On 01/23/2014 11:52 PM, Jason Gunthorpe wrote:
On Thu, Jan 23, 2014 at 11:38:06PM +0100, Sebastian Hesselbarth wrote:
Bridge IRQ_CAUSE bits are asserted regardless of the corresponding bit in
IRQ_MASK register. To avoid interrupt events on stale irqs, we have to clear
them before unmask. This installs an .irq_enable callback to ensure stale
irqs are cleared before initial unmask.
I'm not sure if putting this in irq_enable is correct. I think this
should only happen at irq_startup.
The question boils down to what is supposed to happen with this code
sequence:
disable_irq(..);
write(.. something to cause an interrupt edge ..);
.. synchronize ..
enable_irq(..);
Do we get the interrupt or not?
Jason,
I get the point and actually I'd chosen .irq_enable because using
.irq_startup didn't work. I rechecked this and now it works.. maybe
it is getting too late for me. I'll send a v2 of this patch shortly.
Sebastian
I found this message from Linus long ago:
http://yarchive.net/comp/linux/edge_triggered_interrupts.html
Btw, the "disable_irq()/enable_irq()" subsystem has been written so that
when you disable an edge-triggered interrupt, and the edge happens while
the interrupt is disabled, we will re-play the interrupt at enable time.
Exactly so that drivers can have an easier time and don't have to
normally worry about whether something is edge or level-triggered.
And found this note in Documentation/DocBook/genericirq.tmpl:
This prevents losing edge interrupts on hardware which does
not store an edge interrupt event while the interrupt is disabled at
the hardware level.
So I think it is very clear that the chip driver should not discard
edges that happened while the interrupt was disabled.
Regards,
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
