| Message ID | 20210222130735.1313443-1-djrscally@gmail.com |
|---|---|
| Headers | show |
| Series | Introduce intel_skl_int3472 module | expand |
Hi all On 22/02/2021 13:07, Daniel Scally wrote: > diff --git a/drivers/platform/x86/intel-int3472/Kconfig b/drivers/platform/x86/intel-int3472/Kconfig > new file mode 100644 > index 000000000000..b94622245c21 > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/Kconfig > @@ -0,0 +1,31 @@ > +config INTEL_SKL_INT3472 > + tristate "Intel SkyLake ACPI INT3472 Driver" > + depends on ACPI > + depends on REGULATOR > + depends on GPIOLIB > + depends on COMMON_CLK && CLKDEV_LOOKUP > + depends on I2C > + select MFD_CORE > + select REGMAP_I2C > + help > + This driver adds support for the INT3472 ACPI devices found on some > + Intel SkyLake devices. > + > + The INT3472 is an Intel camera power controller, a logical device > + found on some Skylake-based systems that can map to different > + hardware devices depending on the platform. On machines > + designed for Chrome OS, it maps to a TPS68470 camera PMIC. On > + machines designed for Windows, it maps to either a TP68470 > + camera PMIC, a uP6641Q sensor PMIC, or a set of discrete GPIOs > + and power gates. > + > + If your device was designed for Chrome OS, this driver will provide > + an ACPI OpRegion, which must be available before any of the devices > + using it are probed. For this reason, you should select Y if your > + device was designed for ChromeOS. For the same reason the > + I2C_DESIGNWARE_PLATFORM option must be set to Y too. > + > + Say Y or M here if you have a SkyLake device designed for use > + with Windows or ChromeOS. Say N here if you are not sure. > + > + The module will be named "intel-skl-int3472" The Kconfig option for the existing tps68470 driver is a bool which depends on I2C_DESIGNWARE_PLATFORM=y, giving the following reason: This option is a bool as it provides an ACPI operation region, which must be available before any of the devices using this are probed. This option also configures the designware-i2c driver to be built-in, for the same reason. One problem I've faced is that that scenario only applies to some devices that this new driver can support, so hard-coding it as built in didn't make much sense. For that reason I opted to set it tristate, but of course that issue still exists for ChromeOS devices where the OpRegion will be registered. I opted for simply documenting that requirement, as is done in aaac4a2eadaa6: "mfd: axp20x-i2c: Document that this must be builtin on x86", but that's not entirely satisfactory. Possible alternatives might be setting "depends on I2C_DESIGNWARE_PLATFORM=y if CHROME_PLATFORMS" or something similar, though of course the User would still have to realise they need to build-in the INTEL_SKL_INT3472 Kconfig option too. Feedback around this issue would be particularly welcome, as I'm not sure what the best approach might be.
On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: > > The acpi_walk_dep_device_list() is not as generalisable as its name > implies, serving only to decrement the dependency count for each > dependent device of the input. Extend the function to instead accept > a callback which can be applied to all the dependencies in acpi_dep_list. > Replace all existing calls to the function with calls to a wrapper, passing > a callback that applies the same dependency reduction. The code looks okay to me, if it was the initial idea, feel free to add Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> See a few nit-picks below. > Suggested-by: Rafael J. Wysocki <rafael@kernel.org> > Signed-off-by: Daniel Scally <djrscally@gmail.com> > --- > Changes in v3: > - patch introduced > > drivers/acpi/ec.c | 2 +- > drivers/acpi/pmic/intel_pmic_chtdc_ti.c | 2 +- > drivers/acpi/scan.c | 58 ++++++++++++++++------- > drivers/gpio/gpiolib-acpi.c | 2 +- > drivers/i2c/i2c-core-acpi.c | 2 +- > drivers/platform/surface/surface3_power.c | 2 +- > include/acpi/acpi_bus.h | 7 +++ > include/linux/acpi.h | 4 +- > 8 files changed, 55 insertions(+), 24 deletions(-) > > diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c > index 13565629ce0a..a258db713bd2 100644 > --- a/drivers/acpi/ec.c > +++ b/drivers/acpi/ec.c > @@ -1627,7 +1627,7 @@ static int acpi_ec_add(struct acpi_device *device) > WARN(!ret, "Could not request EC cmd io port 0x%lx", ec->command_addr); > > /* Reprobe devices depending on the EC */ > - acpi_walk_dep_device_list(ec->handle); > + acpi_dev_flag_dependency_met(ec->handle); > > acpi_handle_debug(ec->handle, "enumerated.\n"); > return 0; > diff --git a/drivers/acpi/pmic/intel_pmic_chtdc_ti.c b/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > index a5101b07611a..59cca504325e 100644 > --- a/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > +++ b/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > @@ -117,7 +117,7 @@ static int chtdc_ti_pmic_opregion_probe(struct platform_device *pdev) > return err; > > /* Re-enumerate devices depending on PMIC */ > - acpi_walk_dep_device_list(ACPI_HANDLE(pdev->dev.parent)); > + acpi_dev_flag_dependency_met(ACPI_HANDLE(pdev->dev.parent)); > return 0; > } > > diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c > index 80b668c80073..c9e4190316ef 100644 > --- a/drivers/acpi/scan.c > +++ b/drivers/acpi/scan.c > @@ -49,12 +49,6 @@ static DEFINE_MUTEX(acpi_hp_context_lock); > */ > static u64 spcr_uart_addr; > > -struct acpi_dep_data { > - struct list_head node; > - acpi_handle supplier; > - acpi_handle consumer; > -}; > - > void acpi_scan_lock_acquire(void) > { > mutex_lock(&acpi_scan_lock); > @@ -2099,30 +2093,58 @@ static void acpi_bus_attach(struct acpi_device *device, bool first_pass) > device->handler->hotplug.notify_online(device); > } > > -void acpi_walk_dep_device_list(acpi_handle handle) > +static int __acpi_dev_flag_dependency_met(struct acpi_dep_data *dep, > + void *data) > { > - struct acpi_dep_data *dep, *tmp; > struct acpi_device *adev; > > + acpi_bus_get_device(dep->consumer, &adev); > + if (!adev) > + return 0; > + > + adev->dep_unmet--; > + if (!adev->dep_unmet) > + acpi_bus_attach(adev, true); > + > + list_del(&dep->node); > + kfree(dep); > + return 0; > +} > + > +void acpi_walk_dep_device_list(acpi_handle handle, > + int (*callback)(struct acpi_dep_data *, void *), > + void *data) > +{ > + struct acpi_dep_data *dep, *tmp; > + int ret; > + > mutex_lock(&acpi_dep_list_lock); > list_for_each_entry_safe(dep, tmp, &acpi_dep_list, node) { > if (dep->supplier == handle) { > - acpi_bus_get_device(dep->consumer, &adev); > - if (!adev) > - continue; > - > - adev->dep_unmet--; > - if (!adev->dep_unmet) > - acpi_bus_attach(adev, true); > - > - list_del(&dep->node); > - kfree(dep); > + ret = callback(dep, data); > + if (ret) > + break; > } > } > mutex_unlock(&acpi_dep_list_lock); > } > EXPORT_SYMBOL_GPL(acpi_walk_dep_device_list); > > +/** > + * acpi_dev_flag_dependency_met() - Inform consumers of @handle that the device > + * is now active ... - Inform consumers that the device is now active (and it will be on one line) > + * @handle: acpi_handle for the supplier device ACPI handle > + * > + * This function walks through the dependencies list and informs each consumer > + * of @handle that their dependency upon it is now met. Devices with no more > + * unmet dependencies will be attached to the acpi bus. acpi -> ACPI ? > + */ > +void acpi_dev_flag_dependency_met(acpi_handle handle) > +{ Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? > + acpi_walk_dep_device_list(handle, __acpi_dev_flag_dependency_met, NULL); > +} > +EXPORT_SYMBOL_GPL(acpi_dev_flag_dependency_met); > + > /** > * acpi_bus_scan - Add ACPI device node objects in a given namespace scope. > * @handle: Root of the namespace scope to scan. > diff --git a/drivers/gpio/gpiolib-acpi.c b/drivers/gpio/gpiolib-acpi.c > index e37a57d0a2f0..e4d728fda982 100644 > --- a/drivers/gpio/gpiolib-acpi.c > +++ b/drivers/gpio/gpiolib-acpi.c > @@ -1254,7 +1254,7 @@ void acpi_gpiochip_add(struct gpio_chip *chip) > > acpi_gpiochip_request_regions(acpi_gpio); > acpi_gpiochip_scan_gpios(acpi_gpio); > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > } > > void acpi_gpiochip_remove(struct gpio_chip *chip) > diff --git a/drivers/i2c/i2c-core-acpi.c b/drivers/i2c/i2c-core-acpi.c > index 37c510d9347a..38647cf34bde 100644 > --- a/drivers/i2c/i2c-core-acpi.c > +++ b/drivers/i2c/i2c-core-acpi.c > @@ -283,7 +283,7 @@ void i2c_acpi_register_devices(struct i2c_adapter *adap) > if (!handle) > return; > > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > } > > static const struct acpi_device_id i2c_acpi_force_400khz_device_ids[] = { > diff --git a/drivers/platform/surface/surface3_power.c b/drivers/platform/surface/surface3_power.c > index cc4f9cba6856..ad895285d3e9 100644 > --- a/drivers/platform/surface/surface3_power.c > +++ b/drivers/platform/surface/surface3_power.c > @@ -478,7 +478,7 @@ static int mshw0011_install_space_handler(struct i2c_client *client) > return -ENOMEM; > } > > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > return 0; > } > > diff --git a/include/acpi/acpi_bus.h b/include/acpi/acpi_bus.h > index 02a716a0af5d..91172af3a04d 100644 > --- a/include/acpi/acpi_bus.h > +++ b/include/acpi/acpi_bus.h > @@ -278,6 +278,12 @@ struct acpi_device_power { > struct acpi_device_power_state states[ACPI_D_STATE_COUNT]; /* Power states (D0-D3Cold) */ > }; > > +struct acpi_dep_data { > + struct list_head node; > + acpi_handle supplier; > + acpi_handle consumer; > +}; > + > /* Performance Management */ > > struct acpi_device_perf_flags { > @@ -683,6 +689,7 @@ static inline bool acpi_device_can_poweroff(struct acpi_device *adev) > > bool acpi_dev_hid_uid_match(struct acpi_device *adev, const char *hid2, const char *uid2); > > +void acpi_dev_flag_dependency_met(acpi_handle handle); > struct acpi_device * > acpi_dev_get_next_match_dev(struct acpi_device *adev, const char *hid, const char *uid, s64 hrv); > struct acpi_device * > diff --git a/include/linux/acpi.h b/include/linux/acpi.h > index 2630c2e953f7..2d5e6e88e8a0 100644 > --- a/include/linux/acpi.h > +++ b/include/linux/acpi.h > @@ -655,7 +655,9 @@ extern bool acpi_driver_match_device(struct device *dev, > const struct device_driver *drv); > int acpi_device_uevent_modalias(struct device *, struct kobj_uevent_env *); > int acpi_device_modalias(struct device *, char *, int); > -void acpi_walk_dep_device_list(acpi_handle handle); > +void acpi_walk_dep_device_list(acpi_handle handle, > + int (*callback)(struct acpi_dep_data *, void *), > + void *data); > > struct platform_device *acpi_create_platform_device(struct acpi_device *, > struct property_entry *); > -- > 2.25.1 >
On Mon, Feb 22, 2021 at 01:07:32PM +0000, Daniel Scally wrote: > Some places in the kernel allow users to map resources to a device > using device name (for example, in the struct gpiod_lookup_table). > Currently this involves waiting for the I2C client to have been registered > so we can use dev_name(&client->dev). We want to add a function to allow > users to refer to an I2C device by name before it has been instantiated, > so create a macro for the format that's accessible outside the I2C layer > and use it in i2c_dev_set_name(). > > Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> > Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com> > Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > Signed-off-by: Daniel Scally <djrscally@gmail.com> Acked-by: Wolfram Sang <wsa@kernel.org> # for changing I2C core
On Mon, Feb 22, 2021 at 3:11 PM Daniel Scally <djrscally@gmail.com> wrote: > > v1 for this series was originally 14-18 of this series: > https://lore.kernel.org/linux-media/20201130133129.1024662-1-djrscally@gmail.com/T/#m91934e12e3d033da2e768e952ea3b4a125ee3e67 > > v2 was here: > https://lore.kernel.org/platform-driver-x86/20210118003428.568892-1-djrscally@gmail.com/ > > Series level changelog: > > - Dropped the patch moving acpi_lpss_dep() to utils since it's not used > in acpi_dev_get_dependent_dev() anymore. > - Replaced it with a patch extending acpi_walk_dep_device_list() to be > able to apply a given callback against each device in acpi_dep_list > - Dropped the patch creating acpi_i2c_dev_name() and simply open coded > that functionality. > > This has been tested on a number of devices, but currently **not** on a device > designed for ChromeOS, which we ideally need to do to ensure no regression > caused by replacing the tps68470 MFD driver. Sakari / Tomasz, is there any way > you could help with that? Unfortunately, I don't have a device to test it on > myself. +Cc: Dmitry and Guenter. Guys, do you know by a chance who can help with the above request from Daniel? > Original cover letter: > > At the moment in the kernel the ACPI _HID INT3472 is taken by the tps68470 > MFD driver, but that driver can only handle some of the cases of that _HID > that we see. There are at least these three possibilities: > > 1. INT3472 devices that provide GPIOs through the usual framework and run > power and clocks through an operation region; this is the situation that > the current module handles and is seen on ChromeOS devices > 2. INT3472 devices that provide GPIOs, plus clocks and regulators that are > meant to be driven through the usual frameworks, usually seen on devices > designed to run Windows > 3. INT3472 devices that don't actually represent a physical tps68470, but > are being used as a convenient way of grouping a bunch of system GPIO > lines that are intended to enable power and clocks for sensors which > are called out as dependent on them. Also seen on devices designed to > run Windows. > > This series introduces a new module which registers: > > 1. An i2c driver that determines which scenario (#1 or #2) applies to the > machine and registers platform devices to be bound to GPIO, OpRegion, > clock and regulator drivers as appropriate. > 2. A platform driver that binds to the dummy INT3472 devices described in > #3 > > The platform driver for the dummy device registers the GPIO lines that > enable the clocks and regulators to the sensors via those frameworks so > that sensor drivers can consume them in the usual fashion. The existing > GPIO and OpRegion tps68470 drivers will work with the i2c driver that's > registered. Clock and regulator drivers are available but have not so far been > tested, so aren't part of this series. > > The existing mfd/tps68470.c driver being thus superseded, it is removed. > > Thanks > Dan > > Daniel Scally (6): > ACPI: scan: Extend acpi_walk_dep_device_list() > ACPI: scan: Add function to fetch dependent of acpi device > i2c: core: Add a format macro for I2C device names > gpiolib: acpi: Export acpi_get_gpiod() > platform/x86: Add intel_skl_int3472 driver > mfd: tps68470: Remove tps68470 MFD driver > > MAINTAINERS | 5 + > drivers/acpi/ec.c | 2 +- > drivers/acpi/pmic/Kconfig | 2 +- > drivers/acpi/pmic/intel_pmic_chtdc_ti.c | 2 +- > drivers/acpi/scan.c | 92 ++- > drivers/gpio/Kconfig | 2 +- > drivers/gpio/gpiolib-acpi.c | 38 +- > drivers/i2c/i2c-core-acpi.c | 2 +- > drivers/i2c/i2c-core-base.c | 4 +- > drivers/mfd/Kconfig | 18 - > drivers/mfd/Makefile | 1 - > drivers/mfd/tps68470.c | 97 --- > drivers/platform/surface/surface3_power.c | 2 +- > drivers/platform/x86/Kconfig | 2 + > drivers/platform/x86/Makefile | 1 + > drivers/platform/x86/intel-int3472/Kconfig | 31 + > drivers/platform/x86/intel-int3472/Makefile | 4 + > .../intel-int3472/intel_skl_int3472_common.c | 106 ++++ > .../intel-int3472/intel_skl_int3472_common.h | 110 ++++ > .../intel_skl_int3472_discrete.c | 592 ++++++++++++++++++ > .../intel_skl_int3472_tps68470.c | 113 ++++ > include/acpi/acpi_bus.h | 8 + > include/linux/acpi.h | 4 +- > include/linux/gpio/consumer.h | 7 + > include/linux/i2c.h | 3 + > 25 files changed, 1100 insertions(+), 148 deletions(-) > delete mode 100644 drivers/mfd/tps68470.c > create mode 100644 drivers/platform/x86/intel-int3472/Kconfig > create mode 100644 drivers/platform/x86/intel-int3472/Makefile > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c > > -- > 2.25.1 >
