See https://savannah.gnu.org/bugs/index.php?item_id=66853
Ignoring the suggestion that the malloc.m4 test is wrong (I can't see that it is, at any rate), the reporter says that this comment: On Solaris 11.3, the three functions return NULL with errno set to EAGAIN, not ENOMEM, when the argument is larger than PTRDIFF_MAX. does not match behavior they observe on Solaris 11.3 (in particular, when an allocation of PTRDIFF_MAX + 1 bytes is requested, Solaris returns the allocation successfully, and errno remains 0). Is it possible that the documentation is incomplete, i.e. that Solaris 11.3 *can* return EAGAIN in this situation, but can also simply succeed? -- Web: rrt.sc3d.org
