Adding definition for _SC_AVPHYS_PAGES

[MedourMehdi]

Related to Freemint kernel commit ea227c8

[mikrosk]

Hmm, maybe we should also update posix/sysconf.c? It's quite outdated (see _SC_LAST...) and maybe we should add at least something like Malloch(-1)/PAGE_SIZE there (and something similar for total memory size)

[MedourMehdi]

Yes I'm agree but doesn't it need to be part of an other commit?

I though that missing values ​​here are

    case _SC_CHILD_MAX:
      return CHILD_MAX; /* This is defined to 999 in mintlib but the actual call return UNLIMITED? */

    case _SC_NPROCESSORS_CONF:
      return __get_nprocs_conf (); /* Can check number of line in /kern/stat or just return 1 */

    case _SC_NPROCESSORS_ONLN:
      return __get_nprocs (); /* Can check number of line with CPU in /kern/stat or just return 1 */

    case _SC_PHYS_PAGES:
      return __get_phys_pages (); /* Using memtop and ramtop ? */

    case _SC_AVPHYS_PAGES:
      return __get_avphys_pages (); /*--> Don't really know if Malloc(-1) is ok. See https://www.atari-forum.com/viewtopic.php?p=470691#p470691  */

About SC_LAST I don't know where it comes from and what it does

[mikrosk]

_SC_LAST is, if I understand correctly, the last index of all handled values, like you updated in FreeMiNT.

The proposal with Malloch() was the first thing which came into my mind -- obviously, it's gonna be a hack anyway but we should return at least something so given software will not be working only in FreeMiNT.

And sure, a separate commit is fine.

[MedourMehdi]

Yes I understand that too but it doesn't exist elsewhere (at least no trace in my google searches) and it's not used in mintlib so I was wondering if you'll have some clue about it.

Yes agreed better to give something than crashing :)