Hi Andy - thanks for comments! On 22/02/2021 14:58, Andy Shevchenko wrote: > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: >> ACPI devices with _HID INT3472 are currently matched to the tps68470 >> driver, however this does not cover all situations in which that _HID >> occurs. We've encountered three possibilities: >> >> 1. On Chrome OS devices, an ACPI device with _HID INT3472 (representing >> a physical TPS68470 device) that requires a GPIO and OpRegion driver >> 2. On devices designed for Windows, an ACPI device with _HID INT3472 >> (again representing a physical TPS68470 device) which requires GPIO, >> Clock and Regulator drivers. >> 3. On other devices designed for Windows, an ACPI device with _HID >> INT3472 which does **not** represent a physical TPS68470, and is instead >> used as a dummy device to group some system GPIO lines which are meant >> to be consumed by the sensor that is dependent on this entry. >> >> This commit adds a new module, registering a platform driver to deal >> with the 3rd scenario plus an i2c driver to deal with #1 and #2, by >> querying the CLDB buffer found against INT3472 entries to determine >> which is most appropriate. > Can you split CLK parts (and maybe regulators as well) to something > like intel_skl_int3472_clk.c? Sure, no problem > >> +#include <linux/acpi.h> >> +#include <linux/i2c.h> >> +#include <linux/platform_device.h> >> +#include <linux/slab.h> >> + dev_err(&adev->dev, "%s object is not an ACPI buffer\n", id); > Perhaps acpi_handle_err() et al. instead of dev_*(&adev->dev, ...) > where it's applicable? Ah - yes, ok, thanks. TIL those exist >> + if (obj->buffer.length > sizeof(*cldb)) { >> + dev_err(&adev->dev, "The CLDB buffer is too large\n"); >> + ret = -EINVAL; > ENOSPC? ENOMEM? I still think EINVAL actually, as in this case the problem isn't that space couldn't be allocated but that the buffer in the SSDB is larger than I expect it to be, which means the definition of it has changed / this device isn't actually supported. >> + ret = platform_driver_register(&int3472_discrete); >> + if (ret) >> + return ret; >> + >> + ret = i2c_register_driver(THIS_MODULE, &int3472_tps68470); >> + if (ret) >> + platform_driver_unregister(&int3472_discrete); > Not a fan of the above, but let's see what others will say... Yeah; happy to discuss this more if needed. >> +#include <linux/clk-provider.h> > This is definitely not for *.h. (Not all C files needed this) > >> +#include <linux/gpio/machine.h> > Ditto. > >> +#include <linux/regulator/driver.h> >> +#include <linux/regulator/machine.h> > Ditto. Yep; I'll move them to *_clk.c and *_regulator.c files. >> +static int skl_int3472_clk_prepare(struct clk_hw *hw) >> +{ >> + struct int3472_gpio_clock *clk = to_int3472_clk(hw); >> + >> + gpiod_set_value(clk->ena_gpio, 1); >> + if (clk->led_gpio) > Make it optional and drop this check. Same for other places of use of this GPIO. Oops, of course, thanks >> +static int skl_int3472_clk_enable(struct clk_hw *hw) >> +{ >> + /* >> + * We're just turning a GPIO on to enable, which operation has the >> + * potential to sleep. Given enable cannot sleep, but prepare can, >> + * we toggle the GPIO in prepare instead. Thus, nothing to do here. >> + */ > Missed . and / or () in some words? (Describing callbacks, personally > I use the form "->callback()" in such cases) OK, I'll fix the comment to match that style. >> +static unsigned int skl_int3472_get_clk_frequency(struct int3472_discrete_device *int3472) >> +{ >> + union acpi_object *obj; >> >> + unsigned int ret = 0; > unsigned for ret is unusual. Looking into the code, first of all it > doesn't need this assignment; second, it probably can gain a better > name: "frequency"? Yep ok, I'll rename to freq/frequency >> + if (!IS_ERR_OR_NULL(sensor_config) && sensor_config->function_maps) { > Hmm... > > Would > > if (IS_ERR_OR_NULL(sensor_config)) > return 0; > > if (!_maps) > return 0; > > with respective comments working here? No, because the absence of either sensor_config or sensor_config->function_maps is not a failure mode. We only need to provide sensor_configs for some platforms, and function_maps for even fewer. So if that check is false, the rest of the function should still execute. >> +static int skl_int3472_register_clock(struct int3472_discrete_device *int3472) >> +{ >> + struct clk_init_data init = { >> + .ops = &skl_int3472_clock_ops, >> + .flags = CLK_GET_RATE_NOCACHE, >> + }; >> + int ret = 0; >> + >> + init.name = kasprintf(GFP_KERNEL, "%s-clk", >> + acpi_dev_name(int3472->adev)); > devm_*() ? Or is the lifetime different? No it's not; I'll use devm_*(), thanks >> + sensor_config = int3472->sensor_config; >> + if (IS_ERR_OR_NULL(sensor_config)) { >> + dev_err(int3472->dev, "No sensor module config\n"); >> + return PTR_ERR(sensor_config); > NULL -> 0. Is it okay? Ah, no it's not - good catch thank you. >> + if (ares->type != ACPI_RESOURCE_TYPE_GPIO || >> + ares->data.gpio.connection_type != ACPI_RESOURCE_GPIO_TYPE_IO) >> + return 1; /* Deliberately positive so parsing continues */ > I don't like to lose control over ACPI_RESOURCE_TYPE_GPIO, i.e. > spreading it over kernel code (yes, I know about one existing TS > case). > Consider to provide a helper in analogue to acpi_gpio_get_irq_resource(). Sure, but I probably name it acpi_gpio_is_io_resource() - a function named "get" which returns a bool seems a bit funny to me. >> + if (ret < 0 && ret != -EPROBE_DEFER) >> + dev_err(int3472->dev, err_msg); > dev_err_probe() will make the above conditional go away. And you may even do... Ah-ha - thought that must exist but couldn't find it - thank you. >> + if (int3472->clock.ena_gpio) { > Not sure you need this here. We haven't seen a device that lacks a clock enable GPIO it's true, but since all the other kinds seem optional it didn't seem impossible that that one is optional too. I can remove if you prefer and we can just deal with it when we encounter one like that though? >> + /* Max num GPIOs we've seen plus a terminator */ >> + int3472 = kzalloc(struct_size(int3472, gpios.table, >> + INT3472_MAX_SENSOR_GPIOS + 1), GFP_KERNEL); > Wonder of you can use devm_*() APIs in this function. Yeah I can, I'll switch to that. >> +int skl_int3472_discrete_remove(struct platform_device *pdev) >> +{ >> + struct int3472_discrete_device *int3472 = platform_get_drvdata(pdev); >> + if (int3472->gpios.dev_id) >> + gpiod_remove_lookup_table(&int3472->gpios); > gpiod_remove_lookup_table() is now NULL-aware. > But in any case I guess you don't need the above check. Sorry; forgot to call out that I didn't follow that suggestion; int3472->gpios is a _struct_ rather than a pointer, so &int3472->gpios won't be NULL, even if I haven't filled anything in to there yet because it failed before it got to that point. So, not sure that it quite works there. > >> + if (!IS_ERR(int3472->regulator.rdev)) >> + regulator_unregister(int3472->regulator.rdev); > Shouldn't it be the pointer to the regulator itself? int3472->regulator is type struct int3472_gpio_regulator, the .rdev is the normal regulator_dev > >> + if (!IS_ERR(int3472->clock.clk)) > If you get it optional, you won't need this additional check. Yes - here it will definitely work; thanks, I'll add that patch >> + ret = skl_int3472_fill_cldb(adev, &cldb); >> + if (!ret && cldb.control_logic_type != 2) { >> + dev_err(&client->dev, "Unsupported control logic type %u\n", >> + cldb.control_logic_type); >> + return -EINVAL; >> + } >> + >> + if (ret) >> + cldb_present = false; > if (ret) > ... > else if (...) { > ... > return ...; > } Oh yeah...now you point that out I have no idea what I was thinking there...
Hi Hans, thanks for the input On 22/02/2021 13:27, Hans de Goede wrote: > Hi, > > On 2/22/21 2:19 PM, Daniel Scally wrote: >> Hi all >> >> On 22/02/2021 13:07, Daniel Scally wrote: >>> diff --git a/drivers/platform/x86/intel-int3472/Kconfig b/drivers/platform/x86/intel-int3472/Kconfig >>> new file mode 100644 >>> index 000000000000..b94622245c21 >>> --- /dev/null >>> +++ b/drivers/platform/x86/intel-int3472/Kconfig >>> @@ -0,0 +1,31 @@ >>> +config INTEL_SKL_INT3472 >>> + tristate "Intel SkyLake ACPI INT3472 Driver" >>> + depends on ACPI >>> + depends on REGULATOR >>> + depends on GPIOLIB >>> + depends on COMMON_CLK && CLKDEV_LOOKUP >>> + depends on I2C >>> + select MFD_CORE >>> + select REGMAP_I2C >>> + help >>> + This driver adds support for the INT3472 ACPI devices found on some >>> + Intel SkyLake devices. >>> + >>> + The INT3472 is an Intel camera power controller, a logical device >>> + found on some Skylake-based systems that can map to different >>> + hardware devices depending on the platform. On machines >>> + designed for Chrome OS, it maps to a TPS68470 camera PMIC. On >>> + machines designed for Windows, it maps to either a TP68470 >>> + camera PMIC, a uP6641Q sensor PMIC, or a set of discrete GPIOs >>> + and power gates. >>> + >>> + If your device was designed for Chrome OS, this driver will provide >>> + an ACPI OpRegion, which must be available before any of the devices >>> + using it are probed. For this reason, you should select Y if your >>> + device was designed for ChromeOS. For the same reason the >>> + I2C_DESIGNWARE_PLATFORM option must be set to Y too. >>> + >>> + Say Y or M here if you have a SkyLake device designed for use >>> + with Windows or ChromeOS. Say N here if you are not sure. >>> + >>> + The module will be named "intel-skl-int3472" >> The Kconfig option for the existing tps68470 driver is a bool which >> depends on I2C_DESIGNWARE_PLATFORM=y, giving the following reason: >> >> This option is a bool as it provides an ACPI operation >> region, which must be available before any of the devices >> using this are probed. This option also configures the >> designware-i2c driver to be built-in, for the same reason. >> >> One problem I've faced is that that scenario only applies to some >> devices that this new driver can support, so hard-coding it as built in >> didn't make much sense. For that reason I opted to set it tristate, but >> of course that issue still exists for ChromeOS devices where the >> OpRegion will be registered. I opted for simply documenting that >> requirement, as is done in aaac4a2eadaa6: "mfd: axp20x-i2c: Document >> that this must be builtin on x86", but that's not entirely satisfactory. >> Possible alternatives might be setting "depends on >> I2C_DESIGNWARE_PLATFORM=y if CHROME_PLATFORMS" or something similar, >> though of course the User would still have to realise they need to >> build-in the INTEL_SKL_INT3472 Kconfig option too. >> >> Feedback around this issue would be particularly welcome, as I'm not >> sure what the best approach might be. > This is a tricky area, I actually wrote the "mfd: axp20x-i2c: Document > that this must be builtin on x86" patch you refer to. At first I tried > to express the dependency in Kconfig language but things got too complex > and Kconfig sometimes became unhappy about circular deps (or something > like that). Yes, I had a go too; with similar results > The most important thing here is to make sure that the generic configs > shipped by distros get this right; and we can hope that people creating > those configs at least read the help text... > > So all in all I believe that just documenting the requirement is fine. OK - that's what I'm hoping is the consensus, as I don't think it can be made _entirely_ seamless through dependencies or whatever anyway, in which case documenting it seems like the cleanest approach to me. > > The alternative would be to just change I2C_DESIGNWARE_PLATFORM (and the > core) to a bool, or at least make it not selectable as module when > X86 and ACPI are set... That would be a bit of a big hammer but might > not be the worst idea actually. > > Regards, > > Hans >
On Mon, Feb 22, 2021 at 10:35:44PM +0000, Daniel Scally wrote: > On 22/02/2021 14:58, Andy Shevchenko wrote: > > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: ... > >> + if (obj->buffer.length > sizeof(*cldb)) { > >> + dev_err(&adev->dev, "The CLDB buffer is too large\n"); > >> + ret = -EINVAL; > > ENOSPC? ENOMEM? > > I still think EINVAL actually, as in this case the problem isn't that > space couldn't be allocated but that the buffer in the SSDB is larger > than I expect it to be, which means the definition of it has changed / > this device isn't actually supported. OK! ... > >> + if (!IS_ERR_OR_NULL(sensor_config) && sensor_config->function_maps) { > > Hmm... > > > > Would > > > > if (IS_ERR_OR_NULL(sensor_config)) > > return 0; > > > > if (!_maps) > > return 0; > > > > with respective comments working here? > > No, because the absence of either sensor_config or > sensor_config->function_maps is not a failure mode. We only need to > provide sensor_configs for some platforms, and function_maps for even > fewer. So if that check is false, the rest of the function should still > execute. I see, thanks for elaboration. ... > >> + if (ares->type != ACPI_RESOURCE_TYPE_GPIO || > >> + ares->data.gpio.connection_type != ACPI_RESOURCE_GPIO_TYPE_IO) > >> + return 1; /* Deliberately positive so parsing continues */ > > I don't like to lose control over ACPI_RESOURCE_TYPE_GPIO, i.e. > > spreading it over kernel code (yes, I know about one existing TS > > case). > > Consider to provide a helper in analogue to acpi_gpio_get_irq_resource(). > > Sure, but I probably name it acpi_gpio_is_io_resource() - a function > named "get" which returns a bool seems a bit funny to me. But don't you need the resource itself? You may extract and check resource at the same time as acpi_gpio_get_irq_resource() does. ... > >> + struct int3472_discrete_device *int3472 = platform_get_drvdata(pdev); > >> + if (int3472->gpios.dev_id) > >> + gpiod_remove_lookup_table(&int3472->gpios); > > gpiod_remove_lookup_table() is now NULL-aware. > > But in any case I guess you don't need the above check. > > Sorry; forgot to call out that I didn't follow that suggestion; > int3472->gpios is a _struct_ rather than a pointer, so &int3472->gpios > won't be NULL, even if I haven't filled anything in to there yet because > it failed before it got to that point. So, not sure that it quite works > there. I think if you initialize the ->list member you can remove without check.
