#336 OpenSIPS 1.6.4 fails with Signal 10 (Bus Error) on Solaris

[Sergio Gutierrez]

Trying to run OpenSIPS 1.6.4 on Solaris 10, it is failing with Bus Error (Signal 10)
Attached appear log and backtrace from the core.

[Sergio Gutierrez]

OpenSIPS output and Backtrace

[Bogdan-Andrei Iancu]

In GDB, please print ujiffies :

p ujiffies;

Thanks and regards,
Boogdan

[Sergio Gutierrez]

Hello Bogdan.

This is the print you requested:

(gdb) p ujiffies
$1 = (utime_t *) 0xfcc040ec
(gdb) x 0xfcc040ec
0xfcc040ec: 0x00000000
(gdb) x ujiffies
0xfcc040ec: 0x00000000
(gdb) x *ujiffies
0x0: Cannot access memory at address 0x0
(gdb) p *ujiffies
$2 = 0
(gdb)

[Bogdan-Andrei Iancu]

Please print "jiffies" also,

Another question - are you running on a 32 or 64 arch ?

Thanks and regards,
Bogdan

[Sergio Gutierrez]

Hello Bogdan

I am running on SPARC, but OpenSIPS is compiled with 32 bits default compilation.

(gdb) p ujiffies
$1 = (utime_t *) 0xfcc040ec
(gdb) x 0xfcc040ec
0xfcc040ec: 0x00000000
(gdb) x ujiffies
0xfcc040ec: 0x00000000
(gdb) x *ujiffies
0x0: Cannot access memory at address 0x0
(gdb) p *ujiffies
$2 = 0

[Sergio Gutierrez]

Hello Bogdan.

Wrong copy & paste

(gdb) print jiffies
$1 = (unsigned int *) 0xfcc040d8
(gdb) x 0xfcc040d8
0xfcc040d8: 0x00000000
(gdb) print *jiffies
$2 = 0
(gdb) x *jiffies
0x0: Cannot access memory at address 0x0
(gdb)

[Bogdan-Andrei Iancu]

It seams it is a issue with the arch detection - I guess "sparc" is 64b arch and when opensips compiles as for 32b, some mem alignment is screwed.

Do you use any particular compile options or you let the Makefile to do all the detection ?

Regards,
Bogdan

[Sergio Gutierrez]

Hello Bogdan.

Previously we have not had problems using OpenSIPS compiled for 32bits on Sparc. I am not using any particular option; Makefile is performing the guessing itself.

I will try to compile and test on 64 bits.

Thanks and regards.

[Sergio Gutierrez]

Hello Bogdan.

Today I recompiled OpenSIPS by enforcing 64bits compilation (gcc -m64 flag), and OpenSIPS run without problems.

I think you are right, and the problem is related to variables alignment, but, I am not sure why.

Regards.

[Bogdan-Andrei Iancu]

Hi Sergio,

For your system, can you run the print here:
uname -p
uname -m

Thanks and regards,
Bogdan