On Tue, Jun 12, 2018 at 2:40 PM, Brian Norris [off-list ref] wrote:

Hi Rob,

On Tue, Jun 12, 2018 at 01:10:11PM -0600, Rob Herring wrote:

quoted

On Thu, Jun 07, 2018 at 11:12:13AM -0700, Matthias Kaehlcke wrote:

quoted

The cros_ec_throttler monitors events from the Chrome OS Embedded
Controller to throttle the system if needed, using the mechanisms
provided by the throttler core.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
---
Changes in v2:
- patch added to series

 Documentation/devicetree/bindings/misc/cros_ec_throttler.txt | 4 ++++
 1 file changed, 4 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/misc/cros_ec_throttler.txt

diff --git a/Documentation/devicetree/bindings/misc/cros_ec_throttler.txt b/Documentation/devicetree/bindings/misc/cros_ec_throttler.txt
new file mode 100644
index 000000000000..7316dcc0ef75
--- /dev/null
+++ b/Documentation/devicetree/bindings/misc/cros_ec_throttler.txt

@@ -0,0 +1,4 @@
+* cros_ec_throttler driver

Bindings are for h/w, not drivers.

(OK, sure, don't call this "driver". And maybe this could use some more
description about what kind of events are emitted by this sort of
device.)

quoted

I continue to fail to see why this needs to be in DT. There are other
ways to instantiate drivers.

This is mostly relevant to:
[PATCH v2 08/11] dt-bindings: PM / OPP: add opp-throttlers property

so it's probably good to take a look at that one too.

I did on v1, that's why I say "continue"...

The primary purpose is to have a target to point at when determining who
is the source of throttling. This is similar to other CrOS EC subdevices
(e.g., PWM) where we technically don't require a subnode (the EC
firmware can its own PWM hardware without DT), but it is important that,
e.g., a backlight device has something to point at when it's using a PWM
attached to the EC. So we have a PWM subnode.

Yes, but the PWM consumers could point to the EC node. Or to put it another way the EC node can be a provider of multiple things.

In this case, we're a little vague about what exactly the hardware is
here, but there *is* hardware that's emitting "throttle" events (hint:
here, it's related to sensing too high of system current). This is all
abstracted by firmware, which simply tells us we need to scale back our
power usage.

So, what do you think of patch 8? Should OPPs have phandles to such a
throttler? If so, should the phandle just point at the main EC device
(see mfd/cros-ec.txt), or is it reasonable to have a subnode to
represent something more specific?

Or maybe this is entirely on the wrong track. But this is the resulting
proposal after your comments on v1, so it's probably best we have a
clearer overall review of what makes sense here, so we don't just go in
cycles on new proposals that get rejected.

Brian

quoted

quoted

+
+Required properties:
+- compatible: "google,cros-ec-throttler"
--
2.18.0.rc1.242.g61856ae69a-goog