Re: [gift-users] giftd core dump (ds_data_free(data=0x0))

[Greg W. <gr...@wo...>]

Craig Barnes (hb...@so...) wrote:

> Woah, malloc() must have failed allocating node->key in
> dataset.c:new_node() and then the error cleanup path wasn't prepared to
> receive a NULL ptr. Surprised you hit this. How much memory do you
> have/was giftd using at the time? One solution is to return in
> free_node() if the ptr is null.

I don't know how much it was using at the time (since it died), but
"ulimit -m" is returning 478836 (kbytes).  That's a pretty weird value,
though... I wonder where it's coming from.

> > #3  0x4003a87d in insert (d=3D0x3a9ee10, key=3D0xcfbfc474, data=3D0xcfb=
fc468)
> >     at dataset.c:513
> > #4  0x4003a95b in dataset_insert_ex (d=3D0x3a99f1c, key=3D0xcfbfc474,=
=20
> >     data=3D0xcfbfc468) at dataset.c:544
> > #5  0x4003aa15 in dataset_insert (d=3D0x3a99f1c, key=3D0x3a9eda0, key_l=
en=3D10,=20
> >     value=3D0xcfbfc6c7, value_len=3D6) at dataset.c:560
> > #6  0x4003aaac in dataset_insertstr (d=3D0x3a99f1c, key=3D0x3a9eda0 "fr=
equency",=20
> >     value=3D0xcfbfc6c7 "44100") at dataset.c:568
> > #7  0x400775e1 in share_set_meta (share=3D0x3a99f00, key=3D0xcfbfc6bd "=
frequency",=20
> >     value=3D0xcfbfc6c7 "44100") at share.c:213
> > #8  0xbc8e in read_meta (rec=3D0xcfbfc590, file=3D0x3a99f00) at share_d=
b.c:368
> > #9  0xbdb0 in sdb_read (sdb=3D0xffffffff) at share_db.c:403
>=20
> This sdb=3D0xffffffff looks suspicious, maybe gdb is confused?

More likely something assigned -1 to sdb -- or to the memory region
that sdb was using.  I can't think of how gdb could be confused enough
to report wrong values.  Could have been stack damage, I suppose.

--=20
Greg Wooledge                  |   "Truth belongs to everybody."
gr...@wo...              |    - The Red Hot Chili Peppers
http://wooledge.org/~greg/     |