Introduce PRISM_HAS_MMAP macro

[hasumikin]

This is to know if the memory map interface is available in the target platform by using _POSIX_MAPPED_FILES and _WIN32

ref #2217

[kddnewton]

@hasumikin I think _POSIX_MAPPED_FILES is defined in unistd.h, so you probably need to include that before doing the check.

[hasumikin]

@kddnewton
Thank you.
My problem was fixed but another static check error was revealed that I couldn't figure out how to resolve :(

[kddnewton]

Thanks @hasumikin you should be able to rebase now and it should be fixed.

My feedback on the PR, is that I think we should keep PM_STRING_MAPPED around, even if PRISM_HAS_MMAP is false. Maybe we should rename it to PM_STRING_FILE or something, but that way we could malloc and read the file into a const uint8_t * the same way you would mmap it.

[kddnewton]

I'm also happy to make those changes after we merge this if you'd prefer. No problem! 😀

[hasumikin]

@kddnewton

My feedback on the PR, is that I think we should keep PM_STRING_MAPPED around, even if PRISM_HAS_MMAP is false. Maybe we should rename it to PM_STRING_FILE or something, but that way we could malloc and read the file into a const uint8_t * the same way you would mmap it.

If I didn't get you wrong, we'd better create one in addition to PM_STRING_MAPPED and not include _FILE because (from the embedded system point of view,) we can not assume we have a filesystem.

How about PM_STRING_GENERAL for instance?

/* pm_string.h */
typedef struct {
　(......)
    /** The type of the string. This field determines how the string should be freed. */
    enum {
    　(......)

        /** This string owns its memory, and should be freed using `pm_string_free`. */
        PM_STRING_OWNED,

        /** This is a general string, and should be taken care of by the user. */
        PM_STRING_GENERAL,    // 👈

#ifdef PRISM_HAS_MMAP
        /** This string is a memory-mapped file, and should be freed using `pm_string_free`. */
        PM_STRING_MAPPED
#endif
    } type;
} pm_string_t;

[kddnewton]

Actually I realize now it should simply return a PM_STRING_OWNED if it doesn't have mmap, because that's effectively what it is. I'll merge this as-is and see what it takes to do that.

[kddnewton]

Thank you for the contribution!