On 24.04.19 10:18, Mika Westerberg wrote:
On Wed, Apr 24, 2019 at 10:12:42AM +0200, Jan Kiszka wrote:
On 24.04.19 09:58, Mika Westerberg wrote:
+Rafael and linux-acpi.
On Thu, Apr 18, 2019 at 11:23:49AM +0200, Jan Kiszka wrote:
From: Jan Kiszka <jan.kiszka@xxxxxxxxxxx>
Validated on the Quark platform, this adds interrupt support on rising
and/or falling edges.
The irqchip parts look good to me but but the ACPI SCI handling seems
weird. This is typically handled by ACPI core based on the values read
from FADT ACPI table. What does it contain on this Quark platform?
There is no FADT on the original Quark firmware, nor did we add one. As we
are talking about existing devices, possibly not only Quarks, I was going
down the ACPI-independent way to hook into the interrupt. But I'm open to
learn about better alternatives.
Hmm, if it does not have FADT table why would you need SCI then? Is this
implementing some real use case?
Maybe I'm looking at the wrong place: Where would I find it? There is definitely
no separate entry under /sys/firmware/acpi/tables - but that's also true for my
workstation.
In any case: The hardware defines that the GPIO events are sent via SCIs. That
fact is probably not expressed in ACPI language.
Jan
--
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux
