Skip to contentSkip to conversation

platform/chrome: Create new USB driver for RGB keyboard in ChromeOS devices

PATCH In review Won Chung 3 participants 6 messages Nov 30, 2022 – Dec 2, 2022

Subsystems (2)
lore
Share & tools
lore

Won Chung

author
#0 lore raw

PATCH 1/1 platform/chrome: Create new USB driver for RGB keyboard in ChromeOS devices

Without any driver bound to RGB keyboard, it may not be suspended properly, preventing USB xHCI to be suspended and causing power drain. Create new USB driver for RGB keyboard so that it can be suspended properly.

---
 drivers/platform/chrome/Kconfig             |  8 ++++
 drivers/platform/chrome/Makefile            |  1 +
 drivers/platform/chrome/cros_rgb_keyboard.c | 44 +++++++++++++++++++++
 3 files changed, 53 insertions(+)
 create mode 100644 drivers/platform/chrome/cros_rgb_keyboard.c
drivers/platform/chrome/Kconfig +8 -0
index c45fb376d653..a7c36df99432 100644
@@ -265,6 +265,14 @@ config CHROMEOS_PRIVACY_SCREEN
 	  this should probably always be built into the kernel to avoid or
 	  minimize drm probe deferral.
 
+config CROS_RGB_KEYBOARD
+	tristate "ChromeOS RGB keyboard"
+	depends on USB
+	help
+	  This driver supports RGB keyboard in some ChromeOS devices. This shall be
+	  enabled if RGB keyboard is present, otherwise it may not be suspended
+	  properly.
+
 source "drivers/platform/chrome/wilco_ec/Kconfig"
 
 # Kunit test cases
drivers/platform/chrome/Makefile +1 -0
index f7e74a845afc..e4ffa17c57fc 100644
@@ -28,6 +28,7 @@ obj-$(CONFIG_CROS_EC_SENSORHUB)		+= cros-ec-sensorhub.o
 obj-$(CONFIG_CROS_EC_SYSFS)		+= cros_ec_sysfs.o
 obj-$(CONFIG_CROS_USBPD_LOGGER)		+= cros_usbpd_logger.o
 obj-$(CONFIG_CROS_USBPD_NOTIFY)		+= cros_usbpd_notify.o
+obj-$(CONFIG_CROS_RGB_KEYBOARD)		+= cros_rgb_keyboard.o
 
 obj-$(CONFIG_WILCO_EC)			+= wilco_ec/
 
drivers/platform/chrome/cros_rgb_keyboard.c new file +44 -0

44 lines not on this page · raw message

Signature · 1 line
2.39.0.rc0.267.gcb52ba06e7-goog
#1 lore raw

Hi Won,

Quoted 5 lines · Won Chung

Without any driver bound to RGB keyboard, it may not be suspended properly, preventing USB xHCI to be suspended and causing power drain. Create new USB driver for RGB keyboard so that it can be suspended properly.

This seems like overkill. Can't you set this from USB's sysfs nodes like power/control [1] ?

[1] https://www.kernel.org/doc/html/latest/driver-api/usb/power-management.html#the-user-interface-for-dynamic-pm

Best regards,

-Prashant

maintainer
#2 lore raw

Hi Prashant,

Quoted 13 lines · Prashant Malani

Hi Won,

Quoted 5 lines · Won Chung

Without any driver bound to RGB keyboard, it may not be suspended properly, preventing USB xHCI to be suspended and causing power drain. Create new USB driver for RGB keyboard so that it can be suspended properly.

This seems like overkill. Can't you set this from USB's sysfs nodes like power/control [1] ?

Quoted 6 lines

[1] https://www.kernel.org/doc/html/latest/driver-api/usb/power-management.html#the-user-interface-for-dynamic-pm

Best regards,

-Prashant

We're seeing some behavior where a bound driver is needed in order for this USB device to properly enter suspend state. Just manipulating the power/control and other sysfs nodes for this usb device when there's no driver in the kernel doesn't seem to affect the device's ability to drop into a usb low power state.

Also, I synced with Won about this offline, but the primary concern is not this prism usb device runtime suspending, it's actually it's ability to enter suspend state during system suspend. Right now, this internal usb device is keeping the whole system from entering lower S0iX states because it's not sleeping. This driver patch doesn't address that yet, but I'd like Won to dig down and see if he can get it suspending at suspend time too.