On 23/02/2021 12:01, Andy Shevchenko wrote: >>>> + if (ares->type != ACPI_RESOURCE_TYPE_GPIO || >>>> + ares->data.gpio.connection_type != ACPI_RESOURCE_GPIO_TYPE_IO) >>>> + return 1; /* Deliberately positive so parsing continues */ >>> I don't like to lose control over ACPI_RESOURCE_TYPE_GPIO, i.e. >>> spreading it over kernel code (yes, I know about one existing TS >>> case). >>> Consider to provide a helper in analogue to acpi_gpio_get_irq_resource(). >> Sure, but I probably name it acpi_gpio_is_io_resource() - a function >> named "get" which returns a bool seems a bit funny to me. > But don't you need the resource itself? > > You may extract and check resource at the same time as > acpi_gpio_get_irq_resource() does. Oh! Reading comprehension fail; I didn't notice it was returning the pointer through agpio; you're right of course. > > ... > >>>> + struct int3472_discrete_device *int3472 = platform_get_drvdata(pdev); >>>> + if (int3472->gpios.dev_id) >>>> + gpiod_remove_lookup_table(&int3472->gpios); >>> gpiod_remove_lookup_table() is now NULL-aware. >>> But in any case I guess you don't need the above check. >> Sorry; forgot to call out that I didn't follow that suggestion; >> int3472->gpios is a _struct_ rather than a pointer, so &int3472->gpios >> won't be NULL, even if I haven't filled anything in to there yet because >> it failed before it got to that point. So, not sure that it quite works >> there. > I think if you initialize the ->list member you can remove without check. I'll give that a try - thanks >
Hi Daniel, Thank you for the patch. On Mon, Feb 22, 2021 at 01:07:34PM +0000, Daniel Scally wrote: > ACPI devices with _HID INT3472 are currently matched to the tps68470 > driver, however this does not cover all situations in which that _HID > occurs. We've encountered three possibilities: > > 1. On Chrome OS devices, an ACPI device with _HID INT3472 (representing > a physical TPS68470 device) that requires a GPIO and OpRegion driver > 2. On devices designed for Windows, an ACPI device with _HID INT3472 > (again representing a physical TPS68470 device) which requires GPIO, > Clock and Regulator drivers. > 3. On other devices designed for Windows, an ACPI device with _HID > INT3472 which does **not** represent a physical TPS68470, and is instead > used as a dummy device to group some system GPIO lines which are meant > to be consumed by the sensor that is dependent on this entry. > > This commit adds a new module, registering a platform driver to deal > with the 3rd scenario plus an i2c driver to deal with #1 and #2, by > querying the CLDB buffer found against INT3472 entries to determine > which is most appropriate. > > Suggested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> > Signed-off-by: Daniel Scally <djrscally@gmail.com> > --- > Changes in v3: > > - Added clk_recalc_rate() operation to the clk that's registered, plus > some associated functions / revisions. > - Moved source and header files to their own folder within platform/x86 > - Switched the GPIO toggling for the clock from enable/disable to > prepare/unprepare to avoid any sleep problems > - Switched handling of the privacy LED GPIO from map-to-sensor to being > toggled along with the clk enable GPIO; register the clock in > intel_skl_int3472_parse_crs() instead of during handle_gpio_resources > - Better commenting in a lot of places > - Used the sensor_name formed from acpi_dev_name(int3472->sensor) and > the i2c name format macro as the dev_id in regulator init data rather > than hardcoding an instance name. > - Fetched the sensor module config a single time rather than once per > GPIO > - Switched int3472-tps68470 driver to use MFD framework properly rather > than open coding the same functionality > - A myriad of other fixes too minor to call out. > > Suggested changes (from Andy) that I didn't follow: > > - Using clk-gpio.c as a library: The requirement to add clk_recalc_rate > so that clk_get_rate() would be supported, along with driving the > privacy LED on clk_enable() meant this wouldn't work I think > - Leave the MFD driver in its usual place, and prevent the INT3472 from > probing as an i2c device from ACPI. I definitely see the argument for > this, but in the end I think probably having all the code for the HID > within a single place is probably a bit preferable. > > MAINTAINERS | 5 + > drivers/platform/x86/Kconfig | 2 + > drivers/platform/x86/Makefile | 1 + > drivers/platform/x86/intel-int3472/Kconfig | 31 + > drivers/platform/x86/intel-int3472/Makefile | 4 + > .../intel-int3472/intel_skl_int3472_common.c | 106 ++++ > .../intel-int3472/intel_skl_int3472_common.h | 110 ++++ > .../intel_skl_int3472_discrete.c | 592 ++++++++++++++++++ > .../intel_skl_int3472_tps68470.c | 113 ++++ > 9 files changed, 964 insertions(+) > create mode 100644 drivers/platform/x86/intel-int3472/Kconfig > create mode 100644 drivers/platform/x86/intel-int3472/Makefile > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c > > diff --git a/MAINTAINERS b/MAINTAINERS > index a091b496fdd8..cf44b3e77b90 100644 > --- a/MAINTAINERS > +++ b/MAINTAINERS > @@ -9147,6 +9147,11 @@ S: Maintained > F: arch/x86/include/asm/intel_scu_ipc.h > F: drivers/platform/x86/intel_scu_* > > +INTEL SKYLAKE INT3472 ACPI DEVICE DRIVER > +M: Daniel Scally <djrscally@gmail.com> > +S: Maintained > +F: drivers/platform/x86/intel-int3472/intel_skl_int3472_* > + > INTEL SPEED SELECT TECHNOLOGY > M: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> > L: platform-driver-x86@vger.kernel.org > diff --git a/drivers/platform/x86/Kconfig b/drivers/platform/x86/Kconfig > index 91e6176cdfbd..9739d30951b6 100644 > --- a/drivers/platform/x86/Kconfig > +++ b/drivers/platform/x86/Kconfig > @@ -844,6 +844,8 @@ config INTEL_CHT_INT33FE > device and CONFIG_TYPEC_FUSB302=m and CONFIG_BATTERY_MAX17042=m > for Type-C device. > > +source "drivers/platform/x86/intel-int3472/Kconfig" > + > config INTEL_HID_EVENT > tristate "INTEL HID Event" > depends on ACPI > diff --git a/drivers/platform/x86/Makefile b/drivers/platform/x86/Makefile > index 581475f59819..2293b6c3d1c2 100644 > --- a/drivers/platform/x86/Makefile > +++ b/drivers/platform/x86/Makefile > @@ -86,6 +86,7 @@ obj-$(CONFIG_INTEL_HID_EVENT) += intel-hid.o > obj-$(CONFIG_INTEL_INT0002_VGPIO) += intel_int0002_vgpio.o > obj-$(CONFIG_INTEL_MENLOW) += intel_menlow.o > obj-$(CONFIG_INTEL_OAKTRAIL) += intel_oaktrail.o > +obj-$(CONFIG_INTEL_SKL_INT3472) += intel-int3472/ > obj-$(CONFIG_INTEL_VBTN) += intel-vbtn.o > > # MSI > diff --git a/drivers/platform/x86/intel-int3472/Kconfig b/drivers/platform/x86/intel-int3472/Kconfig > new file mode 100644 > index 000000000000..b94622245c21 > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/Kconfig > @@ -0,0 +1,31 @@ > +config INTEL_SKL_INT3472 > + tristate "Intel SkyLake ACPI INT3472 Driver" > + depends on ACPI > + depends on REGULATOR > + depends on GPIOLIB > + depends on COMMON_CLK && CLKDEV_LOOKUP > + depends on I2C Maybe these could be sorted alphabetically ? > + select MFD_CORE > + select REGMAP_I2C > + help > + This driver adds support for the INT3472 ACPI devices found on some > + Intel SkyLake devices. > + > + The INT3472 is an Intel camera power controller, a logical device > + found on some Skylake-based systems that can map to different > + hardware devices depending on the platform. On machines > + designed for Chrome OS, it maps to a TPS68470 camera PMIC. On > + machines designed for Windows, it maps to either a TP68470 > + camera PMIC, a uP6641Q sensor PMIC, or a set of discrete GPIOs > + and power gates. > + > + If your device was designed for Chrome OS, this driver will provide > + an ACPI OpRegion, which must be available before any of the devices > + using it are probed. For this reason, you should select Y if your > + device was designed for ChromeOS. For the same reason the > + I2C_DESIGNWARE_PLATFORM option must be set to Y too. > + > + Say Y or M here if you have a SkyLake device designed for use > + with Windows or ChromeOS. Say N here if you are not sure. > + > + The module will be named "intel-skl-int3472" > diff --git a/drivers/platform/x86/intel-int3472/Makefile b/drivers/platform/x86/intel-int3472/Makefile > new file mode 100644 > index 000000000000..c887ee7d52ca > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/Makefile > @@ -0,0 +1,4 @@ > +obj-$(CONFIG_INTEL_SKL_INT3472) += intel_skl_int3472.o > +intel_skl_int3472-objs := intel_skl_int3472_common.o \ > + intel_skl_int3472_discrete.o \ > + intel_skl_int3472_tps68470.o > diff --git a/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c b/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c > new file mode 100644 > index 000000000000..f61166b6c497 > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c > @@ -0,0 +1,106 @@ > +// SPDX-License-Identifier: GPL-2.0 > +/* Author: Dan Scally <djrscally@gmail.com> */ > + > +#include <linux/acpi.h> > +#include <linux/i2c.h> > +#include <linux/platform_device.h> > +#include <linux/slab.h> > + > +#include "intel_skl_int3472_common.h" > + > +union acpi_object *skl_int3472_get_acpi_buffer(struct acpi_device *adev, > + char *id) > +{ > + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; > + acpi_handle handle = adev->handle; > + union acpi_object *obj; > + acpi_status status; > + > + status = acpi_evaluate_object(handle, id, NULL, &buffer); > + if (ACPI_FAILURE(status)) > + return ERR_PTR(-ENODEV); > + > + obj = buffer.pointer; > + if (!obj) > + return ERR_PTR(-ENODEV); > + > + if (obj->type != ACPI_TYPE_BUFFER) { > + dev_err(&adev->dev, "%s object is not an ACPI buffer\n", id); > + kfree(obj); > + return ERR_PTR(-EINVAL); > + } > + > + return obj; > +} > + > +int skl_int3472_fill_cldb(struct acpi_device *adev, struct int3472_cldb *cldb) > +{ > + union acpi_object *obj; > + int ret = 0; > + > + obj = skl_int3472_get_acpi_buffer(adev, "CLDB"); > + if (IS_ERR(obj)) > + return PTR_ERR(obj); > + > + if (obj->buffer.length > sizeof(*cldb)) { > + dev_err(&adev->dev, "The CLDB buffer is too large\n"); > + ret = -EINVAL; > + goto out_free_obj; > + } > + > + memcpy(cldb, obj->buffer.pointer, obj->buffer.length); > + > +out_free_obj: > + kfree(obj); > + return ret; > +} > + > +static const struct acpi_device_id int3472_device_id[] = { > + { "INT3472", 0 }, > + { } > +}; > +MODULE_DEVICE_TABLE(acpi, int3472_device_id); > + > +static struct platform_driver int3472_discrete = { > + .driver = { > + .name = "int3472-discrete", > + .acpi_match_table = int3472_device_id, > + }, > + .probe = skl_int3472_discrete_probe, > + .remove = skl_int3472_discrete_remove, > +}; > + > +static struct i2c_driver int3472_tps68470 = { > + .driver = { > + .name = "int3472-tps68470", > + .acpi_match_table = int3472_device_id, > + }, > + .probe_new = skl_int3472_tps68470_probe, > +}; > + > +static int skl_int3472_init(void) > +{ > + int ret = 0; > + > + ret = platform_driver_register(&int3472_discrete); > + if (ret) > + return ret; > + > + ret = i2c_register_driver(THIS_MODULE, &int3472_tps68470); > + if (ret) > + platform_driver_unregister(&int3472_discrete); > + > + return ret; > +} > +module_init(skl_int3472_init); > + > +static void skl_int3472_exit(void) > +{ > + platform_driver_unregister(&int3472_discrete); > + i2c_del_driver(&int3472_tps68470); > +} > +module_exit(skl_int3472_exit); > + > +MODULE_DESCRIPTION("Intel SkyLake INT3472 ACPI Device Driver"); > +MODULE_AUTHOR("Daniel Scally <djrscally@gmail.com>"); > +MODULE_LICENSE("GPL v2"); > diff --git a/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h b/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h > new file mode 100644 > index 000000000000..9169356cd522 > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h > @@ -0,0 +1,110 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +/* Author: Dan Scally <djrscally@gmail.com> */ > + > +#ifndef _INTEL_SKL_INT3472_H > +#define _INTEL_SKL_INT3472_H > + > +#include <linux/clk-provider.h> > +#include <linux/gpio/machine.h> > +#include <linux/regulator/driver.h> > +#include <linux/regulator/machine.h> > +#include <linux/types.h> > + > +/* PMIC GPIO Types */ > +#define INT3472_GPIO_TYPE_RESET 0x00 > +#define INT3472_GPIO_TYPE_POWERDOWN 0x01 > +#define INT3472_GPIO_TYPE_POWER_ENABLE 0x0b > +#define INT3472_GPIO_TYPE_CLK_ENABLE 0x0c > +#define INT3472_GPIO_TYPE_PRIVACY_LED 0x0d > + > +#define INT3472_PDEV_MAX_NAME_LEN 23 > +#define INT3472_MAX_SENSOR_GPIOS 3 > + > +#define GPIO_REGULATOR_NAME_LENGTH 21 > +#define GPIO_REGULATOR_SUPPLY_NAME_LENGTH 9 > + > +#define CIO2_SENSOR_SSDB_MCLKSPEED_OFFSET 86 > + > +#define INT3472_REGULATOR(_name, _supply, _ops) \ > + (const struct regulator_desc) { \ > + .name = _name, \ > + .supply_name = _supply, \ > + .type = REGULATOR_VOLTAGE, \ > + .ops = _ops, \ > + .owner = THIS_MODULE, \ > + } > + > +#define to_int3472_clk(hw) \ > + container_of(hw, struct int3472_gpio_clock, clk_hw) > + > +#define to_int3472_device(clk) \ > + container_of(clk, struct int3472_discrete_device, clock) > + > +struct platform_device; > +struct i2c_client; > +struct acpi_device; Alphabetical order ? > + > +struct int3472_cldb { > + u8 version; > + /* > + * control logic type > + * 0: UNKNOWN > + * 1: DISCRETE(CRD-D) > + * 2: PMIC TPS68470 > + * 3: PMIC uP6641 > + */ > + u8 control_logic_type; > + u8 control_logic_id; > + u8 sensor_card_sku; > + u8 reserved[28]; > +}; > + > +struct int3472_gpio_function_remap { > + char *documented; > + char *actual; const char for both ? > +}; > + > +struct int3472_sensor_config { > + const char *sensor_module_name; > + struct regulator_consumer_supply supply_map; > + const struct int3472_gpio_function_remap *function_maps; > +}; > + > +struct int3472_discrete_device { > + struct acpi_device *adev; > + struct device *dev; > + struct acpi_device *sensor; > + const char *sensor_name; > + > + struct int3472_sensor_config *sensor_config; const > + > + struct int3472_gpio_regulator { > + char regulator_name[GPIO_REGULATOR_NAME_LENGTH]; > + char supply_name[GPIO_REGULATOR_SUPPLY_NAME_LENGTH]; > + struct gpio_desc *gpio; > + struct regulator_dev *rdev; > + struct regulator_desc rdesc; > + } regulator; > + > + struct int3472_gpio_clock { > + struct clk *clk; > + struct clk_hw clk_hw; > + struct clk_lookup *cl; > + struct gpio_desc *ena_gpio; > + struct gpio_desc *led_gpio; > + u32 frequency; > + } clock; > + > + unsigned int n_gpios; /* how many GPIOs have we seen */ > + unsigned int n_sensor_gpios; /* how many have we mapped to sensor */ > + struct gpiod_lookup_table gpios; > +}; > + > +int skl_int3472_discrete_probe(struct platform_device *pdev); > +int skl_int3472_discrete_remove(struct platform_device *pdev); > +int skl_int3472_tps68470_probe(struct i2c_client *client); > +union acpi_object *skl_int3472_get_acpi_buffer(struct acpi_device *adev, > + char *id); > +int skl_int3472_fill_cldb(struct acpi_device *adev, struct int3472_cldb *cldb); > + > +#endif > diff --git a/drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c b/drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c > new file mode 100644 > index 000000000000..40652161bbbf > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c > @@ -0,0 +1,592 @@ > +// SPDX-License-Identifier: GPL-2.0 > +/* Author: Dan Scally <djrscally@gmail.com> */ > + > +#include <linux/acpi.h> > +#include <linux/clkdev.h> > +#include <linux/gpio/consumer.h> > +#include <linux/i2c.h> > +#include <linux/kernel.h> > +#include <linux/module.h> > +#include <linux/overflow.h> > +#include <linux/platform_device.h> > +#include <linux/regulator/driver.h> > +#include <linux/slab.h> > + > +#include "intel_skl_int3472_common.h" > + > +/* > + * 79234640-9e10-4fea-a5c1-b5aa8b19756f > + * This _DSM GUID returns information about the GPIO lines mapped to a > + * discrete INT3472 device. Function number 1 returns a count of the GPIO > + * lines that are mapped. Subsequent functions return 32 bit ints encoding > + * information about the GPIO line, including its purpose. > + */ > +static const guid_t int3472_gpio_guid = > + GUID_INIT(0x79234640, 0x9e10, 0x4fea, > + 0xa5, 0xc1, 0xb5, 0xaa, 0x8b, 0x19, 0x75, 0x6f); > + > +/* > + * 822ace8f-2814-4174-a56b-5f029fe079ee > + * This _DSM GUID returns a string from the sensor device, which acts as a > + * module identifier. > + */ > +static const guid_t cio2_sensor_module_guid = > + GUID_INIT(0x822ace8f, 0x2814, 0x4174, > + 0xa5, 0x6b, 0x5f, 0x02, 0x9f, 0xe0, 0x79, 0xee); > + > +/* > + * Here follows platform specific mapping information that we can pass to > + * the functions mapping resources to the sensors. Where the sensors have > + * a power enable pin defined in DSDT we need to provide a supply name so > + * the sensor drivers can find the regulator. The device name will be derived > + * from the sensor's ACPI device within the code. Optionally, we can provide a > + * NULL terminated array of function name mappings to deal with any platform > + * specific deviations from the documented behaviour of GPIOs. > + * > + * Map a GPIO function name to NULL to prevent the driver from mapping that > + * GPIO at all. > + */ > + > +static const struct int3472_gpio_function_remap ov2680_gpio_function_remaps[] = { > + { "reset", NULL }, > + { "powerdown", "reset" }, > + { } > +}; > + > +static struct int3472_sensor_config int3472_sensor_configs[] = { This should be static const (and there will be some fallout due to that, as skl_int3472_register_regulator() modifies the supply_map, so I think you'll have a copy of supply_map in int3472_discrete_device). > + /* Lenovo Miix 510-12ISK - OV2680, Front */ > + { "GNDF140809R", { 0 }, ov2680_gpio_function_remaps}, Missing space before } (and below too). > + /* Lenovo Miix 510-12ISK - OV5648, Rear */ > + { "GEFF150023R", REGULATOR_SUPPLY("avdd", NULL), NULL}, > + /* Surface Go 1&2 - OV5693, Front */ > + { "YHCU", REGULATOR_SUPPLY("avdd", NULL), NULL}, > +}; > + > +/* > + * The regulators have to have .ops to be valid, but the only ops we actually > + * support are .enable and .disable which are handled via .ena_gpiod. Pass an > + * empty struct to clear the check without lying about capabilities. > + */ > +static const struct regulator_ops int3472_gpio_regulator_ops = { 0 }; > + > +static int skl_int3472_clk_prepare(struct clk_hw *hw) > +{ > + struct int3472_gpio_clock *clk = to_int3472_clk(hw); > + > + gpiod_set_value(clk->ena_gpio, 1); > + if (clk->led_gpio) > + gpiod_set_value(clk->led_gpio, 1); > + > + return 0; > +} > + > +static void skl_int3472_clk_unprepare(struct clk_hw *hw) > +{ > + struct int3472_gpio_clock *clk = to_int3472_clk(hw); > + > + gpiod_set_value(clk->ena_gpio, 0); > + if (clk->led_gpio) > + gpiod_set_value(clk->led_gpio, 0); > +} > + > +static int skl_int3472_clk_enable(struct clk_hw *hw) > +{ > + /* > + * We're just turning a GPIO on to enable, which operation has the > + * potential to sleep. Given enable cannot sleep, but prepare can, > + * we toggle the GPIO in prepare instead. Thus, nothing to do here. > + */ > + return 0; > +} > + > +static void skl_int3472_clk_disable(struct clk_hw *hw) > +{ > + /* Likewise, nothing to do here... */ > +} > + > +static unsigned int skl_int3472_get_clk_frequency(struct int3472_discrete_device *int3472) > +{ > + union acpi_object *obj; > + unsigned int ret = 0; > + > + obj = skl_int3472_get_acpi_buffer(int3472->sensor, "SSDB"); > + if (IS_ERR(obj)) > + return 0; /* report rate as 0 on error */ > + > + if (obj->buffer.length < CIO2_SENSOR_SSDB_MCLKSPEED_OFFSET + sizeof(u32)) { Should we define an ssdb structure instead of peeking into the buffer with an offset ? > + dev_err(int3472->dev, "The buffer is too small\n"); > + goto out_free_buff; > + } > + > + ret = *(u32 *)(obj->buffer.pointer + CIO2_SENSOR_SSDB_MCLKSPEED_OFFSET); > + > +out_free_buff: > + kfree(obj); > + return ret; > +} > + > +static unsigned long skl_int3472_clk_recalc_rate(struct clk_hw *hw, > + unsigned long parent_rate) > +{ > + struct int3472_gpio_clock *clk = to_int3472_clk(hw); > + struct int3472_discrete_device *int3472 = to_int3472_device(clk); > + > + return int3472->clock.frequency; Maybe just struct int3472_gpio_clock *clk = to_int3472_clk(hw); return clk->frequency; > +} > + > +static const struct clk_ops skl_int3472_clock_ops = { > + .prepare = skl_int3472_clk_prepare, > + .unprepare = skl_int3472_clk_unprepare, > + .enable = skl_int3472_clk_enable, > + .disable = skl_int3472_clk_disable, > + .recalc_rate = skl_int3472_clk_recalc_rate, > +}; > + > +static struct int3472_sensor_config * > +skl_int3472_get_sensor_module_config(struct int3472_discrete_device *int3472) > +{ > + struct int3472_sensor_config *ret; > + union acpi_object *obj; > + unsigned int i; > + > + obj = acpi_evaluate_dsm_typed(int3472->sensor->handle, > + &cio2_sensor_module_guid, 0x00, > + 0x01, NULL, ACPI_TYPE_STRING); > + > + if (!obj) { > + dev_err(int3472->dev, > + "Failed to get sensor module string from _DSM\n"); > + return ERR_PTR(-ENODEV); > + } > + > + if (obj->string.type != ACPI_TYPE_STRING) { > + dev_err(int3472->dev, > + "Sensor _DSM returned a non-string value\n"); > + ret = ERR_PTR(-EINVAL); > + goto out_free_obj; > + } > + > + ret = NULL; > + for (i = 0; i < ARRAY_SIZE(int3472_sensor_configs); i++) { > + if (!strcmp(int3472_sensor_configs[i].sensor_module_name, > + obj->string.pointer)) { > + ret = &int3472_sensor_configs[i]; > + break; > + } > + } > + > +out_free_obj: > + ACPI_FREE(obj); > + return ret; > +} > + > +static int skl_int3472_map_gpio_to_sensor(struct int3472_discrete_device *int3472, > + struct acpi_resource *ares, > + char *func, u32 polarity) > +{ > + char *path = ares->data.gpio.resource_source.string_ptr; > + const struct int3472_sensor_config *sensor_config; > + struct gpiod_lookup *table_entry; > + struct acpi_device *adev; > + acpi_handle handle; > + acpi_status status; > + int ret; > + > + if (int3472->n_sensor_gpios >= INT3472_MAX_SENSOR_GPIOS) { > + dev_warn(int3472->dev, "Too many GPIOs mapped\n"); > + return -EINVAL; > + } > + > + sensor_config = int3472->sensor_config; > + if (!IS_ERR_OR_NULL(sensor_config) && sensor_config->function_maps) { > + const struct int3472_gpio_function_remap *remap = > + sensor_config->function_maps; > + > + for (; remap->documented; ++remap) > + if (!strcmp(func, remap->documented)) { > + func = remap->actual; > + break; > + } No strictly required, but I'd use curly braces for the for loop. Up to you. > + } > + > + /* Functions mapped to NULL should not be mapped to the sensor */ > + if (!func) > + return 0; > + > + status = acpi_get_handle(NULL, path, &handle); > + if (ACPI_FAILURE(status)) > + return -EINVAL; > + > + ret = acpi_bus_get_device(handle, &adev); > + if (ret) > + return -ENODEV; > + > + table_entry = &int3472->gpios.table[int3472->n_sensor_gpios]; > + table_entry->key = acpi_dev_name(adev); > + table_entry->chip_hwnum = ares->data.gpio.pin_table[0]; > + table_entry->con_id = func; > + table_entry->idx = 0; > + table_entry->flags = polarity; > + > + int3472->n_sensor_gpios++; > + > + return 0; > +} > + > +static int skl_int3472_map_gpio_to_clk(struct int3472_discrete_device *int3472, > + struct acpi_resource *ares, u8 type) > +{ > + char *path = ares->data.gpio.resource_source.string_ptr; > + struct gpio_desc *gpio; > + > + switch (type) { > + case INT3472_GPIO_TYPE_CLK_ENABLE: > + gpio = acpi_get_gpiod(path, ares->data.gpio.pin_table[0], > + "int3472,clk-enable"); > + if (IS_ERR(gpio)) > + return (PTR_ERR(gpio)); > + > + int3472->clock.ena_gpio = gpio; > + break; > + case INT3472_GPIO_TYPE_PRIVACY_LED: > + gpio = acpi_get_gpiod(path, ares->data.gpio.pin_table[0], > + "int3472,privacy-led"); > + if (IS_ERR(gpio)) > + return (PTR_ERR(gpio)); > + > + int3472->clock.led_gpio = gpio; > + break; > + default: > + dev_err(int3472->dev, "Invalid GPIO type 0x%hx for clock\n", > + type); type being a u8, it should be %hhx, but I'd just use 0x%02x. > + break; > + } > + > + return 0; > +} > + > +static int skl_int3472_register_clock(struct int3472_discrete_device *int3472) > +{ > + struct clk_init_data init = { > + .ops = &skl_int3472_clock_ops, > + .flags = CLK_GET_RATE_NOCACHE, > + }; > + int ret = 0; > + > + init.name = kasprintf(GFP_KERNEL, "%s-clk", > + acpi_dev_name(int3472->adev)); > + if (!init.name) > + return -ENOMEM; > + > + int3472->clock.frequency = skl_int3472_get_clk_frequency(int3472); > + > + int3472->clock.clk_hw.init = &init; > + int3472->clock.clk = clk_register(&int3472->adev->dev, > + &int3472->clock.clk_hw); > + if (IS_ERR(int3472->clock.clk)) { > + ret = PTR_ERR(int3472->clock.clk); > + goto out_free_init_name; > + } > + > + int3472->clock.cl = clkdev_create(int3472->clock.clk, NULL, > + int3472->sensor_name); > + if (!int3472->clock.cl) { > + ret = -ENOMEM; > + goto err_unregister_clk; > + } > + > + goto out_free_init_name; > + > +err_unregister_clk: > + clk_unregister(int3472->clock.clk); > +out_free_init_name: > + kfree(init.name); > + > + return ret; > +} > + > +static int skl_int3472_register_regulator(struct int3472_discrete_device *int3472, > + struct acpi_resource *ares) > +{ > + char *path = ares->data.gpio.resource_source.string_ptr; > + struct int3472_sensor_config *sensor_config; > + struct regulator_init_data init_data = { }; > + struct regulator_config cfg = { }; > + int ret; > + > + sensor_config = int3472->sensor_config; > + if (IS_ERR_OR_NULL(sensor_config)) { > + dev_err(int3472->dev, "No sensor module config\n"); > + return PTR_ERR(sensor_config); > + } > + > + if (!sensor_config->supply_map.supply) { > + dev_err(int3472->dev, "No supply name defined\n"); > + return -ENODEV; > + } > + > + init_data.constraints.valid_ops_mask = REGULATOR_CHANGE_STATUS; > + init_data.num_consumer_supplies = 1; > + sensor_config->supply_map.dev_name = int3472->sensor_name; > + init_data.consumer_supplies = &sensor_config->supply_map; > + > + snprintf(int3472->regulator.regulator_name, > + sizeof(int3472->regulator.regulator_name), "%s-regulator", > + acpi_dev_name(int3472->adev)); > + snprintf(int3472->regulator.supply_name, > + GPIO_REGULATOR_SUPPLY_NAME_LENGTH, "supply-0"); > + > + int3472->regulator.rdesc = INT3472_REGULATOR( > + int3472->regulator.regulator_name, > + int3472->regulator.supply_name, > + &int3472_gpio_regulator_ops); > + > + int3472->regulator.gpio = acpi_get_gpiod(path, > + ares->data.gpio.pin_table[0], > + "int3472,regulator"); > + if (IS_ERR(int3472->regulator.gpio)) { > + dev_err(int3472->dev, "Failed to get regulator GPIO lines\n"); s/lines/line/ (sorry, it was a typo in my review of v2) > + return PTR_ERR(int3472->regulator.gpio); > + } > + > + cfg.dev = &int3472->adev->dev; > + cfg.init_data = &init_data; > + cfg.ena_gpiod = int3472->regulator.gpio; > + > + int3472->regulator.rdev = regulator_register(&int3472->regulator.rdesc, > + &cfg); > + if (IS_ERR(int3472->regulator.rdev)) { > + ret = PTR_ERR(int3472->regulator.rdev); > + goto err_free_gpio; > + } > + > + return 0; > + > +err_free_gpio: > + gpiod_put(int3472->regulator.gpio); > + > + return ret; > +} > + > +/** > + * skl_int3472_handle_gpio_resources: Map PMIC resources to consuming sensor > + * @ares: A pointer to a &struct acpi_resource > + * @data: A pointer to a &struct int3472_discrete_device > + * > + * This function handles GPIO resources that are against an INT3472 > + * ACPI device, by checking the value of the corresponding _DSM entry. > + * This will return a 32bit int, where the lowest byte represents the > + * function of the GPIO pin: > + * > + * 0x00 Reset > + * 0x01 Power down > + * 0x0b Power enable > + * 0x0c Clock enable > + * 0x0d Privacy LED > + * > + * There are some known platform specific quirks where that does not quite > + * hold up; for example where a pin with type 0x01 (Power down) is mapped to > + * a sensor pin that performs a reset function or entries in _CRS and _DSM that > + * do not actually correspond to a physical connection. These will be handled > + * by the mapping sub-functions. > + * > + * GPIOs will either be mapped directly to the sensor device or else used > + * to create clocks and regulators via the usual frameworks. > + * > + * Return: > + * * 0 - When all resources found are handled properly. > + * * -EINVAL - If the resource is not a GPIO IO resource > + * * -ENODEV - If the resource has no corresponding _DSM entry > + * * -Other - Errors propagated from one of the sub-functions. > + */ > +static int skl_int3472_handle_gpio_resources(struct acpi_resource *ares, > + void *data) > +{ > + struct int3472_discrete_device *int3472 = data; > + u16 pin = ares->data.gpio.pin_table[0]; > + union acpi_object *obj; > + char *err_msg; > + int ret = 0; > + u8 type; > + > + if (ares->type != ACPI_RESOURCE_TYPE_GPIO || > + ares->data.gpio.connection_type != ACPI_RESOURCE_GPIO_TYPE_IO) > + return 1; /* Deliberately positive so parsing continues */ > + > + /* > + * n_gpios + 2 because the index of this _DSM function is 1-based and > + * the first function is just a count. > + */ > + obj = acpi_evaluate_dsm_typed(int3472->adev->handle, > + &int3472_gpio_guid, 0x00, > + int3472->n_gpios + 2, > + NULL, ACPI_TYPE_INTEGER); > + > + if (!obj) { > + dev_warn(int3472->dev, "No _DSM entry for GPIO pin %u\n", pin); > + return 1; > + } > + > + type = obj->integer.value & 0xff; > + > + switch (type) { > + case INT3472_GPIO_TYPE_RESET: > + ret = skl_int3472_map_gpio_to_sensor(int3472, ares, "reset", > + GPIO_ACTIVE_LOW); > + if (ret) > + err_msg = "Failed to map reset pin to sensor\n"; > + > + break; > + case INT3472_GPIO_TYPE_POWERDOWN: > + ret = skl_int3472_map_gpio_to_sensor(int3472, ares, > + "powerdown", > + GPIO_ACTIVE_LOW); > + if (ret) > + err_msg = "Failed to map powerdown pin to sensor\n"; > + > + break; > + case INT3472_GPIO_TYPE_CLK_ENABLE: > + case INT3472_GPIO_TYPE_PRIVACY_LED: > + ret = skl_int3472_map_gpio_to_clk(int3472, ares, type); > + if (ret) > + err_msg = "Failed to map GPIO to clock\n"; > + > + break; > + case INT3472_GPIO_TYPE_POWER_ENABLE: > + ret = skl_int3472_register_regulator(int3472, ares); > + if (ret) > + err_msg = "Failed to map regulator to sensor\n"; > + > + break; > + default: > + dev_warn(int3472->dev, > + "GPIO type 0x%02x unknown; the sensor may not work\n", > + type); > + ret = 1; > + break; > + } > + > + if (ret < 0 && ret != -EPROBE_DEFER) > + dev_err(int3472->dev, err_msg); > + > + int3472->n_gpios++; > + ACPI_FREE(obj); > + > + return ret; > +} > + > +static int skl_int3472_parse_crs(struct int3472_discrete_device *int3472) > +{ > + struct list_head resource_list; > + int ret; > + > + INIT_LIST_HEAD(&resource_list); > + > + int3472->sensor_config = skl_int3472_get_sensor_module_config(int3472); I have forgotten some of the context I'm afraid :-/ Are there valid use cases for not checking for an error here, or should we do so and drop the error checks in other functions above ? > + > + ret = acpi_dev_get_resources(int3472->adev, &resource_list, > + skl_int3472_handle_gpio_resources, > + int3472); > + if (ret) > + goto out_free_res_list; > + > + if (int3472->clock.ena_gpio) { > + ret = skl_int3472_register_clock(int3472); > + if (ret) > + goto out_free_res_list; > + } else { > + if (int3472->clock.led_gpio) > + dev_warn(int3472->dev, > + "No clk GPIO. The privacy LED won't work\n"); > + } > + > + int3472->gpios.dev_id = int3472->sensor_name; > + gpiod_add_lookup_table(&int3472->gpios); > + > +out_free_res_list: > + acpi_dev_free_resource_list(&resource_list); > + > + return ret; > +} > + > +int skl_int3472_discrete_probe(struct platform_device *pdev) > +{ > + struct acpi_device *adev = ACPI_COMPANION(&pdev->dev); > + struct int3472_discrete_device *int3472; > + struct int3472_cldb cldb; > + int ret; > + > + ret = skl_int3472_fill_cldb(adev, &cldb); > + if (ret) { > + dev_err(&pdev->dev, "Couldn't fill CLDB structure\n"); > + return ret; > + } > + > + if (cldb.control_logic_type != 1) { > + dev_err(&pdev->dev, "Unsupported control logic type %u\n", > + cldb.control_logic_type); > + return -EINVAL; > + } > + > + /* Max num GPIOs we've seen plus a terminator */ > + int3472 = kzalloc(struct_size(int3472, gpios.table, > + INT3472_MAX_SENSOR_GPIOS + 1), GFP_KERNEL); > + if (!int3472) > + return -ENOMEM; > + > + int3472->adev = adev; > + int3472->dev = &pdev->dev; > + platform_set_drvdata(pdev, int3472); > + > + int3472->sensor = acpi_dev_get_dependent_dev(adev); > + if (IS_ERR_OR_NULL(int3472->sensor)) { > + dev_err(&pdev->dev, > + "INT3472 seems to have no dependents.\n"); > + ret = -ENODEV; > + goto err_free_int3472; > + } > + get_device(&int3472->sensor->dev); I see no corresponding put_device(), am I missing something ? I'm also not sure why this is needed. > + > + int3472->sensor_name = kasprintf(GFP_KERNEL, I2C_DEV_NAME_FORMAT, > + acpi_dev_name(int3472->sensor)); This needs a NULL check. > + > + ret = skl_int3472_parse_crs(int3472); > + if (ret) { > + skl_int3472_discrete_remove(pdev); > + return ret; > + } > + > + return 0; > + > +err_free_int3472: > + kfree(int3472); > + return ret; > +} > + > +int skl_int3472_discrete_remove(struct platform_device *pdev) > +{ > + struct int3472_discrete_device *int3472 = platform_get_drvdata(pdev); > + > + if (int3472->gpios.dev_id) > + gpiod_remove_lookup_table(&int3472->gpios); > + > + if (!IS_ERR(int3472->regulator.rdev)) > + regulator_unregister(int3472->regulator.rdev); > + > + if (!IS_ERR(int3472->clock.clk)) > + clk_unregister(int3472->clock.clk); > + > + if (int3472->clock.cl) > + clkdev_drop(int3472->clock.cl); > + > + gpiod_put(int3472->regulator.gpio); > + gpiod_put(int3472->clock.ena_gpio); > + gpiod_put(int3472->clock.led_gpio); > + > + acpi_dev_put(int3472->sensor); > + > + kfree(int3472->sensor_name); > + kfree(int3472); > + > + return 0; > +} > diff --git a/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c b/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c > new file mode 100644 > index 000000000000..d0d2391e263f > --- /dev/null > +++ b/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c > @@ -0,0 +1,113 @@ > +// SPDX-License-Identifier: GPL-2.0 > +/* Author: Dan Scally <djrscally@gmail.com> */ > + > +#include <linux/i2c.h> > +#include <linux/mfd/core.h> > +#include <linux/mfd/tps68470.h> > +#include <linux/platform_device.h> > +#include <linux/regmap.h> > + > +#include "intel_skl_int3472_common.h" > + > +static const struct mfd_cell tps68470_c[] = { > + { .name = "tps68470-gpio" }, > + { .name = "tps68470_pmic_opregion" }, > +}; > + > +static const struct mfd_cell tps68470_w[] = { Maybe more explicit names than _c and _w could be nice ? > + { .name = "tps68470-gpio" }, > + { .name = "tps68470-clk" }, > + { .name = "tps68470-regulator"}, Missing space before }. > +}; > + > +static const struct regmap_config tps68470_regmap_config = { > + .reg_bits = 8, > + .val_bits = 8, > + .max_register = TPS68470_REG_MAX, > +}; > + > +static int tps68470_chip_init(struct device *dev, struct regmap *regmap) > +{ > + unsigned int version; > + int ret; > + > + /* Force software reset */ > + ret = regmap_write(regmap, TPS68470_REG_RESET, TPS68470_REG_RESET_MASK); > + if (ret) > + return ret; > + > + ret = regmap_read(regmap, TPS68470_REG_REVID, &version); > + if (ret) { > + dev_err(dev, "Failed to read revision register: %d\n", ret); > + return ret; > + } > + > + dev_info(dev, "TPS68470 REVID: 0x%02x\n", version); > + > + return 0; > +} > + > +int skl_int3472_tps68470_probe(struct i2c_client *client) > +{ > + struct acpi_device *adev = ACPI_COMPANION(&client->dev); > + struct int3472_cldb cldb = { 0 }; > + bool cldb_present = true; > + struct regmap *regmap; > + int ret; > + > + regmap = devm_regmap_init_i2c(client, &tps68470_regmap_config); > + if (IS_ERR(regmap)) { > + dev_err(&client->dev, "Failed to create regmap: %ld\n", > + PTR_ERR(regmap)); > + return PTR_ERR(regmap); > + } > + > + i2c_set_clientdata(client, regmap); > + > + ret = tps68470_chip_init(&client->dev, regmap); > + if (ret < 0) { > + dev_err(&client->dev, "TPS68470 init error %d\n", ret); > + return ret; > + } > + > + /* > + * Check CLDB buffer against the PMIC's adev. If present, then we check > + * the value of control_logic_type field and follow one of the > + * following scenarios: > + * > + * 1. No CLDB - likely ACPI tables designed for ChromeOS. We > + * create platform devices for the GPIOs and OpRegion drivers. > + * > + * 2. CLDB, with control_logic_type = 2 - probably ACPI tables > + * made for Windows 2-in-1 platforms. Register pdevs for GPIO, > + * Clock and Regulator drivers to bind to. > + * > + * 3. Any other value in control_logic_type, we should never have > + * gotten to this point; fail probe and return. > + */ > + ret = skl_int3472_fill_cldb(adev, &cldb); > + if (!ret && cldb.control_logic_type != 2) { > + dev_err(&client->dev, "Unsupported control logic type %u\n", > + cldb.control_logic_type); > + return -EINVAL; > + } > + > + if (ret) > + cldb_present = false; > + > + if (cldb_present) > + ret = devm_mfd_add_devices(&client->dev, PLATFORM_DEVID_NONE, > + tps68470_w, ARRAY_SIZE(tps68470_w), > + NULL, 0, NULL); > + else > + ret = devm_mfd_add_devices(&client->dev, PLATFORM_DEVID_NONE, > + tps68470_c, ARRAY_SIZE(tps68470_c), > + NULL, 0, NULL); > + > + if (ret) { > + dev_err(&client->dev, "Failed to add MFD devices\n"); > + return ret; > + } > + > + return 0; > +}
Hi Laurent On 23/02/2021 20:04, Laurent Pinchart wrote: > + > +/* > + * Here follows platform specific mapping information that we can pass to > + * the functions mapping resources to the sensors. Where the sensors have > + * a power enable pin defined in DSDT we need to provide a supply name so > + * the sensor drivers can find the regulator. The device name will be derived > + * from the sensor's ACPI device within the code. Optionally, we can provide a > + * NULL terminated array of function name mappings to deal with any platform > + * specific deviations from the documented behaviour of GPIOs. > + * > + * Map a GPIO function name to NULL to prevent the driver from mapping that > + * GPIO at all. > + */ > + > +static const struct int3472_gpio_function_remap ov2680_gpio_function_remaps[] = { > + { "reset", NULL }, > + { "powerdown", "reset" }, > + { } > +}; > + > +static struct int3472_sensor_config int3472_sensor_configs[] = { > This should be static const (and there will be some fallout due to that, > as skl_int3472_register_regulator() modifies the supply_map, so I think > you'll have a copy of supply_map in int3472_discrete_device). Ack to all of the constness; you mentioned that last time too - not sure how I missed doing those! I think I can just having a local struct regulator_consumer_supply in skl_int3472_register_regulator and fill it from int3472->sensor_config.supply_map >> +static unsigned int skl_int3472_get_clk_frequency(struct int3472_discrete_device *int3472) >> +{ >> + union acpi_object *obj; >> + unsigned int ret = 0; >> + >> + obj = skl_int3472_get_acpi_buffer(int3472->sensor, "SSDB"); >> + if (IS_ERR(obj)) >> + return 0; /* report rate as 0 on error */ >> + >> + if (obj->buffer.length < CIO2_SENSOR_SSDB_MCLKSPEED_OFFSET + sizeof(u32)) { > Should we define an ssdb structure instead of peeking into the buffer > with an offset ? I thought about that, but in the end decided it didn't seem worth defining the whole SSDB structure just to use one field. Particularly since we use it in cio2-bridge already, so if we're going to do that it really ought to just live in a header that's included in both - and that seemed even less worthwhile. I don't have a strong feeling though, so if you think it's better to define the struct I'm happy to. >> +static unsigned long skl_int3472_clk_recalc_rate(struct clk_hw *hw, >> + unsigned long parent_rate) >> +{ >> + struct int3472_gpio_clock *clk = to_int3472_clk(hw); >> + struct int3472_discrete_device *int3472 = to_int3472_device(clk); >> + >> + return int3472->clock.frequency; > Maybe just > > struct int3472_gpio_clock *clk = to_int3472_clk(hw); > > return clk->frequency; Oops, of course. >> +static int skl_int3472_register_regulator(struct int3472_discrete_device *int3472, >> + struct acpi_resource *ares) >> +{ >> + char *path = ares->data.gpio.resource_source.string_ptr; >> + struct int3472_sensor_config *sensor_config; >> + struct regulator_init_data init_data = { }; >> + struct regulator_config cfg = { }; >> + int ret; >> + >> + sensor_config = int3472->sensor_config; >> + if (IS_ERR_OR_NULL(sensor_config)) { >> + dev_err(int3472->dev, "No sensor module config\n"); >> + return PTR_ERR(sensor_config); >> + } >> + >> + if (!sensor_config->supply_map.supply) { >> + dev_err(int3472->dev, "No supply name defined\n"); >> + return -ENODEV; >> + } >> + >> + init_data.constraints.valid_ops_mask = REGULATOR_CHANGE_STATUS; >> + init_data.num_consumer_supplies = 1; >> + sensor_config->supply_map.dev_name = int3472->sensor_name; >> + init_data.consumer_supplies = &sensor_config->supply_map; >> + >> + snprintf(int3472->regulator.regulator_name, >> + sizeof(int3472->regulator.regulator_name), "%s-regulator", >> + acpi_dev_name(int3472->adev)); >> + snprintf(int3472->regulator.supply_name, >> + GPIO_REGULATOR_SUPPLY_NAME_LENGTH, "supply-0"); >> + >> + int3472->regulator.rdesc = INT3472_REGULATOR( >> + int3472->regulator.regulator_name, >> + int3472->regulator.supply_name, >> + &int3472_gpio_regulator_ops); >> + >> + int3472->regulator.gpio = acpi_get_gpiod(path, >> + ares->data.gpio.pin_table[0], >> + "int3472,regulator"); >> + if (IS_ERR(int3472->regulator.gpio)) { >> + dev_err(int3472->dev, "Failed to get regulator GPIO lines\n"); > s/lines/line/ (sorry, it was a typo in my review of v2) No problem! >> +static int skl_int3472_parse_crs(struct int3472_discrete_device *int3472) >> +{ >> + struct list_head resource_list; >> + int ret; >> + >> + INIT_LIST_HEAD(&resource_list); >> + >> + int3472->sensor_config = skl_int3472_get_sensor_module_config(int3472); > I have forgotten some of the context I'm afraid :-/ Are there valid use > cases for not checking for an error here, or should we do so and drop > the error checks in other functions above ? Not all platforms need a sensor_config; only those which have either a regulator pin or need a GPIO function to be remapped; the rest will do without it. So, we need to not check for an error here because the absence of a sensor_config isn't necessarily an error, we won't know till later. > +int skl_int3472_discrete_probe(struct platform_device *pdev) > +{ > + struct acpi_device *adev = ACPI_COMPANION(&pdev->dev); > + struct int3472_discrete_device *int3472; > + struct int3472_cldb cldb; > + int ret; > + > + ret = skl_int3472_fill_cldb(adev, &cldb); > + if (ret) { > + dev_err(&pdev->dev, "Couldn't fill CLDB structure\n"); > + return ret; > + } > + > + if (cldb.control_logic_type != 1) { > + dev_err(&pdev->dev, "Unsupported control logic type %u\n", > + cldb.control_logic_type); > + return -EINVAL; > + } > + > + /* Max num GPIOs we've seen plus a terminator */ > + int3472 = kzalloc(struct_size(int3472, gpios.table, > + INT3472_MAX_SENSOR_GPIOS + 1), GFP_KERNEL); > + if (!int3472) > + return -ENOMEM; > + > + int3472->adev = adev; > + int3472->dev = &pdev->dev; > + platform_set_drvdata(pdev, int3472); > + > + int3472->sensor = acpi_dev_get_dependent_dev(adev); > + if (IS_ERR_OR_NULL(int3472->sensor)) { > + dev_err(&pdev->dev, > + "INT3472 seems to have no dependents.\n"); > + ret = -ENODEV; > + goto err_free_int3472; > + } > + get_device(&int3472->sensor->dev); > I see no corresponding put_device(), am I missing something ? I'm also > not sure why this is needed. > The put is acpi_dev_put() in skl_int3472_discrete_remove(); there seems to be no acpi_dev_get() for some reason. We use the sensor acpi_device to get the clock frequency, and to fetch the sensor module string, so I thought it ought to hold a reference on those grounds. >> diff --git a/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c b/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c >> new file mode 100644 >> index 000000000000..d0d2391e263f >> --- /dev/null >> +++ b/drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c >> @@ -0,0 +1,113 @@ >> +// SPDX-License-Identifier: GPL-2.0 >> +/* Author: Dan Scally <djrscally@gmail.com> */ >> + >> +#include <linux/i2c.h> >> +#include <linux/mfd/core.h> >> +#include <linux/mfd/tps68470.h> >> +#include <linux/platform_device.h> >> +#include <linux/regmap.h> >> + >> +#include "intel_skl_int3472_common.h" >> + >> +static const struct mfd_cell tps68470_c[] = { >> + { .name = "tps68470-gpio" }, >> + { .name = "tps68470_pmic_opregion" }, >> +}; >> + >> +static const struct mfd_cell tps68470_w[] = { > Maybe more explicit names than _c and _w could be nice ? _chrome and _windows was in my mind - sound ok?
On Wed, Feb 24, 2021 at 12:16 PM Laurent Pinchart <laurent.pinchart@ideasonboard.com> wrote: > On Tue, Feb 23, 2021 at 10:36:18PM +0000, Daniel Scally wrote: > > On 23/02/2021 20:04, Laurent Pinchart wrote: ... > > >> + get_device(&int3472->sensor->dev); > > > > > > I see no corresponding put_device(), am I missing something ? I'm also > > > not sure why this is needed. > > > > The put is acpi_dev_put() in skl_int3472_discrete_remove(); there seems > > to be no acpi_dev_get() for some reason. We use the sensor acpi_device > > to get the clock frequency, and to fetch the sensor module string, so I > > thought it ought to hold a reference on those grounds. > > Shouldn't acpi_dev_get_dependent_dev() increase the reference count > then, instead of doing it manually here ? That's what I expected as well. We have plenty of acpi_dev_get_*() and they do increase the reference counter one way or the other.
Hi Andy, Laurent On 24/02/2021 10:18, Andy Shevchenko wrote: > On Wed, Feb 24, 2021 at 12:16 PM Laurent Pinchart > <laurent.pinchart@ideasonboard.com> wrote: >> On Tue, Feb 23, 2021 at 10:36:18PM +0000, Daniel Scally wrote: >>> On 23/02/2021 20:04, Laurent Pinchart wrote: > ... > >>>>> + get_device(&int3472->sensor->dev); >>>> I see no corresponding put_device(), am I missing something ? I'm also >>>> not sure why this is needed. >>> The put is acpi_dev_put() in skl_int3472_discrete_remove(); there seems >>> to be no acpi_dev_get() for some reason. We use the sensor acpi_device >>> to get the clock frequency, and to fetch the sensor module string, so I >>> thought it ought to hold a reference on those grounds. >> Shouldn't acpi_dev_get_dependent_dev() increase the reference count >> then, instead of doing it manually here ? > That's what I expected as well. > We have plenty of acpi_dev_get_*() and they do increase the reference > counter one way or the other. > Okedokey, I'll move the get() to that function and drop it from here.
