Re: [PATCH 6/6] clk: samsung: Introduce Exynos850 clock driver

[Sam Protsenko]

On Wed, 15 Sept 2021 at 11:59, Krzysztof Kozlowski
<krzysztof.kozlowski@xxxxxxxxxxxxx> wrote:
>
> On 14/09/2021 17:56, Sam Protsenko wrote:
> > This is the initial implementation adding only basic clocks like UART,
> > MMC, I2C and corresponding parent clocks. Design is influenced by
> > Exynos7 and Exynos5433 clock drivers.
> >
> > Signed-off-by: Sam Protsenko <semen.protsenko@xxxxxxxxxx>
> > ---
> >  drivers/clk/samsung/Makefile        |   1 +
> >  drivers/clk/samsung/clk-exynos850.c | 700 ++++++++++++++++++++++++++++
> >  2 files changed, 701 insertions(+)
> >  create mode 100644 drivers/clk/samsung/clk-exynos850.c
> >
> > diff --git a/drivers/clk/samsung/Makefile b/drivers/clk/samsung/Makefile
> > index 028b2e27a37e..c46cf11e4d0b 100644
> > --- a/drivers/clk/samsung/Makefile
> > +++ b/drivers/clk/samsung/Makefile
> > @@ -17,6 +17,7 @@ obj-$(CONFIG_EXYNOS_ARM64_COMMON_CLK)       += clk-exynos5433.o
> >  obj-$(CONFIG_EXYNOS_AUDSS_CLK_CON) += clk-exynos-audss.o
> >  obj-$(CONFIG_EXYNOS_CLKOUT)  += clk-exynos-clkout.o
> >  obj-$(CONFIG_EXYNOS_ARM64_COMMON_CLK)        += clk-exynos7.o
> > +obj-$(CONFIG_EXYNOS_ARM64_COMMON_CLK)        += clk-exynos850.o
> >  obj-$(CONFIG_S3C2410_COMMON_CLK)+= clk-s3c2410.o
> >  obj-$(CONFIG_S3C2410_COMMON_DCLK)+= clk-s3c2410-dclk.o
> >  obj-$(CONFIG_S3C2412_COMMON_CLK)+= clk-s3c2412.o
> > diff --git a/drivers/clk/samsung/clk-exynos850.c b/drivers/clk/samsung/clk-exynos850.c
> > new file mode 100644
> > index 000000000000..1028caa2102e
> > --- /dev/null
> > +++ b/drivers/clk/samsung/clk-exynos850.c
> > @@ -0,0 +1,700 @@
> > +// SPDX-License-Identifier: GPL-2.0-only
> > +/*
> > + * Copyright (C) 2021 Linaro Ltd.
> > + * Author: Sam Protsenko <semen.protsenko@xxxxxxxxxx>
> > + *
> > + * Common Clock Framework support for Exynos850 SoC.
> > + */
> > +
> > +#include <linux/clk-provider.h>
> > +#include <linux/of.h>
> > +#include <linux/of_address.h>
> > +
> > +#include <dt-bindings/clock/exynos850.h>
> > +
> > +#include "clk.h"
> > +
> > +/* Gate register bits */
> > +#define GATE_MANUAL          BIT(20)
> > +#define GATE_ENABLE_HWACG    BIT(28)
> > +
> > +/* Gate register offsets range */
> > +#define GATE_OFF_START               0x2000
> > +#define GATE_OFF_END         0x2fff
> > +
> > +/**
> > + * exynos850_init_clocks - Set clocks initial configuration
> > + * @np:                      CMU device tree node with "reg" property (CMU addr)
> > + * @reg_offs:                Register offsets array for clocks to init
> > + * @reg_offs_len:    Number of register offsets in reg_offs array
> > + *
> > + * Set manual control mode for all gate clocks.
> > + */
> > +static void __init exynos850_init_clocks(struct device_node *np,
> > +             const unsigned long *reg_offs, size_t reg_offs_len)
> > +{
> > +     const __be32 *regaddr_p;
> > +     u64 regaddr;
> > +     u32 base;
> > +     size_t i;
> > +
> > +     /* Get the base address ("reg" property in dts) */
> > +     regaddr_p = of_get_address(np, 0, NULL, NULL);
> > +     if (!regaddr_p)
> > +             panic("%s: failed to get reg regaddr\n", __func__);
> > +
> > +     regaddr = of_translate_address(np, regaddr_p);
> > +     if (regaddr == OF_BAD_ADDR || !regaddr)
> > +             panic("%s: bad reg regaddr\n", __func__);
> > +
> > +     base = (u32)regaddr;
> > +
> > +     for (i = 0; i < reg_offs_len; ++i) {
> > +             void __iomem *reg;
> > +             u32 val;
> > +
> > +             /* Modify only gate clock registers */
> > +             if (reg_offs[i] < GATE_OFF_START || reg_offs[i] > GATE_OFF_END)
> > +                     continue;
> > +
> > +             reg = ioremap(base + reg_offs[i], 4);
>
> You first translate the address to CPU physical address and then apply
> offset. This should be equivalent to one of_iomap() of entire range and
> iterate starting from the base pointer.  IOW, I don't get why you have
> to map each register instead of mapping entire SFR/IO range?
>

Thanks, will do in v2.

> > +             val = ioread32(reg);
> > +             val |= GATE_MANUAL;
> > +             val &= ~GATE_ENABLE_HWACG;
> > +             iowrite32(val, reg);
>
> All other drivers use readl/writel, so how about keeping it consistent?
>

Ok. Though io* variants looks better to me (API names consistent with
ioremap/iounmap) :)

> Rest looks good but I did not verify the numbers :)
>
> Best regards,
> Krzysztof