Configuring external XIP

This guide describes the external execute in place (XIP) support for the nRF5340 SoC. The nRF5340 SoC is equipped with a Quad Serial Peripheral Interface (QSPI) memory interface, which is capable of exposing QSPI flash as memory for the CPU used to fetch and execute the program code. The QSPI is available for the application core. Therefore, it is possible to relocate a part of the application’s code to an external memory. NCS supports dividing an application into an internal and external part, along with the MCUboot support.

Enabling configuration options

Application configuration must support an external XIP and image splitting.

Enable the following options:

Additionally, set the following options:

  • CONFIG_CUSTOM_LINKER_SCRIPT to "<linker_file_for_relocation>"

  • CONFIG_UPDATEABLE_IMAGE_NUMBER to 3

  • CONFIG_FLASH_INIT_PRIORITY to 40 - You must ensure the QSPI device initialization priority, as it makes the external XIP code accessible before it is executed. If any initialization code is expected to be run from the QSPI XIP, then its initialization priority value must be lower than the QSPI device initialization priority.

Setting up QSPI flash

The QSPI flash DTS must be available for the application and the MCUboot. You must correctly set up the QSPI flash chip in the board devicetree file, including the operating mode. The flash chip does not have to run in the QSPI mode for XIP to function, but using other modes will reduce the execution speed of the application.

See the following snippet for an example of the Nordic Thingy:53 configuration that supports DSPI:

    &qspi {
    status = "okay";
    pinctrl-0 = <&qspi_default>;
    pinctrl-1 = <&qspi_sleep>;
    pinctrl-names = "default", "sleep";
    mx25r64: mx25r6435f@0 {
        compatible = "nordic,qspi-nor";
        reg = <0>;
        writeoc = "pp2o";
        readoc = "read2io";
        sck-frequency = <8000000>;
        jedec-id = [c2 28 17];
        sfdp-bfp = [
            e5 20 f1 ff  ff ff ff 03  44 eb 08 6b  08 3b 04 bb
            ee ff ff ff  ff ff 00 ff  ff ff 00 ff  0c 20 0f 52
            10 d8 00 ff  23 72 f5 00  82 ed 04 cc  44 83 68 44
            30 b0 30 b0  f7 c4 d5 5c  00 be 29 ff  f0 d0 ff ff
        ];
        size = <67108864>;
        has-dpd;
        t-enter-dpd = <10000>;
        t-exit-dpd = <35000>;
    };
};

Note

Due to QSPI peripheral product anomaly, the QSPI peripheral must be ran with the HFCLK192MCTRL=0 setting. Any other value may cause undefined operation of the device.

Add the following to the DTS overlay for your board:

/ {
    chosen {
            nordic,pm-ext-flash = &mx25r64;
    };
};

Setting up static partition manager

You need to complete the setup in order to use a static partitioning in your project. The configuration must have 3 images with 2 slots each:

Static partitioning slots in the nRF5340 SoC

Static partitioning slots in the nRF5340 SoC.

  • The first set of slots is for the internal flash part of the application. These slots should be named mcuboot_primary and mcuboot_secondary.

  • The second set of slots is for the network core update. These slots should be named mcuboot_primary_1 and mcuboot_secondary_1.

  • The third set of slots is for the QSPI XIP part of the application. These slots should be named mcuboot_primary_2 and mcuboot_secondary_2.

This means a basic dual image configuration for the nRF5340 DK needs to describe an external QSPI XIP code partition as mcuboot_primary_2 partition. Additionally, ensure that:

  • The mcuboot_primary_2 address is expressed as the QSPI flash physical address.

  • The device field is the QSPI device name.

  • The region field is set as external_flash.

See the following snippet for an example of the static configuration for partition manager:

app:
    address: 0x10200
    end_address: 0xe4000
    region: flash_primary
    size: 0xd3e00
external_flash:
    address: 0x120000
    device: MX25R64
    end_address: 0x800000
    region: external_flash
    size: 0x6e0000
mcuboot:
    address: 0x0
    end_address: 0x10000
    region: flash_primary
    size: 0x10000
mcuboot_pad:
    address: 0x10000
    end_address: 0x10200
    region: flash_primary
    size: 0x200
mcuboot_primary:
    address: 0x10000
    end_address: 0xf0000
    orig_span: &id001
    - mcuboot_pad
    - app
    region: flash_primary
    size: 0xe0000
    span: *id001
mcuboot_primary_1:
    address: 0x0
    device: flash_ctrl
    end_address: 0x40000
    region: ram_flash
    size: 0x40000
mcuboot_primary_app:
    address: 0x10200
    end_address: 0xf0000
    orig_span: &id002
    - app
    region: flash_primary
    size: 0xdfe00
    span: *id002
mcuboot_secondary:
    address: 0x0
    device: MX25R64
    end_address: 0xe0000
    region: external_flash
    size: 0xe0000
mcuboot_secondary_1:
    address: 0xe0000
    device: MX25R64
    end_address: 0x120000
    region: external_flash
    size: 0x40000
mcuboot_primary_2:
    address: 0x120000
    device: MX25R64
    end_address: 0x160000
    region: external_flash
    size: 0x40000
mcuboot_secondary_2:
    address: 0x160000
    device: MX25R64
    end_address: 0x1a0000
    region: external_flash
    size: 0x40000
otp:
    address: 0xff8100
    end_address: 0xff83fc
    region: otp
    size: 0x2fc
pcd_sram:
    address: 0x20000000
    end_address: 0x20002000
    region: sram_primary
    size: 0x2000
ram_flash:
    address: 0x40000
    end_address: 0x40000
    region: ram_flash
    size: 0x0
rpmsg_nrf53_sram:
    address: 0x20070000
    end_address: 0x20080000
    placement:
        before:
        - end
    region: sram_primary
    size: 0x10000
settings_storage:
    address: 0xf0000
    end_address: 0x100000
    region: flash_primary
    size: 0x10000
sram_primary:
    address: 0x20002000
    end_address: 0x20070000
    region: sram_primary
    size: 0x6e000

Configuring linker script

To relocate code to the external flash, you need to configure a linker script. The script needs to describe the EXTFLASH flash memory block to which the code will be linked. The ORIGIN of the area can be calculated using following elements:

  • The QSPI memory starting with the 0x10000000 internal memory address.

