| Message ID | 20240124074527.48869-1-krzysztof.kozlowski@linaro.org |
|---|---|
| Headers | show |
| Series | reset: gpio: ASoC: shared GPIO resets | expand |
On 24/01/2024 08:45, Krzysztof Kozlowski wrote: > Devices sharing a reset GPIO could use the reset framework for > coordinated handling of that shared GPIO line. We have several cases of > such needs, at least for Devicetree-based platforms. > > If Devicetree-based device requests a reset line, while "resets" > Devicetree property is missing but there is a "reset-gpios" one, > instantiate a new "reset-gpio" platform device which will handle such > reset line. This allows seamless handling of such shared reset-gpios > without need of changing Devicetree binding [1]. > > To avoid creating multiple "reset-gpio" platform devices, store the > Devicetree "reset-gpios" GPIO specifiers used for new devices on a > linked list. Later such Devicetree GPIO specifier (phandle to GPIO > controller, GPIO number and GPIO flags) is used to check if reset > controller for given GPIO was already registered. > > If two devices have conflicting "reset-gpios" property, e.g. with > different ACTIVE_xxx flags, this would allow to spawn two separate > "reset-gpio" devices, where the second would fail probing on busy GPIO > request. > > Link: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ [1] > Cc: Bartosz Golaszewski <brgl@bgdev.pl> > Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> > Cc: Sean Anderson <sean.anderson@seco.com> > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> > > --- > > Depends on previous of change. > --- > drivers/reset/core.c | 215 +++++++++++++++++++++++++++++-- > include/linux/reset-controller.h | 4 + > 2 files changed, 206 insertions(+), 13 deletions(-) > LKP reported issue when building !GPIOLIB: https://lore.kernel.org/oe-kbuild-all/202401250958.YksQmnWj-lkp@intel.com/ but I intend to solve it providing the stubs. Therefore this patch will not change. Best regards, Krzysztof
On 24/01/2024 08:45, Krzysztof Kozlowski wrote: > Devices sharing a reset GPIO could use the reset framework for > coordinated handling of that shared GPIO line. We have several cases of > such needs, at least for Devicetree-based platforms. > > If Devicetree-based device requests a reset line, while "resets" > Devicetree property is missing but there is a "reset-gpios" one, > instantiate a new "reset-gpio" platform device which will handle such > reset line. This allows seamless handling of such shared reset-gpios > without need of changing Devicetree binding [1]. > > To avoid creating multiple "reset-gpio" platform devices, store the > Devicetree "reset-gpios" GPIO specifiers used for new devices on a > linked list. Later such Devicetree GPIO specifier (phandle to GPIO > controller, GPIO number and GPIO flags) is used to check if reset > controller for given GPIO was already registered. > > If two devices have conflicting "reset-gpios" property, e.g. with > different ACTIVE_xxx flags, this would allow to spawn two separate > "reset-gpio" devices, where the second would fail probing on busy GPIO > request. > > Link: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ [1] > Cc: Bartosz Golaszewski <brgl@bgdev.pl> > Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> > Cc: Sean Anderson <sean.anderson@seco.com> > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> > > --- > > Depends on previous of change. > --- > drivers/reset/core.c | 215 +++++++++++++++++++++++++++++-- > include/linux/reset-controller.h | 4 + > 2 files changed, 206 insertions(+), 13 deletions(-) > > diff --git a/drivers/reset/core.c b/drivers/reset/core.c > index 4d5a78d3c085..60a8a33c4419 100644 > --- a/drivers/reset/core.c > +++ b/drivers/reset/core.c <snip> > + } > + > + ret = __reset_add_reset_gpio_lookup(id, args->np, args->args[0], > + args->args[1]); What would happen with gpio controllers using #gpio-cells = <3> (or more) like allwinner,sun4i-a10-pinctrl.yaml ? On this example the flags are