Re: [PATCH v2 9/9] mfd: mt6360: Merge different sub-devices i2c read/write

From: Gene Chen <gene.chen.richtek@gmail.com>
To: Lee Jones <lee.jones@linaro.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>,
	linux-arm-kernel@lists.infradead.org,
	linux-mediatek@lists.infradead.org, linux-kernel@vger.kernel.org,
	Gene Chen <gene_chen@richtek.com>,
	benjamin.chao@mediatek.com, shufan_lee@richtek.com,
	cy_huang@richtek.com
Subject: Re: [PATCH v2 9/9] mfd: mt6360: Merge different sub-devices i2c read/write
Date: Thu, 30 Jul 2020 10:56:20 +0800	[thread overview]
Message-ID: <CAE+NS37hURYnWqsewnc+T9yn62pFdSHUTqL4BvdsH_3mRf6Yrg@mail.gmail.com> (raw)
In-Reply-To: <20200729101244.GH2419169@dell>

Lee Jones <lee.jones@linaro.org> 於 2020年7月29日 週三 下午6:12寫道：
>
> On Wed, 29 Jul 2020, Gene Chen wrote:
>
> > Lee Jones <lee.jones@linaro.org> 於 2020年7月27日 週一 下午8:43寫道：
> > >
> > > On Fri, 17 Jul 2020, Gene Chen wrote:
> > >
> > > > From: Gene Chen <gene_chen@richtek.com>
> > > >
> > > > Remove unuse register definition.
> > >
> > > This should not be in here.
> > >
> > > > Merge different sub-devices i2c read/write function into one regmap,
> > >
> > > "I2C", "functions", "Regmap".
> > >
> >
> > ACK
> >
> > > > because pmic and ldo part need crc bits for access protection.
> > >
> > > "PMIC", "LDO", "CRC".
> > >
> >
> > ACK
> >
> > > > Signed-off-by: Gene Chen <gene_chen@richtek.com>
> > > > ---
> > > >  drivers/mfd/Kconfig        |   1 +
> > > >  drivers/mfd/mt6360-core.c  | 229 +++++++++++++++++++++++++++++++++++++-----
> > > >  include/linux/mfd/mt6360.h | 240 ---------------------------------------------
> > > >  3 files changed, 204 insertions(+), 266 deletions(-)
> > > >  delete mode 100644 include/linux/mfd/mt6360.h
> > > >
> > > > diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
> > > > index a37d7d1..0684ddc 100644
> > > > --- a/drivers/mfd/Kconfig
> > > > +++ b/drivers/mfd/Kconfig
> > > > @@ -913,6 +913,7 @@ config MFD_MT6360
> > > >       select MFD_CORE
> > > >       select REGMAP_I2C
> > > >       select REGMAP_IRQ
> > > > +     select CRC8
> > > >       depends on I2C
> > > >       help
> > > >         Say Y here to enable MT6360 PMU/PMIC/LDO functional support.
> > > > diff --git a/drivers/mfd/mt6360-core.c b/drivers/mfd/mt6360-core.c
> > > > index 3186a7c..97ef1ad 100644
> > > > --- a/drivers/mfd/mt6360-core.c
> > > > +++ b/drivers/mfd/mt6360-core.c
> > > > @@ -14,7 +14,46 @@
> > > >  #include <linux/interrupt.h>
> > > >  #include <linux/mfd/core.h>
> > > >
> > > > -#include <linux/mfd/mt6360.h>
> > > > +enum {
> > > > +     MT6360_SLAVE_TCPC = 0,
> > > > +     MT6360_SLAVE_PMIC,
> > > > +     MT6360_SLAVE_LDO,
> > > > +     MT6360_SLAVE_PMU,
> > > > +     MT6360_SLAVE_MAX,
> > > > +};
> > > > +
> > > > +struct mt6360_data {
> > > > +     struct i2c_client *i2c[MT6360_SLAVE_MAX];
> > > > +     struct device *dev;
> > > > +     struct regmap *regmap;
> > > > +     struct regmap_irq_chip_data *irq_data;
> > > > +     unsigned int chip_rev;mt6360_data
> > > > +     u8 crc8_tbl[CRC8_TABLE_SIZE];
> > > > +};
> > >
> > > Make sure all of these entries are still used.
> > >
> > > > +#define MT6360_PMU_SLAVEID           0x34
> > > > +#define MT6360_PMIC_SLAVEID          0x1A
> > > > +#define MT6360_LDO_SLAVEID           0x64
> > > > +#define MT6360_TCPC_SLAVEID          0x4E
> > >
> > > Can these be placed into ID order?
> > >
> >
> > ACK
> >
> > > > +#define MT6360_REG_TCPCSTART         0x00
> > > > +#define MT6360_REG_TCPCEND           0xFF
> > > > +#define MT6360_REG_PMICSTART         0x100
> > > > +#define MT6360_REG_PMICEND           0x13B
