[U-Boot] [EXT] Re: [PATCH] i2c: pcf2127: fix bug that read wrong time
Chuanhua Han
chuanhua.han at nxp.com
Wed May 22 12:45:59 UTC 2019
> -----Original Message-----
> From: Lukasz Majewski <lukma at denx.de>
> Sent: 2019年5月22日 19:32
> To: Chuanhua Han <chuanhua.han at nxp.com>
> Cc: hs at denx.de; u-boot at lists.denx.de; Biwen Li <biwen.li at nxp.com>;
> sjg at chromium.org; Stefano Babic <sbabic at denx.de>
> Subject: Re: [EXT] Re: [PATCH] i2c: pcf2127: fix bug that read wrong time
>
> On Wed, 22 May 2019 09:31:35 +0000
> Chuanhua Han <chuanhua.han at nxp.com> wrote:
>
> > > -----Original Message-----
> > > From: Lukasz Majewski <lukma at denx.de>
> > > Sent: 2019年5月22日 16:41
> > > To: Chuanhua Han <chuanhua.han at nxp.com>
> > > Cc: hs at denx.de; u-boot at lists.denx.de; Biwen Li <biwen.li at nxp.com>;
> > > sjg at chromium.org; Stefano Babic <sbabic at denx.de>
> > > Subject: Re: [EXT] Re: [PATCH] i2c: pcf2127: fix bug that read wrong
> > > time
> > >
> > > Hi Chuanhua,
> > >
> > > > > -----Original Message-----
> > > > > From: Lukasz Majewski <lukma at denx.de>
> > > > > Sent: 2019年5月22日 15:16
> > > > > To: Chuanhua Han <chuanhua.han at nxp.com>
> > > > > Cc: hs at denx.de; u-boot at lists.denx.de; Biwen Li
> > > > > <biwen.li at nxp.com>; sjg at chromium.org
> > > > > Subject: [EXT] Re: [PATCH] i2c: pcf2127: fix bug that read wrong
> > > > > time
> > > > >
> > > > > Hi Chuanhua,
> > > > >
> > > > > > Because i2c driver does not generate a stop bit when reading
> > > > > > registers of pcf2127
> > > > >
> > > > > The change (in the common i2c code) is rather large when
> > > > > considering the description above.
> > > > >
> > > > > Could you write a more detailed patch description? Is this only
> > > > > the problem with i2c in DM?
> > > > OK, This is a problem with the i2c transport framework. After
> > > > writing the register address that you want to read, the stop
> > > > signal is missing before reading the register data.
> > > > >
> > > > > Is the same code (as introduced in this commit) available in
> > > > > Linux kernel?
> > > > The kernel does not have such a problem
> > >
> > > Do you know maybe why such issue is not present on Linux kernel?
> > The kernel code is encapsulated when reading the pcf2127 register,
> > which separates the read and write, thus generating a stop signal
> > before reading the register. Eg: Here is the wrapper made by the
> > kernel code: static int pcf2127_i2c_read(void *context, const void
> > *reg, size_t reg_size, void *val, size_t val_size) {
> > struct device *dev = context;
> > struct i2c_client *client = to_i2c_client(dev);
> > int ret;
> >
> > if (WARN_ON(reg_size != 1))
> > return -EINVAL;
> >
> > ret = i2c_master_send(client, reg, 1);
> > if (ret != 1)
> > return ret < 0 ? ret : -EIO;
> >
> > ret = i2c_master_recv(client, val, val_size);
> > if (ret != val_size)
> > return ret < 0 ? ret : -EIO;
> >
> > return 0;
> > }
>
> That was my point - the same shall be done in the pcf2127 driver. This is a slow
> device, so we can afford for this.
There is no similar api in the uboot code to split the read data into the address of the send register and receive the data.
Uboot encapsulates the transmit register address and the read data, so no stop signal is generated.
>
> > >
> > > > > How the error is reproduced? What are the symptoms of it?
