Re: [RFC 2/6] mfd: Support for Ettus Research E31x devices PMU

From: Moritz Fischer
Date: Sun Feb 17 2019 - 11:37:50 EST


Hi Lee,

thanks for your feedback!

On Thu, Feb 14, 2019 at 1:34 AM Lee Jones <lee.jones@xxxxxxxxxx> wrote:
>
> On Mon, 11 Feb 2019, Virendra Kakade wrote:
>
> > Signed-off-by: Virendra Kakade <virendra.kakade@xxxxxx>
> > ---
> > drivers/mfd/Kconfig | 7 +++
> > drivers/mfd/Makefile | 2 +-
> > drivers/mfd/e31x-pmu.c | 89 ++++++++++++++++++++++++++++++++++++
> > include/linux/mfd/e31x-pmu.h | 20 ++++++++
> > 4 files changed, 117 insertions(+), 1 deletion(-)
> > create mode 100644 drivers/mfd/e31x-pmu.c
> > create mode 100644 include/linux/mfd/e31x-pmu.h
> >
> > diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
> > index 8c5dfdce4326..6c4c0747f2a5 100644
> > --- a/drivers/mfd/Kconfig
> > +++ b/drivers/mfd/Kconfig
> > @@ -1916,4 +1916,11 @@ config RAVE_SP_CORE
> > device found on several devices in RAVE line of hardware.
> >
> > endmenu
> > +
> > +config MFD_E31X_PMU
> > + tristate "E31X PMU driver"
> > + select MFD_SYSCON
> > + help
> > + Select this to get support for the Ettus Research E31x devices.
> > +
> > endif
> > diff --git a/drivers/mfd/Makefile b/drivers/mfd/Makefile
> > index 12980a4ad460..e37c2057134b 100644
> > --- a/drivers/mfd/Makefile
> > +++ b/drivers/mfd/Makefile
> > @@ -241,4 +241,4 @@ obj-$(CONFIG_MFD_MXS_LRADC) += mxs-lradc.o
> > obj-$(CONFIG_MFD_SC27XX_PMIC) += sprd-sc27xx-spi.o
> > obj-$(CONFIG_RAVE_SP_CORE) += rave-sp.o
> > obj-$(CONFIG_MFD_ROHM_BD718XX) += rohm-bd718x7.o
> > -
> > +obj-$(CONFIG_MFD_E31X_PMU) += e31x-pmu.o
> > diff --git a/drivers/mfd/e31x-pmu.c b/drivers/mfd/e31x-pmu.c
> > new file mode 100644
> > index 000000000000..383df68a538f
> > --- /dev/null
> > +++ b/drivers/mfd/e31x-pmu.c
> > @@ -0,0 +1,89 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright (c) 2018 National Instruments Corp
>
> This is out of date.
>
> '\n' here.
>
> > + * Author: Virendra Kakade<virendra.kakade@xxxxxx>
>
> You need a space after your name.
>
> > + * Ettus Research E31x PMU MFD driver
>
> Please expand PMU.
>
> > + */
> > +
> > +#include <linux/err.h>
> > +#include <linux/delay.h>
> > +#include <linux/interrupt.h>
> > +#include <linux/kernel.h>
> > +#include <linux/mfd/syscon.h>
> > +#include <linux/module.h>
> > +#include <linux/pm_runtime.h>
> > +#include <linux/regmap.h>
> > +#include <linux/mfd/core.h>
> > +#include <linux/of_device.h>
> > +#include <linux/mfd/e31x-pmu.h>
> > +#include <linux/platform_device.h>
>
> Alphabetical.
>
> > +#define E31X_PMU_MISC_IRQ_MASK BIT(8)
> > +#define E31X_PMU_MISC_IRQ_SHIFT 8
> > +#define E31X_PMU_MISC_VERSION_MIN_MASK GENMASK(3, 0)
> > +#define E31X_PMU_MISC_VERSION_MIN_SHIFT 0
> > +#define E31X_PMU_MISC_VERSION_MAJ_MASK GENMASK(7, 4)
> > +#define E31X_PMU_MISC_VERSION_MAJ_SHIFT 4
> > +
> > +struct e31x_pmu {
> > + struct regmap *regmap;
> > +};
>
> A whole struct for one attribute?
>
> > +static int e31x_pmu_check_version(struct e31x_pmu *pmu)
> > +{
> > + int timeout = 100;
>
> Where does 100 come from?
>
> > + u32 misc, maj;
> > + int err;
>
> 'ret' is more common.
>
> > + /* we need to wait a bit for firmware to populate the fields */
>
> "a bit" ?
>
> What does the datasheet say?
>
> Please use proper grammar. Capital letters.
>
> > + while (timeout--) {
> > + err = regmap_read(pmu->regmap, E31X_PMU_REG_MISC, &misc);
> > + if (err)
> > + return err;
> > + if (misc)
> > + break;
> > +
> > + usleep_range(2500, 5000);
>
> Formatting.
>
> > + }
> > +
> > + /* only firmware versions above 2.0 are supported */
>
> "Only supports firmware version 2.0 and above"
>
> > + maj = E31X_PMU_GET_FIELD(MISC_VERSION_MAJ, misc);
> > + if (maj < 2)
> > + return -ENOTSUPP;
>
> '\n' here.
>
> > + return 0;
> > +}
> > +
> > +static int e31x_pmu_probe(struct platform_device *pdev)
> > +{
> > + struct e31x_pmu *pmu;
> > +
> > + pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
> > + if (!pmu)
> > + return -ENOMEM;
>
> '\n' here.
>
> > + pmu->regmap = syscon_regmap_lookup_by_phandle(\
>
> Why are you storing regmap?

See my comment below.
>
> Just pass it directly to e31x_pmu_check_version().
>
> > + pdev->dev.of_node, "regmap");
> > +
>
> Remove this line.
>
> > + if (IS_ERR(pmu->regmap))
> > + return PTR_ERR(pmu->regmap);
>
> '\n' here.
>
> > + if (e31x_pmu_check_version(pmu))
>
> Please split out the function from the if.
>
> Use a variable 'ret' to feed into and check that.
>
> > + return -ENOTSUPP;
>
> '\n' here.
>
> > + return devm_of_platform_populate(&pdev->dev);
> > +}
> > +
> > +static const struct of_device_id e31x_pmu_id[] = {
> > + { .compatible = "ni,e31x-pmu" },
> > + {},
> > +};
> > +
> > +static struct platform_driver e31x_pmu_driver = {
> > + .driver = {
> > + .name = "e31x-pmu",
> > + .of_match_table = e31x_pmu_id,
>
> These 2 lines need indenting.
>
> > + },
> > + .probe = e31x_pmu_probe,
> > +};
> > +module_platform_driver(e31x_pmu_driver);
> > +
> > +MODULE_DESCRIPTION("E31x PMU driver");
> > +MODULE_AUTHOR("Virendra Kakade <virendra.kakade@xxxxxx>");
> > +MODULE_LICENSE("GPL");
>
> So the whole purpose of this driver is to do a version check.
>
> Seems like over-kill!
>
> Probably better to do the version check in an inline function stored
> in a header file, then call it from each of the children's .probe()
> function.

A bit of context here, this is based on an in-house driver that we had that does
a bunch of other stuff (e.g. it controls a flag on whether the device auto-boots
on power being plugged in etc.). These functions will use the regmap.

I had advised Virendra to submit a base version and follow up with patches
that would add these functions one by one as he figures out how to expose them
in a proper way to the kernel.

Cheers,
Moritz