args->args[2] so this would probably fail. This would also fails badly with #gpio-cells = <1>, args->args[1] value would be undefined. You should probably limit to args->args_count == 2 for now. Neil > + if (ret < 0) > + goto err_kfree; > + > + rgpio_dev->of_args = *args; > + /* > + * We keep the device_node reference, but of_args.np is put at the end > + * of __of_reset_control_get(), so get it one more time. > + * Hold reference as long as rgpio_dev memory is valid. > + */ > + of_node_get(rgpio_dev->of_args.np); > + pdev = platform_device_register_data(NULL, "reset-gpio", id, > + &rgpio_dev->of_args, > + sizeof(rgpio_dev->of_args)); <snip>
On Thu, Jan 25, 2024 at 9:02 AM Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> wrote: > > On 24/01/2024 08:45, Krzysztof Kozlowski wrote: > > Devices sharing a reset GPIO could use the reset framework for > > coordinated handling of that shared GPIO line. We have several cases of > > such needs, at least for Devicetree-based platforms. > > > > If Devicetree-based device requests a reset line, while "resets" > > Devicetree property is missing but there is a "reset-gpios" one, > > instantiate a new "reset-gpio" platform device which will handle such > > reset line. This allows seamless handling of such shared reset-gpios > > without need of changing Devicetree binding [1]. > > > > To avoid creating multiple "reset-gpio" platform devices, store the > > Devicetree "reset-gpios" GPIO specifiers used for new devices on a > > linked list. Later such Devicetree GPIO specifier (phandle to GPIO > > controller, GPIO number and GPIO flags) is used to check if reset > > controller for given GPIO was already registered. > > > > If two devices have conflicting "reset-gpios" property, e.g. with > > different ACTIVE_xxx flags, this would allow to spawn two separate > > "reset-gpio" devices, where the second would fail probing on busy GPIO > > request. > > > > Link: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ [1] > > Cc: Bartosz Golaszewski <brgl@bgdev.pl> > > Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> > > Cc: Sean Anderson <sean.anderson@seco.com> > > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> > > > > --- > > > > Depends on previous of change. > > --- > > drivers/reset/core.c | 215 +++++++++++++++++++++++++++++-- > > include/linux/reset-controller.h | 4 + > > 2 files changed, 206 insertions(+), 13 deletions(-) > > > > LKP reported issue when building !GPIOLIB: > https://lore.kernel.org/oe-kbuild-all/202401250958.YksQmnWj-lkp@intel.com/ > > but I intend to solve it providing the stubs. Therefore this patch will > not change. > > Best regards, > Krzysztof > Ah, so this is why you sent the patches. I don't like stubs in gpio/driver.h but I get why they're needed here. Maybe we should consider adding gpio/misc.h for that kind of stuff. Bart
On Mi, 2024-01-24 at 08:45 +0100, Krzysztof Kozlowski wrote: > Add a simple driver to control GPIO-based resets using the reset > controller API for the cases when the GPIOs are shared and reset should > be coordinated. The driver is expected to be used by reset core > framework for ad-hoc reset controllers. > > Cc: Bartosz Golaszewski <brgl@bgdev.pl> > Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> > Cc: Sean Anderson <sean.anderson@seco.com> > Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org> > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de> regards Philipp