> > > > You can use the i2c command to read the register data of pcf2127.
> > >
> > > So the problem is with using "i2c ..." commands from U-Boot prompt?
> > Yes,but adding debugging to the rtc driver is also the same problem
> > >
> > > > You will find that you want to read the address 0 of the register,
> > > > but the data of the 1 address is read, and the data read later
> > > > will be offset.
> > >
> > > As fair as I can tell the pcf2127 has its own DM aware driver at
> > > driver/rtc/pcf2127.c.
> > >
> > > Is this driver broken so you need to modify the generic i.MX I2C
> > > code? Have you tried to test this driver on your setup?
> > Pcf2127 driver also has problems, has been modified, need i2c general
> > code to support
>
> Just one remark the mxc_i2c.c is IMX specific I2C code. Not the generic one.
ok
>
> Moreover, it looks like the same approach (first send, then read) is performed
> in the pcf2127 driver at pcf2127_rtc_get() function.
>
> I think that the driver shall be first thoroughly checked, then fixes shall be
> added to it.
>
> > >
> > > > >
> > > > > In which way the pcf2127 needs special treatment (non I2C
> > > > > compatible)? How does it differ from other RTC devices?
> > > > According to the pcf2127 chip manual, when you need to read the
> > > > internal register data, you need to write the register address
> > > > first and then need stop bit(it is necessary).
> > > > >
> > > > > >
> > > > > > Signed-off-by: Biwen Li <biwen.li at nxp.com>
> > > > > > Signed-off-by: Chuanhua Han <chuanhua.han at nxp.com>
> > > > > > ---
> > > > > > drivers/i2c/i2c-uclass.c | 2 ++
> > > > > > drivers/i2c/mxc_i2c.c | 70
> > > > > > +++++++++++++++++++++++++++++++++++-----
> drivers/rtc/pcf2127.c
> > > > > > | 18 ++++++----- include/i2c.h | 2 ++
> > > > > > 4 files changed, 76 insertions(+), 16 deletions(-)
> > > > > >
> > > > > > diff --git a/drivers/i2c/i2c-uclass.c
> > > > > > b/drivers/i2c/i2c-uclass.c index e47abf1833..18f7364d72 100644
> > > > > > --- a/drivers/i2c/i2c-uclass.c
> > > > > > +++ b/drivers/i2c/i2c-uclass.c
> > > > > > @@ -141,6 +141,8 @@ int dm_i2c_read(struct udevice *dev, uint
> > > > > > offset, uint8_t *buffer, int len) if (len) {
> > > > > > ptr->addr = chip->chip_addr;
> > > > > > ptr->flags = chip->flags &
> > > > > > DM_I2C_CHIP_10BIT ? I2C_M_TEN : 0;
> > > > > > + ptr->flags |= chip->flags &
> > > > > > DM_I2C_CHIP_RD_NEED_STOP_BIT ?
> > > > > > + I2C_M_RD_NEED_STOP_BIT : 0;
> > > > > > ptr->flags |= I2C_M_RD;
> > > > > > ptr->len = len;
> > > > > > ptr->buf = buffer;
> > > > > > diff --git a/drivers/i2c/mxc_i2c.c b/drivers/i2c/mxc_i2c.c
> > > > > > index 73b9807598..241367bac8 100644
> > > > > > --- a/drivers/i2c/mxc_i2c.c
> > > > > > +++ b/drivers/i2c/mxc_i2c.c
> > > > > > @@ -962,6 +962,8 @@ static int mxc_i2c_xfer(struct udevice
> > > > > > *bus, struct i2c_msg *msg, int nmsgs) int reg_shift =
> > > > > > i2c_bus->driver_data & I2C_QUIRK_FLAG ? VF610_I2C_REGSHIFT :
> > > > > > IMX_I2C_REGSHIFT; int read_mode;
> > > > > > + bool quirk = i2c_bus->driver_data & I2C_QUIRK_FLAG ?
