1 Jul
2012
1 Jul
'12
3:54 a.m.
Just to let you know
I made a lot of tests using different modules including carrierroute, I do have the bus
error also after the first call and core dump.
thank you
O
On Jun 27, 2012, at 11:45 PM, Akan wrote:
Daniel,
i did a repull from git, compiled, installed and re-test of Kamailio and got the same
error in the same program's location as previous (in tm_funcs.c). Here are the
commands that I entered to make sure I did everything correctly:
mkdir -p /usr/local/src/kamailio/kamailio-3.2-noalignment
cd /usr/local/src/kamailio/kamailio-3.2-noalignment
git clone --depth 1 git://git.sip-router.org/sip-router kamailio
cd kamailio
|gmake proper
gmake prefix=/opt/kamailio-3.2 FLAVOUR=kamailio cfg
vi modules.lst
| gmake prefix=/opt/kamailio-3.2 Q=0 CC_EXTRA_OPTS=|-mno-faster-structs all
|gmake prefix=/opt/kamailio-3.2 Q=0 CC_EXTRA_OPTS=|-mno-faster-structs install
Makefile:363: config.mak: No such file or directory
normal Makefile.defs exec
target architecture <sparc64>, host architecture <sparc64>
generating autover.h ...
making config...
Makefile.defs defs skipped
gcc -g -O9 -funroll-loops -m64 -mcpu=ultrasparc -mtune=ultrasparc -fno-strict-overflow
-ftree-vectorize -Wall -mno-faster-structs -DNAME='"ser"'
-DVERSION='"3.2.3"' -DARCH='"sparc64"'
-DOS='solaris_' -DOS_QUOTED='"solaris"'
-DCOMPILER='"gcc 4.6.2"' -D__CPU_sparc64 -D__OS_solaris
-DSER_VER=3002003 -DCFG_DIR='"/opt/kamailio-3.2/etc/ser/"' -DPKG_MALLOC
-DSHM_MEM -DSHM_MMAP -DDNS_IP_HACK -DUSE_IPV6 -DUSE_MCAST -DUSE_TCP -DDISABLE_NAGLE
-DHAVE_RESOLV_RES -DUSE_DNS_CACHE -DUSE_DNS_FAILOVER -DUSE_DST_BLACKLIST -DUSE_NAPTR
-DWITH_XAVP -DDBG_QM_MALLOC -DUSE_TLS -DTLS_HOOKS -DFAST_LOCK -DADAPTIVE_WAIT
-DADAPTIVE_WAIT_LOOPS=1024 -DCC_GCC_LIKE_ASM -DSPARC64_MODE -DHAVE_GETIPNODEBYNAME
-DHAVE_SYS_SOCKIO_H -DHAVE_SCHED_YIELD -DHAVE_ALLOCA_H -DUSE_SIGACTION -DHAVE_DEVPOLL
-DHAVE_SELECT -DHAVE_FILIO_H -c endianness.c -o endianness.o -MMD -MP
gcc -g -O9 -funroll-loops -m64 -mcpu=ultrasparc -mtune=ultrasparc -fno-strict-overflow
-ftree-vectorize -Wall -mno-faster-structs -DNAME='"ser"'
-DVERSION='"3.2.3"' -DARCH='"sparc64"'
-DOS='solaris_' -DOS_QUOTED='"solaris"'
-DCOMPILER='"gcc 4.6.2"' -D__CPU_sparc64 -D__OS_solaris
-DSER_VER=3002003 -DCFG_DIR='"/opt/kamailio-3.2/etc/ser/"' -DPKG_MALLOC
-DSHM_MEM -DSHM_MMAP -DDNS_IP_HACK -DUSE_IPV6 -DUSE_MCAST -DUSE_TCP -DDISABLE_NAGLE
-DHAVE_RESOLV_RES -DUSE_DNS_CACHE -DUSE_DNS_FAILOVER -DUSE_DST_BLACKLIST -DUSE_NAPTR
-DWITH_XAVP -DDBG_QM_MALLOC -DUSE_TLS -DTLS_HOOKS -DFAST_LOCK -DADAPTIVE_WAIT
-DADAPTIVE_WAIT_LOOPS=1024 -DCC_GCC_LIKE_ASM -DSPARC64_MODE -DHAVE_GETIPNODEBYNAME
-DHAVE_SYS_SOCKIO_H -DHAVE_SCHED_YIELD -DHAVE_ALLOCA_H -DUSE_SIGACTION -DHAVE_DEVPOLL
-DHAVE_SELECT -DHAVE_FILIO_H -c select_buf.c -o select_buf.o -MMD -MP
||
Thanks
Nathaniel
|
On 6/25/2012 3:05 AM, Daniel-Constantin Mierla wrote:
Hello,
thanks for troubleshooting further -- as I expected, it is a memalign problem, but some
confusing reports by not using always the patches I made to registrar module to align the
structure, made thinking is something else. Now is no longer in registrar, but in another
module.
