t-forward-control on slow machines

[Damien Miller]

On Sun, 17 Nov 2013, Christian Weisgerber wrote:

> On how slow a machine do we want the regression tests, specifically
> forward-control.sh, to still work?
> 
> On my Alpha PC164 (21164-500, ~1997), the forward-control test
> produces erratic results.  It eventually dawned on me that this is
> just a timing problem.  Bumping the timeouts from 5 to 10s makes
> it run reliably.  But then I'm sure Miod has even slower machines,
> so I don't know what we want to accommodate.

I think waiting 10s or longer is okay, since in the normal case we
should never have to wait that long. I'd love a way to make this test
work without polling, but I couldn't figure out how when I wrote the
test.

Your diff is ok djm@

-d