> > > > > > true : false;
> > > > > > + unsigned int temp;
> > > > > >
> > > > > > /* Here address len is set to -1 to not send any address at
> > > > > > first.
> > > > > > * Otherwise i2c_init_transfer will send the chip address
> > > > > > with write @@ -976,6 +978,7 @@ static int mxc_i2c_xfer(struct
> > > > > > udevice *bus, struct i2c_msg *msg, int
> > > > > > nmsgs) read_mode = -1; /* So it's always different on the
> > > > > > first message */ for (; nmsgs > 0; nmsgs--, msg++) { const int
> > > > > > msg_is_read = !!(msg->flags & I2C_M_RD);
> > > > > > + bool next_is_read = nmsgs > 1 &&
> > > > > > (msg[1].flags & I2C_M_RD);
> > > > > > debug("i2c_xfer: chip=0x%x, len=0x%x, dir=%c\n",
> msg->addr,
> > > > > > msg->len, msg_is_read ? 'R' : 'W'); @@ -983,13 +986,16 @@
> > > > > > static int mxc_i2c_xfer(struct udevice *bus, struct i2c_msg
> > > > > > *msg, int nmsgs) if (msg_is_read !=
> > > > > > read_mode) { /* Send repeated start if not 1st message */ if
> > > > > > (read_mode != -1) {
> > > > > > - debug("i2c_xfer:
> > > > > > [RSTART]\n");
> > > > > > - ret = readb(base + (I2CR <<
> > > > > > reg_shift));
> > > > > > - ret |= I2CR_RSTA;
> > > > > > - writeb(ret, base + (I2CR <<
> > > > > > reg_shift));
> > > > > > + if (!(msg[1].flags &
> > > > > > I2C_M_RD_NEED_STOP_BIT)) {
> > > > > > + debug("i2c_xfer:
> > > > > > [RSTART]\n");
> > > > > > + ret = readb(base +
> > > > > > (I2CR << reg_shift));
> > > > > > + ret |= I2CR_RSTA;
> > > > > > + writeb(ret, base +
> > > > > > (I2CR << reg_shift));
> > > > > > + }
> > > > > > }
> > > > > > debug("i2c_xfer: [ADDR %02x | %c]\n",
> > > > > > msg->addr, msg_is_read ? 'R' : 'W');
> > > > > > +
> > > > > > ret = tx_byte(i2c_bus, (msg->addr <<
> > > > > > 1) | msg_is_read); if (ret < 0) {
> > > > > > debug("i2c_xfer: [STOP]\n"); @@ -999,16
> +1005,64 @@
> > > > > > static int mxc_i2c_xfer(struct udevice *bus, struct i2c_msg
> > > > > > *msg, int nmsgs) read_mode = msg_is_read;
> > > > > > }
> > > > > >
> > > > > > - if (msg->flags & I2C_M_RD)
> > > > > > + if (msg->flags & I2C_M_RD) {
> > > > > > ret = i2c_read_data(i2c_bus,
> > > > > > msg->addr, msg->buf, msg->len, nmsgs == 1 ||
> > > > > > (msg->flags
> > > > > > & I2C_M_STOP));
> > > > > > - else
> > > > > > - ret = i2c_write_data(i2c_bus,
> > > > > > msg->addr, msg->buf,
> > > > > > - msg->len);
> > > > > > + if (ret < 0)
> > > > > > + break;
> > > > > > + continue;
> > > > > > + }
> > > > > >
> > > > > > + /* Write message */
> > > > > > + ret = i2c_write_data(i2c_bus, msg->addr,
> > > > > > msg->buf,
> > > > > > + msg->len);
> > > > > > if (ret < 0)
> > > > > > break;
> > > > > > +
> > > > > > + if (!next_is_read)
> > > > > > + continue;
> > > > > > +
> > > > > > + /* Read message following write message */
> > > > > > + if (msg[1].flags & I2C_M_RD_NEED_STOP_BIT) {
> > > > > > + /* Generate a stop bit */