Hi, On 2/22/21 2:07 PM, Daniel Scally wrote: > v1 for this series was originally 14-18 of this series: > https://lore.kernel.org/linux-media/20201130133129.1024662-1-djrscally@gmail.com/T/#m91934e12e3d033da2e768e952ea3b4a125ee3e67 > > v2 was here: > https://lore.kernel.org/platform-driver-x86/20210118003428.568892-1-djrscally@gmail.com/ > > Series level changelog: > > - Dropped the patch moving acpi_lpss_dep() to utils since it's not used > in acpi_dev_get_dependent_dev() anymore. > - Replaced it with a patch extending acpi_walk_dep_device_list() to be > able to apply a given callback against each device in acpi_dep_list > - Dropped the patch creating acpi_i2c_dev_name() and simply open coded > that functionality. > > This has been tested on a number of devices, but currently **not** on a device > designed for ChromeOS, which we ideally need to do to ensure no regression > caused by replacing the tps68470 MFD driver. Sakari / Tomasz, is there any way > you could help with that? Unfortunately, I don't have a device to test it on > myself. > > Original cover letter: > > At the moment in the kernel the ACPI _HID INT3472 is taken by the tps68470 > MFD driver, but that driver can only handle some of the cases of that _HID > that we see. There are at least these three possibilities: > > 1. INT3472 devices that provide GPIOs through the usual framework and run > power and clocks through an operation region; this is the situation that > the current module handles and is seen on ChromeOS devices > 2. INT3472 devices that provide GPIOs, plus clocks and regulators that are > meant to be driven through the usual frameworks, usually seen on devices > designed to run Windows > 3. INT3472 devices that don't actually represent a physical tps68470, but > are being used as a convenient way of grouping a bunch of system GPIO > lines that are intended to enable power and clocks for sensors which > are called out as dependent on them. Also seen on devices designed to > run Windows. > > This series introduces a new module which registers: > > 1. An i2c driver that determines which scenario (#1 or #2) applies to the > machine and registers platform devices to be bound to GPIO, OpRegion, > clock and regulator drivers as appropriate. > 2. A platform driver that binds to the dummy INT3472 devices described in > #3 > > The platform driver for the dummy device registers the GPIO lines that > enable the clocks and regulators to the sensors via those frameworks so > that sensor drivers can consume them in the usual fashion. The existing > GPIO and OpRegion tps68470 drivers will work with the i2c driver that's > registered. Clock and regulator drivers are available but have not so far been > tested, so aren't part of this series. > > The existing mfd/tps68470.c driver being thus superseded, it is removed. Thank you for this patch series. Since there have already been a whole bunch of review-comments, I've not taken a detailed look at this yet. I do wonder if you have thought about how this series should be merged? This series is spread over quite a few subsytems and since there are various interdependencies in the patches it is probably best if it gets merged in its entirety through a single tree. I guess that merging though either Rafael's (drivers/acpi) tree or Lee's (drivers/mfd) tree makes the most sense. As drivers/platform/x86 maintainer I'm happy with whatever solution works for the other subsystem maintainers. Regards, Hans > > Thanks > Dan > > Daniel Scally (6): > ACPI: scan: Extend acpi_walk_dep_device_list() > ACPI: scan: Add function to fetch dependent of acpi device > i2c: core: Add a format macro for I2C device names > gpiolib: acpi: Export acpi_get_gpiod() > platform/x86: Add intel_skl_int3472 driver > mfd: tps68470: Remove tps68470 MFD driver > > MAINTAINERS | 5 + > drivers/acpi/ec.c | 2 +- > drivers/acpi/pmic/Kconfig | 2 +- > drivers/acpi/pmic/intel_pmic_chtdc_ti.c | 2 +- > drivers/acpi/scan.c | 92 ++- > drivers/gpio/Kconfig | 2 +- > drivers/gpio/gpiolib-acpi.c | 38 +- > drivers/i2c/i2c-core-acpi.c | 2 +- > drivers/i2c/i2c-core-base.c | 4 +- > drivers/mfd/Kconfig | 18 - > drivers/mfd/Makefile | 1 - > drivers/mfd/tps68470.c | 97 --- > drivers/platform/surface/surface3_power.c | 2 +- > drivers/platform/x86/Kconfig | 2 + > drivers/platform/x86/Makefile | 1 + > drivers/platform/x86/intel-int3472/Kconfig | 31 + > drivers/platform/x86/intel-int3472/Makefile | 4 + > .../intel-int3472/intel_skl_int3472_common.c | 106 ++++ > .../intel-int3472/intel_skl_int3472_common.h | 110 ++++ > .../intel_skl_int3472_discrete.c | 592 ++++++++++++++++++ > .../intel_skl_int3472_tps68470.c | 113 ++++ > include/acpi/acpi_bus.h | 8 + > include/linux/acpi.h | 4 +- > include/linux/gpio/consumer.h | 7 + > include/linux/i2c.h | 3 + > 25 files changed, 1100 insertions(+), 148 deletions(-) > delete mode 100644 drivers/mfd/tps68470.c > create mode 100644 drivers/platform/x86/intel-int3472/Kconfig > create mode 100644 drivers/platform/x86/intel-int3472/Makefile > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c > create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c >
Hi Hans On 04/03/2021 13:37, Hans de Goede wrote: > Hi, > > On 2/22/21 2:07 PM, Daniel Scally wrote: >> v1 for this series was originally 14-18 of this series: >> https://lore.kernel.org/linux-media/20201130133129.1024662-1-djrscally@gmail.com/T/#m91934e12e3d033da2e768e952ea3b4a125ee3e67 >> >> v2 was here: >> https://lore.kernel.org/platform-driver-x86/20210118003428.568892-1-djrscally@gmail.com/ >> >> Series level changelog: >> >> - Dropped the patch moving acpi_lpss_dep() to utils since it's not used >> in acpi_dev_get_dependent_dev() anymore. >> - Replaced it with a patch extending acpi_walk_dep_device_list() to be >> able to apply a given callback against each device in acpi_dep_list >> - Dropped the patch creating acpi_i2c_dev_name() and simply open coded >> that functionality. >> >> This has been tested on a number of devices, but currently **not** on a device >> designed for ChromeOS, which we ideally need to do to ensure no regression >> caused by replacing the tps68470 MFD driver. Sakari / Tomasz, is there any way >> you could help with that? Unfortunately, I don't have a device to test it on >> myself. >> >> Original cover letter: >> >> At the moment in the kernel the ACPI _HID INT3472 is taken by the tps68470 >> MFD driver, but that driver can only handle some of the cases of that _HID >> that we see. There are at least these three possibilities: >> >> 1. INT3472 devices that provide GPIOs through the usual framework and run >> power and clocks through an operation region; this is the situation that >> the current module handles and is seen on ChromeOS devices >> 2. INT3472 devices that provide GPIOs, plus clocks and regulators that are >> meant to be driven through the usual frameworks, usually seen on devices >> designed to run Windows >> 3. INT3472 devices that don't actually represent a physical tps68470, but >> are being used as a convenient way of grouping a bunch of system GPIO >> lines that are intended to enable power and clocks for sensors which >> are called out as dependent on them. Also seen on devices designed to >> run Windows. >> >> This series introduces a new module which registers: >> >> 1. An i2c driver that determines which scenario (#1 or #2) applies to the >> machine and registers platform devices to be bound to GPIO, OpRegion, >> clock and regulator drivers as appropriate. >> 2. A platform driver that binds to the dummy INT3472 devices described in >> #3 >> >> The platform driver for the dummy device registers the GPIO lines that >> enable the clocks and regulators to the sensors via those frameworks so >> that sensor drivers can consume them in the usual fashion. The existing >> GPIO and OpRegion tps68470 drivers will work with the i2c driver that's >> registered. Clock and regulator drivers are available but have not so far been >> tested, so aren't part of this series. >> >> The existing mfd/tps68470.c driver being thus superseded, it is removed. > Thank you for this patch series. Since there have already been a whole > bunch of review-comments, I've not taken a detailed look at this yet. No problem, I'm hoping to do a v3 over the weekend anyway. > I do wonder if you have thought about how this series should be merged? > This series is spread over quite a few subsytems and since there are > various interdependencies in the patches it is probably best if it gets > merged in its entirety through a single tree. > > I guess that merging though either Rafael's (drivers/acpi) tree or > Lee's (drivers/mfd) tree makes the most sense. > > As drivers/platform/x86 maintainer I'm happy with whatever solution > works for the other subsystem maintainers. I also think it's a good idea to go through a single tree, and my plan was to raise that probably after the next review round or so, but I hadn't gotten as far as thinking about whos tree it should be or anything yet. To be honest I'm not sure what factors dictate which choice is best in that regard; handling complex git merges is a bit outside my experience. > > Regards, > > Hans > > > > >> Thanks >> Dan >> >> Daniel Scally (6): >> ACPI: scan: Extend acpi_walk_dep_device_list() >> ACPI: scan: Add function to fetch dependent of acpi device >> i2c: core: Add a format macro for I2C device names >> gpiolib: acpi: Export acpi_get_gpiod() >> platform/x86: Add intel_skl_int3472 driver >> mfd: tps68470: Remove tps68470 MFD driver >> >> MAINTAINERS | 5 + >> drivers/acpi/ec.c | 2 +- >> drivers/acpi/pmic/Kconfig | 2 +- >> drivers/acpi/pmic/intel_pmic_chtdc_ti.c | 2 +- >> drivers/acpi/scan.c | 92 ++- >> drivers/gpio/Kconfig | 2 +- >> drivers/gpio/gpiolib-acpi.c | 38 +- >> drivers/i2c/i2c-core-acpi.c | 2 +- >> drivers/i2c/i2c-core-base.c | 4 +- >> drivers/mfd/Kconfig | 18 - >> drivers/mfd/Makefile | 1 - >> drivers/mfd/tps68470.c | 97 --- >> drivers/platform/surface/surface3_power.c | 2 +- >> drivers/platform/x86/Kconfig | 2 + >> drivers/platform/x86/Makefile | 1 + >> drivers/platform/x86/intel-int3472/Kconfig | 31 + >> drivers/platform/x86/intel-int3472/Makefile | 4 + >> .../intel-int3472/intel_skl_int3472_common.c | 106 ++++ >> .../intel-int3472/intel_skl_int3472_common.h | 110 ++++ >> .../intel_skl_int3472_discrete.c | 592 ++++++++++++++++++ >> .../intel_skl_int3472_tps68470.c | 113 ++++ >> include/acpi/acpi_bus.h | 8 + >> include/linux/acpi.h | 4 +- >> include/linux/gpio/consumer.h | 7 + >> include/linux/i2c.h | 3 + >> 25 files changed, 1100 insertions(+), 148 deletions(-) >> delete mode 100644 drivers/mfd/tps68470.c >> create mode 100644 drivers/platform/x86/intel-int3472/Kconfig >> create mode 100644 drivers/platform/x86/intel-int3472/Makefile >> create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.c >> create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_common.h >> create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_discrete.c >> create mode 100644 drivers/platform/x86/intel-int3472/intel_skl_int3472_tps68470.c >>
Hi Andy On 22/02/2021 13:34, Andy Shevchenko wrote: > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: >> The acpi_walk_dep_device_list() is not as generalisable as its name >> implies, serving only to decrement the dependency count for each >> dependent device of the input. Extend the function to instead accept >> a callback which can be applied to all the dependencies in acpi_dep_list. >> Replace all existing calls to the function with calls to a wrapper, passing >> a callback that applies the same dependency reduction. > The code looks okay to me, if it was the initial idea, feel free to add > Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> Thank you! >> + */ >> +void acpi_dev_flag_dependency_met(acpi_handle handle) >> +{ > Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? I can do this, but I avoided it because in most of the uses in the kernel currently there's no struct acpi_device, they're just passing ACPI_HANDLE(dev) instead, so I'd need to get the adev with ACPI_COMPANION() in each place. It didn't seem worth it...but happy to do it if you'd prefer it that way?
On Sun, Mar 7, 2021 at 3:36 PM Daniel Scally <djrscally@gmail.com> wrote: > On 22/02/2021 13:34, Andy Shevchenko wrote: > > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: > >> The acpi_walk_dep_device_list() is not as generalisable as its name > >> implies, serving only to decrement the dependency count for each > >> dependent device of the input. Extend the function to instead accept > >> a callback which can be applied to all the dependencies in acpi_dep_list. > >> Replace all existing calls to the function with calls to a wrapper, passing > >> a callback that applies the same dependency reduction. > > The code looks okay to me, if it was the initial idea, feel free to add > > Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> > > > Thank you! > > > >> + */ > >> +void acpi_dev_flag_dependency_met(acpi_handle handle) > >> +{ > > Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? > > > I can do this, but I avoided it because in most of the uses in the > kernel currently there's no struct acpi_device, they're just passing > ACPI_HANDLE(dev) instead, so I'd need to get the adev with > ACPI_COMPANION() in each place. It didn't seem worth it...but happy to > do it if you'd prefer it that way? I see, let Rafael decide then. I'm not pushing here. -- With Best Regards, Andy Shevchenko
On Sun, Mar 7, 2021 at 9:39 PM Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > > On Sun, Mar 7, 2021 at 3:36 PM Daniel Scally <djrscally@gmail.com> wrote: > > On 22/02/2021 13:34, Andy Shevchenko wrote: > > > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: > > >> The acpi_walk_dep_device_list() is not as generalisable as its name > > >> implies, serving only to decrement the dependency count for each > > >> dependent device of the input. Extend the function to instead accept > > >> a callback which can be applied to all the dependencies in acpi_dep_list. > > >> Replace all existing calls to the function with calls to a wrapper, passing > > >> a callback that applies the same dependency reduction. > > > The code looks okay to me, if it was the initial idea, feel free to add > > > Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> > > > > > > Thank you! > > > > > > >> + */ > > >> +void acpi_dev_flag_dependency_met(acpi_handle handle) > > >> +{ > > > Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? > > > > > > I can do this, but I avoided it because in most of the uses in the > > kernel currently there's no struct acpi_device, they're just passing > > ACPI_HANDLE(dev) instead, so I'd need to get the adev with > > ACPI_COMPANION() in each place. It didn't seem worth it... It may not even be possible sometimes, because that function may be called before creating all of the struct acpi_device objects (like in the case of deferred enumeration). > > but happy to > > do it if you'd prefer it that way? > > I see, let Rafael decide then. I'm not pushing here. Well, it's a matter of correctness.