On Mi, 2024-01-24 at 08:45 +0100, Krzysztof Kozlowski wrote: > Devices sharing a reset GPIO could use the reset framework for > coordinated handling of that shared GPIO line. We have several cases of > such needs, at least for Devicetree-based platforms. > > If Devicetree-based device requests a reset line, while "resets" > Devicetree property is missing but there is a "reset-gpios" one, > instantiate a new "reset-gpio" platform device which will handle such > reset line. This allows seamless handling of such shared reset-gpios > without need of changing Devicetree binding [1]. > > To avoid creating multiple "reset-gpio" platform devices, store the > Devicetree "reset-gpios" GPIO specifiers used for new devices on a > linked list. Later such Devicetree GPIO specifier (phandle to GPIO > controller, GPIO number and GPIO flags) is used to check if reset > controller for given GPIO was already registered. > > If two devices have conflicting "reset-gpios" property, e.g. with > different ACTIVE_xxx flags, this would allow to spawn two separate > "reset-gpio" devices, where the second would fail probing on busy GPIO > request. > > Link: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ [1] > Cc: Bartosz Golaszewski <brgl@bgdev.pl> > Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> > Cc: Sean Anderson <sean.anderson@seco.com> > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de> regards Philipp
On 25/01/2024 09:27, neil.armstrong@linaro.org wrote: > On 24/01/2024 08:45, Krzysztof Kozlowski wrote: >> Devices sharing a reset GPIO could use the reset framework for >> coordinated handling of that shared GPIO line. We have several cases of >> such needs, at least for Devicetree-based platforms. >> >> If Devicetree-based device requests a reset line, while "resets" >> Devicetree property is missing but there is a "reset-gpios" one, >> instantiate a new "reset-gpio" platform device which will handle such >> reset line. This allows seamless handling of such shared reset-gpios >> without need of changing Devicetree binding [1]. >> >> To avoid creating multiple "reset-gpio" platform devices, store the >> Devicetree "reset-gpios" GPIO specifiers used for new devices on a >> linked list. Later such Devicetree GPIO specifier (phandle to GPIO >> controller, GPIO number and GPIO flags) is used to check if reset >> controller for given GPIO was already registered. >> >> If two devices have conflicting "reset-gpios" property, e.g. with >> different ACTIVE_xxx flags, this would allow to spawn two separate >> "reset-gpio" devices, where the second would fail probing on busy GPIO >> request. >> >> Link: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ [1] >> Cc: Bartosz Golaszewski <brgl@bgdev.pl> >> Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> >> Cc: Sean Anderson <sean.anderson@seco.com> >> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> >> >> --- >> >> Depends on previous of change. >> --- >> drivers/reset/core.c | 215 +++++++++++++++++++++++++++++-- >> include/linux/reset-controller.h | 4 + >> 2 files changed, 206 insertions(+), 13 deletions(-) >> >> diff --git a/drivers/reset/core.c b/drivers/reset/core.c >> index 4d5a78d3c085..60a8a33c4419 100644 >> --- a/drivers/reset/core.c >> +++ b/drivers/reset/core.c > > <snip> > >> + } >> + >> + ret = __reset_add_reset_gpio_lookup(id, args->np, args->args[0], >> + args->args[1]); > > What would happen with gpio controllers using #gpio-cells = <3> (or more) like allwinner,sun4i-a10-pinctrl.yaml ? > > On this example the flags are args->args[2] so this would probably fail. > > This would also fails badly with #gpio-cells = <1>, args->args[1] value would be undefined. > > You should probably limit to args->args_count == 2 for now. Hm, good point. Both cells are actually possible, so I need to check it. Thanks. Best regards, Krzysztof
Hi, Patch #2 (cpufreq: do not open-code of_phandle_args_equal()) and patch #4 (reset: Instantiate reset GPIO controller for shared reset-gpios) depend on OF change (patch #1). Changes in v5 ============= 1. Minor comments from Philipp: missing cleanup.h, correct indentation of pr_err(), shorten gpio_device_find_by_fwnode() line. 2. Acks/Rbs. Changes in v4 ============= 1. New patches: of: add of_phandle_args_equal() helper cpufreq: do not open-code of_phandle_args_equal() 2. reset-gpio.c: - Drop unneeded comment (Bartosz), add Rb tag. - Do not assign of_node. 3. reset/core.c: - Implement most of Bartosz feedback (I responded to one which I did not implement) and comments from Philipp. - Expect either rcdev->of_args or rcdev->of_node. - Drop __reset_gpios_args_match() and use common helper (Philipp). - Move declarations of automatic-cleanup variables in __reset_add_reset_gpio_lookup() to place of use (Bartosz). - Separate gpio_device_get_label() and kstrdup() (Philipp). - Correct doc for __reset_add_reset_gpio_device(), rewrite few comments. - Drop unneeded "r" variable in __reset_find_rcdev() (Philipp). - Drop of_phandle_args initialization in __of_reset_control_get (Philipp). - Check if CONFIG_RESET_GPIO is enabled before trying to look up reset-gpios. 