On Mon, Feb 24, 2020 at 11:24 AM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> I don't know. This does not seem to be a particularly serious load.
> But it does feel like it should be possible to combine the two atomic
> accesses into one, where you don't need to do the refcount thing
> except for the case where sigcount goes from zero to non-zero (and
> back to zero again).

Ok, that looks just as simple as I thought it would be.

TOTALLY UNTESTED patch attached. It may be completely buggy garbage,
but it _looks_ trivial enough. Just make the rule be that "if we have
any user->sigpending cases, we'll get a ref to the user for the first
one, and drop it only when getting rid of the last one".

So it might be worth testing this. But again: I have NOT done so.

There might be some silly reason why this doesn't work because I just
did the tests wrong or missed some case.

Or there might be some subtle reason why it doesn't work because I
didn't think this through properly.

But it _looks_ obvious and simple enough. And it compiles for me. So
maybe it works.

              Linus