[ask for help] threaded app crashes with pretty null backtrace.

[PaweÅ Sikora <pluto at agmk dot net>]

Hi,

i'd like to ask glibc's hackers if the glibc has some internal machinery
to debug threads or guard stacks? i have a buggy application which crashes
in ugly way. gdb and valgrind didn't help  :(

(gdb) r
Starting program: /home/users/pawels/6107/testbench64.x
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7fffecfe5700 (LWP 850)]
[New Thread 0x7fffec7e4700 (LWP 851)]
[New Thread 0x7fffebfe3700 (LWP 852)]
[New Thread 0x7fffeb7e2700 (LWP 871)]
[New Thread 0x7fffeafe1700 (LWP 872)]
[New Thread 0x7fffea7e0700 (LWP 873)]
[New Thread 0x7fffe9f54700 (LWP 1031)]

Program received signal SIGSEGV, Segmentation fault.
0x0000000000000000 in ?? ()
(gdb) bt
#0  0x0000000000000000 in ?? ()
#1  0x0000000000000000 in ?? ()

other threads sleep on select(), pthread_cond_wait() and nanosleep().

valgrind notices some stack changes (probably due to http://en.wikipedia.org/wiki/SystemC
kernel activity).

==24140== Warning: client switching stacks?  SP change: 0x13c675d8 --> 0x8baaac8
==24140==          to suppress, use: --max-stackframe=185322256 or greater
==24140== Warning: client switching stacks?  SP change: 0x7fefcfce8 --> 0x13c88f70
==24140==          to suppress, use: --max-stackframe=34010852728 or greater
==24140==          further instances of this message will not be shown.
==24140== Jump to the invalid address stated on the next line
==24140==    at 0x0: ???
==24140==  Address 0x0 is not stack'd, malloc'd or (recently) free'd

program was compiled with -g2 -O{0,1} -fno-inline -fstack-protector-all -D_FORTIY_SOURCE=2
used software: gcc-4.6, glibc-2.15, gdb-7.4.50-20120120, kernel-3.2.12.


is there any other way to catch the first 0x0 code access?

thanks in advance for any hints.

BR,
PaweÅ.