Known issues¶

Known issues listed on this page and tagged with the latest official release version are valid for the current state of development. Use the drop-down filter to see known issues for previous releases and check if they are still valid.

Filter by versions:

nRF9160 ¶

Asset tracker ¶

NCSDK-6898: Setting CONFIG_SECURE_BOOT does not work

The immutable bootloader is not able to find the required metadata in the MCUboot image. See the related NCSDK-6898 known issue in Build system for more details.

Workaround: Set CONFIG_FW_INFO in MCUboot.

External antenna performance setting

The preprogrammed Asset Tracker does not come with the best external antenna performance.

Workaround: If you are using nRF9160 DK v0.15.0 or higher and Thingy:91 v1.4.0 or higher, set CONFIG_NRF9160_GPS_ANTENNA_EXTERNAL to y. Alternatively, for nRF9160 DK v0.15.0, you can set the CONFIG_NRF9160_GPS_COEX0_STRING option to AT%XCOEX0 when building the preprogrammed Asset Tracker to achieve the best external antenna performance.

NCSDK-5574: Warnings during FOTA: The nRF9160: Asset Tracker application prints warnings and error messages during successful FOTA.

NCSDK-6689: High current consumption in Asset Tracker: The nRF9160: Asset Tracker application might show up to 2.5 mA current consumption in idle mode with CONFIG_POWER_OPTIMIZATION_ENABLE=y.

Sending data before connecting to nRF Connect for Cloud: The nRF9160: Asset Tracker application does not wait for connection to nRF Connect for Cloud before trying to send data. This causes the application to crash if the user toggles one of the switches before the kit is connected to the cloud.

IRIS-2676: Missing support for FOTA on nRF Connect for Cloud

The nRF9160: Asset Tracker application does not support the nRF Connect for Cloud FOTA_v2 protocol.

Workaround: The implementation for supporting the nRF Connect for Cloud FOTA_v2 can be found in the following commits:

cef289b559b92186cc54f0257b8c9adc0997f334

156d4cf3a568869adca445d43a786d819ae10250

f520159f0415f011ae66efb816384a8f7bade83d

Other issues ¶

NCSDK-7856: Faulty indirection on nrf_cc3xx memory slab when freeing the platform mutex

The mutex_free_platform() function has a bug where a call to k_mem_slab_free() provides wrong indirection on a parameter to free the platform mutex.

Workaround: Write the call to free the mutex in the following way: k_mem_slab_free(&mutex_slab, &mutex->mutex). The change adds & before the parameter mutex->mutex.

NCSDK-7914: The nrf_cc3xx RSA implementation does not deduce missing parameters

The calls to mbedtls_rsa_complete() will not deduce all types of missing RSA parameters when using nrf_cc3xx v0.9.6 or earlier.

Workaround: Calculate the missing parameters outside of this function or update to nrf_cc3xx v0.9.7 or later.

NCSDK-8075: Invalid initialization of mbedtls_entropy_context mutex type

The calls to mbedtls_entropy_init() do not zero-initialize the member variable mutex when nrf_cc3xx is enabled.

Workaround: Zero-initialize the structure type before using it or make it a static variable to ensure that it is zero-initialized.

NRF91-989: Unable to bootstrap after changing SIMs: In some cases, swapping the SIM card may trigger the bootstrap Pre-Shared Key to be deleted from the device. This can prevent future bootstraps from succeeding.

NCSDK-5666: LTE Sensor Gateway: The nRF9160: LTE Sensor Gateway sample crashes when Thingy:52 is flipped.

NCSDK-6073: nrf_send is blocking

The nrf_send() function in the Modem library might be blocking for several minutes, even if the socket is configured for non-blocking operation. The behavior depends on the cellular network connection.

Workaround: For nRF Connect SDK v1.4.0, set the non-blocking mode for a partial workaround for non-blocking operation.

GPS sockets and SUPL client library stops working

The nRF9160: GPS sockets and SUPL client library sample stops working if SUPL client and SUPL client OS integration support is enabled, but the SUPL host name cannot be resolved.

Workaround: Insert a delay (k_sleep()) of a few seconds after the printf on line 294 in main.c.

Calling nrf_connect immediately causes fail: nrf_connect fails if called immediately after initialization of the device. A delay of 1000 ms is required for this to work as intended.

Problems with RTT Viewer/Logger

The SEGGER Control Block cannot be found by automatic search by the RTT Viewer/Logger.

Workaround: Set the RTT Control Block address to 0 and it will try to search from address 0 and upwards. If this does not work, look in the builddir/zephyr/zephyr.map file to find the address of the _SEGGER_RTT symbol in the map file and use that as input to the viewer/logger.

Receive error with large packets: nRF91 fails to receive large packets (over 4000 bytes).

Modem FW reset on debugger connection through SWD: If a debugger (for example, J-Link) is connected via SWD to the nRF9160, the modem firmware will reset. Therefore, the LTE modem cannot be operational during debug sessions.

nRF5 ¶

nRF5340 ¶

KRKNWK-6756: 802.15.4 Service Layer (SL) library support for the nRF53: The binary variant of the 802.15.4 Service Layer (SL) library for the nRF53 does not support such features as synchronization of TIMER with RTC or timestamping of received frames. For this reason, 802.15.4 features like delayed transmission or delayed reception are not available for the nRF53.

