Re: kdestroy coredump (fcc_close)

[Assar Westerlund <assar@sics.se>]

Leif Johansson <leifj@matematik.su.se> writes:
> fcc_close or some routine close to it does something nasty
> with memory causing a __libc_free segfault. Probably some
> memory gets freed twice or not allocated properly or something.
> This did not happen in 0.0t so it must be due to some recent
> change in 0.0u. I enclose the usual gdb sample session showing
> exactly where the error occurs. However as you see there does
> not seem to be any reason for crashing on this free... Strange!

fcc_close doesn't look like it's doing anything wrong, so it's
probably somewhere else where the error is really happening and it's
manifesting itself here.  Running with electric fence or purify is
probably the best way of catching this.  I'm afraid I haven't had the
time to do this recently.  If you have occasion to do this before I do
it, feel free the send me the output.

/assar