On Wed, 15 Dec 2010, Jan Kiszka wrote: > Am 15.12.2010 14:04, Thomas Gleixner wrote: > > On Wed, 15 Dec 2010, Jan Kiszka wrote: > >> Am 14.12.2010 21:54, Thomas Gleixner wrote: > >>> On Mon, 13 Dec 2010, Jan Kiszka wrote: > >>>> @@ -943,6 +950,9 @@ static struct irqaction *__free_irq(unsigned int irq, void *dev_id) > >>>> /* Make sure it's not being used on another CPU: */ > >>>> synchronize_irq(irq); > >>>> > >>>> + if (single_handler) > >>>> + desc->irq_data.drv_status &= ~IRQS_SHARED; > >>>> + > >>> > >>> What's the reason to clear this flag outside of the desc->lock held > >>> region. > >> > >> We need to synchronize the irq first before clearing the flag. > >> > >> The problematic scenario behind this: An IRQ started in shared mode, > >> this the line was unmasked after the hardirq. Now we clear IRQS_SHARED > >> before calling into the threaded handler. And that handler may now think > >> that the line is still masked as IRQS_SHARED is set. > > > > That should read "not set" I guess. > > Can't remember who wrote this, but that guy might have been too tired > for clear sentences: Yes, of course, we could run into troubles, if > IRQS_SHARED was _not_ set while the IRQ line is unmasked between hard > and threaded handler. Right. As a side note, the current implementation requires that you lookup irq_data.drv_status for every invocation of the handler or have a reference to irq_data.drv_status somewhere locally, which I don't like either. I have an evil and nasy idea how to avoid that, need to look how ugly it gets. Worst case we need to go back to that notification thing which I wanted really avoid in the first place. Though I like the register_mutex idea which came out of this a lot as it allows us to reduce desc->lock held and interrupt disabled regions quite nicely. /me goes back to stare at the code > > Hmm, needs more thoughts :( > > Be warned, might be painful. Bah, my brain became pain resistant when I started hacking that code. Thanks, tglx -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html