Re: assertion: z->z_Magic == ZALLOC_SLAB_MAGIC in _slabfree

["Magliano Andre'" <masterblaster@xxxxxxxxxx>]

Hi Peter,

On 3/24/2011, "Peter Avalos" <peter@theshell.com> wrote:

>On Thu, Mar 24, 2011 at 10:00:59AM +0100, Magliano Andre' wrote:
>> Hi Peter,
>> 
>> it seems to me (if i don't see ghosts) that the problem is at:
>> 
>> #6 0x2ac85455 in operator delete (ptr=0x0)
>> 
>> but if i look in
>> /usr/src/contrib/gcc-4.4/libstdc++-v3/libsupc++/del_op.cc:44
>> i see:
>> 
>> _GLIBCXX_WEAK_DEFINITION void
>> operator delete(void* ptr) throw ()
>> {
>>   if (ptr)
>>     std::free(ptr);
>> }
>> 
>> i.e. checking against NULL pointer is done, so it should'nt happen.
>> 
>
>I'm not sure what's going on there...
>
>#5  0x2821bd7b in free (ptr=0x2abd81bc) at /usr/src/lib/libc/../libc/stdlib/nmalloc.c:774
>#6  0x2ac93455 in operator delete (ptr=0x0)
>    at /usr/src/gnu/lib/gcc44/libstdc++/../../../usr.bin/cc44/cc_tools/../../../../contrib/gcc-4.4/libstdc++-v3/libsupc++/del_op.cc:44
>#7  0x2ac27385 in __gnu_cxx::new_allocator<char>::deallocate (this=0x2abd81bc, __a=...) at /usr/obj/usr/src/world_i386/usr/include/c++/4.4/ext/new_allocator.h:95
>
>Notice that in #7, you have this=0x2abd81bc and in #5 you have
>ptr=0x2abd81bc.  Not sure how to explain that...

this seems to me procedure call stack corruption, which i experienced
some times in case of

- writing data out of bounds (array index out of range for example)
- binary mismatch (this case?)

>> What happens if you recompile vlc with gcc 4.4?
>> 
>
>I can't compile vlc any more.  It fails during the build.  The 2nd
>backtrace is when I attempt to build vlc (it tries to run some program
>as part of the build that winds up hitting the assertion).

Well, maybe it would be worth fixing vlc compilation with gcc44 if
there's even the suspect of hunting a ghost...


ByE!