public inbox for devicetree@vger.kernel.org
 help / color / mirror / Atom feed
From: "Crystal Guo (郭晶)" <Crystal.Guo@mediatek.com>
To: "robh@kernel.org" <robh@kernel.org>,
	"matthias.bgg@gmail.com" <matthias.bgg@gmail.com>,
	AngeloGioacchino Del Regno
	<angelogioacchino.delregno@collabora.com>,
	"krzk@kernel.org" <krzk@kernel.org>,
	"conor+dt@kernel.org" <conor+dt@kernel.org>
Cc: "linux-arm-kernel@lists.infradead.org"
	<linux-arm-kernel@lists.infradead.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-mediatek@lists.infradead.org"
	<linux-mediatek@lists.infradead.org>,
	"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
	Project_Global_Chrome_Upstream_Group
	<Project_Global_Chrome_Upstream_Group@mediatek.com>
Subject: Re: [v3,2/2] memory/mediatek: Add an interface to get current DDR data rate
Date: Wed, 2 Apr 2025 03:36:38 +0000	[thread overview]
Message-ID: <76d4cd09c609ad4b5de2dc859d0499a512df3790.camel@mediatek.com> (raw)
In-Reply-To: <ae533d2e-49ed-4be1-89de-a55b21f4d30c@collabora.com>

On Wed, 2025-03-26 at 11:27 +0100, AngeloGioacchino Del Regno wrote:
> External email : Please do not click links or open attachments until
> you have verified the sender or the content.
> 
> 
> Il 26/03/25 07:30, Crystal Guo ha scritto:
> > Add MediaTek DRAMC driver to provide an interface that can
> > obtain current DDR data rate.
> > 
> > Signed-off-by: Crystal Guo <crystal.guo@mediatek.com>
> > ---
> >   drivers/memory/Kconfig              |   1 +
> >   drivers/memory/Makefile             |   1 +
> >   drivers/memory/mediatek/Kconfig     |  21 +++
> >   drivers/memory/mediatek/Makefile    |   2 +
> >   drivers/memory/mediatek/mtk-dramc.c | 232
> > ++++++++++++++++++++++++++++
> >   5 files changed, 257 insertions(+)
> >   create mode 100644 drivers/memory/mediatek/Kconfig
> >   create mode 100644 drivers/memory/mediatek/Makefile
> >   create mode 100644 drivers/memory/mediatek/mtk-dramc.c
> > 
> > diff --git a/drivers/memory/Kconfig b/drivers/memory/Kconfig
> > index c82d8d8a16ea..b1698549ff81 100644
> > --- a/drivers/memory/Kconfig
> > +++ b/drivers/memory/Kconfig
> > @@ -227,5 +227,6 @@ config STM32_FMC2_EBI
> > 
> >   source "drivers/memory/samsung/Kconfig"
> >   source "drivers/memory/tegra/Kconfig"
> > +source "drivers/memory/mediatek/Kconfig"
> > 
> >   endif
> > diff --git a/drivers/memory/Makefile b/drivers/memory/Makefile
> > index d2e6ca9abbe0..c0facf529803 100644
> > --- a/drivers/memory/Makefile
> > +++ b/drivers/memory/Makefile
> > @@ -27,6 +27,7 @@ obj-$(CONFIG_STM32_FMC2_EBI)        += stm32-
> > fmc2-ebi.o
> > 
> >   obj-$(CONFIG_SAMSUNG_MC)    += samsung/
> >   obj-$(CONFIG_TEGRA_MC)              += tegra/
> > +obj-$(CONFIG_MEDIATEK_MC)    += mediatek/
> >   obj-$(CONFIG_TI_EMIF_SRAM)  += ti-emif-sram.o
> >   obj-$(CONFIG_FPGA_DFL_EMIF) += dfl-emif.o
> > 
> > diff --git a/drivers/memory/mediatek/Kconfig
> > b/drivers/memory/mediatek/Kconfig
> > new file mode 100644
> > index 000000000000..3f238e0d9647
> > --- /dev/null
> > +++ b/drivers/memory/mediatek/Kconfig
> > @@ -0,0 +1,21 @@
> > +# SPDX-License-Identifier: GPL-2.0-only
> > +config MEDIATEK_MC
> > +     bool "MediaTek Memory Controller support"
> 
> default y
> depends on ARCH_MEDIATEK || COMPILE_TEST
> 
> > +     help
> > +       This option allows to enable MediaTek memory controller
> > drivers,
> > +       which may include controllers for DRAM or others.
> > +       Select Y here if you need support for MediaTek memory
> > controller.
> > +       If you don't need, select N.
> > +
> > +if MEDIATEK_MC
> > +
> > +config MTK_DRAMC
> > +     tristate "MediaTek DRAMC driver"
> > +     default y
> > +     help
> > +       This option selects the MediaTek DRAMC driver, which
> > provides
> 
>            This driver is for the DRAM Controller found in MediaTek
> SoCs
>            and provides a sysfs interface for reporting the current
> DRAM
>            data rate.
> 
> The part saying select y or n can be avoided.

