Fabian Keil <freebsd-listen_at_fabiankeil.de> wrote:

> "Jonathan C." <well.bad_at_hotmail.com> wrote:
>
> > I think this must be somewhat related to what I reported some time
> > ago http://curl.haxx.se/mail/lib-2014-06/0062.html
>
> Thanks.
>
> My second debug patch adds three assertions to lib1513.c
>
> | assert(processKiller_calls != 0);
> | assert(processKiller_calls == 1);
> | assert(res == 42);
>
> If the callback was called more than once, the second assertion should
> be triggered.
>
> At least for me this never happened so there doesn't seem to be any
> indication that the callback is "called once again after returning
> non-zero value" on my system (or at least it's not reproducible
> with test 1513).

Never mind, I can confirm that 907520c4b9361 (also mentioned by Ray)
fixes the problem you reported. Reverting it doesn't make the test
more reliably for me, though.

Fabian

-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html