FOTA does not work: FOTA with the SMP Server Sample does not work.

nRF52820 ¶

Missing CMakeLists.txt

The CMakeLists.txt file for developing applications that emulate nRF52820 on the nRF52833 DK is missing.

Workaround: Create a CMakeLists.txt file in the ncs/zephyr/boards/arm/nrf52833dk_nrf52820 folder with the following content:

zephyr_compile_definitions(DEVELOP_IN_NRF52833)
zephyr_compile_definitions(NRFX_COREDEP_DELAY_US_LOOP_CYCLES=3)

You can download this file from the upstream Zephyr repository. After you add it, the file is automatically included by the build system.

Thread ¶

KRKNWK-9461 / KRKNWK-9596Multiprotocol sample crashes with some smartphones: With some smartphones, the multiprotocol sample crashes on the nRF5340 due to timer timeout inside the 802.15.4 radio driver logic.

KRKNWK-9094: Possible deadlock in shell subsystem

Issuing OpenThread commands too fast might cause a deadlock in the shell subsystem.

Workaround: If possible, avoid invoking a new command before execution of the previous one has completed.

KRKNWK-6848: Reduced throughput: Performance testing for NCP sample shows a decrease of throughput of around 10-20% compared with the standard OpenThread.

KRKNWK-7885: Throughput is lower when using CC310 nrf_security backend

A decrease of throughput of around 5-10% has been observed for the CC310 nrf_security backend when compared with nrf_oberon or the standard mbedtls backend. CC310 nrf_security backend is used by default for nRF52840 boards. The source of throughput decrease is coupled to the cost of RTOS mutex locking when using the CC310 nrf_security backend when the APIs are called with shorter inputs.

Workaround: Use AES-CCM ciphers from the nrf_oberon backend by setting the following options:

KRKNWK-7721: MAC counter updating issue

The RxDestAddrFiltered MAC counter is not being updated. This is because the PENDING_EVENT_RX_FAILED event is not implemented in Zephyr.

Workaround: To fix the error, cherry-pick commits from the upstream Zephyr PR #29226.

KRKNWK-7962: Logging interferes with shell output

CONFIG_LOG_MINIMAL is configured by default for most OpenThread samples. It accesses the UART independently from the shell backend, which sometimes leads to malformed output.

Workaround: Disable logging or enable a more advanced logging option.

KRKNWK-7803: Automatically generated libraries are missing otPlatLog for NCP

When building OpenThread libraries using a different sample than the Thread NCP sample, the ncp_base.cpp is not compiled with the otPlatLog() function. This results in a linking failure when building the NCP with these libraries.

Workaround: Use the Thread NCP sample to create OpenThread libraries.

NCSDK-5014: Building with SES not possible: It is not possible to build Thread samples using SEGGER Embedded Studio (SES). SES does not support .cpp files in nRF Connect SDK projects.

KRKNWK-6358: CoAP client sample provisioning issues: It is not possible to provision the Thread: CoAP Client sample to servers that it cannot directly communicate with. This is because Link Local Address is used for communication.

KRKNWK-6408: diag command not supported: The diag command is not yet supported by Thread in the nRF Connect SDK.

Zigbee ¶

KRKNWK-8478: NCP host application crash on exceeding TX_BUFFERS_POOL_SIZE

If the NCP host application exceeds the TX_BUFFERS_POOL_SIZE pending requests, the application will crash on an assertion.

Workaround: Increase the value of TX_BUFFERS_POOL_SIZE or define shorter polling interval (NCP_TRANSPORT_REFRESH_TIME).

KRKNWK-8200: Sleepy End Device halts during the commissioning

If the turbo poll is disabled in the ZB_BDB_SIGNAL_DEVICE_FIRST_START signal, SED halts during the commissioning.

Workaround: Use the development libraries link or use ZB_BDB_SIGNAL_STEERING signal with successful status to disable turbo poll. See the following snippet for an example:

/* Workaround for KRKNWK-8200 (turbo poll) */
switch(sig)
{
case ZB_BDB_SIGNAL_DEVICE_REBOOT:
case ZB_BDB_SIGNAL_STEERING:
        if (status == RET_OK) {
                zb_zdo_pim_permit_turbo_poll(0);
                zb_zdo_pim_set_long_poll_interval(2000);
        }
        break;
}

KRKNWK-8200: Successful signal on commissioning fail

A successful steering signal is generated if the commissioning fails during TCLK exchange.

Workaround: Use the development libraries link or check for Extended PAN ID in the steering signal handler. If it is equal to zero, handle the signal as if it had unsuccessful status. See the following snippet for an example:

/* Workaround for KRKNWK-8200 (signal status) */
switch(sig)
{
case ZB_BDB_SIGNAL_STEERING:
        if (status == RET_OK) {
                zb_ext_pan_id_t extended_pan_id;
                zb_get_extended_pan_id(extended_pan_id);
                if (!(ZB_IEEE_ADDR_IS_VALID(extended_pan_id))) {
                       zb_buf_set_status(bufid, -1);
                       status = -1;
                }
        }
        break;
}

KRKNWK-9461 / KRKNWK-9596Multiprotocol sample crashes with some smartphones: With some smartphones, the multiprotocol sample crashes on the nRF5340 due to timer timeout inside the 802.15.4 radio driver logic.

