SIGSEGV in GC_register_finalizer_inner

Rutger Ovidius r_ovidius@eml.cc
Mon Nov 1 17:27:00 GMT 2004


Tuesday, October 26, 2004, 10:15:00 AM, you wrote:

BH> I think it's important to track this down.

BH> But this isn't very much to go on.  Is there a reason we
BH> can't get line number information, and perhaps the value
BH> of relevant local variables?

I've relinked without stripping symbols this time. It really does take
a few days to reproduce.

The result is:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18266


BH> I think the finalizer registration code is probably being
BH> invoked from an inlined maybe_remove_all_heavy() call.
BH> But it would be good to confirm or refute that.

BH> Failing that, the disassembly of the 100 instructions or so
BH> surrounding the faulting instruction (0x40523321 in this case),
BH> together with the value of the base register for the faulting
BH> load or store, might help in manually mapping this back to
BH> the source.

I've tried to include this in the report.  If there is anything else I
can do just let me know.

BH> It would also be good to file this in bugzilla, with any
BH> additional information you can get.  Please include me on the
BH> cc: list.

Done.


Regards,
Rutger Ovidius



More information about the Java mailing list