Hi All,
When support for the "charge_behaviour" property was added the list of
available values was made a bitmask in power_supply_desc.
"usb_types" is very similar in that:
1. It is an enum
2. The list of available values is stored in power_supply_desc
3. When shown it shows all available values, with the active one surrounded
by square brackets.
But "usb_types" uses an array with valid enum values instead of a bitmask.
This uses more memory then the bitmap approach and it makes it impossible
to have a shared generic show() function for properties which show
available values, with the active one surrounded by square brackets.
This patch-set moves "usb_types" over to a bitmask in power_supply_desc
to indicate the available values.
Patches 1 - 3:
It turns out that the ucs1002-power driver contained a surprise in that
it supports writing to "usb_type" even though the ABI doc says it is
read-only. Since we cannot break shipped userspace API, the ship has sailed
on this one. The first patch documents that writing "usb_type" is allowed,
but only for power-supply devices which provide USB power rather then
consume it.
Enum properties accept writing the FOO_TEXT[] string values, passing
the enum value matching the FOO_TEXT entry to set_property(), the second
patch adjusts ucs1002_set_usb_type() to directly accept enum values.
The rt9467 driver was another driver which allowed writing to "usb_type"
but there the use made no sense, so it is simply dropped.
Patches 4 - 6:
These patches implement the actual moving of usb_types to a bitmask.
Patch 6 is a bit of a bigbang patch moving all drivers over in one go,
touching a couple of drivers outside drivers/power/supply: 1 in
drivers/extcon/ 1 in drivers/phy/ and 5 in drivers/usb/typec/ since
the changes outside of drivers/power/supply are small I've chosen to
make all the changes in one go rather then have some sort of
intermediate state where both ways are supported.
For merging this I believe it would be best for an immutable branch / tag
to be created on the linux-power-supply tree and then send a pull-request
to the extcon, phy and usb-typec maintainers to merge the tag.
extcon, phy and typec maintainers can you please give your Acked-by for
patch 6/6 for merging these changes through the linux-power-supply tree?
This set is based on top of the latest linux-power-supply/for-next.
Regards,
Hans
Hans de Goede (6):
power: supply: "usb_type" property may be written to
power: supply: ucs1002: Adjust ucs1002_set_usb_type() to accept string
values
power: supply: rt9467-charger: Remove "usb_type" property write
support
power: supply: sysfs: Add power_supply_show_enum_with_available()
helper
power: supply: sysfs: Move power_supply_show_enum_with_available() up
power: supply: Change usb_types from an array into a bitmask
Documentation/ABI/testing/sysfs-class-power | 7 +-
drivers/extcon/extcon-intel-cht-wc.c | 15 ++---
drivers/phy/ti/phy-tusb1210.c | 11 +---
drivers/power/supply/axp20x_usb_power.c | 13 ++--
drivers/power/supply/bq256xx_charger.c | 15 ++---
drivers/power/supply/cros_usbpd-charger.c | 22 +++----
.../power/supply/lenovo_yoga_c630_battery.c | 7 +-
drivers/power/supply/mp2629_charger.c | 15 ++---
drivers/power/supply/mt6360_charger.c | 13 ++--
drivers/power/supply/mt6370-charger.c | 13 ++--
drivers/power/supply/power_supply_core.c | 4 --
drivers/power/supply/power_supply_sysfs.c | 66 ++++++-------------
drivers/power/supply/qcom_battmgr.c | 37 ++++++-----
drivers/power/supply/qcom_pmi8998_charger.c | 13 ++--
drivers/power/supply/rk817_charger.c | 9 +--
drivers/power/supply/rn5t618_power.c | 13 ++--
drivers/power/supply/rt9467-charger.c | 16 ++---
drivers/power/supply/rt9471.c | 15 ++---
drivers/power/supply/ucs1002_power.c | 26 ++++----
drivers/usb/typec/anx7411.c | 11 +---
drivers/usb/typec/rt1719.c | 11 +---
drivers/usb/typec/tcpm/tcpm.c | 11 +---
drivers/usb/typec/tipd/core.c | 9 +--
drivers/usb/typec/ucsi/psy.c | 11 +---
include/linux/power_supply.h | 3 +-
25 files changed, 132 insertions(+), 254 deletions(-)
--
2.46.0