KRKNWK-6348: ZCL Occupancy Sensing cluster is not complete

The ZBOSS stack provides only definitions of constants and an abstract cluster definition (sensing cluster without sensors).

Workaround: To use the sensing cluster with physical sensor, copy the implementation and extend it with the selected sensor logic and properties. For more information, see the declaring custom cluster guide.

KRKNWK-6336: OTA transfer may be aborted after the MAC-level packet retransmission

If the device receives the APS ACK for a packet that was not successfully acknowledged on the MAC level, the OTA client cluster implementation stops the image transfer.

Workaround: Add a watchdog timer that will restart the OTA image transfer.

KRKNWK-7831: Factory reset broken on coordinator with Zigbee shell

A coordinator with the Zigbee shell component enabled could assert after executing the bdb factory_reset command.

Workaround: Call the bdb_reset_via_local_action function twice to remove all the network information.

KRKNWK-7723: OTA upgrade process restarting after client reset: After the reset of OTA Upgrade Client, the client will start the OTA upgrade process from the beginning instead of continuing the previous process.

KRKNWK-6318: Device assert after multiple Leave requests: If a device that rejoins the network receives Leave requests several times in a row, the device could assert.

KRKNWK-6071: ZBOSS alarms inaccurate

On average, ZBOSS alarms last longer by 6.4 percent than Zephyr alarms.

Workaround: Use Zephyr alarms.

KRKNWK-5535: Device assert if flooded with multiple Network Address requests: The device could assert if it receives Network Address requests every 0.2 second or more frequently.

KRKNWK-9119: Zigbee shell does not work with ZBOSS development libraries

Because of changes to the ZBOSS API, the Zigbee shell library cannot be enabled when Zigbee samples are built with the ZBOSS Zigbee stack development libraries.

Workaround: Use only the production version of ZBOSS Zigbee stack when using Zigbee shell.

KRKNWK-9145: Corrupted payload in commands of the Scenes cluster

When receiving Scenes cluster commands, the payload is corrupted when using the ZBOSS Zigbee stack production libraries.

Workaround: Use the development version of ZBOSS Zigbee stack.

KRKNWK-7836: Coordinator asserting when flooded with ZDO commands: Executing a high number of ZDO commands can cause assert on the coordinator with the Zigbee shell component enabled.

KRKNWK-6073: Potential delay during FOTA: There might be a noticeable delay (~220 ms) between calling the ZBOSS API and on-the-air activity.

Project Connected Home over IP (Project CHIP)¶

KRKNWK-9214: Pigweed submodule may not be accessible from some regions

The west update command may generate log notifications about the failure to access the pigweed submodule. As a result, the Project CHIP samples will not build.

Workaround: Execute the following commands in the root folder:

git -C modules/lib/connectedhomeip submodule set-url third_party/pigweed/repo https://github.com/google/pigweed.git
git -C modules/lib/connectedhomeip submodule sync third_party/pigweed/repo
west update

nRF Desktop ¶

DESK-978: Directed advertising issues with SoftDevice Link Layer

Directed advertising (CONFIG_DESKTOP_BLE_DIRECT_ADV) should not be used by the nRF Desktop application when the SoftDevice Controller is in use, because that leads to reconnection problems. For more detailed information, see the Known issues and limitations section of the SoftDevice Controller’s Changelog.

Workaround: Directed advertising is disabled by default for nRF Desktop.

NCSDK-8304: HID configurator issues for peripherals connected over Bluetooth LE to Linux host

Using HID configurator for nRF Desktop for peripherals connected to host directly over Bluetooth LE may result in receiving improper HID feature report ID. In such case, the device will provide HID input reports, but it cannot be configured with the HID configurator.

Workaround: Connect the nRF Desktop peripheral through USB or using the nRF Desktop dongle.

Subsystems ¶

Bluetooth LE ¶

DRGN-15435: GATT notifications and Writes Without Response might be sent out of order

GATT notifications and Writes Without Response might be sent out of order when not using a complete callback.

Workaround: Always set a callback for notifications and Writes Without Response.

DRGN-15448: Incomplete bond overwrite during pairing procedure when peer is not using the IRK stored in the bond: When pairing with a peer that has deleted its bond information and is using a new IRK to establish the connection, the existing bond is not overwritten during the pairing procedure. This can lead to MIC errors during reconnection if the old LTK is used instead.

NCSDK-8321: NUS shell transport sample does not display the initial shell prompt uart:~$ on the remote terminal.

Also, few logs with sending errors are displayed on the terminal connected directly to the DK. This issue is caused by the shell being enabled before turning on the notifications for the NUS service by the remote peer.

Workaround: Enable the shell after turning on the NUS notifications or block it until turning on the notifications.

NCSDK-8224: Callbacks for “security changed” and “pairing failed” are not always called

The pairing failed and security changed callbacks are not called when the connection is disconnected during the pairing procedure or the required security is not met.

Workaround: Application should use the disconnected callback to handle pairing failed.

NCSDK-8223: GATT requests might deadlock RX thread

GATT requests might deadlock the RX thread when all TX buffers are taken by GATT requests and the RX thread tries to allocate a TX buffer for a response. This causes a deadlock because only the RX thread releases the TX buffers for the GATT requests. The deadlock is resolved by a 30 second timeout, but the ATT bearer cannot transmit without reconnecting.

