> > > Mobile +43 676 514 7797
> > > Fax +423 232 97 01
> > > Email as...@is...
> > > Web www.isi-it.com
> > >
> > > ----------------------
> > > CONFIDENTIALITY NOTICE
> > > This communication contains information which is confidential and may
> > > also be privileged. It is for the exclusive use of the intended
> > > recipient(s).
> > > If you are not the intended recipient(s), please note that any
> > > distribution, copying or use of this communication or the information
> > > in it is strictly prohibited. If you have received this communication
> > > in error, please notify us immediately by returned mail or
> > > telephone number above and destroy the email and any copies of it.
> > >
> > >
> > >
> > >
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:ThinkGeek
> > > Welcome to geek heaven.
> > > http://thinkgeek.com/sf
> > > _______________________________________________
> > > Openh323gk-developer mailing list
> > > Ope...@li...
> > > https://lists.sourceforge.net/lists/listinfo/openh323gk-developer
> > >
> >
> >
> >
> > -------------------------------------------------------
> > This sf.net email is sponsored by:ThinkGeek
> > Welcome to geek heaven.
> > http://thinkgeek.com/sf
> > _______________________________________________
> > Openh323gk-developer mailing list
> > Ope...@li...
> > https://lists.sourceforge.net/lists/listinfo/openh323gk-developer
> >
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Openh323gk-developer mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openh323gk-developer
>
|