Re: Re: Upgrading expat for apache question

On 23 March 2010 05:20, wayne collier <[email protected]> wrote:
> Okay Graham.=A0 I ran the "thread apply all bt" that dicuss in the link=
 and
> got the following:

That looks to be showing where the process got interrupted when you
attached gdb from what I can tell as got nothing to do with your ldap
code.

If you are going to access running process. You need to configure
Apache to only start a single child process.

Once attached to the child process, go 'cont' in gdb and then trigger
the request which causes it to crash. This will interrupt the gdb
session again and then you can generate the stack trace using 'where'
for current thread, or 'thread apply all bt' if need be to dump stack
for all threads.

Is this what you posted the stack trace from after doing the request
and it crashing?

Graham

> (gdb) thread apply all bt
>
> Thread 1 (Thread 0x135700 (LWP 28732)):
> #0=A0 0x001ef402 in __kernel_vsyscall ()
> #1=A0 0x00c735bb in semop () from /lib/libc.so.6
> #2=A0 0x0012420a in proc_mutex_sysv_acquire (mutex=3D0x134934)
> =A0=A0=A0 at locks/unix/proc_mutex.c:226
> #3=A0 0x00123242 in apr_proc_mutex_lock (mutex=3D0x9123098)
> =A0=A0=A0 at locks/unix/proc_mutex.c:878
> ---Type <return> to continue, or q <return> to quit---
> #4=A0 0x081ac7a0 in child_main (child_num_arg=3D<value optimized out>)
> =A0=A0=A0 at prefork.c:205
> #5=A0 0x081acb77 in make_child (s=3D0x909d9a8, slot=3D12) at prefork.c:=
746
> #6=A0 0x081ad4d0 in ap_mpm_run (_pconf=3D0x90960a8, plog=3D0x90d41a0, s=
=3D0x909d9a8)
> =A0=A0=A0 at prefork.c:881
> #7=A0 0x08088945 in main (argc=3D151601312, argv=3D0x91d5338) at main.c=
:740
> (gdb)
>
>
> The line 740 and 741 of=A0 main.c is as follows:
>
> 740 =A0=A0=A0=A0 if (ap_mpm_run(pconf, plog, server_conf))
> 741 =A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 break;
>
> Here are the lines from prefolk.c:
>
> 881=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 make_child(ap_server_c=
onf, free_slots[i]);
> 746=A0=A0=A0=A0=A0=A0=A0 child_main(slot);
> 205=A0=A0=A0 apr_status_t rv =3D apr_proc_mutex_lock(accept_mutex);
>
>
> Here are the lines from locks/unix/proc_mutex.c :
>
> 876=A0 APR_DECLARE(apr_status_t) apr_proc_mutex_lock(apr_proc_mutex_t *=
mutex)
> 877=A0 {
> 878=A0=A0=A0=A0=A0 return mutex->meth->acquire(mutex);=A0=A0=A0 #<=3D=3D=
=3D=3D=3Dline 878
> 879=A0 }
>
>
> static apr_status_t proc_mutex_sysv_acquire(apr_proc_mutex_t *mutex)
> {
> =A0=A0=A0 int rc;
>
> =A0=A0=A0 do {
> =A0=A0=A0=A0=A0=A0=A0 rc =3D semop(mutex->interproc->filedes, &proc_mut=
ex_op_on, 1);
> #<=3D=3D=3D=3D=3Dline 226
> =A0=A0=A0 } while (rc < 0 && errno =3D=3D EINTR);
> =A0=A0=A0 if (rc < 0) {
> =A0=A0=A0=A0=A0=A0=A0 return errno;
> =A0=A0=A0 }
> =A0=A0=A0 mutex->curr_locked =3D 1;
> =A0=A0=A0 return APR_SUCCESS;
> }
>
>
> Can you detect what might be an issue from?
>
> Wayne
>
> Graham Dumpleton wrote:
>
> On 19 March 2010 07:48, wayne collier <[email protected]> wrote:
>
>
> Graham -my understanding of RH apache=A0 is=A0 modules have to=A0=A0 ex=
plicitly
> loaded=A0 in httpd.con file.=A0 I check all the snippet=A0 files=A0 and=
=A0 no=A0 php
> shared object file is being loaded.=A0 I noticed before that did not se=
em to
> have any mod_authz_ldap shared object loaded in my configuration file. =
Do I
> need to set things up like i am doing authorization and authentication =
thru
> apache?=A0 Is possible that needs to be included?
>
> Also, I went thru some of the troubleshooting steps. I ran my web code =
to
> reproduce the seg fault and then gbc to attach the process id.=A0 I got=
 the
> following before getting really lost here. Any suggestions here would