Workaround: Set CONFIG_BT_L2CAP_TX_BUF_COUNT >= CONFIG_BT_ATT_TX_MAX + 2.

NCSDK-6845: Pairing failure with simultaneous pairing on multiple connections

When using LE Secure Connections pairing, the pairing fails with simultaneous pairing on multiple connections. The failure reason is unspecified.

Workaround: Retry the pairing on the connections that failed one by one after the pairing procedure has finished.

NCSDK-6844: Security procedure failure can terminate GATT client request

A security procedure terminates the GATT client request that is currently in progress, unless the request was the reason to initiate the security procedure. If a new GATT client request is queued at this time, this might potentially cause a GATT transaction violation and fail as well.

Workaround: Do not initiate a security procedure in parallel with GATT client requests.

NCSDK-5711: High-throughput transmission can deadlock the receive thread: High-throughput transmission can deadlock the receive thread if the connection is suddenly disconnected.

Only secure applications can use Bluetooth LE

Bluetooth LE cannot be used in a non-secure application, for example, an application built for the nrf5340_dk_nrf5340_cpuappns build target.

Workaround: Use the nrf5340_dk_nrf5340_cpuapp build target instead.

Peripheral HIDS keyboard sample cannot be used with nRF Bluetooth LE Controller: The Bluetooth: Peripheral HIDS keyboard sample cannot be used with the SoftDevice Controller because the NFC subsystem does not work with the controller library. The library uses the MPSL Clock driver, which does not provide an API for asynchronous clock operation. NFC requires this API to work correctly.

Peripheral HIDS mouse sample advertising issues: When the Bluetooth: Peripheral HIDS mouse sample is used with the Zephyr Bluetooth LE Controller, directed advertising does not time out and the regular advertising cannot be started.

Central HIDS sample issues with directed advertising: The Bluetooth: Central HIDS sample cannot connect to a peripheral that uses directed advertising.

Unstable samples: Bluetooth Low Energy peripheral samples are unstable in some conditions (when pairing and bonding are performed and then disconnections/re-connections happen).

CONFIG_BT_SMP alignment requirement: When running the Bluetooth: Central DFU SMP sample, the CONFIG_BT_SMP configuration must be aligned between this sample and the Zephyr counterpart (SMP Server Sample). However, security is not enabled by default in the Zephyr sample.

Reconnection issues on some operating systems: On some operating systems, the nRF Desktop application is unable to reconnect to a host.

Bluetooth: Central UART cannot handle long strings: A too long 212-byte string cannot be handled when entered to the console to send to Bluetooth: Peripheral UART.

Bluetooth: Central HIDS loses UART connectivity: After programming a HEX file to the nrf52_pca10040 board, UART connectivity is lost when using the Bluetooth LE Controller. The board must be reset to get UART output.

Samples crashing on nRF51 when using GPIO: On nRF51 devices, Bluetooth LE samples that use GPIO might crash when buttons are pressed frequently. In such case, the GPIO ISR introduces latency that violates real-time requirements of the Radio ISR. nRF51 is more sensitive to this issue than nRF52 (faster core).

GATT Discovery Manager missing support: The GATT Discovery Manager is not supported on nRF51 devices.

Samples do not work with SD Controller v0.1.0: Bluetooth LE samples cannot be built with the SoftDevice Controller v0.1.0.

LED Button Service reporting issue: Bluetooth: Peripheral LBS does not report the Button 1 state correctly.

MITM protection missing for central samples: The central samples (Bluetooth: Central UART, Bluetooth: Central HIDS) do not support any pairing methods with MITM protection.

Peripheral UART string size issue: Bluetooth: Peripheral UART cannot handle the corner case that a user attempts to send a string of more than 211 bytes.

Reconnection issues after bonding: The peripheral samples (Bluetooth: Peripheral UART, Bluetooth: Peripheral LBS, Bluetooth: Peripheral HIDS mouse) have reconnection issues after performing bonding (LE Secure Connection pairing enable) with nRF Connect for Desktop. These issues result in disconnection.

Bluetooth mesh ¶

NCSDK-5580: nRF5340 only supports SoftDevice Controller: On nRF5340, only the SoftDevice Controller is supported for Bluetooth mesh.

Bootloader ¶

NCSDK-7173: nRF5340 network core bootloader cannot be built stand-alone

The nRF5340: Network core bootloader sample does not compile when built stand-alone. It compiles without problems when included as a child image.

Workaround: Include the nRF5340: Network core bootloader sample as child image instead of compiling it stand-alone.

Public keys revocation: Public keys are not revoked when subsequent keys are used.

Incompatibility with nRF51: The bootloader does not work properly on nRF51.

Immutable bootloader not supported in SES: Building and programming the immutable bootloader (see Secure bootloader chain) is not supported in SEGGER Embedded Studio.

Immutable bootloader board restrictions

The immutable bootloader can only be used with the following boards:

nrf52840_pca10056
nrf9160_pca10090

Immutable bootloader and netboot can overwrite non-OTP provisioning data: In architectures that do not have OTP regions, b0 and b0n images incorrectly linked to the size of their container can overwrite provisioning partition data from their image sizes. Issue related to NCSDK-7982.