> > > > +#define MT6360_REG_LDOSTART          0x200
> > > > +#define MT6360_REG_LDOEND            0x21C
> > > > +#define MT6360_REG_PMUSTART          0x300
> > > > +#define MT6360_PMU_DEV_INFO          0x300
> > > > +#define MT6360_PMU_CHG_IRQ1          0x3D0
> > > > +#define MT6360_PMU_CHG_MASK1         0x3F0
> > > > +#define MT6360_REG_PMUEND            0x3FF
> > > > +
> > > > +/* from 0x3D0 to 0x3DF */
> > >
> > > We don't need this in here.
> > >
> >
> > ACK
> >
> > > > +#define MT6360_PMU_IRQ_REGNUM                16
> > > > +
> > > > +#define CHIP_VEN_MASK                0xF0
> > > > +#define CHIP_VEN_MT6360              0x50
> > > > +#define CHIP_REV_MASK                0x0F
> > > >
> > > >  /* reg 0 -> 0 ~ 7 */
> > > >  #define MT6360_CHG_TREG_EVT          4
> > > > @@ -220,12 +259,6 @@ static const struct regmap_irq_chip mt6360_irq_chip = {
> > > >       .use_ack = true,
> > > >  };
> > > >
> > > > -static const struct regmap_config mt6360_pmu_regmap_config = {
> > > > -     .reg_bits = 8,
> > > > -     .val_bits = 8,
> > > > -     .max_register = MT6360_PMU_MAXREG,
> > > > -};
> > > > -
> > > >  static const struct resource mt6360_adc_resources[] = {
> > > >       DEFINE_RES_IRQ_NAMED(MT6360_ADC_DONEI, "adc_donei"),
> > > >  };
> > > > @@ -310,11 +343,153 @@ static int mt6360_check_vendor_info(struct mt6360_data *data)
> > > >       return 0;
> > > >  }
> > > >
> > > > -static const unsigned short mt6360_slave_addr[MT6360_SLAVE_MAX] = {
> > > > -     MT6360_PMU_SLAVEID,
> > > > +static const u16 mt6360_slave_addrs[MT6360_SLAVE_MAX] = {
> > > > +     MT6360_TCPC_SLAVEID,
> > > >       MT6360_PMIC_SLAVEID,
> > > >       MT6360_LDO_SLAVEID,
> > > > -     MT6360_TCPC_SLAVEID,
> > > > +     MT6360_PMU_SLAVEID,
> > > > +};
> > > > +
> > > > +static int mt6360_xlate_pmicldo_addr(u8 *addr, int rw_size)
> > > > +{
> > > > +     u8 flags[4] = { 0x00, 0x40, 0x80, 0xc0 };
> > > > +
> > > > +     if (rw_size < 1 || rw_size > 4)
> > > > +             return -EINVAL;
> > > > +
> > > > +     *addr &= 0x3f;
> > > > +     *addr |= flags[rw_size - 1];
> > > > +
> > > > +     return 0;
> > > > +}
> > >
> > > You need some comments in here to explain what's going on.
> > >
> >
> > ACK
> >
> > Is this comment readable?
> >
> > /*
> >  * When access sud-device PMIC and LDO part which only addressed
> > 0x00~0x3F, read and write action need crc for protection.
> >
> >  * Addr[5:0] is real access real register address.
> >  * Addr[7:6] use to store size, maximum 4 bytes.
> >
> >  * When received the Addr, ic can interpret real register address and size to calculate or check crc
> >  * /
>
> Don't you think this reads better?
>
> No need for comments then:
>
>  #define MT6360_ADDRESS_MASK 0x3f
>  #define MT6360_DATA_SIZE_1_BYTE  0x00
>  #define MT6360_DATA_SIZE_2_BYTES 0x40
>  #define MT6360_DATA_SIZE_3_BYTES 0x80
>  #define MT6360_DATA_SIZE_4_BYTES 0xC0
>
>  static int mt6360_xlate_pmicldo_addr(u8 *addr, int rw_size)
>  {
>         /* Address is already in encoded [5:0] */
>         *addr &= MT6360_ADDRESS_MASK;
>
>         /* Encode size [7:6] */
>         switch (rw_size) {
>         case 1:
>                 *addr |= MT6360_DATA_SIZE_1_BYTE
>                 break;
>         case 2:
>                 *addr |= MT6360_DATA_SIZE_2_BYTES
>                 break;
>         case 3:
>                 *addr |= MT6360_DATA_SIZE_3_BYTES
>                 break;
>         case 4:
>                 *addr |= MT6360_DATA_SIZE_4_BYTES
>                 break;
>         default:
>                 return -EINVAL;
>         }
>
>         return 0;
>  }
>

