On Thu, Mar 31, 2016 at 3:25 PM, Eric W. Biederman
<ebiederm@xmission.com> wrote:
>
> I suspect we want this to be a per-mm attribute rather than a per-thread
> attribute.
>
> Otherwise you are breaking anything that uses a N-M threading model.
> Which I suspect means that this implementation choice breaks all go
> programs on linux.

That sounds like a good point, but wouldn't it make more conceptual
sense to make it part of "struct sighand_struct" instead?

That is also shared for threads.

                Linus