hang in device_instance_find() in SDK7.2

Question

Hi, 
 I recently tried SDK7.2 on PCA10028 board in order to make sure that SDK7.2 works well before I migrate my project to SDK7.2 codebase. My project code didn't work as expected once I resolved all the compilation issues. The issue was in the bonding procedure, the firmware was hung somewhere (the bonding was established between PCA10028 board and iPhone4S). I observed that the DM Log was on by default; when I disabled DM Logs the issue went away. 
 Later on I enabled DM Logs to further debug and after putting several debug printfs in device_manager_peripheral.c, I saw that the issue happened at the following DM_TRC whenever p_addr was NULL. If I put a condition that the DM_TRC is executed only when p_addr is not NULL, the issues goes away. 
 static api_result_t device_instance_find(ble_gap_addr_t const * p_addr, uint32_t * p_device_index, uint16_t div)
{
api_result_t err_code;
uint32_t index; 
 err_code = NRF_ERROR_NOT_FOUND;

if(p_addr) // My change to avoid the hang issue
DM_TRC("[DM]: Searching for device 0x%02X 0x%02X 0x%02X 0x%02X 0x%02X 0x%02X.
",
 p_addr->addr[0], p_addr->addr[1], p_addr->addr[2], p_addr->addr[3],
 p_addr->addr[4], p_addr->addr[5]);

for (index = 0; index < DEVICE_MANAGER_MAX_BONDS; index++)
{
 DM_LOG("[DM]:[DI 0x%02X]: Device type 0x%02X.
",
 index, m_peer_table[index].peer_id.id_addr_info.addr_type);
 
 Any idea why NULL pointer referencing would cause this issue? Has anybody else seen this as well? Any pointers would be a great help. 
 Thanks,
Sam

Susheel Nuguru · Accepted Answer

Hi Sam, 
 Thanks for reporting this issue 
 the function where this problem occurs device_instance_find is used at two places. 
 
 when we have get a BLE_GAP_EVT_CONNECTED event and 
 when we get BLE_GAP_EVT_SEC_INFO_REQUEST event 
 
 when the latter occurs the the p_addr is NULL as we try to find the instance using evt.gap_evt.params.sec_info_request.div
That check for p_addr which you have added to fix this issue is a proper way to fix this issue. This is a bug which I have reported to the team. Hopefully it will be fixed in next release.