Setting up LittleFS¶

What you provide¶

A LittleFS checkout, a struct lfs_config carrying your block-device callbacks, and a mounted lfs_t. Mount it before creating any file, and keep it mounted for as long as the library holds one.

Build¶

Compile LittleFS's lfs.c and lfs_util.c into your target and link SolidSyslog::LittleFs, which carries the adapter's own sources.

LittleFS's headers are built upstream with a narrower warning set than this library's: lfs_util.h does not compile under -Wsign-conversion. Add the tree as a system include so its warnings do not reach your build, rather than relaxing your own:

target_include_directories(your_target SYSTEM PRIVATE ${LITTLEFS_PATH})

Creating a file¶

static uint8_t recordFileCache[LFS_CACHE_SIZE];

struct SolidSyslogFile* file =
    SolidSyslogLittleFsFile_Create(&lfs, recordFileCache, sizeof(recordFileCache));

The buffer must be at least the cache_size you mounted with, and must outlive the file. It is the file's own cache, so give each file its own.

What will catch you out¶

The cache buffer is per file, not per filesystem. LittleFS already has a read cache and a program cache of its own; this is a third, and every open file needs one. Creating two files from one buffer corrupts both, and nothing reports it - the adapter cannot tell two callers apart.

Getting the size wrong is refused, not tolerated. A buffer smaller than cache_size makes Create return the Null file and raise a CRITICAL. Size it from the same constant you mount with rather than a number typed twice.

block_cycles defaults to no wear levelling. LittleFS disables dynamic wear levelling unless you set it, and a store that rewrites records in rotation is exactly the workload that needs it. Set it to a few hundred.

A mount is not the adapter's to make. If the filesystem is unmounted while a file is open, every later call fails through LittleFS rather than through the library, and the library has nothing to report.