On 12/31/06, Brian May <[EMAIL PROTECTED]> wrote:
Zack> 2) Run the crashing command under valgrind - valgrind does support PPC
Zack> now, I believe - and get us a backtrace from that.
==6046== Memcheck, a memory error detector.
==6046== Copyright (C) 2002-2006, and GNU GPL'd, by Julian Seward et al.
==6046== Using LibVEX rev 1658, a library for dynamic binary translation.
==6046== Copyright (C) 2004-2006, and GNU GPL'd, by OpenWorks LLP.
==6046== Using valgrind-3.2.1-Debian, a dynamic binary instrumentation
framework.
==6046== Copyright (C) 2000-2006, and GNU GPL'd, by Julian Seward et al.
==6046== For more details, rerun with: -v
... libc noise ...
==6046==
==6046== Invalid read of size 4
==6046== at 0x101CFEA0: boost::detail::shared_count::~shared_count() (in
/usr/bin/mtn)
==6046== by 0x10226778: (within /usr/bin/mtn)
==6046== by 0x1022ACE0: (within /usr/bin/mtn)
==6046== by 0x10225C78: (within /usr/bin/mtn)
==6046== by 0x1022B06C: (within /usr/bin/mtn)
==6046== by 0x10208094: (within /usr/bin/mtn)
==6046== by 0x102091A0: (within /usr/bin/mtn)
==6046== by 0x1012A670: (within /usr/bin/mtn)
==6046== by 0x1012B05C: (within /usr/bin/mtn)
==6046== by 0x1012B278: (within /usr/bin/mtn)
==6046== by 0x100F6ED4: (within /usr/bin/mtn)
==6046== by 0x100F70E4: (within /usr/bin/mtn)
==6046== Address 0x454BDF4 is 4 bytes inside a block of size 16 free'd
==6046== at 0xFFBB1D0: operator delete(void*) (vg_replace_malloc.c:244)
==6046== by 0x10226158: (within /usr/bin/mtn)
==6046== by 0x100103C0: boost::detail::sp_counted_base::destroy() (in
/usr/bin/mtn)
==6046== by 0x1022B0A4: (within /usr/bin/mtn)
==6046== by 0x10208094: (within /usr/bin/mtn)
==6046== by 0x102091A0: (within /usr/bin/mtn)
==6046== by 0x1012A670: (within /usr/bin/mtn)
==6046== by 0x1012B05C: (within /usr/bin/mtn)
==6046== by 0x1012B278: (within /usr/bin/mtn)
==6046== by 0x100F6ED4: (within /usr/bin/mtn)
==6046== by 0x100F70E4: (within /usr/bin/mtn)
==6046== by 0x100F7640: (within /usr/bin/mtn)
This is where it goes off the rails. All the valgrind chatter after
this point is secondary damage - uninteresting. Unfortunately, you
used binaries with no debugging symbols so the backtrace is less than
useful. Is there any way you can build a local binary with debug
symbols - from the debian source package, if you can't get a checkout
of monotone HEAD - repeat the test and send us a more useful
backtrace?
thanks,
zw
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
