Re: core dumping inside Performer

New Message Reply Date view Thread view Subject view Author view

John Rohlf (jrohlf++at++tubes)
Wed, 16 Nov 94 11:41:47 PST

>
> Hi
>
> I tried to link my Performer application with malloc_db.o from the Dev. Toolbox
> (in search of some malloc problems).
> Suprisingly, Performer crashed inside pfAllocChanData, upon tying to
> free (NULL). It doesn't do so when I link without malloc_db.o .
>
> This is the 'where' output of dbx :
>
> > 0 db_unlink(ptr = 0xffffffec) [0x43786c]
> 1 free(vptr = (nil)) ["malloc_db.c":503, 0x43780c]
> 2 operator delete(void*)(d = (nil)) ["../../../lib/libpf/pfProcess.C":3442,
> 0x4dd69c]
> 3 pfChannel::newPassThruData(long)(this = 0x20f76a0, size = 8)
> ["../../../lib/libpf/pfChannel.C":1291, 0x4f0ae8]
> 4 pfAllocChanData(_chan = 0x20f76a0, size = 8)
> ["../../../lib/libpf/cChannel.C":226, 0x4cd440]
> 5 nova_new_channel(pipe = 0x1f75c10, type = 2)
> ["/tmp_mnt/usr/login/otw/nova/nova_interface.c":3875, 0x454fdc]
>
> What it seems to say is that newPassThruData does delete of a nil pointer,
> which in turn calls free with that nil pointer.

This looks like a bug in malloc_db since the malloc man page says:

     Undefined results will occur if the space assigned by malloc is overrun
     or if some random number is handed to free. It is always permitted to
     pass NULL to free.

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:50:40 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.