Ok, I will remove "select y or n" in the next version.

> 
> > +       an interface for reporting the current data rate of DRAM.
> > +       Select Y here if you need support for the MediaTek DRAMC
> > driver.
> > +       If you don't need, select N.
> > +
> > +endif
> > diff --git a/drivers/memory/mediatek/Makefile
> > b/drivers/memory/mediatek/Makefile
> > new file mode 100644
> > index 000000000000..a1395fc55b41
> > --- /dev/null
> > +++ b/drivers/memory/mediatek/Makefile
> > @@ -0,0 +1,2 @@
> > +# SPDX-License-Identifier: GPL-2.0
> > +obj-$(CONFIG_MTK_DRAMC)              += mtk-dramc.o
> > diff --git a/drivers/memory/mediatek/mtk-dramc.c
> > b/drivers/memory/mediatek/mtk-dramc.c
> > new file mode 100644
> > index 000000000000..22042c9d8e42
> > --- /dev/null
> > +++ b/drivers/memory/mediatek/mtk-dramc.c
> > @@ -0,0 +1,232 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright (c) 2025 MediaTek Inc.
> > + */
> > +#include <linux/bitops.h>
> > +#include <linux/bitfield.h>
> > +#include <linux/device.h>
> > +#include <linux/io.h>
> > +#include <linux/kernel.h>
> > +#include <linux/module.h>
> > +#include <linux/of.h>
> > +#include <linux/of_address.h>
> > +#include <linux/of_platform.h>
> > +#include <linux/platform_device.h>
> > +#include <linux/printk.h>
> > +
> > +static unsigned int read_reg_field(void __iomem *base, unsigned
> > int offset, unsigned int mask)
> 
> Please move this function to before mtk_dramc_probe()

Ok, it will be updated in the next version.

> 
> > +{
> > +     unsigned int val = readl(base + offset);
> > +     unsigned int shift = __ffs(mask);
> > +
> > +     return (val & mask) >> shift;
> > +}
> > +
> > +struct mtk_dramc_pdata {
> > +     u8 fmeter_version;
> > +     u8 ref_freq_mhz;
> > +     const u16 *regs;
> > +     const u32 *masks;
> > +     u32 posdiv_purify;
> > +     u8 prediv;
> > +     u16 shuffle_offset;
> > +};
> > +
> > +struct mtk_dramc_dev_t {
> 
> That may be confused with a typedef so, please, just "struct
> mtk_dramc"
> 

Ok, it will be updated in the next version.


