Segfault in org::w3c::dom::Node::class

David Daney ddaney@avtrex.com
Thu Oct 6 15:44:00 GMT 2005


Khiraly wrote:
> Hi!
> 
> I have a small bot, and I wanted to compile with gcj. 
> I managed it with the following command:
> gcj-4.0 --main=org.epointsystem.puli.bot.Watch @files -o lol.out
> 
> Normally the program have only one parameter: the path of the config
> file (config.xml)
> 
> ./lol.out config.xml
> 
> And the program abort:
> hub@elite:~/dcw/dcwforras$ ./lol.out config.xml
> *** DCWatch 0.1.2d
> Aborted
> hub@elite:~/dcw/dcwforras$
> 
> I can run with java (sdk 1.4) without problem.
> Here is the output of ldd:
> hub@elite:~/dcw/dcwforras$ ldd lol.out
>         linux-gate.so.1 =>  (0xffffe000)
>         libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x4001f000)
>         libgcj.so.6 => /usr/lib/libgcj.so.6 (0x4002a000)
>         libm.so.6 => /lib/tls/libm.so.6 (0x41108000)
>         libpthread.so.0 => /lib/tls/libpthread.so.0 (0x4112e000)
>         libz.so.1 => /usr/lib/libz.so.1 (0x41141000)
>         libdl.so.2 => /lib/tls/libdl.so.2 (0x41153000)
>         libc.so.6 => /lib/tls/libc.so.6 (0x41157000)
>         /lib/ld-linux.so.2 (0x40000000)
> 
> Here is the output of gdb:
> (gdb) run config.xml
> Starting program: /home/hub/dcw/dcwforras/lol.out config.xml
> [Thread debugging using libthread_db enabled]
> [New Thread 1093204192 (LWP 13521)]
> [New Thread 1103940528 (LWP 13525)]
> *** DCWatch 0.1.2d
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 1093204192 (LWP 13521)]
> 0x0805f465 in org::w3c::dom::Node::class$ ()

SIGSEGV is usually a "normal" occurrence in the gcj runtime.  On most 
platforms it is converted into a NullPointerException and then handled 
by the exception handling framework.

You should try to determine if this is just a normal instance of 
NullPointerException or if something went wrong (i.e. is the faulting 
address close to 0 which is a normal NPE).  You program is aborting 
which could indicate a problem encountered by the exception processing 
code in libgcj.  This type of problem is exceptionally fun to debug (no 
pun intended).

David Daney.



More information about the Java mailing list