Linux or Proprietary Real Time Operating System?

Although it is a fine choice for some applications, Linux is a hefty piece of software that can be confusing for developers who are not familiar with its intricacies.  Your project only calls for USB, Ethernet, and LCD interfaces, but you'll typically get much more than those components from Linux.  Even if memory is not a major issue for you, the presence of additional components could be a problem.  For instance, because Linux utilizes memory management hardware, the application code that you would write for a Linux-based system would likely be much more complex than that needed for a system incorporating a leaner RTOS.  If you don't actually require memory management support, then this added complexity would just represent wasted effort.

The bulkiness of Linux, in addition to unnecessarily adding to your project's complexity, can create performance issues.  If you require real-time performance (even what is typically considered soft real-time performance) an RTOS may be a better answer.  The best RTOSes use efficient, preemptive schedulers, helping to ensure that the important parts of your application perform according to expectation.

Another attribute of a good RTOS that could prove beneficial to your project is stability.  The releases for top-tier commercial RTOSes are carefully managed; updated versions of the software are made available only when there are bug fixes or new features.  Linux, on the other hand, is a moving target.  The kernel and the other pieces of the operating system change frequently, so the Linux that you download today is likely not the same Linux that you downloaded last week.

Ultimately, your decision of whether or not to use Linux should be made with the needs of your particular project in mind.  As I noted above, Linux is not without its advantages.  However, for many embedded systems like the one you outlined, a proprietary RTOS is a much more efficient and easy-to-use solution.

Although it is a fine choice for some applications, Linux is a hefty piece of software that can be confusing for developers who are not familiar with its intricacies.  Your project only calls for USB, Ethernet, and LCD interfaces, but you'll typically get much more than those components from Linux.  Even if memory is not a major issue for you, the presence of additional components could be a problem.  For instance, because Linux utilizes memory management hardware, the application code that you would write for a Linux-based system would likely be much more complex than that needed for a system incorporating a leaner RTOS.  If you don't actually require memory management support, then this added complexity would just represent wasted effort.

The bulkiness of Linux, in addition to unnecessarily adding to your project's complexity, can create performance issues.  If you require real-time performance (even what is typically considered soft real-time performance) an RTOS may be a better answer.  The best RTOSes use efficient, preemptive schedulers, helping to ensure that the important parts of your application perform according to expectation.

Another attribute of a good RTOS that could prove beneficial to your project is stability.  The releases for top-tier commercial RTOSes are carefully managed; updated versions of the software are made available only when there are bug fixes or new features.  Linux, on the other hand, is a moving target.  The kernel and the other pieces of the operating system change frequently, so the Linux that you download today is likely not the same Linux that you downloaded last week.

Ultimately, your decision of whether or not to use Linux should be made with the needs of your particular project in mind.  As I noted above, Linux is not without its advantages.  However, for many embedded systems like the one you outlined, a proprietary RTOS is a much more efficient and easy-to-use solution.