> > +     void __iomem *anaphy_base;
> > +     void __iomem *ddrphy_base;
> > +     const struct mtk_dramc_pdata *pdata;
> > +};
> > +
> > +enum mtk_dramc_reg_index {
> > +     DRAMC_DPHY_DVFS_STA,
> > +     DRAMC_APHY_SHU_PHYPLL2,
> > +     DRAMC_APHY_SHU_CLRPLL2,
> > +     DRAMC_APHY_SHU_PHYPLL3,
> > +     DRAMC_APHY_SHU_CLRPLL3,
> > +     DRAMC_APHY_SHU_PHYPLL4,
> > +     DRAMC_APHY_ARPI0,
> > +     DRAMC_APHY_CA_ARDLL1,
> > +     DRAMC_APHY_B0_TX0,
> > +};
> > +
> > +enum mtk_dramc_mask_index {
> > +     DRAMC_DPHY_DVFS_SHU_LV,
> > +     DRAMC_DPHY_DVFS_PLL_SEL,
> > +     DRAMC_APHY_PLL2_SDMPCW,
> > +     DRAMC_APHY_PLL3_POSDIV,
> > +     DRAMC_APHY_PLL4_FBKSEL,
> > +     DRAMC_APHY_ARPI0_SOPEN,
> > +     DRAMC_APHY_ARDLL1_CK_EN,
> > +     DRAMC_APHY_B0_TX0_SER_MODE,
> > +};
> > +
> > +static const u16 mtk_dramc_regs_mt8196[] = {
> > +     [DRAMC_DPHY_DVFS_STA] = 0xe98,
> > +     [DRAMC_APHY_SHU_PHYPLL2] = 0x908,
> > +     [DRAMC_APHY_SHU_CLRPLL2] = 0x928,
> > +     [DRAMC_APHY_SHU_PHYPLL3] = 0x90c,
> > +     [DRAMC_APHY_SHU_CLRPLL3] = 0x92c,
> > +     [DRAMC_APHY_SHU_PHYPLL4] = 0x910,
> > +     [DRAMC_APHY_ARPI0] = 0x0d94,
> > +     [DRAMC_APHY_CA_ARDLL1] = 0x0d08,
> > +     [DRAMC_APHY_B0_TX0] = 0x0dc4,
> > +};
> > +
> > +static const u32 mtk_dramc_masks_mt8196[] = {
> > +     [DRAMC_DPHY_DVFS_SHU_LV] = GENMASK(15, 14),
> > +     [DRAMC_DPHY_DVFS_PLL_SEL] = GENMASK(25, 25),
> > +     [DRAMC_APHY_PLL2_SDMPCW] = GENMASK(18, 3),
> > +     [DRAMC_APHY_PLL3_POSDIV] = GENMASK(13, 11),
> > +     [DRAMC_APHY_PLL4_FBKSEL] = GENMASK(6, 6),
> > +     [DRAMC_APHY_ARPI0_SOPEN] = GENMASK(26, 26),
> > +     [DRAMC_APHY_ARDLL1_CK_EN] = GENMASK(0, 0),
> > +     [DRAMC_APHY_B0_TX0_SER_MODE] = GENMASK(4, 3),
> > +};
> > +
> 
> function read_reg_field goes here.
> 

Got it, thanks.

