Non-Volatile Storage (NVS)

Elements, represented as id-data pairs, are stored in flash using a FIFO-managed circular buffer. The flash area is divided into sectors. Elements are appended to a sector until storage space in the sector is exhausted. Then a new sector in the flash area is prepared for use (erased). Before erasing the sector it is checked that identifier - data pairs exist in the sectors in use, if not the id-data pair is copied.

The id is a 16-bit unsigned number. NVS ensures that for each used id there is at least one id-data pair stored in flash at all time.

NVS allows storage of binary blobs, strings, integers, longs, and any combination of these.

Each element is stored in flash as metadata (8 byte) and data. The metadata is written in a table starting from the end of a nvs sector, the data is written one after the other from the start of the sector. The metadata consists of: id, data offset in sector, data length, part (unused), and a CRC. This CRC is only calculated over the metadata and only ensures that a write has been completed. The actual data of the element can be protected by a different (and optional) CRC-32. Use the CONFIG_NVS_DATA_CRC configuration item to enable the data part CRC. Note: the data CRC is checked only when the whole data of the element is read. The data CRC is not checked for a partial read, as it is stored at the end of the element data area. Note 2: enabling the data CRC feature on a previously existing NVS content without data CRC will make all existing data invalid.

A write of data to nvs always starts with writing the data, followed by a write of the metadata. Data that is written in flash without metadata is ignored during initialization.

During initialization NVS will verify the data stored in flash, if it encounters an error it will ignore any data with missing/incorrect metadata.

NVS checks the id-data pair before writing data to flash. If the id-data pair is unchanged no write to flash is performed.

To protect the flash area against frequent erases it is important that there is sufficient free space. NVS has a protection mechanism to avoid getting in a endless loop of flash page erases when there is limited free space. When such a loop is detected NVS returns that there is no more space available.

For NVS the file system is declared as:

static struct nvs_fs fs = {
.flash_device = NVS_FLASH_DEVICE,
.sector_size = NVS_SECTOR_SIZE,
.sector_count = NVS_SECTOR_COUNT,
.offset = NVS_STORAGE_OFFSET,
};

where

  • NVS_FLASH_DEVICE is a reference to the flash device that will be used. The device needs to be operational.

  • NVS_SECTOR_SIZE is the sector size, it has to be a multiple of the flash erase page size and a power of 2.

  • NVS_SECTOR_COUNT is the number of sectors, it is at least 2, one sector is always kept empty to allow copying of existing data.

  • NVS_STORAGE_OFFSET is the offset of the storage area in flash.

Flash wear

When writing data to flash a study of the flash wear is important. Flash has a limited life which is determined by the number of times flash can be erased. Flash is erased one page at a time and the pagesize is determined by the hardware. As an example a nRF51822 device has a pagesize of 1024 bytes and each page can be erased about 20,000 times.

Calculating expected device lifetime

Suppose we use a 4 bytes state variable that is changed every minute and needs to be restored after reboot. NVS has been defined with a sector_size equal to the pagesize (1024 bytes) and 2 sectors have been defined.

Each write of the state variable requires 12 bytes of flash storage: 8 bytes for the metadata and 4 bytes for the data. When storing the data the first sector will be full after 1024/12 = 85.33 minutes. After another 85.33 minutes, the second sector is full. When this happens, because we’re using only two sectors, the first sector will be used for storage and will be erased after 171 minutes of system time. With the expected device life of 20,000 writes, with two sectors writing every 171 minutes, the device should last about 171 * 20,000 minutes, or about 6.5 years.

More generally then, with

  • NS as the number of storage requests per minute,

  • DS as the data size in bytes,

  • SECTOR_SIZE in bytes, and

  • PAGE_ERASES as the number of times the page can be erased,

the expected device life (in minutes) can be calculated as:

SECTOR_COUNT * SECTOR_SIZE * PAGE_ERASES / (NS * (DS+8)) minutes

From this formula it is also clear what to do in case the expected life is too short: increase SECTOR_COUNT or SECTOR_SIZE.

Flash write block size migration

It is possible that during a DFU process, the flash driver used by the NVS changes the supported minimal write block size. The NVS in-flash image will stay compatible unless the physical ATE size changes. Especially, migration between 1,2,4,8-bytes write block sizes is allowed.

Sample

A sample of how NVS can be used is supplied in samples/subsys/nvs.

Troubleshooting

MPU fault while using NVS, or -ETIMEDOUT error returned

NVS can use the internal flash of the SoC. While the MPU is enabled, the flash driver requires MPU RWX access to flash memory, configured using CONFIG_MPU_ALLOW_FLASH_WRITE. If this option is disabled, the NVS application will get an MPU fault if it references the internal SoC flash and it’s the only thread running. In a multi-threaded application, another thread might intercept the fault and the NVS API will return an -ETIMEDOUT error.

API Reference

The NVS subsystem APIs are provided by nvs.h:

Non-volatile Storage Data Structures
Non-volatile Storage APIs