Datalogger example

From Miosix Wiki
Jump to navigation Jump to search
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

You can find this example here https://github.com/fedetft/miosix-kernel/tree/testing/miosix/_examples/datalogger

Some applications - possibly real-time - require to log data, and do so from multiple threads. While a simple library that opens a file and writes to is sufficient for simple applications, there are some problems:

  • the Miosix filesystem uses typically an SD card as storage, and SD and other FLASH based storage devices may pause to do wear leveling. SD cards, even good quality class 10 ones, may pause for up to 1 second.
  • the Miosix OS, contrary to Linux, does not do much buffering at the filesystem layer. This is desirable, since when an OS runs on as little as a few tens of KB of RAM, the last thing you want is an OS that uses an unquatifiable amount of memory.

This example code shows a high-performance logging class that

  • has a nonblocking log() member function, which can be called concurrently from multiple threads, to log a user-defined class or struct. Tscpp (https://github.com/fedetft/tscpp) is used to serialize data. Being nonblocking, it can be called also in real-time threads of your codebase with confidence.
  • buffers data to compensate for the delays of the storage medium

To configure the logger for your application, to trade off buffer space vs write data rate, uoy can edit Logger.h

   static const unsigned int filenameMaxRetry = 100; ///< Limit on new filename
   static const unsigned int maxRecordSize    = 128; ///< Limit on logged data
   static const unsigned int numRecords       = 128; ///< Size of record queues
   static const unsigned int bufferSize       = 4096;///< Size of each buffer
   static const unsigned int numBuffers       = 4;   ///< Number of buffers
   static constexpr bool logStatsEnabled      = true;///< Log logger stats?