[PATCH v3 3/6] drivers: mux: mmio-based syscon mux controller

[Simon Glass]

Hi Pratyush,

On Thu, 11 Jun 2020 at 13:45, Pratyush Yadav <p.yadav at ti.com> wrote:
>
> From: Jean-Jacques Hiblot <jjhiblot at ti.com>
>
> This adds a driver for mmio-based syscon multiplexers controlled by
> bitfields in a syscon register range.
> This is heavily based on the linux mmio-mux driver.
>
> Signed-off-by: Jean-Jacques Hiblot <jjhiblot at ti.com>
> Reviewed-by: Simon Glass <sjg at chromium.org>
> Signed-off-by: Pratyush Yadav <p.yadav at ti.com>
> ---
>  drivers/mux/Kconfig  |  14 +++++
>  drivers/mux/Makefile |   1 +
>  drivers/mux/mmio.c   | 143 +++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 158 insertions(+)
>  create mode 100644 drivers/mux/mmio.c

Reviewed-by: Simon Glass <sjg at chromium.org>

nit below

>
> diff --git a/drivers/mux/Kconfig b/drivers/mux/Kconfig
> index 35c1c5673c..f15ee4f833 100644
> --- a/drivers/mux/Kconfig
> +++ b/drivers/mux/Kconfig
> @@ -8,4 +8,18 @@ config MULTIPLEXER
>          controllers. It provides the same API as Linux and mux drivers should
>          be portable with a minimum effort.
>
> +if MULTIPLEXER
> +
> +config MUX_MMIO
> +       bool "MMIO register bitfield-controlled Multiplexer"
> +       depends on MULTIPLEXER && SYSCON
> +       help
> +         MMIO register bitfield-controlled Multiplexer controller.
> +
> +         The driver builds multiplexer controllers for bitfields in a syscon
> +         register. For N bit wide bitfields, there will be 2^N possible
> +         multiplexer states.
> +
> +endif
> +
>  endmenu
> diff --git a/drivers/mux/Makefile b/drivers/mux/Makefile
> index 351e4363d3..78ebf04c7a 100644
> --- a/drivers/mux/Makefile
> +++ b/drivers/mux/Makefile
> @@ -4,3 +4,4 @@
>  # Jean-Jacques Hiblot <jjhiblot at ti.com>
>
>  obj-$(CONFIG_$(SPL_)MULTIPLEXER) += mux-uclass.o
> +obj-$(CONFIG_$(SPL_)MUX_MMIO) += mmio.o
> diff --git a/drivers/mux/mmio.c b/drivers/mux/mmio.c
> new file mode 100644
> index 0000000000..573e599dd1
> --- /dev/null
> +++ b/drivers/mux/mmio.c
> @@ -0,0 +1,143 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * MMIO register bitfield-controlled multiplexer driver
> + * Based on the linux mmio multiplexer driver
> + *
> + * Copyright (C) 2017 Pengutronix, Philipp Zabel <kernel at pengutronix.de>
> + * Copyright (C) 2019 Texas Instrument, Jean-jacques Hiblot <jjhiblot at ti.com>
> + */
> +#include <common.h>
> +#include <dm.h>
> +#include <mux-internal.h>
> +#include <regmap.h>
> +#include <syscon.h>
> +#include <dm/device.h>
> +#include <dm/read.h>
> +#include <dm/devres.h>
> +#include <dt-bindings/mux/mux.h>
> +#include <linux/bitops.h>
> +
> +static int mux_mmio_set(struct mux_control *mux, int state)
> +{
> +       struct regmap_field **fields = dev_get_priv(mux->dev);
> +
> +       return regmap_field_write(fields[mux_control_get_index(mux)], state);
> +}
> +
> +static const struct mux_control_ops mux_mmio_ops = {
> +       .set = mux_mmio_set,
> +};
> +
> +static const struct udevice_id mmio_mux_of_match[] = {
> +       { .compatible = "mmio-mux" },
> +       { /* sentinel */ },
> +};
> +
> +static int mmio_mux_probe(struct udevice *dev)
> +{
> +       struct regmap_field **fields;
> +       struct mux_chip *mux_chip = dev_get_uclass_priv(dev);
> +       struct regmap *regmap;
> +       u32 *mux_reg_masks;
> +       u32 *idle_states;
> +       int num_fields;
> +       int ret;
> +       int i;
> +
> +       regmap = syscon_node_to_regmap(dev_ofnode(dev->parent));
> +       if (IS_ERR(regmap)) {
> +               ret = PTR_ERR(regmap);
> +               dev_err(dev, "failed to get regmap: %d\n", ret);
> +               return ret;
> +       }
> +
> +       num_fields = dev_read_size(dev, "mux-reg-masks");
> +       if (num_fields < 0)
> +               return log_msg_ret("mux-reg-masks missing or invalid", -EINVAL);

It occurred to me at some point that all we really need here is a
short string, like "mux-reg-masks" since we only need to find the
unique string within the function (which logging automatically
records). If we use very long strings then enabling the message
logging will cause a very large increase in code size.

Regards,
Simon