> greatly appreciated. Thanks.
>
> Wayne
>
> GBD WORK .....
>
> Detaching from program: /usr/local/httpd-2.2.11/bin/httpd, process 1842=
1
> [root@mingus psp]# gdb /usr/local/httpd-2.2.11/bin/httpd 18421
> GNU gdb Fedora (6.8-37.el5)
> Copyright (C) 2008 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later
> <http://gnu.org/licenses/gpl.html>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law.=A0 Type "show cop=
ying"
> and "show warranty" for details.
> This GDB was configured as "i386-redhat-linux-gnu"...
> Attaching to program: /usr/local/httpd-2.2.11/bin/httpd, process 18421
>
> warning: .dynamic section for "/usr/lib/libgssapi_krb5.so.2" is not at =
the
> expected address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/usr/lib/libkrb5.so.3" is not at the exp=
ected
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/lib/libcom_err.so.2" is not at the expe=
cted
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/usr/lib/libk5crypto.so.3" is not at the
> expected address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/lib/libresolv.so.2" is not at the expec=
ted
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/usr/lib/libkrb5support.so.0" is not at =
the
> expected address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/lib/libkeyutils.so.1" is not at the exp=
ected
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/lib/libselinux.so.1" is not at the expe=
cted
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/lib/libsepol.so.1" is not at the expect=
ed
> address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
>
> warning: .dynamic section for "/usr/lib/liblber-2.3.so.0" is not at the
> expected address
>
> warning: difference appears to be caused by prelink, adjusting expectat=
ions
> Reading symbols from /lib/libm.so.6...done.
> Loaded symbols for /lib/libm.so.6
> Reading symbols from /usr/local/httpd-2.2.11/lib/libaprutil-1.so.0...do=
ne.
> Loaded symbols for /usr/local/httpd-2.2.11/lib/libaprutil-1.so.0
> Reading symbols from /lib/libexpat.so.0...done.
> Loaded symbols for /lib/libexpat.so.0
> Reading symbols from /usr/local/httpd-2.2.11/lib/libapr-1.so.0...done.
> Loaded symbols for /usr/local/httpd-2.2.11/lib/libapr-1.so.0
> Reading symbols from /lib/libuuid.so.1...done.
> Loaded symbols for /lib/libuuid.so.1
> Reading symbols from /lib/librt.so.1...done.
> Loaded symbols for /lib/librt.so.1
> Reading symbols from /lib/libcrypt.so.1...done.
> Loaded symbols for /lib/libcrypt.so.1
> Reading symbols from /lib/libpthread.so.0...done.
> [Thread debugging using libthread_db enabled]
> [New Thread 0x135700 (LWP 18421)]
> Loaded symbols for /lib/libpthread.so.0
> Reading symbols from /lib/libdl.so.2...done.
> Loaded symbols for /lib/libdl.so.2
> Reading symbols from /lib/libc.so.6...done.
> Loaded symbols for /lib/libc.so.6
> Reading symbols from /lib/ld-linux.so.2...done.
> Loaded symbols for /lib/ld-linux.so.2
> Reading symbols from /lib/libnss_files.so.2...done.
> Loaded symbols for /lib/libnss_files.so.2
>
>
> (gdb) print PyRun_SimpleString("import traceback; traceback.print_stact=
()")
>
> Program received signal SIGSEGV, Segmentation fault.
> PyImport_AddModule (name=3D0x5596b3 "__main__") at Python/import.c:366
> 366=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 PyInterpreterState *interp =3D =
PyThreadState_GET()->interp;
> The program being debugged was signaled while in a function called from=
 GDB.
> GDB remains in the frame where the signal was received.
> To change this behavior use "set unwindonsignal on"
> Evaluation of the expression containing the function (PyRun_SimpleStrin=
g)
> will be abandoned.
> (gdb)
>
>
> I don't want a Python stack trace, I want a C stack trace as
