Re: [ 57/72] genirq: Unmask oneshot irqs when thread was not woken

[Linus Torvalds]

On Tue, Mar 6, 2012 at 12:26 PM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
> Subject: genirq: Clear action->thread_mask if IRQ_ONESHOT is not set

Umm. Apparently this patch fixes the bug, but the patch itself is just insane.

> -       if (new->flags & IRQF_ONESHOT && thread_mask == ~0UL) {
> -               ret = -EBUSY;
> -               goto out_mask;
> +       if (new->flags & IRQF_ONESHOT) {
> +               if (thread_mask == ~0UL) {
> +                       ret = -EBUSY;
> +                       goto out_mask;
> +               }
> +               new->thread_mask = new->flags & IRQF_ONESHOT;
>        }
> -       new->thread_mask = 1 << ffz(thread_mask);

WHAT?

You just checked that "new->flags & IRQF_ONESHOT" nonzero, and inside
that if-statement, you then do

  new->thread_mask = new->flags & IRQF_ONESHOT;

which is just crazy. Why don't you just do

    new->thread_mask = IRQF_ONESHOT;

if that is what you actually meant?

What is that code actually *supposed* to do?

Also, what was the meaning of that old insane line:

   new->thread_mask = 1 << ffz(thread_mask);

which you removed? It was crap, I agree, but what was the thinking
behind it? And the reason it was crap is because that's a crazy
expression that could be written better ways (*), and it needs a
comment on what the heck the point of it was..

So stop with these "random code" snippets, and explain what the f*&^
the code is meant to do, AND THEN WRITE THE CODE IN A SANE MANNER
instead of posting these kinds of insane patches.

Because right now it really looks like the "random monkey" approach to
programming.

                   Linus

(*) "1 << ffz(a)" can be written as

   a = ~a;        /* Turn the zero bits into 1 bits */
   a &= -a;      /* .. and find the first one. */

without ever doing any insane bit scanning.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/