ACK. Thanks for your suggestions.

> > /*
> >  * CRC calculation
> >  * total size is 2 byte and number of access bytes
> >  * 2 bytes include i2c device address, r/w bit and address which want to access
> >  * others for read or write data
> >  * /
> >
> > > > +static int mt6360_regmap_read(void *context, const void *reg, size_t reg_size,
> > > > +                           void *val, size_t val_size)
> > > > +{
> > > > +     struct mt6360_data *data = context;
> > > > +     u8 bank = *(u8 *)reg, reg_addr = *(u8 *)(reg + 1);
> > > > +     struct i2c_client *i2c = data->i2c[bank];
> > > > +     bool crc_needed = false;
> > > > +     u8 *buf;
> > > > +     /* first two is i2c_addr + reg_addr , last is crc8 */
> > > > +     int alloc_size = 2 + val_size + 1, read_size = val_size;
> > > > +     u8 crc;
> > > > +     int ret;
> > > > +
> > > > +     if (bank == MT6360_SLAVE_PMIC || bank == MT6360_SLAVE_LDO) {
> > > > +             crc_needed = true;
> > > > +             ret = mt6360_xlate_pmicldo_addr(&reg_addr, val_size);
> > > > +             if (ret < 0)
> > > > +                     return ret;
> > > > +             read_size += 1;
> > > > +     }
> > > > +
> > > > +     buf = kzalloc(alloc_size, GFP_KERNEL);
> > > > +     if (!buf)
> > > > +             return -ENOMEM;
> > > > +
> > > > +     /* 7 bit slave addr + read bit */
> > > > +     buf[0] = ((i2c->addr & 0x7f) << 1) + 1;
> > > > +     buf[1] = reg_addr;
> > > > +
> > > > +     ret = i2c_smbus_read_i2c_block_data(i2c, reg_addr, read_size, buf + 2);
> > > > +
> > > > +     if (ret == read_size) {
> > > > +             memcpy(val, buf + 2, val_size);
> > > > +             if (crc_needed) {
> > > > +                     crc = crc8(data->crc8_tbl, buf, val_size + 2, 0);
> > > > +                     if (crc != buf[val_size + 2])
> > > > +                             ret = -EIO;
> > > > +             }
> > > > +     }
> > > > +
> > > > +     kfree(buf);
> > > > +
> > > > +     if (ret < 0)
> > > > +             return ret;
> > > > +     else if (ret != read_size)
> > > > +             return -EIO;
> > > > +
> > > > +     return 0;
> > > > +}
> > > > +
> > > > +static int mt6360_regmap_write(void *context, const void *val, size_t val_size)
> > > > +{
> > > > +     struct mt6360_data *data = context;
> > > > +     u8 bank = *(u8 *)val, reg_addr = *(u8 *)(val + 1);
> > > > +     struct i2c_client *i2c = data->i2c[bank];
> > > > +     bool crc_needed = false;
> > > > +     u8 *buf;