  • The offset of an external application part image within the QSPI flash. The external application code partition is mapped by the mcuboot_primary_2 PM partition.

  • The image header size of the MCUboot image (0x200).

See the following example of the calculation:

#include <zephyr/linker/sections.h>
#include <zephyr/devicetree.h>
#include <zephyr/linker/linker-defs.h>
#include <zephyr/linker/linker-tool.h>

MEMORY
{
    /* This maps in mcuboot_primary_2 partition defined in pm_static.yaml
    * components for ORIGIN calculation:
    *  - 0x10000000: offset of QSPI external memory in SoC memory mapping.
    *  - 0x120000: mcuboot_primary_2 offset in QSPI external memory
    *  - 0x200: image header size.
    * The size of this region is size of mcuboot_primary_2 reduced by the
    * image header size.
    */
    EXTFLASH (wx) : ORIGIN = 0x10120200, LENGTH = 0x3FE00
}

#include <zephyr/arch/arm/cortex_m/scripts/linker.ld>

Setting up code relocation

Relocating code to QSPI XIP is a part of the project’s CMakeLists.txt file. You can set up the relocation on a file or library basis using the zephyr_code_relocate() function. For example, to relocate a file in the application, use the following configuration:

zephyr_code_relocate(FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/bluetooth.c LOCATION EXTFLASH_TEXT NOCOPY)
zephyr_code_relocate(FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/bluetooth.c LOCATION RAM_DATA)

where the first line relocates the XIP code (.text) and the second line relocates the data initialization content section (.data).

Similarly, it is possible to relocate certain libraries, for example:

zephyr_code_relocate(LIBRARY subsys__mgmt__mcumgr__mgmt LOCATION EXTFLASH_TEXT NOCOPY)
zephyr_code_relocate(LIBRARY subsys__mgmt__mcumgr__mgmt LOCATION RAM_DATA)

Flashing the project

For the nRF5340 DK and other boards equipped with flash working in the QSPI mode, use the west flash command. For other cases, flashing needs to be done manually.

Flashing to external flash in SPI/DSPI mode

Flashing an application with west triggers the nrfjprog runner. The runner uses the default system settings that configure the application in the QSPI mode when flashing the external flash. You can change this behavior by using a custom Qspi.ini configuration file, however, it will prevent flashing through west.

Note

The Qspi.ini file is required to work on the Nordic Thingy:53.

If you wish to use the Qspi.ini file, you will need to manually flash the HEX files in the repository. For example, for the nRF5340: SMP Server with external XIP sample, you need to flash the following files (paths are relative to the build directory):

  • <cpunet_build_subdirectory>/zephyr/merged_CPUNET.hex * For Bluetooth stack application the path is <cpunet_build_subdirectory> hci_ipc.

  • mcuboot/zephyr/zephyr.hex

  • zephyr/internal_flash_signed.hex

  • zephyr/qspi_flash_signed.hex

Use the following commands to flash and verify the Simple Management Protocol (SMP) server sample:

nrfjprog -f NRF53 --coprocessor CP_NETWORK --sectorerase --program hci_ipc/zephyr/merged_CPUNET.hex --verify
nrfjprog -f NRF53 --sectorerase --program mcuboot/zephyr/zephyr.hex --verify
nrfjprog -f NRF53 --sectorerase --program zephyr/internal_flash_signed.hex --verify
nrfjprog -f NRF53 --qspisectorerase --program zephyr/qspi_flash_signed.hex --qspiini <path_to>/Qspi.ini --verify
nrfjprog -f NRF53 --reset

Note

The external flash chip must be connected to the dedicated QSPI peripheral port pins of the nRF5340 SoC. It is not possible to program an external flash chip that is connected to different pins using nrfjprog.

Troubleshooting

Refer to the following sections for information on how to solve the most common issues.

Module does not appear to start

When using QSPI XIP, a frequent issue is the module not starting or crashing before the application runs. This often results from a mismatch in init priorities between the code on QSPI flash and the QSPI flash device.

To debug this issue, you can use a debugger such as GNU Debugger (GDB) to single-step through the application code until a QSPI address is encountered. The backtrace functionality can then show which part of the code is responsible for the issue, and you can adjust the init priority of that module accordingly.

Given that the QSPI flash init priority defaults to 41 at the POST_KERNEL level, take into account the following points:

  • There should be no QSPI flash residing code that has an init priority value that is less than or equal to the POST_KERNEL level 41.

  • No interrupt handlers in the QSPI flash should be enabled until the QSPI flash driver has been initialized.

Module does not boot after update

This issue can occur if there is a mismatch between the internal flash code and the QSPI XIP code. Both slots must be running the same build to successfully boot. The application will fail to boot in the following cases:

  • If one of the updates is not loaded.

  • If a different build is loaded to one of the slots.

  • If one of the loaded updates is corrupt and deleted.