Re: [Gpsbabel-code] PATCH: Realtime NMEA output

[Jon M. <ja...@we...>]

On Tue, Mar 13, 2007 at 09:08:02PM -0700, Jon McClintock wrote:
> Also, one other thing I noticed: Because the only logic for ending the
> realtime tracking is looking for a request_terminate from the GPS, if
> you kill gpsbabel with a Ctrl-C, the cleanup code never gets called,
> which causes the GPS to crash the next time you run gpsbabel. I've added
> a SIGINT handler, and it's much nicer to the GPS now, but I'm getting
> double-free errors somewhere in the cleanup code. My test platform is
> OSX 10.4 on an Intel Core Duo with an etrex Legend CX. 

Found the problem here. gusb_teardown() was being added as an atexit()
routine. It calls xfree() on it's arg, but atexit handlers take no args,
which means it was freeing an invalid memory location. Attached is a fix
which wraps gusb_teardown(), passing it a NULL address.

-Jon