Build system ¶

Missing files or permissions when building on Windows

Because of the Windows path length limitations, the build can fail with errors related to permissions or missing files if some paths in the build are too long.

Workaround: Shorten the build folder name, for example, from “build_nrf5340dk_nrf5340_cpuappns” to “build”, or shorten the path to the build folder in some other way.

NCSDK-6898: Overriding child images

Adding child image overlay from the CMakeLists.txt top-level file located in the samples directory overrides the existing child image overlay.

Workaround: Apply the configuration from the overlay to the child image manually.

NCSDK-6777: Project out of date when CONFIG_SECURE_BOOT is set

The DFU .zip file is regenerated even when no changes are made to the files it depends on. As a consequence, SES displays a “Project out of date” message even when the project is not out of date.

Workaround: Apply the fix from sdk-nrf PR #3241.

NCSDK-6848: MCUboot must be built from source when included

The build will fail if either CONFIG_MCUBOOT_BUILD_STRATEGY_SKIP_BUILD or CONFIG_MCUBOOT_BUILD_STRATEGY_USE_HEX_FILE is set.

Workaround: Set CONFIG_MCUBOOT_BUILD_STRATEGY_FROM_SOURCE instead.

KRKNWK-7827: Application build system is not aware of the settings partition

The application build system is not aware of partitions, including the settings partition, which can result in application code overlapping with other partitions. As a consequence, writing to overlapping partitions might remove or damage parts of the firmware, which can lead to errors that are difficult to debug.

Workaround: Define and use a code partition to shrink the effective flash memory available for the application. You can use one of the following solutions:

Partition Manager from nRF Connect SDK - see the page for all configuration options. For example, for single image (without bootloader and with the settings partition used), set the CONFIG_PM_SINGLE_IMAGE Kconfig option to y and define the value for CONFIG_PM_PARTITION_SIZE_SETTINGS_STORAGE to the required settings storage size.
Devicetree code partition from Zephyr. Set CONFIG_USE_DT_CODE_PARTITION Kconfig option to y. Make sure that the code partition is defined and chosen correctly (offset and size).

NCSDK-6117: Build configuration issues

The build configuration consisting of Immutable bootloader, Secure Partition Manager, and application does not work.

Workaround: Either include MCUboot in the build or use MCUboot instead of the immutable bootloader.

Flash commands only program one core

west flash and ninja flash only program one core, even if multiple cores are included in the build.

Workaround: Execute the flash command from inside the build directory of the child image that is placed on the other core (for example, build/hci_rpmsg).

NCSDK-8232: Secure Partition Manager and application building together: It is not possible to build and program Secure Partition Manager and the application individually.

NCSDK-7982: partition manager: Incorrect partition size linkage from name conflict

Partition manager will incorrectly link a partition’s size to the size of its container if the container partition’s name matches its child image’s name in CMakeLists.txt. This can cause the inappropriately-sized partition to overwrite another partition beyond its intended boundary.

Workaround: Rename the container partitions in the pm.yml and pm_static.yml files to something that does not match the child images’ names, and rename the child images’ main image partition to its name in CMakeLists.txt.

DFU and FOTA ¶

NCSDK-6238: Socket API calls may hang when using Download client

When using the Download client library with HTTP (without TLS), the application might not process incoming fragments fast enough, which can starve the Modem library buffers and make calls to the Modem library hang. Samples and applications that are affected include those that use Download client to download files through HTTP, or those that use FOTA download with modem updates enabled.

Workaround: Set CONFIG_DOWNLOAD_CLIENT_RANGE_REQUESTS.

Jobs not received after reset

When using AWS FOTA, no new jobs are received on the device if the device is reset during a firmware upgrade or loses the MQTT connection.

Workaround: Delete the stalled in progress job from AWS IoT.

Stalled download

FOTA download does not resume a download if the device loses the connection.

Workaround: Call fota_download_start() again with the same arguments when the connection is re-established to resume the download.

Offset not retained with an MCUboot target: When using the MCUboot target in DFU target, the write/downloaded offset is not retained when the device is reset.

Download stopped on socket connection timeout

In the nRF9160: AWS FOTA and nRF9160: HTTP application update samples, the download is stopped if the socket connection times out before the modem can delete the modem firmware. A fix for this issue is available in commit 38625ba7.

Workaround: Call fota_download_start() again with the same arguments.

Update event triggered by an error event: If the last fragment of a FOTA download is received but is corrupted, or if the last write is unsuccessful, the library emits an error event as expected. However, it also emits an apply/request update event, even though the downloaded data is invalid.

FW upgrade is broken for multi-image builds

Firmware upgrade using mcumgr or USB DFU is broken for multi-image builds, because the devicetree configuration is not used. Therefore, it is not possible to upload the image.

Workaround: Build MCUboot and the application separately.

NFC ¶

Sample incompatibility with the nRF5340 PDK: The NFC: TNEP poller and NFC: Tag reader samples cannot be run on the nRF5340 PDK. There is an incorrect number of pins defined in the MDK files, and the pins required for NFC Reader ST25R3911B cannot be configured properly.

Unstable NFC tag samples: NFC tag samples are unstable when exhaustively tested (performing many repeated read and/or write operations). NFC tag data might be corrupted.

