FPC402: Does FPC402 can support or compatible with I3C bus?

Hi team, 

Add more info.  

In our case, we have a Designware I3C controller (master) and in this case there's only FPC402 on the I3C bus.

The OS is Linux. The FPC402 is defined in the DTS as static legacy (i2c) device (slave) with address 0xf (e.g. reg = <0x0f 0x00 0x30>;)

As a test, we've tried different I3C DTS configurations, including telling the I3C controller that the I2C device has/has no spike filter etc. but they all fail.

The I3C controller is used with other I2C devices and I3C devices, too, and it works OK.

The I3C controller:

  * Issues RSTDAA successfully

  * Issues DISEC successfully

  * Does DAA successfully

No I3C devices found, as expected.

After each of the above steps, the "response queue" of the I3C controller gets cleared.

However, the next step then is to communicate with the only I2C device on the bus i.e. FPC402.

Here, the first message is to send the address to tell FPC402 which address it should use.

The FPC402 is in 7-bit address 0xf and we tell it to use 7-bit 0xf -- this is just a test, we want to see a successful I2C message.

However:

  * When trying to send an I2C message to FPC402, the I3C controller fails with error 5

Here error 5 is not Linux errno, it's internal controller error from HW level, and means "Address NACK'd. This bit is set in case the Slave NACK's for Dynamic Address Assignment during ENTDAA process".

It's strange that when sending I2C messages the "Address NACKd" error case is triggered, especially since the response queue was cleared immediately prior.

My questions:

  * Is FPC402 somehow incompatible with I3C?

  * The spike filter of FPC402 seems to "filter under 30 nsec and pass over 50 nsec" but what happens between 30..50 nsec?

  * Is using 0xf as the unique address valid? (I think it is?)

FYI the documentation I have is SNLU227/JUNE2017 and SNLS582/MAY2017.

Thanks. 

B.R.

Lucas