> documented in that page I sent you to. Use:
>
>   (gdb) thread apply all bt
>
> Graham
>
>
>
> Graham Dumpleton wrote:
>
> On 13 March 2010 11:28, wayne collier <[email protected]> wrote:
>
>
> Okay I am running RH linux and I don't seem to have mod_ldap or
> mod_authnz_ldap shared objects loaded in my httpd.conf file.
>
>
> They may not be loaded from httpd.conf but from a configuration file
> snippet loaded into Apache. I don't know RH Apache configuration
> layout, but check all snippet files in subdirectories to the
> httpd.conf directory.
>
> The conflict may also be because you are loading mod_php.
>
> Ultimately you may have to use gdb to work out exactly where it
> crashes. Hopefully not too much symbol information has been stripped
> and gdb will be able to tell you from which library LDAP code that is
> crashing is coming from.
>
> Use first documented procedure for running Apache in single process
> under gdb in:
>
>
> http://code.google.com/p/modwsgi/wiki/DebuggingTechniques#Debugging_Cra=
shes_With_GDB
>
> This is for mod_wsgi, but same can be done for mod_python.
>
> Graham
>
>
>
> Also I ran ldd
> and got the following results:
>
>
>
>
> ldd /usr/local/httpd-2.2.11/bin/httpd
>
>
> =A0=A0=A0 linux-gate.so.1 =3D>=A0 (0x00157000)
> =A0=A0=A0 libm.so.6 =3D> /lib/libm.so.6 (0x00ce8000)
> =A0=A0=A0 libaprutil-1.so.0 =3D> /usr/local/httpd-2.2.11/lib/libaprutil=
-1.so.0
> (0x00315000)
> =A0=A0=A0 libexpat.so.0 =3D> /lib/libexpat.so.0 (0x00d70000)
> =A0=A0=A0 libapr-1.so.0 =3D> /usr/local/httpd-2.2.11/lib/libapr-1.so.0 =
(0x00844000)
> =A0=A0=A0 libuuid.so.1 =3D> /lib/libuuid.so.1 (0x002f2000)
> =A0=A0=A0 librt.so.1 =3D> /lib/librt.so.1 (0x00d65000)
> =A0=A0=A0 libcrypt.so.1 =3D> /lib/libcrypt.so.1 (0x07e07000)
> =A0=A0=A0 libpthread.so.0 =3D> /lib/libpthread.so.0 (0x00d17000)
> =A0=A0=A0 libdl.so.2 =3D> /lib/libdl.so.2 (0x00d11000)
> =A0=A0=A0 libc.so.6 =3D> /lib/libc.so.6 (0x00ba0000)
> =A0=A0=A0 /lib/ld-linux.so.2 (0x00b82000)
>
>
>
> ldd /usr/local/httpd-2.2.11/modules/mod_python.so
>
>
> =A0=A0=A0 linux-gate.so.1 =3D>=A0 (0x00aec000)
> =A0=A0=A0 libpthread.so.0 =3D> /lib/libpthread.so.0 (0x00730000)
> =A0=A0=A0 libdl.so.2 =3D> /lib/libdl.so.2 (0x0027f000)
> =A0=A0=A0 libutil.so.1 =3D> /lib/libutil.so.1 (0x00f70000)
> =A0=A0=A0 libm.so.6 =3D> /lib/libm.so.6 (0x009b1000)
> =A0=A0=A0 libc.so.6 =3D> /lib/libc.so.6 (0x00110000)
> =A0=A0=A0 /lib/ld-linux.so.2 (0x00b82000)
>
> I don't seem to have anything with ldap listed. Am I on the right track=
.?
>
>
> Wayne
>
> Graham Dumpleton wrote:
>
> A segmentation fault can occur for other reasons. If it is happening
> when you use LDAP, it could be a conflict in versions of LDAP
> libraries being used by Apache or PHP and Python LDAP module.
>
> You should use the 'ldd' command to determine what LDAP shared library
> is used by mod_authnz_ldap and compare that to the LDAP library being
> used by the C extension module component of the Python module you are
> using to access LDAP. Similarly find any PHP extension modules which
> use LDAP and look at those.
>
> If mod_authnz_ldap is loaded into Apache and you don't need it, then
> don't load it and see if problem goes away. Also try not loading PHP
> if it is being loaded. Narrow down which is in conflict.
>
> As far as I remember, Apache itself doesn't link to LDAP library and
> even possible that mod_authnz_ldap uses an internal LDAP library
> rather than linking external library. All the same even use ldd on
> Apache binary itself to see which LDAP library may be linked.
>
> Graham
>
> On 10 March 2010 04:18, wayne collier <[email protected]> wrote:
>
>
> Okay I guess I will look else where. However, I keep getting segmentati=
