I have several applications that continuously scan on Bluetooth that have started exhibiting a new failure mode after upgrading nrfxlib from v2.9.0 to v3.0.2 (https://github.com/nrfconnect/sdk-nrfxlib/tree/v3.0.2).
After a variable time (but usually several hours) of continuous scanning on the extended advertising channels, no new advertising reports are generated. No errors are reported (at LOG_WRN or LOG_ERR).
static const struct bt_le_scan_param scan_param = {
.type = BT_LE_SCAN_TYPE_PASSIVE,
.options = BT_LE_SCAN_OPT_NONE,
.interval = BT_GAP_SCAN_FAST_INTERVAL_MIN,
.window = BT_GAP_SCAN_FAST_WINDOW,
};
void scan_start(void) {
bt_le_scan_start(&scan_param, scan_cb);
}
This error state persists even if the scanning is restarted (below was my attempt at a watchdog to recover from this error). No errors are reported, but advertising reports do no restart.
static void scan_cb(const bt_addr_le_t *addr, int8_t rssi, uint8_t adv_type,
struct net_buf_simple *buf)
{
/* Advertising packet observed, reset watchdog */
k_work_reschedule(&scan_watchdog_work, K_SECONDS(60));
...
}
static void scan_rx_watchdog_expired(struct k_work *work)
{
int rc;
LOG_WRN("Scan RX watchdog expired, restarting scan");
rc = bt_le_scan_stop();
if (rc != 0) {
LOG_ERR("Failed to stop scanning (%d)", rc);
}
rc = bt_le_scan_start(&scan_param, scan_cb);
if (rc != 0) {
LOG_ERR("Failed to restart scanning (%d)", rc);
}
/* Restart the watchdog */
k_work_reschedule(&scan_watchdog_work, K_SECONDS(60));
}
[02:42:42.781,341] <wrn> epacket_bt_adv: Scan RX watchdog expired, restarting scan [02:43:42.785,308] <wrn> epacket_bt_adv: Scan RX watchdog expired, restarting scan [02:44:42.787,445] <wrn> epacket_bt_adv: Scan RX watchdog expired, restarting scan
This exact same code flow (minus the watchdog) previously worked without error for days/weeks of operation.
The most obvious example is from running on a nRF52840DK forwarding packets over USB, but I have also observed the same behavior on a nRF9151/nRF54L15 split host/controller application.
None of the most recent bugfixes on main look to address this, since the application is only scanning: https://github.com/nrfconnect/sdk-nrfxlib/blob/main/softdevice_controller/CHANGELOG.rst#main-branch
All the SDC Kconfig options from the application
#define CONFIG_BT_LL_SOFTDEVICE_HEADERS_INCLUDE 1 #define CONFIG_BT_LL_SOFTDEVICE 1 #define CONFIG_BT_LL_SOFTDEVICE_BUILD_TYPE_LIB 1 #define CONFIG_BT_LL_SOFTDEVICE_MULTIROLE 1 #define CONFIG_BT_LL_SOFTDEVICE_INIT_PRIORITY 40 #define CONFIG_DT_HAS_NORDIC_BT_HCI_SDC_ENABLED 1 #define CONFIG_BT_CTLR_SDC_ALLOW_PARALLEL_SCANNING_AND_INITIATING 1 #define CONFIG_BT_CTLR_SDC_PERIPHERAL_COUNT 1 #define CONFIG_BT_CTLR_SDC_MAX_CONN_EVENT_LEN_DEFAULT 7500 #define CONFIG_BT_CTLR_SDC_CONN_EVENT_EXTEND_DEFAULT 1 #define CONFIG_BT_CTLR_SDC_CENTRAL_ACL_EVENT_SPACING_DEFAULT 7500 #define CONFIG_BT_CTLR_SDC_TX_PACKET_COUNT 3 #define CONFIG_BT_CTLR_SDC_RX_PACKET_COUNT 2 #define CONFIG_BT_CTLR_SDC_SCAN_BUFFER_COUNT 3 #define CONFIG_BT_CTLR_SDC_PERIODIC_SYNC_BUFFER_COUNT 3 #define CONFIG_BT_CTLR_SDC_RX_PRIO 6 #define CONFIG_BT_CTLR_SDC_LE_POWER_CLASS_1 1
