Re: [PATCH] mm,hwpoison: return -EBUSY when page already poisoned

So we can send a generic SIGBUS, but not one with the usual extra
information about the location of the error.

+1, without virtual address of the error, it's not quite recoverable
from the user space' point of view.

Is it possible to add a best-effort VMA walk to kill_me_now() and
signal via force_sig_mceerr when viable?