Layerscape LS1046 FRWY Board GPIO Signals

[matthej]

Hi,

I purchased the LS1046 FRWY board and I am trying to use the GPIO signals. It looks like I have been able to successfully enable them since I am seeing them on the J67 GPIO header. The problem I am having is the mapping of the GPIO signals to the J67 header.

Here is the mapping from the Reference Board Manual.

The problem is that when I write the data register to set a particular GPIO, it seems to be coming out on the wrong pin. For example:

I can write 0x200 to the GPIO3 data register (GPDAT), I then see the signal  appear on J67-11 which corresponds to GPIO3_7. I expect that by writing a 0x200 to the GPDAT register, I would see the signal on  GPIO3_9 or J67-19.

This happens for multiple GPIOs. Is the mapping to the J67 incorrect? Am I not writing to the GDAT register correctly?

Thanks!

[Bulat]

First of all, your understanding of GPDAT bit ordering is not correct. Take a look at the description in the manual, bit ordering is GPDAT[0:31], not GPDAT[31:0]. So to set bit 9 you need to write value 0x00400000. In turn, GPIO block is big-endian, accessing from little-endian ARM core requires to do byte swapping. So actually you need to write 0x4000 to set bit #9. Please try.

I would add that results of your test are really strange. Writing 0x200 to GPDAT should set bit #14, not #7. GPIO3 bit #14 is connected to pin 12 of the header, that is close to pin 11 but  in different row. Please check once again.

[matthej]

Thanks for the response! That helps a lot. The pinout of the J67 seems to be contradictory between the spec and the silkscreen on our board.

I have attached the silkscreen photo which is backwards from the pinout specified in above.

Thanks!

[Bulat]

Can you clarify what is "contradictory between the spec and the silkscreen"? The spec does not illustrate pin topology of the header, simply lists connected signals.

[matthej]

I agree that there is no topology listed in the spec, but it says for example:

GPIO3_6 goes to J67-1.

But for the board I have it actually goes to J67-2 according to the way it is labeled on my board.

My board has silkscreen labeled in this way:

1 2

3  4

5  6

7   8

etc.

But I believe the silkscreen is backwards for this board rev.

Thanks

[Bulat · Accepted solution]

Ok, now I see the problem. In fact the board's silkscreen is correct, while description in the manual is wrong. Attached  schematics with actual J67 routing, please use it as a reference.

The root of the problem is hystorical, J67 description in the manual corresponds to initial board revision that has been changed later. As a result schematics and layout around J67 were changed, while the table in the manual was somehow not fixed.

[matthej]

Thanks!

Will the manual be updated such that this doesn't affect other people as well?

Just curious

[Bulat]

I will notify documentation team, hopefully they will fix the mistake soon.