TMS320C6678: I2C transfer blocks on a semaphore pend

Hi Fabrizio,

The team is notified. They will post their feedback directly here.

BR
Tsvetolin Shulev

Fabrizio,

Your problem seems to be similar to this issue reported in MArch 2017:
e2e.ti.com/.../585284

Can you check if the fix discussed in that E2E thread resolves your issue.

Regards,
Rahul

Hi,

It seems that the code in pdk_2_0_6 is already fixed and different from the 2.0.3 version.

in this, it looks like the timeout fix is already present:

           if (object->operMode == I2C_OPER_MODE_BLOCKING) {
                I2C_drv_log1("\n I2C:(0x%x) Pending on transferComplete semaphore \n",
                    hwAttrs->baseAddr);

                /*
                 * Wait for the transfer to complete here.
                 * It's OK to block from here because the I2C's Hwi will unblock
                 * upon errors
                 */
                semStatus = I2C_osalPendLock(object->transferComplete, transaction->timeout);
				/* If interrupt is enabled and no specific call back routine is supplied by
				   the application, the error status is stored in object->intStatusErr.
				   Read it and return the error status accordingly */
		        if (object->intStatusErr & I2C_INT_NO_ACK)
                {
                   retVal = I2C_STS_ERR_NO_ACK;
		        } else if (object->intStatusErr & I2C_INT_ARBITRATION_LOST)
                { 
                   retVal = I2C_STS_ERR_ARBITRATION_LOST;
                } else {
                   retVal = I2C_STS_SUCCESS;
		        }
				
				if (semStatus == SemaphoreP_TIMEOUT) {
                 retVal = (int16_t)I2C_STS_ERR;
				}
				else{
					I2C_drv_log1("\n I2C:(0x%x) Transaction completed \n", hwAttrs->baseAddr);

					/* Hwi handle has posted a 'transferComplete' check for Errors */
					if (object->mode == I2C_IDLE_MODE) {
						I2C_drv_log1("\n I2C:(0x%x) Transfer OK \n", hwAttrs->baseAddr);
					}
				}
            }

From what I can understand, in the link you kindly shared the error was generated when giving an incorrect slave address.

In my case the address should be correct (0x50) and taken straight from the I2C example define for the I2C EEPROM address:

    I2C_init();

    I2C_Params_init(&i2cParams);
//    i2cParams.transferMode = I2C_MODE_CALLBACK;

    handle = I2C_open(I2C_EEPROM_INSTANCE, &i2cParams);

    I2C_transactionInit(&i2cTransaction);
    i2cTransaction.slaveAddress = I2C_EEPROM_ADDR; //0x50
    i2cTransaction.writeBuf = (uint8_t *)&txBuf[0];
    i2cTransaction.writeCount = I2C_EEPROM_ADDR_SIZE; // 2 bytes
    i2cTransaction.readBuf = (uint8_t *)&rxBuf[0];
    i2cTransaction.readCount = I2C_EEPROM_TEST_LENGTH; // 10 bytes
    i2cTransaction.timeout   = I2C_TRANSACTION_TIMEOUT; // 1000U
    transferStatus = I2C_transfer(handle, &i2cTransaction);

do you think is it the same issue?

Thank you very much for your answer.

Best Regards,

Fabrizio

Here is a snapshot of the I2C registers

any idea of what it can be happened?

Thanks,

Fabrizio

Hi,

any news on this? It would be pretty urgent, please.

Thank you in advance,
Best regards

Fabrizio