Secure Partition Manager (SPM)¶

CIA-248: Samples with default SPM config fails to build for thingy91_nrf9160ns

All samples using the default SPM config fails to build for the thingy91_nrf9160ns build target if the sample is not set up with MCUboot.

Workaround: Use the master branch.

NCSIDB-114: Default logging causes crash

Enabling default logging in the Secure Partition Manager sample makes it crash if the sample logs any data after the application has booted (for example, during a SecureFault, or in a secure service). At that point, RTC1 and UARTE0 are non-secure.

Workaround: Do not enable logging and add a breakpoint in the fault handling, or try a different logging backend.

MCUboot ¶

Recovery with the USB does not work: The MCUboot recovery feature using the USB interface does not work.

nrfxlib ¶

Crypto ¶

NCSDK-5883: CMAC behavior issues

CMAC glued with multiple backends may behave incorrectly due to memory allocation issues.

Workaround: Disable glued CMAC and use only one of the enabled backends.

NCSDK-5546: Oberon missing symbols for HKDF

nRF Oberon v3.0.5 is missing symbols for HKDF using SHA1, which will be fixed in an upcoming version of the library.

Workaround: Use a different backend (for example, vanilla mbed TLS) for HKDF/HMAC using SHA1.

Limited support for Nordic Security Module

The Nordic Security Module is currently only fully supported on nRF52840 and nRF9160 devices. It gives compile errors on nRF52832, nRF52833, nRF52820, nRF52811, and nRF52810.

Workaround: To fix the errors, cherry-pick commits in nrfxlib PR #205.

Glue layer symbol renaming issue: The Nordic Security Module glue layer is broken because symbol renaming is not handled correctly. Therefore, the behavior is undefined when selecting multiple back-ends for the same algorithm (for example, AES).

GNSS sockets ¶

Cold start and A-GPS data not supported: Forcing a cold start and writing A-GPS data is not yet supported.

Hard-fault with GPS in running mode: Implementation might hard-fault when GPS is in running mode and messages are not read fast enough.

NMEA strings might return wrong length: NMEA strings are valid c-strings (0-terminated), but the read function might return wrong length.

Closing sockets: Sockets can only be closed when GPS is in stopped mode. Moreover, closing a socket does not properly clean up all memory resources. If a socket is opened and closed multiple times, this might starve the system.

Multiprotocol Service Layer (MPSL)¶

DRGN-15223: CONFIG_SYSTEM_CLOCK_NO_WAIT is not supported for nRF5340: Using CONFIG_SYSTEM_CLOCK_NO_WAIT with nRF5340 devices might not work as expected.

DRGN-15176: CONFIG_SYSTEM_CLOCK_NO_WAIT is ignored when Low Frequency Clock is started before initializing MPSL

If the application starts the Low Frequency Clock before calling mpsl_init(), the clock configuration option CONFIG_SYSTEM_CLOCK_NO_WAIT has no effect. MPSL will wait for the Low Frequency Clock to start.

Workaround: When CONFIG_SYSTEM_CLOCK_NO_WAIT is set, do not start the Low Frequency Clock.

DRGN-15064: External Full swing and External Low swing not working: Even though the MPSL Clock driver accepts a Low Frequency Clock source configuration for External Full swing and External Low swing, the clock control system is not configured correctly. For this reason, do not use CLOCK_CONTROL_NRF_K32SRC_EXT_FULL_SWING and CLOCK_CONTROL_NRF_K32SRC_EXT_LOW_SWING.

DRGN-6362: Do not use the synthesized low frequency clock source: The synthesized low frequency clock source is neither tested nor intended for usage with MPSL.

DRGN-14153: Radio Notification power performance penalty: The Radio Notification feature has a power performance penalty proportional to the notification distance. This means an additional average current consumption of about 600 µA for the duration of the radio notification.

DRGN-11059: Front-end module API not implemented for SoftDevice Controller: Front-end module API is currently not implemented for SoftDevice Controller. It is only available for 802.15.4.

KRKNWK-8842: MPSL does not support nRF21540 revision 1 or older

The nRF21540 revision 1 or older is not supported by MPSL. This also applies to kits that contain this device.

Workaround: Check nordicsemi.com for the latest information on availability of the product version of nRF21540.

802.15.4 Radio driver ¶

KRKNWK-6255: RSSI parameter adjustment is not applied: The RADIO: RSSI parameter adjustment errata (153 for nRF52840, 225 for nRF52833 and nRF52820, 87 for nRF5340) are not applied for RSSI, LQI, Energy Detection, and CCA values used by the 802.15.4 protocol. There is an expected offset up to +/- 6 dB in extreme temperatures of values based on RSSI measurement.

Workaround: To apply RSSI parameter adjustments, cherry-pick the commits in hal_nordic PR #88, sdk-nrfxlib PR #381, and sdk-zephyr PR #430.

KRKNWK-8133: CSMA-CA issues: Using CSMA-CA with the open-source variant of the 802.15.4 Service Layer (SL) library causes an assertion fault. CSMA-CA support is currently not available in the open-source SL library.

