On 1/20/23 11:17, Alistair Francis wrote:
On Tue, Jan 10, 2023 at 5:13 AM Matti Vaittinen
<mazziesaccount@xxxxxxxxx> wrote:
ma 9. tammik. 2023 klo 14.32 Alistair Francis (alistair23@xxxxxxxxx) kirjoitti:
On Tue, Dec 20, 2022 at 11:17 PM Marco Felsch <m.felsch@xxxxxxxxxxxxxx> wrote:
Hi Alistair,
thanks for your patch, please see below.
On 22-12-20, Alistair Francis wrote:
Add support for the rohm,bd71815 power controller controller for the
reMarkable 2.
Signed-off-by: Alistair Francis <alistair@xxxxxxxxxxxxx>
---
arch/arm/boot/dts/imx7d-remarkable2.dts | 159 ++++++++++++++++++++++++
1 file changed, 159 insertions(+)
diff --git a/arch/arm/boot/dts/imx7d-remarkable2.dts b/arch/arm/boot/dts/imx7d-remarkable2.dts
index a138b292ec6a..4387d30d6180 100644
--- a/arch/arm/boot/dts/imx7d-remarkable2.dts
+++ b/arch/arm/boot/dts/imx7d-remarkable2.dts
@@ -92,6 +92,10 @@ wifi_pwrseq: wifi_pwrseq {
};
};
+&cpu0 {
+ cpu-supply = <&buck1_reg>;
+};
+
&clks {
assigned-clocks = <&clks IMX7D_CLKO2_ROOT_SRC>,
<&clks IMX7D_CLKO2_ROOT_DIV>;
@@ -119,6 +123,148 @@ wacom_digitizer: digitizer@9 {
};
};
+&i2c2 {
+ clock-frequency = <100000>;
We can set this now to 400kHz since the driver can handle quirks now
internal.
After a discussion on the first patch I'm going to leave this as is.
Let me know if you do want me to change it.
+ pinctrl-names = "default", "sleep";
+ pinctrl-0 = <&pinctrl_i2c2>;
+ pinctrl-1 = <&pinctrl_i2c2>;
Do we need a "sleep" config here since this will be the same
configuration as the "default" one.
Nope, dropped.
+ status = "okay";
+
+ bd71815: pmic@4b {
+ compatible = "rohm,bd71815";
+ reg = <0x4b>;
+ pinctrl-names = "default";
+ pinctrl-0 = <&pinctrl_bd71815>;
+ interrupt-parent = <&gpio6>; /* PMIC_INT_B GPIO6_IO16 */
+ interrupts = <16 IRQ_TYPE_LEVEL_LOW>;
+ gpio-controller;
+ clocks = <&clks IMX7D_CLKO2_ROOT_SRC>;
+ clock-output-names = "bd71815-32k-out";
+ #clock-cells = <0>;
+ #gpio-cells = <1>;
+
+ regulators {
+ #address-cells = <1>;
+ #size-cells = <0>;
+
+ buck1_reg: regulator@0 {
+ reg = <0>;
+ regulator-compatible = "buck1";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <2000000>;
+ regulator-boot-on;
+ regulator-always-on;
+ regulator-ramp-delay = <1250>;
+ };
+
+ buck2_reg: regulator@1 {
+ reg = <1>;
+ regulator-compatible = "buck2";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <2000000>;
+ regulator-boot-on;
+ regulator-always-on;
+ regulator-ramp-delay = <1250>;
+ };
+
+ buck3_reg: regulator@2 {
+ reg = <2>;
+ regulator-compatible = "buck3";
+ regulator-min-microvolt = <1200000>;
+ regulator-max-microvolt = <2700000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ buck4_reg: regulator@3 {
+ reg = <3>;
+ regulator-compatible = "buck4";
+ regulator-min-microvolt = <1100000>;
+ regulator-max-microvolt = <1850000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ buck5_reg: regulator@4 {
+ reg = <4>;
+ regulator-compatible = "buck5";
+ regulator-min-microvolt = <1800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ ldo1_reg: regulator@5 {
+ reg = <5>;
+ regulator-compatible = "ldo1";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ ldo2_reg: regulator@6 {
+ reg = <6>;
+ regulator-compatible = "ldo2";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ ldo3_reg: regulator@7 {
+ reg = <7>;
+ regulator-compatible = "ldo3";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ ldo4_reg: regulator@8 {
+ reg = <8>;
+ regulator-compatible = "ldo4";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ ldo5_reg: regulator@9 {
+ reg = <9>;
+ regulator-compatible = "ldo5";
+ regulator-min-microvolt = <800000>;
+ regulator-max-microvolt = <3300000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ dvref_reg: regulator@a {
+ reg = <0xa>;
+ regulator-compatible = "dvref";
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ lpsr_reg: regulator@b {
+ reg = <0xb>;
+ regulator-compatible = "lpsr";
+ regulator-boot-on;
+ regulator-always-on;
+ };
+
+ wled_reg: regulator@c {
+ reg = <0xc>;
+ regulator-compatible = "wled";
+ regulator-min-microamp = <10>;
+ regulator-max-microamp = <25000>;
+ regulator-boot-on;
+ regulator-always-on;
+ };
Note: You have marked all regulators as always on, this is rather
suboptimal due to power consumption.
Good point. The power consumption is already pretty bad, as mainline
doesn't support the i.MX deep sleep. As I know this works I'd like to
keep it as is. I'll test to see if removing them doesn't break
anything and then send a follow up patch.
I'd like to suggest also seeing what removing the "regulator-boot-on"
causes. As far as I understand, the "regulator-boot-on" mainly (only?)
intended to be used in situations where the regulator state can not be
reliably read from the hardware at the boot-up. AFAIR, unlike the
BD71837 and BD71847,
the BD71815 does not have this limitation. (This is just my
understanding - which is based on some age-old discussions with Mark
Brown - hence this comment is just a suggestion to try this while
testing, not a comment intended to be demanding for a change)
I have tried removing both "regulator-boot-on" and "regulator-always-on".
After one boot to userspace (where the display didn't work correctly)
the device now seems to not power up at all. Even u-boot is failing to
start. I'm not sure if the regulators are now in some strange state or
if something is fried.
Ouch. This does not sound good at all. Sorry for a bad advice. :(
I can only assume the "power dependencies" are not modelled correctly -
Eg. some required power-rails are not marked to be used by devices that
actually need them. This is probably being worked around using the
regulator-boot-on / regulator-always-on - which will cause the Linux to
turn on these rails when the driver is load. My understanding is that
this is not the intended use-case for the regulator-boot-on flag though.
Also, as was said - the side-effect of always-on is that there will be
no power-saving options - which is still better for an user than a fried
device would be :|
I'm currently waiting for the battery to discharge to see if that helps.
/me keeps fingers crossed. You wouldn't guess how hard it was to type this!
Yours,
-- Matti
--
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland
~~ When things go utterly wrong vim users can always type :help! ~~
