rt/tst-aio7 weirdness on ppc

[Roland McGrath]

> Roland,
>    Both of these machines are single processor machines. The glibc
> builds were from earlier today before the setpgid change in test-skeleton.c
> and tst-aio7 passed both in make check or run manually with test-run.sh.

Did you try these in the background?  i.e. make check &

Anyway, from the nature of what I saw it seems likely that the problems
involve SMP races.