on
> faults when try an execute the ldap command
>
> t=3Dl.simple_bind_s(dn,pw). It is only line of code that crashes. If th=
e have
> the classic expat problem, would that mean I would get a segmentation
> fault problem from the start.  Also, ldap script works okay as a stand =
alone
> script. Thanks.
>
> Wayne
>
>
> Graham Dumpleton wrote:
>
> That is not your problem, quoting:
>
> http://code.google.com/p/modwsgi/wiki/IssuesWithExpatLibrary
>
> which is more up to date in respect of this issue, even if the
> mod_wsgi variant of that document:
>
> """Note that this only applies to Python versions prior to Python 2.5.
> >From Python 2.5 onwards, the copy of the "expat" library bundled in
> with Python is name space prefixed, thereby avoid name clashes with an
> "expat" library which has previously been loaded."""
>
> You are using Python 2.6, so you cannot be affected by this problem.
>
> Graham
>
> On 9 March 2010 05:54, wayne collier <[email protected]> wrote:
>
>
> Hello all. I need some help solving the dreaded segmentation fault that=
 has
> cropped several time on this board. I have read the link
> http://www.dscpl.com.au/wiki/ModPython/Articles/ExpatCausingApacheCrash=
 and
> followed the step and have uncovered the following.
>
>
>
>
> ldd=A0=A0=A0 /usr/local/httpd-2.2.11/bin/httpd | grep expat
>
>
> =A0=A0=A0 libexpat.so.0 =3D> /lib/libexpat.so.0 (0x00d70000)
>
>
>
> strings /lib/libexpat.so.0=A0 | grep expat_
>
>
> expat_1.95.8
>
>
>
> =A0 ps aux | grep http | head -3
>
>
> webuser=A0 14986=A0 0.0=A0 2.1=A0 54236 19048 ?=A0=A0=A0=A0=A0=A0=A0 S=A0=
=A0=A0 Mar04=A0=A0 0:01
> /usr/local/httpd-2.2.11/bin/httpd -d /usr/local/httpd-2.2.11 -f
> /usr/local/httpd-2.2.11/conf/httpd.conf
> webuser=A0 14987=A0 0.0=A0 3.2 144128 29540 ?=A0=A0=A0=A0=A0=A0=A0 S=A0=
=A0=A0 Mar04=A0=A0 0:01
> /usr/local/httpd-2.2.11/bin/httpd -d /usr/local/httpd-2.2.11 -f
> /usr/local/httpd-2.2.11/conf/httpd.conf
> webuser=A0 14998=A0 0.0=A0 3.2 144212 29660 ?=A0=A0=A0=A0=A0=A0=A0 S=A0=
=A0=A0 Mar04=A0=A0 0:01
> /usr/local/httpd-2.2.11/bin/httpd -d /usr/local/httpd-2.2.11 -f
> /usr/local/httpd-2.2.11/conf/httpd.conf
>
>
>
> =A0 /usr/sbin/lsof -p 14986 | grep expat
>
>
> httpd=A0=A0 14986 webuser=A0 mem=A0=A0=A0 REG=A0=A0=A0=A0 3,5=A0=A0 572=
313=A0 972637
> /opt/Python/lib/python2.6/lib-dynload/pyexpat.so
> httpd=A0=A0 14986 webuser=A0 mem=A0=A0=A0 REG=A0=A0=A0=A0 3,5=A0=A0 133=
120 1883791
> /lib/libexpat.so.0.5.0
>
> I am not sure why I get 2 shared libaries here.
>
>
>
> =A0 strings=A0 /lib/libexpat.so.0.5.0 | grep expat_
>
>
> expat_1.95.8
>
>
>
>
> strings /opt/Python/lib/python2.6/lib-dynload/pyexpat.so | grep expat_
>
>
> pyexpat.expat_CAPI 1.0
> expat_CAPI
> expat_2.0.0
>
>
>
>
> import pyexpat
> pyexpat.version_info
>
>
> (2, 0, 0)
>
>
> So it seems like I need get expat_1.95.8 apache is using up to expat_2.=
0.0.
> Can I simply try replacing the 1.95.8 library with 2.0.0. Or do I have
> recompile with Apache? If I do recompile I need to recompile apache wit=
h
> mod_python as well?
>
>
>
>
>
>
> _______________________________________________
> Mod_python mailing list
> [email protected]
> http://mailman.modpython.org/mailman/listinfo/mod_python
>
>
>
>
>
> _______________________________________________
> Mod_python mailing list
> [email protected]
> http://mailman.modpython.org/mailman/listinfo/mod_python
>
>
>
>
>
> _______________________________________________
> Mod_python mailing list
> [email protected]
> http://mailman.modpython.org/mailman/listinfo/mod_python
>
>
>
>
>
> _______________________________________________
> Mod_python mailing list
> [email protected]
> http://mailman.modpython.org/mailman/listinfo/mod_python
>
>
>