[Toybox] [PATCH 1/2] killall.test: fix flakiness.

[enh]

I actually think it's mksh... Using -s 9 also makes the flake go away. So
my guess is that we're actually waiting for the sleep in the child to
finish, whereas bash isn't?

On Thu, Jul 25, 2019, 19:30 Rob Landley <rob at landley.net> wrote:

> On 7/25/19 5:31 PM, enh via Toybox wrote:
> > And when I say "fix", I mean "hack around in the worst possible way".
> >
> > Android+mksh is often slow enough that we can do the pgrep before the
> > killed process has actually exited. (I'm seeing roughly 100% failure
> > rate on cloud x86 emulators, and about 75% on a real Pixel 2.)
>
> Nah, that's not bad, that's just "force a schedule". If the process's
> parent
> hasn't scheduled yet to receive the SIGCHLD then it'll still be a zombie,
> that's
> understandable if your scheduler gives the new process scheduling priority
> because shortest job first is likely. The tiny sleep _should_ be there.
>
> Rob
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.landley.net/pipermail/toybox-landley.net/attachments/20190725/59d92658/attachment-0001.htm>