HID and NUS

RE: HID and NUS

Laurent Bourassa — Thu, 14 Mar 2019 16:57:56 GMT

Thank you so much for your help. I initiated the NUS service before advertising it, as you said. I had another problem that followed where I did'nt have enough RAM. I was able to change it and now I have successfully implemented NUS with HID. Again, thank you.

RE: HID and NUS

Einar Thorsrud — Thu, 14 Mar 2019 11:33:29 GMT

I see. Then I assume the problem is that the SoftDevice has not been configured with the custom base UUID. This is done by ble_nus_init(), so this has to be called before you configure the advertisement packet. You can see the same issue with the NUS example from the SDK if you change the order of the calls to services_init() and advertising_init().

RE: HID and NUS

Laurent Bourassa — Mon, 11 Mar 2019 21:34:35 GMT

Ok, so I went further with the debugger. Inside uuid_list_encode(), there is two calls for the function uuid_list_sized_encode(). In that function, there is a call to another function, sd_ble_uuid_encode() that I'm pretty positive is the source of my problem. Looking in that last function, you see that it returns NRF_ERROR_INVALID_PARAM when the UUID type is invalid.

You can see that the type for NUS is 0x02 (BLE_UUID_TYPE_VENDOR_BEGIN). Somehow it doesn't seem to recognize 0x02 as a valid type. I changed it to 0x01 (BLE_UUID_TYPE_BLE) with the debugger and the error was gone, but of course, I know 0x02 should work, normally.

Now I need help resolving why it doesn't accept 0x02 as a valid type...

RE: HID and NUS

Laurent Bourassa — Sun, 10 Mar 2019 17:27:46 GMT

Hello Einar, yes, that seems to be the function that returns the error. More specifically, it's the If Statement that encodes the complete UUID list.

This statement calls the function uuid_list_encode() so I went further to see what could be the problem. This function is in two parts, one for encoding 16 bits UUIDs and the other for encoding 128 bits UUIDs.

When encoding my first service, HID, the debugger will enter this function, execute the 16 bits and 128 bits sections with no failure. However, when encoding the NUS service, it enters the same function but fails at the 16 bits encoding and skips over the 128 bits encoding, resulting in the error code 0x07.

Is there a way to specify which encoding to use for NUS without impacting the encoding for HID?

RE: HID and NUS

Einar Thorsrud — Wed, 06 Mar 2019 10:20:52 GMT

OK, so you get NRF_ERROR_INVALID_PARAM returned from ble_advertising_init()? Which SDK are you using?

Assuming SDK 15.2, this means that the error probably comes from ble_advdata_encode(). That function is responsible for encoding the advertisement packet and is quite easy to understand. Essentially there is an if statement for each field, so you can check with a debugger which fails (which is invalid).

RE: HID and NUS

Laurent Bourassa — Tue, 05 Mar 2019 15:16:43 GMT

Hello, yes it is still the same error and it is the same function as before that returns it.

I have shortened the advertising packet as much as I think I could :

No device name = 0 byte

-BLE_GAP_ADV_FLAGS_LE_ONLY_GENERAL_DISC_MODE flag = 3 bytes(?)

-HID = 4 bytes (Altough I have two reports for this profile, could it make it longer? Let's assume 8)

-BAS = Not sure about this one, I did not find where it says it is advertised, but I would assume it takes 4 more bytes...

-NUS = For the sake of testing, I have reduced BLE_NUS_MAX_DATA_LEN to 1. I don't know if the fact that there is both RX and TX extends the space taken by the NUS, I assume it does. For my application I don't need TX. (data(1) + manufacturer ID(2) + length(1) + type(2?) = 6 bytes?)

In total I think that would make 17-21 bytes, no? I must be wrong because if so it should work if it's under 31 bytes...

RE: HID and NUS

Einar Thorsrud — Tue, 05 Mar 2019 14:12:33 GMT

Hi,

Which function returns the error code now? and is it still NRF_ERROR_INVALID_PARAM?

RE: HID and NUS

Laurent Bourassa — Tue, 05 Mar 2019 02:26:46 GMT

Sorry, I see now what you mean. I have found out here what is the advertising packet. I know how to reduce the size taken by my device name, but it's not enough. I would need to reduce the data field of the NUS, but I'm not sure where to edit that. Probably in ble_nus.h. I will continue to look.

RE: HID and NUS

Laurent Bourassa — Tue, 05 Mar 2019 01:57:00 GMT

Hello, the error code is 0x07, which if I get right is NRF_ERROR_INVALID_PARAM. If I am not mistaken, since the NUS is a vendor defined UUID, it has a length of 128 bits, while the HID has 16 bits, right? I'm not quite sure how I am supposed to merge these two.

I attempted to replicate the ble_app_uart example where it uses a NUS profile, and temporarily commented out the HID part to see if I could make it work alone, but with no success. When you say "Too much in the advertising packet", you're talking about the advertising uuid list, correct?

RE: HID and NUS

RK — Sun, 03 Mar 2019 22:50:15 GMT

well you could just debug through to find out where the error is returned or look at the error code to see which one it is. Most likely you've put too much in the advertising packet and overfilled it so you're getting an error that there's no space.