> > > > > > + i2c_imx_stop(i2c_bus);
> > > > > > + /* Reset i2c slave */
> > > > > > + i2c_force_reset_slave();
> > > > > > +
> > > > > > + /* Enable I2C controller */
> > > > > > + if (quirk)
> > > > > > + ret = readb(base + (I2CR <<
> > > > > > reg_shift))
> > > > > > + & I2CR_IDIS;
> > > > > > + else
> > > > > > + ret = !(readb(base + (I2CR <<
> > > > > > reg_shift))
> > > > > > + & I2CR_IEN);
> > > > > > + if (ret) {
> > > > > > + writeb(I2CR_IEN, base +
> > > > > > (I2CR << reg_shift));
> > > > > > + /* Wait for controller to be
> > > > > > stable */
> > > > > > + udelay(50);
> > > > > > + }
> > > > > > +
> > > > > > + /* Clear interrupt bit */
> > > > > > + writeb(I2SR_IIF_CLEAR, base + (I2SR
> > > > > > << reg_shift));
> > > > > > + ret = wait_for_sr_state(i2c_bus,
> > > > > > ST_BUS_IDLE);
> > > > > > + if (ret < 0)
> > > > > > + return ret;
> > > > > > +
> > > > > > + /* Start I2C transaction */
> > > > > > + temp = readb(base + (I2CR <<
> > > > > > reg_shift));
> > > > > > + temp |= I2CR_MSTA;
> > > > > > + writeb(temp, base + (I2CR <<
> > > > > > reg_shift)); +
> > > > > > + ret = wait_for_sr_state(i2c_bus,
> > > > > > ST_BUS_BUSY);
> > > > > > + if (ret < 0)
> > > > > > + return ret;
> > > > > > +
> > > > > > + /* Enter transfer mode */
> > > > > > + temp |= I2CR_MTX | I2CR_TX_NO_AK;
> > > > > > + writeb(temp, base + (I2CR <<
> > > > > > reg_shift));
> > > > > > + udelay(50);
> > > > > > + }
> > > > > > }
> > > > > >
> > > > > > if (ret)
> > > > > > diff --git a/drivers/rtc/pcf2127.c b/drivers/rtc/pcf2127.c
> > > > > > index dcf0340b4d..010c45ecbf 100644
> > > > > > --- a/drivers/rtc/pcf2127.c
> > > > > > +++ b/drivers/rtc/pcf2127.c
> > > > > > @@ -24,12 +24,9 @@
> > > > > >
> > > > > > static int pcf2127_rtc_set(struct udevice *dev, const struct
> > > > > > rtc_time *tm) {
> > > > > > - uchar buf[8];
> > > > > > + uchar buf[7] = {0};
> > > > > > int i = 0, ret;
> > > > > >
> > > > > > - /* start register address */
> > > > > > - buf[i++] = PCF2127_REG_SC;
> > > > > > -
> > > > > > /* hours, minutes and seconds */
> > > > > > buf[i++] = bin2bcd(tm->tm_sec);
> > > > > > buf[i++] = bin2bcd(tm->tm_min); @@ -44,7 +41,7 @@ static int
> > > > > > pcf2127_rtc_set(struct udevice *dev, const struct rtc_time
> > > > > > *tm) buf[i++] = bin2bcd(tm->tm_year % 100);
> > > > > >
> > > > > > /* write register's data */
> > > > > > - ret = dm_i2c_write(dev, PCF2127_REG_CTRL1, buf,
> > > > > > sizeof(buf));
> > > > > > + ret = dm_i2c_write(dev, PCF2127_REG_SC, buf, i);
> > > > > >
> > > > > > return ret;
> > > > > > }
> > > > > > @@ -54,9 +51,6 @@ static int pcf2127_rtc_get(struct udevice
> > > > > > *dev, struct rtc_time *tm) int ret = 0;
> > > > > > uchar buf[10] = { PCF2127_REG_CTRL1 };
> > > > > >
> > > > > > - ret = dm_i2c_write(dev, PCF2127_REG_CTRL1, buf, 1);
> > > > > > - if (ret < 0)
> > > > > > - return ret;