> > +static int mtk_dramc_probe(struct platform_device *pdev)
> > +{
> > +     struct mtk_dramc_dev_t *dramc;
> > +     const struct mtk_dramc_pdata *pdata;
> > +
> > +     dramc = devm_kzalloc(&pdev->dev, sizeof(struct
> > mtk_dramc_dev_t), GFP_KERNEL);
> > +     if (!dramc)
> > +             return dev_err_probe(&pdev->dev, -ENOMEM, "Failed to
> > allocate memory\n");
> > +
> > +     pdata = of_device_get_match_data(&pdev->dev);
> > +     if (!pdata)
> > +             return dev_err_probe(&pdev->dev, -EINVAL, "No
> > platform data available\n");
> > +
> > +     dramc->pdata = pdata;
> > +
> > +     dramc->anaphy_base = devm_platform_ioremap_resource(pdev, 0);
> > +     if (IS_ERR(dramc->anaphy_base))
> > +             return dev_err_probe(&pdev->dev, PTR_ERR(dramc-
> > >anaphy_base),
> > +                                  "Unable to map ANAPHY base\n");
> > +
> > +     dramc->ddrphy_base = devm_platform_ioremap_resource(pdev, 1);
> > +     if (IS_ERR(dramc->ddrphy_base))
> > +             return dev_err_probe(&pdev->dev, PTR_ERR(dramc-
> > >ddrphy_base),
> > +                                  "Unable to map DDRPHY base\n");
> > +
> > +     platform_set_drvdata(pdev, dramc);
> > +     return 0;
> > +}
> > +
> > +static unsigned int mtk_fmeter_v1(struct mtk_dramc_dev_t *dramc)
> > +{
> > +     const struct mtk_dramc_pdata *pdata = dramc->pdata;
> > +     unsigned int shu_level, pll_sel, offset;
> > +     unsigned int sdmpcw, posdiv, clkdiv, fbksel, sopen, async_ca,
> > ser_mode;
> > +     unsigned int prediv_freq, posdiv_freq, vco_freq;
> > +     unsigned int final_rate;
> > +
> > +     shu_level = read_reg_field(dramc->ddrphy_base, pdata-
> > >regs[DRAMC_DPHY_DVFS_STA],
> > +                                pdata-
> > >masks[DRAMC_DPHY_DVFS_SHU_LV]);
> > +     pll_sel = read_reg_field(dramc->ddrphy_base, pdata-
> > >regs[DRAMC_DPHY_DVFS_STA],
> > +                              pdata-
> > >masks[DRAMC_DPHY_DVFS_PLL_SEL]);
> > +     offset = pdata->shuffle_offset * shu_level;
> > +
> > +     sdmpcw = read_reg_field(dramc->anaphy_base,
> > +                             ((pll_sel == 0) ?
> > +                             pdata->regs[DRAMC_APHY_SHU_PHYPLL2] :
> > +                             pdata->regs[DRAMC_APHY_SHU_CLRPLL2])
> > + offset,
> > +                             pdata-
> > >masks[DRAMC_APHY_PLL2_SDMPCW]);
> > +     posdiv = read_reg_field(dramc->anaphy_base,
> > +                             ((pll_sel == 0) ?
> > +                             pdata->regs[DRAMC_APHY_SHU_PHYPLL3] :
> > +                             pdata->regs[DRAMC_APHY_SHU_CLRPLL3])
> > + offset,
> > +                             pdata-
> > >masks[DRAMC_APHY_PLL3_POSDIV]);
> > +     fbksel = read_reg_field(dramc->anaphy_base, pdata-
> > >regs[DRAMC_APHY_SHU_PHYPLL4] + offset,
> > +                             pdata-
> > >masks[DRAMC_APHY_PLL4_FBKSEL]);
> > +     sopen = read_reg_field(dramc->anaphy_base, pdata-
> > >regs[DRAMC_APHY_ARPI0] + offset,
> > +                            pdata->masks[DRAMC_APHY_ARPI0_SOPEN]);
> > +     async_ca = read_reg_field(dramc->anaphy_base, pdata-
> > >regs[DRAMC_APHY_CA_ARDLL1] + offset,
> > +                               pdata-
> > >masks[DRAMC_APHY_ARDLL1_CK_EN]);
> > +     ser_mode = read_reg_field(dramc->anaphy_base, pdata-
> > >regs[DRAMC_APHY_B0_TX0] + offset,
> > +                               pdata-
> > >masks[DRAMC_APHY_B0_TX0_SER_MODE]);
> > +
> > +     clkdiv = (ser_mode == 1) ? 1 : 0;
> > +     posdiv &= ~(pdata->posdiv_purify);
> > +
> > +     prediv_freq = pdata->ref_freq_mhz * (sdmpcw >> pdata-
> > >prediv);
> > +     posdiv_freq = (prediv_freq >> posdiv) >> 1;
> > +     vco_freq = posdiv_freq << fbksel;
> > +     final_rate = vco_freq >> clkdiv;
> > +
> > +     if (sopen == 1 && async_ca == 1)
> > +             final_rate >>= 1;
> > +
> > +     return final_rate;
> > +}
> > +
> > +/*
> 
> You're so near to kerneldoc that you might just as well use it.
> 

Ok, I will update the comments in the next version.

> /**
>   * mtk_dramc_get_data_rate - Calculate DRAM data rate
>   * @dev - Device pointer
>   *
>   * Return: DRAM Data Rate in MB/s or negative number for error
>   */
> 
> > + * mtk_dramc_get_data_rate - calculate DRAM data rate
> > + *
> > + * Returns DRAM data rate (MB/s)
> > + */
> > +static unsigned int mtk_dramc_get_data_rate(struct device *dev)
> > +{
> > +     struct mtk_dramc_dev_t *dramc_dev = dev_get_drvdata(dev);
> > +
> > +     if (!dramc_dev) {
> > +             dev_err(dev, "DRAMC device data not found\n");
> 
> That can't happen, because you're assigning drvdata in the probe
> function
> 
> > +             return -EINVAL;
> > +     }
> > +
> > +     if (dramc_dev->pdata) {
> > +             if (dramc_dev->pdata->fmeter_version == 1)
> > +                     return mtk_fmeter_v1(dramc_dev);
> > +
> > +             dev_err(dev, "Unsupported fmeter version\n");
> > +             return -EINVAL;
> > +     }
> > +     dev_err(dev, "DRAMC platform data not found\n");
> 
> That also can't happen.
> 
> > +     return -EINVAL;
> > +}
> 
> /**
>   * mtk_dramc_get_data_rate - Calculate DRAM data rate
>   * @dev - Device pointer
>   *
>   * Return: DRAM Data Rate in MB/s or negative number for error
>   */
> static unsigned int mtk_dramc_get_data_rate(struct device *dev)
> {
>         struct mtk_dramc *dramc = dev_get_drvdata(dev);
> 
>         if (dramc_dev->pdata->fmeter_version == 1)
>                 return mtk_fmeter_v1(dramc_dev);
> 
>         dev_err(dev, "Frequency meter version %u not supported\n");
>         return -EINVAL;
> };
> 

Thanks for the reminder, this part will be updated in the next version.

