[PATCH v24 04/12] mfd: psci-mfd: Add PSCI MFD driver for cpuidle-psci-domain cell
Shivendra Pratap
shivendra.pratap at oss.qualcomm.com
Mon Aug 17 22:34:15 PDT 2026
On 8/13/2026 6:08 PM, Lee Jones wrote:
> On Mon, 03 Aug 2026, Shivendra Pratap wrote:
>
>> PSCI has multiple kernel consumers, such as cpuidle-psci-domain.
>> Currently, both the PSCI core driver and cpuidle-psci-domain bind
>> directly to the same PSCI node "arm,psci-1.0". Additional consumers, if
>> introduced, would also need to bind in the same way, leading to several
>> drivers attached to a single device node.
>>
>> Introduce a PSCI MFD driver that binds to "arm,psci-1.0" and registers
>> PSCI child cells. As the first user, register cpuidle-psci-domain as a
>> child cell.
>>
>> Update cpuidle-psci-domain to probe as an MFD child and use the parent
>> PSCI node for power-domain traversal.
>>
>> Suggested-by: Bartosz Golaszewski <bartosz.golaszewski at oss.qualcomm.com>
>> Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski at oss.qualcomm.com>
>> Signed-off-by: Shivendra Pratap <shivendra.pratap at oss.qualcomm.com>
>> ---
>> MAINTAINERS | 1 +
>> drivers/cpuidle/Kconfig.arm | 1 +
>> drivers/cpuidle/cpuidle-psci-domain.c | 9 +-------
>> drivers/mfd/Kconfig | 12 ++++++++++
>> drivers/mfd/Makefile | 2 ++
>> drivers/mfd/psci-mfd.c | 43 +++++++++++++++++++++++++++++++++++
>
> We don't usually allow the use of the term MFD in devices.
>
> What does this do? Is it a mux? psci-mux?
>
> If not, think about what it is and call it that instead.
will rename it.
>
>> 6 files changed, 60 insertions(+), 8 deletions(-)
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index fe67f7bfa44c..0b6e4f89f319 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -21671,6 +21671,7 @@ L: linux-arm-kernel at lists.infradead.org (moderated for non-subscribers)
>> S: Maintained
>> F: Documentation/devicetree/bindings/arm/psci.yaml
>> F: drivers/firmware/psci/
>> +F: drivers/mfd/psci-mfd.c
>> F: include/linux/psci.h
>> F: include/uapi/linux/psci.h
>>
>> diff --git a/drivers/cpuidle/Kconfig.arm b/drivers/cpuidle/Kconfig.arm
>> index b88b01aa5829..c5ecca0d7397 100644
>> --- a/drivers/cpuidle/Kconfig.arm
>> +++ b/drivers/cpuidle/Kconfig.arm
>> @@ -36,6 +36,7 @@ config ARM_PSCI_CPUIDLE_DOMAIN
>> bool "PSCI CPU idle Domain"
>> depends on ARM_PSCI_CPUIDLE
>> depends on PM_GENERIC_DOMAINS_OF
>> + depends on MFD_PSCI
>> select DT_IDLE_GENPD
>> default y
>> help
>> diff --git a/drivers/cpuidle/cpuidle-psci-domain.c b/drivers/cpuidle/cpuidle-psci-domain.c
>> index b9e4ad7d43a3..a2d01810f295 100644
>> --- a/drivers/cpuidle/cpuidle-psci-domain.c
>> +++ b/drivers/cpuidle/cpuidle-psci-domain.c
>> @@ -17,7 +17,6 @@
>> #include <linux/pm_runtime.h>
>> #include <linux/psci.h>
>> #include <linux/slab.h>
>> -#include <linux/string.h>
>>
>> #include "cpuidle-psci.h"
>> #include "dt_idle_genpd.h"
>> @@ -122,14 +121,9 @@ static void psci_pd_remove(void)
>> }
>> }
>>
>> -static const struct of_device_id psci_of_match[] = {
>> - { .compatible = "arm,psci-1.0" },
>> - {}
>> -};
>> -
>> static int psci_cpuidle_domain_probe(struct platform_device *pdev)
>> {
>> - struct device_node *np = pdev->dev.of_node;
>> + struct device_node *np = pdev->dev.parent->of_node;
>
> dev_of_node(pdev->dev.parent);
Ack.
>
>> bool use_osi = psci_has_osi_support();
>> int ret = 0, pd_count = 0;
>>
>> @@ -181,7 +175,6 @@ static struct platform_driver psci_cpuidle_domain_driver = {
>> .probe = psci_cpuidle_domain_probe,
>> .driver = {
>> .name = "psci-cpuidle-domain",
>> - .of_match_table = psci_of_match,
>> },
>> };
>>
>> diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
>> index e4fd4572472f..d18655d76172 100644
>> --- a/drivers/mfd/Kconfig
>> +++ b/drivers/mfd/Kconfig
>> @@ -2374,6 +2374,18 @@ config MFD_KHADAS_MCU
>> additional drivers must be enabled in order to use the functionality
>> of the device.
>>
>> +config MFD_PSCI
>> + bool "PSCI MFD for psci child cells"
>
> "PSCI mux for registering PSCI component devices" ?
>
>> + depends on ARM_PSCI_FW
>> + depends on OF
>> + select MFD_CORE
>> + default y
>> + help
>> + PSCI MFD registers PSCI child cells and exposes them as
>> + platform devices.
>
> This is what all DT related MFD devices do.
>
>> + Child drivers are probed only if enabled in the
>> + kernel configuration.
>
> Nothing new here either. This is the norm.
>
> "PSCI Mux is a Linux specific device splitter designed to register
> multiple PSCI related components via a single Device Tree compatible.
> Without it, devices would each bind to a single node ... <which causes
> problems with x, y and z>"
Ack. Will make it more relevant.
>
>> + Select this option whenever a supported PSCI
>> + child driver is selected.
>> +
>> config MFD_ACER_A500_EC
>> tristate "Support for Acer Iconia Tab A500 Embedded Controller"
>> depends on I2C
>> diff --git a/drivers/mfd/Makefile b/drivers/mfd/Makefile
>> index 72d3944b0ad8..f3edfad702ff 100644
>> --- a/drivers/mfd/Makefile
>> +++ b/drivers/mfd/Makefile
>> @@ -24,6 +24,8 @@ obj-$(CONFIG_MFD_EXYNOS_LPASS) += exynos-lpass.o
>> obj-$(CONFIG_MFD_GATEWORKS_GSC) += gateworks-gsc.o
>> obj-$(CONFIG_MFD_MACSMC) += macsmc.o
>>
>> +obj-$(CONFIG_MFD_PSCI) += psci-mfd.o
>> +
>> obj-$(CONFIG_MFD_TI_LP873X) += lp873x.o
>> obj-$(CONFIG_MFD_TI_LP87565) += lp87565.o
>> obj-$(CONFIG_MFD_TI_AM335X_TSCADC) += ti_am335x_tscadc.o
>> diff --git a/drivers/mfd/psci-mfd.c b/drivers/mfd/psci-mfd.c
>> new file mode 100644
>> index 000000000000..7affd6bb09dd
>> --- /dev/null
>> +++ b/drivers/mfd/psci-mfd.c
>> @@ -0,0 +1,43 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * Copyright (c) Qualcomm Technologies, Inc. and/or its subsidiaries.
>> + */
>> +
>> +#include <linux/mfd/core.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/platform_device.h>
>> +
>> +static const struct mfd_cell psci_cells[] = {
>> + {
>> + .name = "psci-cpuidle-domain",
>> + },
>> +};
>
> MFD_CELL_NAME() ?
>
> MFD drivers have to support more than one device from the get-go.
> Please add your other device here in this patch.
Ack.
>
>> +static int psci_mfd_probe(struct platform_device *pdev)
>
> No to _mfd_.
Ack.
>> +{
>> + return devm_mfd_add_devices(&pdev->dev, PLATFORM_DEVID_AUTO, psci_cells,
>> + ARRAY_SIZE(psci_cells), NULL, 0, NULL);
>> +}
>> +
>> +static const struct of_device_id psci_mfd_of_match[] = {
>> + { .compatible = "arm,psci-1.0" },
>> + { }
>> +};
>> +
>> +static struct platform_driver psci_mfd_driver = {
>> + .probe = psci_mfd_probe,
>> + .driver = {
>> + .name = "psci-mfd",
>> + .of_match_table = psci_mfd_of_match,
>> + },
>> +};
>> +
>> +static int __init psci_mfd_init(void)
>> +{
>> + return platform_driver_register(&psci_mfd_driver);
>> +}
>> +
>
> Remove this line.
Ack.
>
>> +core_initcall(psci_mfd_init);
>
> This is usually a hack. Why can't you use -EPROBE_DEFER?
Ack.
thanks,
Shivendra
More information about the linux-arm-kernel
mailing list