> > > > > > ret = dm_i2c_read(dev, PCF2127_REG_CTRL1, buf, sizeof(buf));
> > > > > > if (ret < 0)
> > > > > > return ret;
> > > > > > @@ -90,6 +84,13 @@ static int pcf2127_rtc_reset(struct udevice
> > > > > > *dev) return 0;
> > > > > > }
> > > > > >
> > > > > > +static int pcf2127_probe(struct udevice *dev) {
> > > > > > + i2c_set_chip_flags(dev,
> > > > > > DM_I2C_CHIP_RD_NEED_STOP_BIT); +
> > > > > > + return 0;
> > > > > > +}
> > > > > > +
> > > > > > static const struct rtc_ops pcf2127_rtc_ops = {
> > > > > > .get = pcf2127_rtc_get,
> > > > > > .set = pcf2127_rtc_set,
> > > > > > @@ -104,6 +105,7 @@ static const struct udevice_id
> > > > > > pcf2127_rtc_ids[] = { U_BOOT_DRIVER(rtc_pcf2127) = {
> > > > > > .name = "rtc-pcf2127",
> > > > > > .id = UCLASS_RTC,
> > > > > > + .probe = pcf2127_probe,
> > > > > > .of_match = pcf2127_rtc_ids,
> > > > > > .ops = &pcf2127_rtc_ops,
> > > > > > };
> > > > > > diff --git a/include/i2c.h b/include/i2c.h index
> > > > > > a5c760c711..beaa028349 100644
> > > > > > --- a/include/i2c.h
> > > > > > +++ b/include/i2c.h
> > > > > > @@ -28,6 +28,7 @@ enum dm_i2c_chip_flags {
> > > > > > DM_I2C_CHIP_10BIT = 1 << 0, /* Use 10-bit
> > > > > > addressing */ DM_I2C_CHIP_RD_ADDRESS = 1 << 1, /* Send
> > > > > > address for each read byte */ DM_I2C_CHIP_WR_ADDRESS =
> > > > > > 1 << 2, /* Send address for each write byte */
> > > > > > + DM_I2C_CHIP_RD_NEED_STOP_BIT = 1 << 3, /* Need
> > > > > > generate stop bit */ };
> > > > > >
> > > > > > struct udevice;
> > > > > > @@ -87,6 +88,7 @@ enum dm_i2c_msg_flags {
> > > > > > I2C_M_IGNORE_NAK = 0x1000, /* continue after
> > > > > > NAK */ I2C_M_NO_RD_ACK = 0x0800, /* skip the
> > > > > > Ack bit on reads */ I2C_M_RECV_LEN =
> > > > > > 0x0400, /* length is first received byte */
> > > > > > + I2C_M_RD_NEED_STOP_BIT = 0x0002, /* need generate
> > > > > > stop bit */ };
> > > > > >
> > > > > > /**
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Best regards,
> > > > >
> > > > > Lukasz Majewski
> > > > >
> > > > > --
> > > > >
> > > > > DENX Software Engineering GmbH, Managing Director: Wolfgang
> > > > > Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194
> > > > > Groebenzell, Germany Phone: (+49)-8142-66989-59 Fax:
> > > > > (+49)-8142-66989-80 Email: lukma at denx.de
> > >
> > >
> > >
> > >
> > > Best regards,
> > >
> > > Lukasz Majewski
> > >
> > > --
> > >
> > > DENX Software Engineering GmbH, Managing Director: Wolfgang
> > > Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell,
> > > Germany Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email:
> > > lukma at denx.de
>
>
>
>
> Best regards,
>
> Lukasz Majewski
>
> --
>
> DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
> HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
> Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email:
> lukma at denx.de
More information about the U-Boot
mailing list