> > > > +     /* first two is i2c_addr + reg_addr , last crc8 + dummy */
> > > > +     int alloc_size = 2 + val_size + 2, write_size = val_size - 2;
> > > > +     int ret;
> > > > +
> > > > +     if (bank == MT6360_SLAVE_PMIC || bank == MT6360_SLAVE_LDO) {
> > > > +             crc_needed = true;
> > > > +             ret = mt6360_xlate_pmicldo_addr(&reg_addr, val_size - 2);
> > > > +             if (ret < 0)
> > > > +                     return ret;
> > > > +     }
> > > > +
> > > > +     buf = kzalloc(alloc_size, GFP_KERNEL);
> > > > +     if (!buf)
> > > > +             return -ENOMEM;
> > > > +
> > > > +     /* 7 bit slave addr + write bit */
> > > > +     buf[0] = ((i2c->addr & 0x7f) << 1);
> > > > +     buf[1] = reg_addr;
> > > > +     /* val need to minus regaddr 16bit */
> > > > +     memcpy(buf + 2, val + 2, write_size);
> > > > +
> > > > +     if (crc_needed) {
> > > > +             buf[val_size] = crc8(data->crc8_tbl, buf, val_size, 0);
> > > > +             write_size += 2;
> > > > +     }
> > > > +
> > > > +     ret = i2c_smbus_write_i2c_block_data(i2c,
> > > > +                                          reg_addr, write_size, buf + 2);
> > > > +
> > > > +     kfree(buf);
> > > > +
> > > > +     if (ret < 0)
> > > > +             return ret;
> > > > +
> > > > +     return 0;
> > > > +}
> > > > +
> > > > +static const struct regmap_bus mt6360_regmap_bus = {
> > > > +     .read           = mt6360_regmap_read,
> > > > +     .write          = mt6360_regmap_write,
> > > > +
> > > > +     /* due to pmic and ldo crc access size limit */
> > > > +     .max_raw_read   = 4,
> > > > +     .max_raw_write  = 4,
> > > > +};
> > >
> > > Why isn't all of the above in a Regmap driver?
> > >
> >
> > Do you means split out like drivers/base/regmap/regmap-ac97.c?
>
> Yes, I do.
>
> [...]
>

ACK

> > > > +     i2c_set_clientdata(client, data);
> > >
> > > Where is this used?
> >
> > I can use device to get chip_rev from dev_get_drvdata.
> > According to different chip_rev, I may need apply different way to do.
> >
> > > Didn't you just move the definition into this file?
> >
> > ACK, I will seperate move definition into this file to new patch
>
> That's not the point I'm making.
>
> You can't use 'data' outside of this file, so why are you setting it
> inside the clientdata area?
>

I see. It's my logical defect
I will remove set clientdata.

> --
> Lee Jones [李琼斯]
> Senior Technical Lead - Developer Services
> Linaro.org │ Open source software for Arm SoCs
> Follow Linaro: Facebook | Twitter | Blog