vm_page: Fix PG_ZERO handling in vm_page_grab_*()

[markj]

Summary

vm_page_grab_pages() and vm_page_grab_pages_unlocked() were honouring
PG_ZERO for resident, invalid pages, but this isn't really sound.
PG_ZERO should only be considered at allocation time. Unconditionally
already-resident pages if VM_ALLOC_ZERO was specified.

[alc]

sys/vm/vm_page.c

5235–5236

Just as a matter style, for consistency, could you swap the order of the tests (vm_page_none_valid(m), (allocflags & VM_ALLOC_ZERO) != 0) either here or in vm_page_grab_pages_unlocked().

[markj]

Make tests more consistent.

[kib]

Do we have any caller using vm_page_grab_pages(_unlocked)(VM_ALLOC_ZERO)? I did not found any.

I thought what could be a use for such call, and I really do not see how it can be useful. Both vnode and swap objects might have the page content on the volume, so the invalid page we found on the queue cannot be safely zeroed without consulting pager first. In the end, the decision to zero such page belongs to the caller.