> > +
> > +static ssize_t dram_data_rate_show(struct device *dev,
> > +                                struct device_attribute *attr,
> > char *buf)
> > +{
> > +     return snprintf(buf, PAGE_SIZE, "DRAM data rate = %u\n",
> > +                     mtk_dramc_get_data_rate(dev));
> > +}
> > +
> > +static DEVICE_ATTR_RO(dram_data_rate);
> > +
> > +static struct attribute *mtk_dramc_attrs[] = {
> > +     &dev_attr_dram_data_rate.attr,
> > +     NULL
> > +};
> > +ATTRIBUTE_GROUPS(mtk_dramc);
> > +
> > +static const struct mtk_dramc_pdata dramc_pdata_mt8196 = {
> > +     .fmeter_version = 1,
> > +     .ref_freq_mhz = 26,
> > +     .regs = mtk_dramc_regs_mt8196,
> > +     .masks = mtk_dramc_masks_mt8196,
> > +     .posdiv_purify = BIT(2),
> > +     .prediv = 7,
> > +     .shuffle_offset = 0x700,
> > +};
> > +
> > +static const struct of_device_id mtk_dramc_of_ids[] = {
> > +     { .compatible = "mediatek,mt8196-dramc", .data =
> > &dramc_pdata_mt8196 },
> > +     {}
> 
> { /* sentinel */ }
> 
> > +};
> > +MODULE_DEVICE_TABLE(of, mtk_dramc_of_ids);
> > +
> > +static struct platform_driver mtk_dramc_driver = {
> > +     .probe = mtk_dramc_probe,
> > +     .driver = {
> > +             .name = "mtk_dramc_drv",
> 
> .name = "mtk-dramc"
> 
> > +             .of_match_table = mtk_dramc_of_ids,
> > +             .dev_groups = mtk_dramc_groups,
> > +     },
> > +};
> > +
> 
> No blank line here:
> 
> };
> module_platform_driver(....)
> 
> I think that v4 will be the final version :-)
> 
> Cheers,
> Angelo
> 

Thanks for your suggestions. I will make updates based on the above
comments in the next version.

Best regards,
Crystal

> > +module_platform_driver(mtk_dramc_driver);
> > +
> > +MODULE_AUTHOR("Crystal Guo <crystal.guo@mediatek.com>");
> > +MODULE_DESCRIPTION("MediaTek DRAM Controller Driver");
> > +MODULE_LICENSE("GPL");
> 
> 

      reply	other threads:[~2025-04-02  3:36 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-26  6:30 [v3,0/2] Add an interface to get current DDR data rate Crystal Guo
2025-03-26  6:30 ` [v3,1/2] dt-bindings: memory-controllers: Add MediaTek DRAM controller interface Crystal Guo
2025-03-26  7:56   ` Krzysztof Kozlowski
2025-03-26 10:17     ` AngeloGioacchino Del Regno
2025-03-26 10:27       ` Krzysztof Kozlowski
2025-03-26 10:18   ` AngeloGioacchino Del Regno
2025-04-02  3:51     ` Crystal Guo (郭晶)
2025-04-02  9:24       ` AngeloGioacchino Del Regno
2025-03-26  6:30 ` [v3,2/2] memory/mediatek: Add an interface to get current DDR data rate Crystal Guo
2025-03-26 10:27   ` AngeloGioacchino Del Regno
2025-04-02  3:36     ` Crystal Guo (郭晶) [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=76d4cd09c609ad4b5de2dc859d0499a512df3790.camel@mediatek.com \
    --to=crystal.guo@mediatek.com \
    --cc=Project_Global_Chrome_Upstream_Group@mediatek.com \
    --cc=angelogioacchino.delregno@collabora.com \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=krzk@kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mediatek@lists.infradead.org \
    --cc=matthias.bgg@gmail.com \
    --cc=robh@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox