pwritev2 regression test for invalid flags

[Florian Weimer]

On 06/21/2017 03:11 PM, Adhemerval Zanella wrote:

> Yes, we will need to update the testcase in case of newer flags additions. 
> But since the flags are currently defined within GLIBC not pulling from 
> kernel header I think we will have more control over it.

But that doesn't really matter because we need something to trigger
EINVAL from the running kernel.  -1 should be safe for now, though.

Florian