SoftDevice Controller ¶

DRGN-15465: Corrupted advertising data when CONFIG_BT_EXT_ADV is set

Setting scan response data for a legacy advertiser on a build with extended advertising support corrupts parts of the advertising data. When using BT_LE_ADV_OPT_USE_NAME (which is the default configuration in most samples), the device name is put in the scan response. This corrupts the advertising data.

Workaround: Do not set scan response data. That implies not using the BT_LE_ADV_OPT_USE_NAME option, or the BT_LE_ADV_CONN_NAME macro when initializing Bluetooth. Instead, use BT_LE_ADV_CONN, and if necessary set the device name in the advertising data manually.

DRGN-15475: Samples might not initialize the SoftDevice Controller HCI driver correctly

Samples using both the advertising and the scanning state, but not the connected state, fail to initialize the SoftDevice Controller HCI driver. As a result, the function bt_enable() returns an error code.

Workaround: Manually enable CONFIG_SOFTDEVICE_CONTROLLER_MULTIROLE for the project configuration.

DRGN-15382: The SoftDevice Controller cannot be qualified on nRF52832

The SoftDevice Controller cannot be qualified on nRF52832.

Workaround: Upgrade to v1.5.1 or use the master branch.

DRGN-15226: Link disconnects with reason “LMP Response Timeout (0x22)”: If the slave receives an encryption request while the “HCI LE Long Term Key Request” event is disabled, the link disconnects with the reason “LMP Response Timeout (0x22)”. The event is disabled when CONFIG_BT_SMP and/or CONFIG_BT_CTLR_LE_ENC is disabled.

DRGN-11963: LL control procedures cannot be initiated at the same time

The LL control procedures (LE start encryption and LE connection parameter update) cannot be initiated at the same time or more than once. The controller will return an HCI error code “Controller Busy (0x3a)”, as per specification’s chapter 2.55.

Workaround: Do not initiate these procedures at the same time.

DRGN-8476: Long packets not supported in connections on Coded PHY: In connections, the Link Layer payload size is limited to 27 bytes on LE Coded PHY.

DRGN-9083: AAR populated with zero IRK: If the application has set an all zeroes IRK for a device in the resolving list, then a resolvable address that can be resolved with the all zeroes IRK will be reported to the application as that device in the advertisement report or the connected event.

DRGN-13921: Directed advertising issues using RPA in TargetA

The SoftDevice Controller will generate a resolvable address for the TargetA field in directed advertisements if the target device address is in the resolving list with a non-zero IRK, even if privacy is not enabled and the local device address is set to a public address.

Workaround: Remove the device address from the resolving list.

DRGN-11297: Maximum CI before entering LLPM-mode

The maximum connection interval that can be active when switching to a connection interval of 1 ms is 10 ms.

Workaround: An application that needs to use a higher interval than 10 ms needs to perform two connection updates to use 1 ms connection interval:

A first update to 10 ms connection interval.
A second update to 1 ms connection interval.

DRGN-10305: Scanner can’t have more than 16 seconds scan window: If the scanner is configured with a scan window larger than 16 seconds, the scanner will truncate the scan window to 16 seconds.

DRGN-8569: SEVONPEND flag must not be modified: Applications must not modify the SEVONPEND flag in the SCR register when running in priority levels higher than 6 (priority level numerical values lower than 6) as this can lead to undefined behavior.

DRGN-6362: Synthesized low frequency clock source not tested: Synthesized low frequency clock source is not tested or intended for use with the Bluetooth LE stack.

DRGN-10367: Advertiser times out earlier than expected: If an extended advertiser is configured with limited duration, it will time out after the first primary channel packet in the last advertising event.

DRGN-12259: HCI Receiver and Transmitter Test commands not available

The HCI Receiver and Transmitter Test commands are not available.

Workaround: To perform a radio test, use a DTM application:

For nRF52, use the DTM application in the nRF5 SDK.
For nRF53, use Bluetooth: Direct Test Mode.

CONFIG_BT_HCI_TX_STACK_SIZE requires specific value: CONFIG_BT_HCI_TX_STACK_SIZE must be set to 1536 when selecting CONFIG_BT_LL_SOFTDEVICE.

CONFIG_SYSTEM_WORKQUEUE_STACK_SIZE requires specific value: CONFIG_SYSTEM_WORKQUEUE_STACK_SIZE must be set to 2048 when selecting CONFIG_BT_LL_SOFTDEVICE on Bluetooth: Central UART and Bluetooth: Central BAS.

CONFIG_NFCT_IRQ_PRIORITY requires specific value: CONFIG_NFCT_IRQ_PRIORITY must be set to 5 or less when selecting CONFIG_BT_LL_SOFTDEVICE on Bluetooth: Peripheral HIDS keyboard.

Several issues for nRF5340

The following issues can occur when using SoftDevice Controller with nRF5340:

Poor performance when performing active scanning.
The controller could assert when receiving extended advertising packets.
The T_IFS could in certain conditions be off by 5 us.
The radio could stay in the TX state longer than expected. This issue can only occur when sending a packet on either LE 1M or LE 2M PHY after receiving or transmitting a packet on LE Coded PHY. If this occurs while performing a Link Layer Control Procedure, the controller could end up retransmitting an acknowledged packet, resulting in a disconnect.

Sending control packet twice: A control packet could be sent twice even after the packet was acknowledged. This would only occur if the radio was forced off due to an unforeseen condition.