There might be many of these, can you try to compile first without strict alignment to 8
bytes? Checking quickly to gcc, the option should be |-mno-faster-structs
So try:
make proper
make FLAVOUR=kamailio cfg|
make Q=0 CC_EXTRA_OPTS=|-mno-faster-structs all|
...
By providing Q=0, you will see all compile flags, verify that |-mno-faster-structs is
there.
Cheers,
Daniel
|
On 6/22/12 3:26 PM, Akan wrote:
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users(a)lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
Hello,
After doing some research, this is what I found out. On Solaris Sparc 64bit system, there
is a mandatory alignment of memory accesses and also for data types. I went thru the core
dump, disassembled the code and located the instruction that produced the error. The
registers addresses in questioned are on a 4 byte alignment but not an 8 byte or 16 byte
alignment. .The earlier patch must have forced the alignment which is why the error did
not occur in the program common.c. It looks like there is a similar situation in
t_funcs.c.
Here is a link that I found that can better explain:
http://blog.jgc.org/2007/04/debugging-solaris-bus-error-caused-by.html
Core was generated by `/opt/kamailio-3.2/sbin/kamailio'.
Program terminated with signal 10, Bus error.
#0 0xffffffff7bd2b7bc in t_relay_to (p_msg=0x10047c698, proxy=0x0, proto=0, replicate=0)
at t_funcs.c:352
352 if (!t_reply( t, p_msg , 100 ,
0xffffffff7bd2b7b0 <+976>: ldx [ %l7 + %g1 ], %g1
0xffffffff7bd2b7b4 <+980>: ldx [ %g1 ], %g1
0xffffffff7bd2b7b8 <+984>: call 0xffffffff7be9bc80 <t_reply@plt>
*=> 0xffffffff7bd2b7bc <+988>: ldx [ %g1 + 0x38 ], %o3*
0xffffffff7bd2b7c0 <+992>: cmp %o0, 0
0xffffffff7bd2b7c4 <+996>: be,pn %icc, 0xffffffff7bd2bb14
<t_relay_to+1844>
0xffffffff7bd2b7c8 <+1000>: ldx [ %fp + 0x7f7 ], %o0
0xffffffff7bd2b7cc <+1004>: b %xcc, 0xffffffff7bd2b460 <t_relay_to+128>
End of assembler dump.
(gdb) info registers g1
g1 0xffffffff7666c3e4 -2308520988
(gdb) info registers o3
o3 0x18f3d0 1635280
I hope this helps in trying to resolve this problem and to find a solution. Also, if you
need a sparc system to test with, just let me know. We can help from a sparc perspective.
Thanks
Nathaniel
--
Daniel-Constantin Mierla -http://www.asipto.com
http://twitter.com/#!/miconda -http://www.linkedin.com/in/miconda
Kamailio Advanced Training, Seattle, USA, Sep 23-26, 2012 -http://asipto.com/u/katu
Kamailio Practical Workshop, Netherlands, Sep 10-12, 2012 -http://asipto.com/u/kpw
No virus found in this message.
Checked by AVG - www.avg.com <http://www.avg.com>
Version: 2012.0.2180 / Virus Database: 2437/5091 - Release Date: 06/24/12