STM BlueNRG-MS Max Characteristic length limited to 255 due to aci_gatt_add_char function limitations.

[aj.sk]

My general understanding is that the BLE standard supports characteristics with a lenght of up to 512 bytes. I'm using the BlueNRG-MS chip from STM but there the function call to add a characteristic has a uint8_t value for the length parameter

tBleStatus aci_gatt_add_char(uint16_t serviceHandle,

uint8_t charUuidType,

const uint8_t* charUuid,

uint8_t charValueLen,

uint8_t charProperties,

uint8_t secPermissions,

uint8_t gattEvtMask,

uint8_t encryKeySize,

uint8_t isVariable,

uint16_t* charHandle);

So this would allow only a maximum characteristic length of 255 bytes. According to this document the stack itself supports 2 bytes with FW 7.2 or higher. I have 7.23 so this should be fine but I cannot find any reference or example of a BlueNRG-MS middleware that would support a call with

charValueLen

of type

uint16_t

. I also downloaded the latest STSW-BLUENRG-DK and the examples also only support

uint8_t charValueLen

.

[Winfred LU]

Please manually modify the parameter accordingly.

in bluenrg_gatt_aci.h:

tBleStatus aci_gatt_add_char(uint16_t serviceHandle,
			 uint8_t charUuidType,
			 const uint8_t* charUuid, 
			 uint16_t charValueLen, 
			 uint8_t charProperties,
			 uint8_t secPermissions,
			 uint8_t gattEvtMask,
			 uint8_t encryKeySize,
			 uint8_t isVariable,
			 uint16_t* charHandle);

in bluenrg_gatt_aci.c:

tBleStatus aci_gatt_add_char(uint16_t serviceHandle,
 uint8_t charUuidType,
 const uint8_t* charUuid, 
 uint16_t charValueLen, 
 uint8_t charProperties,
 uint8_t secPermissions,
 uint8_t gattEvtMask,
 uint8_t encryKeySize,
 uint8_t isVariable,
 uint16_t* charHandle) 
{
 struct hci_request rq;
 gatt_add_serv_rp resp;
 uint8_t buffer[26];
 uint8_t uuid_len;
 uint8_t indx = 0;
 
 serviceHandle = htobs(serviceHandle);
 Osal_MemCpy(buffer + indx, &serviceHandle, 2);
 indx += 2;
 
 buffer[indx] = charUuidType;
 indx++;
 
 if(charUuidType == UUID_TYPE_16){
 uuid_len = 2;
 }
 else {
 uuid_len = 16;
 } 
 Osal_MemCpy(buffer + indx, charUuid, uuid_len);
 indx += uuid_len;
 
 charValueLen = htobs(charValueLen);
 Osal_MemCpy(buffer + indx, &charValueLen, 2);
 indx += 2;
 
 buffer[indx] = charProperties;
 indx++;
 
 buffer[indx] = secPermissions;
 indx++;
 
 buffer[indx] = gattEvtMask;
 indx++;
 
 buffer[indx] = encryKeySize;
 indx++;
 
 buffer[indx] = isVariable;
 indx++;
 
 Osal_MemSet(&resp, 0, sizeof(resp));
 
 Osal_MemSet(&rq, 0, sizeof(rq));
 rq.ogf = OGF_VENDOR_CMD;
 rq.ocf = OCF_GATT_ADD_CHAR;
 rq.cparam = (void *)buffer;
 rq.clen = indx;
 rq.rparam = &resp;
 rq.rlen = GATT_ADD_CHAR_RP_SIZE;
 
 if (hci_send_req(&rq, FALSE) < 0)
 return BLE_STATUS_TIMEOUT;
 
 if (resp.status) {
 return resp.status;
 }
 
 *charHandle = btohs(resp.handle);
 
 return 0;
}

[aj.sk]

I did the modification but when I do

ret = aci_gatt_add_char(serviceHandle, UUID_TYPE_128, uuid, 470, CHAR_PROP_READ, ATTR_PERMISSION_NONE, GATT_NOTIFY_READ_REQ_AND_WAIT_FOR_APPL_RESP, 16, 0 , &charHandle);

I get return value 0x1F, so it fails to add the characteristic.

Did I miss something? What is the origin of this Osal_MemSet? My code uses BLUENRG_memcpy instead, is that a problem?

[Winfred LU]

0x1F would be out of memory.

Please check the the "Max_Attribute_Records" value in your previous call to aci_gatt_add_serv().

[aj.sk]

Thanks, I'll give it a try! I assume the same modification applies for aci_gatt_update_char_value()?

What confuses me a bit though is that if this is the fix, then I assume the buffer content would have been incorrect so far right? Or does it distinguish between buffer of size 25 and size 26 to determine whether I have 1 or 2 byte charValueLen?