IMX6ULL I2C

[Lizh]

hello：

  When I was using imx6ull's I2C feature recently, I noticed a strange problem with I2C continuous writing: When IMX6ULL continuously writes data to the slave machine, IMX6ULL waits for more than 20 us after each byte is written, and then writes another byte. I recently in the use of imx6ull's I2C function, I found that I2C continuous write has a very strange problem: when IMX6ULL continuous to write data to the slave machine, IMX6ULL every write a byte will wait for more than 20 us, and then write a byte. My request is: is there a way to put this more than 20 us without waiting, have been continuously written. Is that OK? Wait, keep writing. Is that OK?

[Chavira]

Hi @Lizh!

Thank you for contacting NXP Support!

Maybe the problem is in your code.

How you are writing on the i2c bus?

I recommend trying with i2c-tools

Best Regards!

Chavira

[Lizh]

int cs_i2c_byte_read(struct haptic *cs_haptic, uint8_t reg_addr, uint8_t * buf, uint32_t len)

{

int ret = -1;

int i = CS_RETRY_TIMES;

struct i2c_msg msg[2];

struct i2c_adapter *adapter = cs_haptic->i2c->adapter;

if (buf == NULL || len == 0) {

log_err(" input param is invalid");

return -1;

}

/*send reg addr*/

msg[0].addr = cs_haptic->i2c->addr;

msg[0].flags = 0;

msg[0].buf = &reg_addr;

msg[0].len = 1;

/*read reg*/

msg[1].addr = cs_haptic->i2c->addr;

msg[1].flags = I2C_M_RD;

msg[1].buf = buf;

msg[1].len = len;

mutex_lock(&i2c_rw_lock);

do {

ret = i2c_transfer(adapter, msg, sizeof(msg) / sizeof(struct i2c_msg));

if (ret < 0)

log_err("read failed: %d", ret);

else

break;

i--;

} while (i > 0);

mutex_unlock(&i2c_rw_lock);

return ret;

}

int cs_i2c_byte_write(struct haptic *cs_haptic, uint8_t reg_addr, uint8_t * buf, uint32_t len)

{

int32_t ret = 0;

uint8_t *data = NULL;

int i = CS_RETRY_TIMES;

struct i2c_msg msg;

struct i2c_adapter *adapter = cs_haptic->i2c->adapter;

if (buf == NULL || len == 0) {

log_err("input param is invalid");

return -1;

}

data = kmalloc(len + 1, GFP_KERNEL);

if (data == NULL) {

return -ENOMEM;

}

data[0] = reg_addr;

memcpy(&data[1], buf, len);

msg.addr = cs_haptic->i2c->addr;

msg.flags = 0;

msg.buf = data;

msg.len = len + 1;

mutex_lock(&i2c_rw_lock);

do {

ret = i2c_transfer(adapter, &msg, 1);

if (ret < 0)

log_err("write failed: %d", ret);

else

break;

i--;

} while (i > 0);

mutex_unlock(&i2c_rw_lock);

kfree(data);

return ret;

}

int cs_i2c_one_byte_write(struct haptic *cs_haptic, uint8_t reg_addr, uint8_t data)

{

uint8_t temp = 0;

temp = data;

return cs_i2c_byte_write(cs_haptic,reg_addr, &temp, 0x01);

}

int cs_write_sram(struct haptic *cs_haptic, uint8_t reg_addr, uint8_t *buf, uint32_t len){

uint8_t *data = NULL;

int ret = -1;

if (buf == NULL || len == 0) {

log_err("input param is invalid");

return -1;

}

data = kmalloc(len + 1, GFP_KERNEL);

data[0] = reg_addr;

memcpy(&data[1], buf, len);

ret = i2c_master_send(cs_haptic->i2c, data, len + 1);

if (ret < 0)

log_err("i2c write sram err:%d",ret);

kfree(data);

return ret;

}

The i2c_transfer function is actually implemented by your I2C controller driver，It's not my code. I'm calling your underlying interface. The i2c_transfer function actually calls your driver function i2c_imx_xfer, so this should not be my code problem, please help to look at it, thank you.