splinter_close() : Closes A Splinter Memory Store

The splinter_close() function closes the current process' view of a shared Splinter store. When all processes have closed their view of the store, the kernel handles housekeeping.

In shared-memory configuration (not persistent), programs only need to call splinter_close() if they intend to open a different store. Smaller stores are likely to persist in /dev/shm until reboot, even if all processes have closed their view of it, just because the kernel won't reclaim them until it needs to (like any other paging).

However, if using persistent mode, you should call splinter_close() as soon as you can know (in code) when you won't need access to the store any longer. It helps to put it in your clean-up / exit handlers, just so hanging processes don't keep large chunks of contiguous memory busy any longer than absolutely necessary.

Basically, if persisting, treat splinter_close() like close(), but it takes no arguments, and you only have to call it once at the end.

Prototype & Rationale:

The function splinter_close() is prototyped in splinter.h as:

int splinter_close(void);

If / when Splinter allows for multiple file-backed stores to be open simultaneously, splinter_close() will accept a file descriptor as an argument.

Return Value:

splinter_close() returns void and does not set errno.

Example:

The following closes a store at the end of work:

#include "splinter.h"

int i = -1;

i = splinter_open("debug_bus");
if (i != 0) {
    exit(127);
}
splinter_close();
// any splinter calls (except to open) after this will fail

CLI & Tooling Example Uses:

To see the splinter_close() in-use within the tools that ship with Splinter, see the following source files:

Due to the simplicity of the function, no further examples are warranted.

See Also:

The following functions are also relevant: splinter_create(), splinter_open_or_create(), splinter_create_or_open(), splinter_open_numa() and splinter_open().