@@ -51,10 +51,12 @@ pm8916_rpm_regulators: regulators {
pm8916_s3: s3 {
regulator-min-microvolt = <1250000>;
regulator-max-microvolt = <1350000>;
+ regulator-always-on; /* Needed for L2 */
};
pm8916_s4: s4 {
regulator-min-microvolt = <1850000>;
regulator-max-microvolt = <2150000>;
+ regulator-always-on; /* Needed for L5/L7 */
};
/*
@@ -71,6 +73,7 @@ pm8916_s4: s4 {
pm8916_l2: l2 {
regulator-min-microvolt = <1200000>;
regulator-max-microvolt = <1200000>;
+ regulator-always-on; /* Needed for LPDDR RAM */
};
/* pm8916_l3 is managed by rpmpd (MSM8939_VDDMX) */
@@ -80,6 +83,7 @@ pm8916_l2: l2 {
pm8916_l5: l5 {
regulator-min-microvolt = <1800000>;
regulator-max-microvolt = <1800000>;
+ regulator-always-on; /* Needed for most digital I/O */
};
pm8916_l6: l6 {
@@ -90,6 +94,7 @@ pm8916_l6: l6 {
pm8916_l7: l7 {
regulator-min-microvolt = <1800000>;
regulator-max-microvolt = <1800000>;
+ regulator-always-on; /* Needed for CPU PLL */
};
pm8916_l8: l8 {
Some of the regulators must be always-on to ensure correct operation of the system, e.g. PM8916 L2 for the LPDDR RAM, L5 for most digital I/O and L7 for the CPU PLL (strictly speaking the CPU PLL might only need an active-only vote but this is not supported for regulators in mainline currently). The RPM firmware seems to enforce that internally, these supplies stay on even if we vote for them to power off (and there is no other processor running). This means it's pointless to keep sending enable/disable requests because they will just be ignored. Also, drivers are much more likely to get a wrong impression of the regulator status, because regulator_is_enabled() will return false when there are no users, even though the regulator is always on. Describe this properly by marking the regulators as always-on. The same changes was already made for MSM8916 in commit 8bbd35771f90 ("arm64: dts: qcom: msm8916-pm8916: Mark always-on regulators"). Signed-off-by: Stephan Gerhold <stephan@gerhold.net> --- arch/arm64/boot/dts/qcom/msm8939-pm8916.dtsi | 5 +++++ 1 file changed, 5 insertions(+)