arm64: dts: rockchip: Correct regulator for USB host on Odroid-Go2
When writing a battery driver, I noticed that the USB voltage was ~3.7V while running off of battery on a mainline kernel. After consulting the schematics for the Odroid Go Advance, it appears that the BOOST regulator is involved in the process of powering the USB host. Power for the USB host goes from the vccsys regulator into the PMIC, then out from the PMIC BOOST regulator into the FC9516A (which is controlled by GPIO), which then feeds power into the USB host. I named the regulator usb_midu because on the datasheet the pin is described as "MIDU/BOOST - middle point of USB power supply / boost output". Making these changes solved the USB power issue on battery and I'm now reading approximately 5v. Note that on my board at least there is a difference in time from the USB PHY probing and the regulators being powered on. This causes the USB port to be undervolted for a few seconds during boot up. The solutions to this problem are either 1) to add the proper phy-supply on the host port, or to 2) add regulator-boot-on to the regulator. I chose to add regulator-boot-on because there is an issue with the phy clk that causes a warning when booting (see v1 of this patch series). Basically the clock usb480m is a child of the usb480m_phy clock (used by the USB PHY) and also a critical clock. Setting the phy-supply causes this driver to be EPROBE_DEFERed until the regulator is ready, however upon unregistering the driver to be probed later the system cannot remove the usb480m_phy clock due to a child being marked critical. Changes since v2: - Added notes about clk problem and regulator voltage at boot. - Added regulator-boot-on as a workaround for the voltage at boot. - Removed note about fixed regulator warning, as that has been fixed upstream. Changes since v1: - Removed phy-supply, as this generated a warning in dmesg. Signed-off-by: Chris Morgan <macromorgan@hotmail.com> Link: https://lore.kernel.org/r/20210916190938.6175-1-macroalpha82@gmail.comSigned-off-by: Heiko Stuebner <heiko@sntech.de>
Showing
Please register or sign in to comment