4. Drop Chris' patch: "i2c: muxes: pca954x: Allow sharing reset GPIO", because discussion is on going. Changes in v3 ============= 1. reset-gpio.c: - Add reset_gpio_of_xlate (Philipp). - reset_gpio_of_args_put->reset_gpio_of_node_put (Philipp). - Expect via platdata of_phandle_args. - Do not call device_set_node() to attach itself to reset consumer (the final device). This was questionable idea in the first place. Bartosz suggested to use GPIO_LOOKUP to solve this. 2. reset/core.c, implement Philipp's feedback. That was a lot: - Commit msg fixes. - Add new platform_device earlier, when reset core found "reset-gpios" but not "resets". - Do not overwrite of_phandle_args. - Expect matching .of_reset_n_cells. - Pass of_phandle_args as platdata to reset-gpio. - Rename reset_gpio_device->reset_gpio_lookup and others. Fix few comments and code cleanup pointed on review. - From Bartosz: Use GPIO_LOOKUP and a lot of cleanup.h in __reset_add_reset_gpio_lookup(). 3. Include here Chris' patch: "i2c: muxes: pca954x: Allow sharing reset GPIO". Changes in v2 ============= 1. wsa884x.c: add missing return in wsa884x_get_reset(), correct comment. 2. qcom,wsa8840.yaml: fix oneOf syntax. 3. reset-gpio.c: - Fix smatch warning on platdata evaluation. - Parse GPIO args and store them in rc.of_args. 4. reset/core.c: - Revise approach based on Bartosz comments: parse the reset-gpios phandle with arguments, do not use deprecated API and do not rely on gpio_desc pointer. - Create a list of instantiated platform devices to avoid any duplicates. - After creating reset-gpio platform device, try to get new reset controller or return EPROBE_DEFER. - Drop the "cookie" member and add new "of_args" to "struct reset_controller_dev". Description =========== We have at least few cases where hardware engineers decided to use one powerdown/shutdown/reset GPIO line for multiple devices: 1. WSA884x (this and previous patch): https://lore.kernel.org/all/b7aeda24-d638-45b7-8e30-80d287f498f8@sirena.org.uk/ 2. https://lore.kernel.org/all/20231027033104.1348921-1-chris.packham@alliedtelesis.co.nz/ 3. https://lore.kernel.org/lkml/20191030120440.3699-1-peter.ujfalusi@ti.com/ 4. https://lore.kernel.org/all/20211018234923.1769028-1-sean.anderson@seco.com/ 5. https://social.treehouse.systems/@marcan/111268780311634160 I try to solve my case, hopefuly Chris' (2), partially Sean's (4) and maybe Hectors (5), using Rob's suggestion: https://lore.kernel.org/all/YXi5CUCEi7YmNxXM@robh.at.kernel.org/ Best regards, Krzysztof Cc: Chris Packham <chris.packham@alliedtelesis.co.nz> Cc: Bartosz Golaszewski <brgl@bgdev.pl> Cc: Sean Anderson <sean.anderson@seco.com> Krzysztof Kozlowski (6): of: Add of_phandle_args_equal() helper cpufreq: do not open-code of_phandle_args_equal() reset: gpio: Add GPIO-based reset controller reset: Instantiate reset GPIO controller for shared reset-gpios ASoC: dt-bindings: qcom,wsa8840: Add reset-gpios for shared line ASoC: codecs: wsa884x: Allow sharing reset GPIO .../bindings/sound/qcom,wsa8840.yaml | 11 +- MAINTAINERS | 5 + drivers/reset/Kconfig | 9 + drivers/reset/Makefile | 1 + drivers/reset/core.c | 215 ++++++++++++++++-- drivers/reset/reset-gpio.c | 119 ++++++++++ include/linux/cpufreq.h | 3 +- include/linux/of.h | 16 ++ include/linux/reset-controller.h | 4 + sound/soc/codecs/wsa884x.c | 53 ++++- 10 files changed, 410 insertions(+), 26 deletions(-) create mode 100644 drivers/reset/reset-gpio.c