Wait time required after a directed high duty cycle advertiser times out

The application is unable to restart a connectable advertiser right after a high-duty-cycle advertiser times out.

Workaround: Wait 500 ms before restarting a connectable advertiser

Assert risk after performing a DLE procedure: The controller could assert when receiving a packet with a CRC error on LE Coded PHY after performing a DLE procedure where RX Octets is changed to a value above 140.

Assert when using HCI LE Set Extended Advertising Parameters: The controller will assert when setting secondary PHY to 0 when using HCI LE Set Extended Advertising Parameters and the advertising type is set to legacy advertising.

HCI issues with duplicate filtering: HCI LE Set Extended Scan Enable returns UNSUPPORTED_FEATURE when duplicate filtering is enabled.

HCI issues with secondary_max_skip: HCI LE Set Advertising Parameters returns UNSUPPORTED_FEATURE when secondary_max_skip is set to a non-zero value.

No data issue when connected to multiple devices: hci_data_get() may return “No data available” when there is data available. This issue will only occur when connected to multiple devices at the same time.

Assert on LE Write Suggested Default Data Length: The controller will assert if the host issues LE Write Suggested Default Data Length.

HCI LE Set Privacy Mode appears as not supported: The controller does not indicate support for HCI LE Set Privacy Mode although it is supported.

Assert if advertising data is set after HCI Reset: The controller will assert if advertising data is set after HCI Reset without first setting advertising parameters.

Assert on writing to flash: The controller may assert when writing to flash.

Timeout without sending packet: A directed advertiser may time out without sending a packet on air.

nrfx ¶

MDK ¶

Incorrect pin definition for nRF5340: For nRF5340, the pins P1.12 to P1.15 are unavailable due to an incorrect pin number definition in the MDK.

nrfx_saadc driver ¶

Samples might be swapped: Samples might be swapped when buffer is set after starting the sample process, when more than one channel is sampled. This can happen when the sample task is connected using PPI and setting buffers and sampling are not synchronized.

nrfx_uarte driver ¶

RX and TX not disabled in uninit: The driver does not disable RX and TX in uninit, which can cause higher power consumption.

nrfx_uart driver ¶

tx_buffer_length set incorrectly: The nrfx_uart driver might incorrectly set the internal tx_buffer_length variable when high optimization level is set during compilation.

Zephyr ¶

NCSDK-6330: USB Mass Storage Sample Application fails MSC Tests from USB3CV test tool: USB Mass Storage Sample Application fails the USB3CV compliance Command Set Test from the MSC Tests suite.

NCSDK-6328: USB CDC ACM Composite Sample Application fails Chapter 9 Tests from USB3CV test tool: USB CDC ACM Sample Composite Application fails the USB3CV compliance TD 9.1: Device Descriptor Test from the Chapter 9 Test suite.

NCSDK-6331: WebUSB sample application fails Chapter 9 Tests from USB3CV test tool: WebUSB sample application fails the USB3CV compliance TD 9.21: LPM L1 Suspend Resume Test from the Chapter 9 Test suite.

NCSIDB-108: Thread context switch might lead to a kernel fault: If the Zephyr kernel preempts the current thread and performs a context switch to a new thread while the current thread is executing a secure service, the behavior is undefined and might lead to a kernel fault. To prevent this situation, a thread that aims to call a secure service must temporarily lock the kernel scheduler (k_sched_lock()) and unlock the scheduler (k_sched_unlock()) after returning from the secure call.

Counter Alarm sample does not work: The Counter Alarm Sample does not work. A fix can be found in Pull Request #16736.

USB Mass Storage Sample Application compilation issues: USB Mass Storage Sample Application does not compile.

NCSDK-6832: SMP Server sample fails upon initialization

The SMP Server Sample will fail upon initialization when using the bt-overlay.conf Kconfig overlay file. This happens because of a stack overflow.

Workaround: Set CONFIG_MAIN_STACK_SIZE to 2048.

SEGGER Embedded Studio Nordic Edition ¶

NCSDK-6852: Extra CMake options might not be applied in version 5.10d

If you specify Extra CMake Build Options in the Open nRF Connect SDK Project dialog and at the same time select an nRF Connect Toolchain Version of the form X.Y.Z, the additional CMake options are discarded.

Workaround: Select NONE (Use SES settings / environment PATH) from the nRF Connect Toolchain Version drop-down if you want to specify Extra CMake Build Options.

NCSDK-8372: Project name collision causes SES Nordic Edition to load the wrong project

Some samples that are located in different folders use the same project name. For example, there is a light_switch project both in the samples/bluetooth/mesh/ folder and in the samples/zigbee/ folder. When you select one of these samples from the project list in the Open nRF Connect SDK Project dialog, the wrong sample might be selected. Check the Build Directory field in the dialog to see if this is the case. The field indicates the path to the project that SES Nordic Edition will load.

Workaround: If the path in Build Directory points to the wrong project, select the correct project by using the … button for Projects and navigating to the correct project location. The build directory will update automatically.

In addition to these known issues, check the current issues in the official Zephyr repository, since these might apply to the nRF Connect SDK fork of the Zephyr repository as well. To get help and report issues that are not related to Zephyr but to the nRF Connect SDK, go to Nordic’s DevZone.