Background of Linux's "file-max" and "nr_open" limits on file descriptors (2021)

The article discusses the Unix background of Linux's 'file-max' and 'nr_open' kernel limits on file descriptors. It explains that these limits trace back to the early implementations of Unix, such as V7, which used fixed-size arrays for open files and file descriptors due to the simple kernel designs at the time. The sizes of these arrays were set during kernel compilation and influenced memory usage. Over time, as kernels evolved to dynamically allocate resources, limits were introduced to control the number of file descriptors per process and globally. The separate limits serve different purposes, with the per-process limit preventing resource leaks and the global limit restricting kernel memory usage. The article speculates on the naming conventions of these limits in Linux, suggesting historical reasons for the use of underscores and dashes in their names. The discussion provides insights into the evolution of file descriptor limits in Unix-based systems, shedding light on the rationale behind their design and implementation.