On Mon, Mar 8, 2021 at 2:57 PM Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > > On Mon, Mar 08, 2021 at 02:36:27PM +0100, Rafael J. Wysocki wrote: > > On Sun, Mar 7, 2021 at 9:39 PM Andy Shevchenko > > <andy.shevchenko@gmail.com> wrote: > > > On Sun, Mar 7, 2021 at 3:36 PM Daniel Scally <djrscally@gmail.com> wrote: > > > > On 22/02/2021 13:34, Andy Shevchenko wrote: > > > > > On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: > > > > >> The acpi_walk_dep_device_list() is not as generalisable as its name > > > > >> implies, serving only to decrement the dependency count for each > > > > >> dependent device of the input. Extend the function to instead accept > > > > >> a callback which can be applied to all the dependencies in acpi_dep_list. > > > > >> Replace all existing calls to the function with calls to a wrapper, passing > > > > >> a callback that applies the same dependency reduction. > > > > > The code looks okay to me, if it was the initial idea, feel free to add > > > > > Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> > > ... > > > > > >> +void acpi_dev_flag_dependency_met(acpi_handle handle) > > > > > > Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? > > > > > > > > I can do this, but I avoided it because in most of the uses in the > > > > kernel currently there's no struct acpi_device, they're just passing > > > > ACPI_HANDLE(dev) instead, so I'd need to get the adev with > > > > ACPI_COMPANION() in each place. It didn't seem worth it... > > > > It may not even be possible sometimes, because that function may be > > called before creating all of the struct acpi_device objects (like in > > the case of deferred enumeration). > > > > > > but happy to > > > > do it if you'd prefer it that way? > > > > > > I see, let Rafael decide then. I'm not pushing here. > > > > Well, it's a matter of correctness. > > Looking at your above comment it is indeed. Thanks for clarification! Well, actually, the struct device for the object passed to this function should be there already, because otherwise it wouldn't make sense to update the list. So my comment above is not really applicable to this particular device and the function could take a struct acpi_device pointer argument. Sorry for the confusion. > But should we have acpi_dev_*() namespace for this function if it takes handle? It takes a device object handle. Anyway, as per the above, it can take a struct acpi_device pointer argument in which case the "acpi_dev_" prefix should be fine. > For time being nothing better than following comes to my mind: > > __acpi_dev_flag_dependency_met() => __acpi_flag_device_dependency_met() > acpi_dev_flag_dependency_met() => acpi_flag_device_dependency_met() The above said, the name is somewhat confusing overall IMV. Something like acpi_dev_clear_dependencies() might be better. So lets make it something like void acpi_dev_clear_dependencies(struct acpi_device *supplier);
On Mon, Feb 22, 2021 at 2:07 PM Daniel Scally <djrscally@gmail.com> wrote: > > The acpi_walk_dep_device_list() is not as generalisable as its name > implies, serving only to decrement the dependency count for each > dependent device of the input. Extend the function to instead accept > a callback which can be applied to all the dependencies in acpi_dep_list. > Replace all existing calls to the function with calls to a wrapper, passing > a callback that applies the same dependency reduction. > > Suggested-by: Rafael J. Wysocki <rafael@kernel.org> > Signed-off-by: Daniel Scally <djrscally@gmail.com> > --- > Changes in v3: > - patch introduced > > drivers/acpi/ec.c | 2 +- > drivers/acpi/pmic/intel_pmic_chtdc_ti.c | 2 +- > drivers/acpi/scan.c | 58 ++++++++++++++++------- > drivers/gpio/gpiolib-acpi.c | 2 +- > drivers/i2c/i2c-core-acpi.c | 2 +- > drivers/platform/surface/surface3_power.c | 2 +- > include/acpi/acpi_bus.h | 7 +++ > include/linux/acpi.h | 4 +- > 8 files changed, 55 insertions(+), 24 deletions(-) > > diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c > index 13565629ce0a..a258db713bd2 100644 > --- a/drivers/acpi/ec.c > +++ b/drivers/acpi/ec.c > @@ -1627,7 +1627,7 @@ static int acpi_ec_add(struct acpi_device *device) > WARN(!ret, "Could not request EC cmd io port 0x%lx", ec->command_addr); > > /* Reprobe devices depending on the EC */ > - acpi_walk_dep_device_list(ec->handle); > + acpi_dev_flag_dependency_met(ec->handle); > > acpi_handle_debug(ec->handle, "enumerated.\n"); > return 0; > diff --git a/drivers/acpi/pmic/intel_pmic_chtdc_ti.c b/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > index a5101b07611a..59cca504325e 100644 > --- a/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > +++ b/drivers/acpi/pmic/intel_pmic_chtdc_ti.c > @@ -117,7 +117,7 @@ static int chtdc_ti_pmic_opregion_probe(struct platform_device *pdev) > return err; > > /* Re-enumerate devices depending on PMIC */ > - acpi_walk_dep_device_list(ACPI_HANDLE(pdev->dev.parent)); > + acpi_dev_flag_dependency_met(ACPI_HANDLE(pdev->dev.parent)); > return 0; > } > > diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c > index 80b668c80073..c9e4190316ef 100644 > --- a/drivers/acpi/scan.c > +++ b/drivers/acpi/scan.c > @@ -49,12 +49,6 @@ static DEFINE_MUTEX(acpi_hp_context_lock); > */ > static u64 spcr_uart_addr; > > -struct acpi_dep_data { > - struct list_head node; > - acpi_handle supplier; > - acpi_handle consumer; > -}; > - > void acpi_scan_lock_acquire(void) > { > mutex_lock(&acpi_scan_lock); > @@ -2099,30 +2093,58 @@ static void acpi_bus_attach(struct acpi_device *device, bool first_pass) > device->handler->hotplug.notify_online(device); > } > > -void acpi_walk_dep_device_list(acpi_handle handle) > +static int __acpi_dev_flag_dependency_met(struct acpi_dep_data *dep, > + void *data) > { > - struct acpi_dep_data *dep, *tmp; > struct acpi_device *adev; > > + acpi_bus_get_device(dep->consumer, &adev); > + if (!adev) > + return 0; > + > + adev->dep_unmet--; > + if (!adev->dep_unmet) > + acpi_bus_attach(adev, true); > + > + list_del(&dep->node); > + kfree(dep); > + return 0; > +} > + > +void acpi_walk_dep_device_list(acpi_handle handle, > + int (*callback)(struct acpi_dep_data *, void *), > + void *data) > +{ > + struct acpi_dep_data *dep, *tmp; > + int ret; > + > mutex_lock(&acpi_dep_list_lock); > list_for_each_entry_safe(dep, tmp, &acpi_dep_list, node) { > if (dep->supplier == handle) { > - acpi_bus_get_device(dep->consumer, &adev); > - if (!adev) > - continue; > - > - adev->dep_unmet--; > - if (!adev->dep_unmet) > - acpi_bus_attach(adev, true); The above code in the mainline has changed recently, so you need to rebase the above and adjust for the change of behavior. > - > - list_del(&dep->node); > - kfree(dep); > + ret = callback(dep, data); > + if (ret) > + break; > } > } > mutex_unlock(&acpi_dep_list_lock); > } > EXPORT_SYMBOL_GPL(acpi_walk_dep_device_list); > > +/** > + * acpi_dev_flag_dependency_met() - Inform consumers of @handle that the device > + * is now active No parens here, please, and make it fit one line. Also the description should be something like "Clear dependencies on the given device." > + * @handle: acpi_handle for the supplier device > + * > + * This function walks through the dependencies list and informs each consumer > + * of @handle that their dependency upon it is now met. Devices with no more > + * unmet dependencies will be attached to the acpi bus. > + */ > +void acpi_dev_flag_dependency_met(acpi_handle handle) > +{ > + acpi_walk_dep_device_list(handle, __acpi_dev_flag_dependency_met, NULL); > +} > +EXPORT_SYMBOL_GPL(acpi_dev_flag_dependency_met); > + > /** > * acpi_bus_scan - Add ACPI device node objects in a given namespace scope. > * @handle: Root of the namespace scope to scan. > diff --git a/drivers/gpio/gpiolib-acpi.c b/drivers/gpio/gpiolib-acpi.c > index e37a57d0a2f0..e4d728fda982 100644 > --- a/drivers/gpio/gpiolib-acpi.c > +++ b/drivers/gpio/gpiolib-acpi.c > @@ -1254,7 +1254,7 @@ void acpi_gpiochip_add(struct gpio_chip *chip) > > acpi_gpiochip_request_regions(acpi_gpio); > acpi_gpiochip_scan_gpios(acpi_gpio); > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > } > > void acpi_gpiochip_remove(struct gpio_chip *chip) > diff --git a/drivers/i2c/i2c-core-acpi.c b/drivers/i2c/i2c-core-acpi.c > index 37c510d9347a..38647cf34bde 100644 > --- a/drivers/i2c/i2c-core-acpi.c > +++ b/drivers/i2c/i2c-core-acpi.c > @@ -283,7 +283,7 @@ void i2c_acpi_register_devices(struct i2c_adapter *adap) > if (!handle) > return; > > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > } > > static const struct acpi_device_id i2c_acpi_force_400khz_device_ids[] = { > diff --git a/drivers/platform/surface/surface3_power.c b/drivers/platform/surface/surface3_power.c > index cc4f9cba6856..ad895285d3e9 100644 > --- a/drivers/platform/surface/surface3_power.c > +++ b/drivers/platform/surface/surface3_power.c > @@ -478,7 +478,7 @@ static int mshw0011_install_space_handler(struct i2c_client *client) > return -ENOMEM; > } > > - acpi_walk_dep_device_list(handle); > + acpi_dev_flag_dependency_met(handle); > return 0; > } > > diff --git a/include/acpi/acpi_bus.h b/include/acpi/acpi_bus.h > index 02a716a0af5d..91172af3a04d 100644 > --- a/include/acpi/acpi_bus.h > +++ b/include/acpi/acpi_bus.h > @@ -278,6 +278,12 @@ struct acpi_device_power { > struct acpi_device_power_state states[ACPI_D_STATE_COUNT]; /* Power states (D0-D3Cold) */ > }; > > +struct acpi_dep_data { > + struct list_head node; > + acpi_handle supplier; > + acpi_handle consumer; > +}; > + > /* Performance Management */ > > struct acpi_device_perf_flags { > @@ -683,6 +689,7 @@ static inline bool acpi_device_can_poweroff(struct acpi_device *adev) > > bool acpi_dev_hid_uid_match(struct acpi_device *adev, const char *hid2, const char *uid2); > > +void acpi_dev_flag_dependency_met(acpi_handle handle); > struct acpi_device * > acpi_dev_get_next_match_dev(struct acpi_device *adev, const char *hid, const char *uid, s64 hrv); > struct acpi_device * > diff --git a/include/linux/acpi.h b/include/linux/acpi.h > index 2630c2e953f7..2d5e6e88e8a0 100644 > --- a/include/linux/acpi.h > +++ b/include/linux/acpi.h > @@ -655,7 +655,9 @@ extern bool acpi_driver_match_device(struct device *dev, > const struct device_driver *drv); > int acpi_device_uevent_modalias(struct device *, struct kobj_uevent_env *); > int acpi_device_modalias(struct device *, char *, int); > -void acpi_walk_dep_device_list(acpi_handle handle); > +void acpi_walk_dep_device_list(acpi_handle handle, > + int (*callback)(struct acpi_dep_data *, void *), > + void *data); > > struct platform_device *acpi_create_platform_device(struct acpi_device *, > struct property_entry *); > -- > 2.25.1 >
Hi Rafael On 08/03/2021 17:46, Rafael J. Wysocki wrote: >> +void acpi_walk_dep_device_list(acpi_handle handle, >> + int (*callback)(struct acpi_dep_data *, void *), >> + void *data) >> +{ >> + struct acpi_dep_data *dep, *tmp; >> + int ret; >> + >> mutex_lock(&acpi_dep_list_lock); >> list_for_each_entry_safe(dep, tmp, &acpi_dep_list, node) { >> if (dep->supplier == handle) { >> - acpi_bus_get_device(dep->consumer, &adev); >> - if (!adev) >> - continue; >> - >> - adev->dep_unmet--; >> - if (!adev->dep_unmet) >> - acpi_bus_attach(adev, true); > The above code in the mainline has changed recently, so you need to > rebase the above and adjust for the change of behavior. Yeah, I'll rebase onto 5.12-rc2 before next submission. > >> - >> - list_del(&dep->node); >> - kfree(dep); >> + ret = callback(dep, data); >> + if (ret) >> + break; >> } >> } >> mutex_unlock(&acpi_dep_list_lock); >> } >> EXPORT_SYMBOL_GPL(acpi_walk_dep_device_list); >> >> +/** >> + * acpi_dev_flag_dependency_met() - Inform consumers of @handle that the device >> + * is now active > No parens here, please, and make it fit one line. > > Also the description should be something like "Clear dependencies on > the given device." OK - no problem
Hi Rafael On 08/03/2021 17:23, Rafael J. Wysocki wrote: > On Mon, Mar 8, 2021 at 4:45 PM Rafael J. Wysocki <rafael@kernel.org> wrote: >> On Mon, Mar 8, 2021 at 2:57 PM Andy Shevchenko >> <andy.shevchenko@gmail.com> wrote: >>> On Mon, Mar 08, 2021 at 02:36:27PM +0100, Rafael J. Wysocki wrote: >>>> On Sun, Mar 7, 2021 at 9:39 PM Andy Shevchenko >>>> <andy.shevchenko@gmail.com> wrote: >>>>> On Sun, Mar 7, 2021 at 3:36 PM Daniel Scally <djrscally@gmail.com> wrote: >>>>>> On 22/02/2021 13:34, Andy Shevchenko wrote: >>>>>>> On Mon, Feb 22, 2021 at 3:12 PM Daniel Scally <djrscally@gmail.com> wrote: >>>>>>>> The acpi_walk_dep_device_list() is not as generalisable as its name >>>>>>>> implies, serving only to decrement the dependency count for each >>>>>>>> dependent device of the input. Extend the function to instead accept >>>>>>>> a callback which can be applied to all the dependencies in acpi_dep_list. >>>>>>>> Replace all existing calls to the function with calls to a wrapper, passing >>>>>>>> a callback that applies the same dependency reduction. >>>>>>> The code looks okay to me, if it was the initial idea, feel free to add >>>>>>> Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> >>> ... >>> >>>>>>>> +void acpi_dev_flag_dependency_met(acpi_handle handle) >>>>>>> Since it's acpi_dev_* namespace, perhaps it should take struct acpi_device here? >>>>>> I can do this, but I avoided it because in most of the uses in the >>>>>> kernel currently there's no struct acpi_device, they're just passing >>>>>> ACPI_HANDLE(dev) instead, so I'd need to get the adev with >>>>>> ACPI_COMPANION() in each place. It didn't seem worth it... >>>> It may not even be possible sometimes, because that function may be >>>> called before creating all of the struct acpi_device objects (like in >>>> the case of deferred enumeration). >>>> >>>>>> but happy to >>>>>> do it if you'd prefer it that way? >>>>> I see, let Rafael decide then. I'm not pushing here. >>>> Well, it's a matter of correctness. >>> Looking at your above comment it is indeed. Thanks for clarification! >> Well, actually, the struct device for the object passed to this >> function should be there already, because otherwise it wouldn't make >> sense to update the list. So my comment above is not really >> applicable to this particular device and the function could take a >> struct acpi_device pointer argument. Sorry for the confusion. >> >>> But should we have acpi_dev_*() namespace for this function if it takes handle? >> It takes a device object handle. >> >> Anyway, as per the above, it can take a struct acpi_device pointer >> argument in which case the "acpi_dev_" prefix should be fine. OK, so the conclusion there is change the argument to a struct acpi_device pointer and update all the uses. >>> For time being nothing better than following comes to my mind: >>> >>> __acpi_dev_flag_dependency_met() => __acpi_flag_device_dependency_met() >>> acpi_dev_flag_dependency_met() => acpi_flag_device_dependency_met() >> The above said, the name is somewhat confusing overall IMV. >> >> Something like acpi_dev_clear_dependencies() might be better. >> >> So lets make it something like >> >> void acpi_dev_clear_dependencies(struct acpi_device *supplier); > To be precise, there are two functions in the patch, > acpi_dev_flag_dependency_met() which invokes > acpi_walk_dep_device_list() and __acpi_dev_flag_dependency_met() > invoked by the latter as a callback. > > Above I was talking about the first one. > > The callback should still take a struct acpi_dep_data pointer argument > and I would call it acpi_scan_clear_dep() or similar. OK, works for me, I'll make those changes - thanks
On Thu, Mar 04, 2021 at 01:49:14PM +0000, Daniel Scally wrote: > On 04/03/2021 13:37, Hans de Goede wrote: > > On 2/22/21 2:07 PM, Daniel Scally wrote: ... > >> The existing mfd/tps68470.c driver being thus superseded, it is removed. > > Thank you for this patch series. Since there have already been a whole > > bunch of review-comments, I've not taken a detailed look at this yet. > > No problem, I'm hoping to do a v3 over the weekend anyway. Do you mean v4? I'm just wondering if you need any help.
Hi Andy On 29/03/2021 16:03, Andy Shevchenko wrote: > On Thu, Mar 04, 2021 at 01:49:14PM +0000, Daniel Scally wrote: >> On 04/03/2021 13:37, Hans de Goede wrote: >>> On 2/22/21 2:07 PM, Daniel Scally wrote: > ... > >>>> The existing mfd/tps68470.c driver being thus superseded, it is removed. >>> Thank you for this patch series. Since there have already been a whole >>> bunch of review-comments, I've not taken a detailed look at this yet. >> No problem, I'm hoping to do a v3 over the weekend anyway. > Do you mean v4? Oops, I do indeed. > I'm just wondering if you need any help. Thanks - I don't think so; I've just not been working on it very much lately. I got sidetracked with a sensor driver [1] that was pretty fun, so I've been focused on that instead. I'm just finishing up a v2 for that, and then I'll come back to this. [1] https://lore.kernel.org/linux-media/20210312103239.279523-2-djrscally@gmail.com/
Hi Andy On 22/02/2021 14:58, Andy Shevchenko wrote >> +#include <linux/clk-provider.h> > > This is definitely not for *.h. (Not all C files needed this) > >> +#include <linux/gpio/machine.h> > > Ditto. > >> +#include <linux/regulator/driver.h> >> +#include <linux/regulator/machine.h> > > Ditto. Bit more delayed than I wanted to be, but I'm just finishing off the v4 of this. For these includes, I'm using the actual structs from them rather than pointers, so removing these would mean moving the definition of struct int3472_discrete_device into one of the source files; you're happy with that?