| Message ID | 20240831142039.28830-1-hdegoede@redhat.com |
|---|---|
| Headers | show |
| Series | power: supply: Change usb_types from an array into a bitmask | expand |
On Sat, 31 Aug 2024 16:20:33 +0200, Hans de Goede wrote: > 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. > > [...] Applied, thanks! [1/6] power: supply: "usb_type" property may be written to commit: 0d9af1e1c93b6a89f3fb6dcbafa5bc78892cb94f [2/6] power: supply: ucs1002: Adjust ucs1002_set_usb_type() to accept string values commit: 83a4c42df75e8f6ff671fa9fbe7d4c79b98626de [3/6] power: supply: rt9467-charger: Remove "usb_type" property write support commit: 03ec41c1670aedfbd126f541c4acbb8e69d4cd0c [4/6] power: supply: sysfs: Add power_supply_show_enum_with_available() helper commit: a6456d43e9abebb5d7866e5edae3024188273306 [5/6] power: supply: sysfs: Move power_supply_show_enum_with_available() up commit: 322900ac7d82be0475466f81946b6484cd1936bd [6/6] power: supply: Change usb_types from an array into a bitmask commit: 364ea7ccaef917a3068236a19a4b31a0623b561a Best regards,
Hi, On Sat, Aug 31, 2024 at 04:20:33PM GMT, Hans de Goede wrote: > 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. As we are quite close before the merge window I just took it with just the Ack from Greg. The changes in PHY and Extcon are quite small. I did prepare an immutable branch in case it is needed by any of the other involved trees: The following changes since commit 8400291e289ee6b2bf9779ff1c83a291501f017b: Linux 6.11-rc1 (2024-07-28 14:19:55 -0700) are available in the Git repository at: https://git.kernel.org/pub/scm/linux/kernel/git/sre/linux-power-supply.git tags/ib-psy-usb-types-signed for you to fetch changes up to 364ea7ccaef917a3068236a19a4b31a0623b561a: power: supply: Change usb_types from an array into a bitmask (2024-09-03 23:20:28 +0200) ---------------------------------------------------------------- Immutable branch for usb_types change for v6.12 Changing usb_types type from array to bitmap in the power_supply_desc struct requires updating power-supply drivers living in different subsystem, so it is handled via an immutable branch. ---------------------------------------------------------------- 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 +++------ drivers/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(-) -- Sebastian
Hi, On 9/4/24 12:04 AM, Sebastian Reichel wrote: > Hi, > > On Sat, Aug 31, 2024 at 04:20:33PM GMT, Hans de Goede wrote: >> 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. > > As we are quite close before the merge window I just took it with > just the Ack from Greg. The changes in PHY and Extcon are quite > small. Great, thank you. Regards, Hans > I did prepare an immutable branch in case it is needed by > any of the other involved trees: > > The following changes since commit 8400291e289ee6b2bf9779ff1c83a291501f017b: > > Linux 6.11-rc1 (2024-07-28 14:19:55 -0700) > > are available in the Git repository at: > > https://git.kernel.org/pub/scm/linux/kernel/git/sre/linux-power-supply.git tags/ib-psy-usb-types-signed > > for you to fetch changes up to 364ea7ccaef917a3068236a19a4b31a0623b561a: > > power: supply: Change usb_types from an array into a bitmask (2024-09-03 23:20:28 +0200) > > ---------------------------------------------------------------- > Immutable branch for usb_types change for v6.12 > > Changing usb_types type from array to bitmap in the power_supply_desc > struct requires updating power-supply drivers living in different > subsystem, so it is handled via an immutable branch. > > ---------------------------------------------------------------- > 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 +++------ > drivers/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(-) > > -- Sebastian
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(-)