Signature · 7 lines
Benson Leung
Staff Software Engineer
Chrome OS Kernel
Google Inc.
[email protected]
Chromium OS Project
[email protected]
#3 lore raw

Hey Benson,

Quoted 23 lines · Benson Leung

Hi Prashant,

Quoted 18 lines · Prashant Malani

Hi Won,

Quoted 5 lines · Won Chung

Without any driver bound to RGB keyboard, it may not be suspended properly, preventing USB xHCI to be suspended and causing power drain. Create new USB driver for RGB keyboard so that it can be suspended properly.

This seems like overkill. Can't you set this from USB's sysfs nodes like power/control [1] ?

[1] https://www.kernel.org/doc/html/latest/driver-api/usb/power-management.html#the-user-interface-for-dynamic-pm

Best regards,

Quoted 7 lines
Quoted 1 line

-Prashant

We're seeing some behavior where a bound driver is needed in order for this USB device to properly enter suspend state. Just manipulating the power/control and other sysfs nodes for this usb device when there's no driver in the kernel doesn't seem to affect the device's ability to drop into a usb low power state.

That seems like an issue with the device then, which should be debugged from the device side and/or its interaction with the USB subsystem.

Quoted 8 lines

Also, I synced with Won about this offline, but the primary concern is not this prism usb device runtime suspending, it's actually it's ability to enter suspend state during system suspend. Right now, this internal usb device is keeping the whole system from entering lower S0iX states because it's not sleeping. This driver patch doesn't address that yet, but I'd like Won to dig down and see if he can get it suspending at suspend time too.

Auto-suspend / dynamic PM should take care of that FWIU (but I may be mistaken).

author
#4 lore raw
Quoted 29 lines · Prashant Malani

Hey Benson,

Quoted 25 lines · Benson Leung

Hi Prashant,

Quoted 19 lines · Prashant Malani

Hi Won,

Quoted 5 lines · Won Chung

Without any driver bound to RGB keyboard, it may not be suspended properly, preventing USB xHCI to be suspended and causing power drain. Create new USB driver for RGB keyboard so that it can be suspended properly.

This seems like overkill. Can't you set this from USB's sysfs nodes like power/control [1] ?

[1] https://www.kernel.org/doc/html/latest/driver-api/usb/power-management.html#the-user-interface-for-dynamic-pm

Best regards,

-Prashant

Quoted 9 lines
Quoted 5 lines

We're seeing some behavior where a bound driver is needed in order for this USB device to properly enter suspend state. Just manipulating the power/control and other sysfs nodes for this usb device when there's no driver in the kernel doesn't seem to affect the device's ability to drop into a usb low power state.

That seems like an issue with the device then, which should be debugged from the device side and/or its interaction with the USB subsystem.

Hi Prashant,

As Benson mentioned, I can check on my test Vell device that changing power/control does not suspend the device.
Should it be controllable even without a driver bound?

The RGB keyboard seems to be able to enter sleep with a suspend signal. Fyi, here are the related bugs for more context: b/242087721 and b/249173368

Thank you,
Won

Quoted 10 lines
Quoted 8 lines

Also, I synced with Won about this offline, but the primary concern is not this prism usb device runtime suspending, it's actually it's ability to enter suspend state during system suspend. Right now, this internal usb device is keeping the whole system from entering lower S0iX states because it's not sleeping. This driver patch doesn't address that yet, but I'd like Won to dig down and see if he can get it suspending at suspend time too.

Auto-suspend / dynamic PM should take care of that FWIU (but I may be mistaken).

#5 lore raw
Quoted 13 lines · Won Chung
Quoted 10 lines · Prashant Malani
Quoted 6 lines

We're seeing some behavior where a bound driver is needed in order for this USB device to properly enter suspend state. Just manipulating the power/control and other sysfs nodes for this usb device when there's no driver in the kernel doesn't seem to affect the device's ability to drop into a usb low power state.

That seems like an issue with the device then, which should be debugged from the device side and/or its interaction with the USB subsystem.

Quoted 5 lines

Hi Prashant,

As Benson mentioned, I can check on my test Vell device that changing power/control does not suspend the device.
Should it be controllable even without a driver bound?

Um, it should be(?); usb_enable_autosuspend() [1] and echo auto > power/control [2]
both seem to call the same runtime PM function. I've not looked into runtime PM code
deeper than that, but it may be worthwhile to examine what's causing the suspend to get blocked.

[1] https://elixir.bootlin.com/linux/latest/source/drivers/usb/core/driver.c#L1637
[2] https://elixir.bootlin.com/linux/latest/source/drivers/base/power/sysfs.c#L113