Hi DanB,

Good to hear it fixed your problem. I'm currently fighting another crash in the same area. Currently SEMS segfaults when XmlRpc calls close() on the XmlRpcSource object:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1217008720 (LWP 12051)]
0xb7c8c07a in SSL_shutdown () from /usr/lib/i686/cmov/libssl.so.0.9.8
(gdb) bt
#0  0xb7c8c07a in SSL_shutdown () from /usr/lib/i686/cmov/libssl.so.0.9.8
#1 0xb7cca05c in XmlRpc::XmlRpcSource::close (this=0x746e6f43) at src/XmlRpcSource.cpp:35 #2 0xb7cc3269 in XmlRpc::XmlRpcDispatch::waitForAndProcessEvents (this=0x8128490, timeout=-1) at src/XmlRpcDispatch.cpp:260 #3 0xb7cc34f2 in XmlRpc::XmlRpcDispatch::work (this=0x8128490, timeout=-1) at src/XmlRpcDispatch.cpp:87 #4 0xb7cb1256 in XmlRpc::WorkerThread::run (this=0x8128400) at MultithreadXmlRpcServer.cpp:34
#5  0x080eec15 in AmThread::_start (_t=0x8128400) at AmThread.cpp:70
#6  0xb7f07240 in start_thread () from /lib/tls/i686/cmov/libpthread.so.0
#7  0xb7dac49e in clone () from /lib/tls/i686/cmov/libc.so.6

For some reason it evaluates (_ssl_ssl != (SSL *) NULL) as true in XmlRpcSource.cpp:34 to be able to reach line the SSL_shutdown at line 35. In the constructor of XmlRpcSource the _ssl_ssl parameter is set to NULL (see revision r1249) and I can't find any other place where _ssl_ssl is changed to another value.

This happened somewhere in the middle of a test run of 40.000 calls. But it also happened after approx. 1500 calls. Does anybody have a clue what's going on here? Is there anything I can do to further pinpoint the problem?

TIA,

Tom

-------- Originele bericht --------
Onderwerp: Re: [Sems] Xmlrpc causing random crash
Van: Dan-Cristian Bogos <[email protected]>
Aan: [email protected]
CC: [email protected]
Datum: 2-6-2009 9:53
Hi Tom,

many thanks for the tip.

I have checked out the r1269 and installed xmlrpc2di on top of the 1.1.0
engine I had from the packages and it looks like a successful recipe: I
have no more crashes. You saved me a loot of time for
troubleshooting ;-).

Since it looks like a fixed bug, is there any plan to release a package
with the fix included anytime soon?

Ta,
DanB


On Mon, 2009-06-01 at 22:29 +0200, Tom van der Geer wrote:
Hi DanB,

I had the same, or at least a very similar issue a while ago. In my case
it turned out that I was missing a fix which is not in the 1.1.0 tarbal,
but it's available when you checkout branches/1.1 from subversion. The
relevant change set is r1269:
http://svn.berlios.de/wsvn/sems/branches/1.1/apps/xmlrpc2di/?op=log&rev=1269&sc=1&isdir=1

Might be the same problem that you're having now...
Best regards,

Tom


Dan-Cristian Bogos schreef:
Guys,

I have recently upgraded from sems 1.0 running own compiled xmlrpc to
the package provided from iptel.org website (running debian lenny).

After update, I cannot longer run it in production since sems is
randomly crashing (quite often I would say, but without being able to
spot any pattern).
I was running both in debug and daemon mode, and noticed that the
crashes are more often in daemon than debug one.
All I could spot was in my syslog the following sentence(when in debug):
"kernel: [4822269.946788] sems[27927]: segfault at 76726573 ip b7a6d935
sp b76e7010 error 4 in libssl.so.0.9.8[b7a3e000+42000]"

I run the following versions:
ii  sems                              1.1.0-1                   SIP
Express Media Server, very fast and flexible SIP media ser
ii  sems-python-modules               1.1.0-1                   contains
the python modules for SEMS
ii  libssl0.9.8                       0.9.8g-16                 SSL
shared libraries

Anyone there having same issues?

I would appreciate any kind of tips.

Ta,
DanB

_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems



_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems

Reply via email to