[Toybox] tests.sh unkillable?

[enh]

(fwiw, i've not seen this myself. i've seen spinning yes(1) processes
in the past, but haven't even seen that in a while.)

On Wed, Mar 11, 2020 at 8:27 PM Rob Landley <rob at landley.net> wrote:
>
> I did "make tests" without VERBOSE=fail and didn't want to worry about missing
> something so I hit CTRL-C to restart it (somewhere in the endless chmod tests)
> and... it refused to die? I did a dozen ctrl-C and they were showing up on the
> console but instead I had one CPU pegged, and top blamed tests.sh for having
> turned into a CPU-eating loop, which I killed from the command line...
>
> And now I can't get it to do it AGAIN. (This is host bash, not my shell yet!)
>
> In tests.sh there's:
>
>   trap 'kill $(jobs -p) 2>/dev/null; exit 1' INT
>
> But I don't see how that's supposed to hang? Even if the kill fails, it should
> continue to exit?
>
> Anybody else spotting how this could have happened?
>
> Rob
>
> P.S. The jobs -p is cleaning up child processes, and this is one of the things
> that needs to be fixed so tests can be children rather than sourced. A child of
> a child doesn't show up in jobs -p, so things like the pkill/renice tests' "yes
> &" only get killed if the test is sourced. I should really use containers here,
> but my todo list runneth over served by limited hobby time, as always...
> _______________________________________________
> Toybox mailing list
> Toybox at lists.landley.net
> http://lists.landley.net/listinfo.cgi/toybox-landley.net