> A search of the GNATS database for closed PR's with that string should
> yield more details.

I found pr #3703 and compared your fix for pmap.c with the current
version. The difference I found was in pv_unlink4m (and similar
code for sun4c):

After patch from pr #3703:

                       FREE(npv, M_VMPVENT);
               } else {
                       pv->pv_pmap = NULL;
                       pv->pv_va = NULL;
                       pv->pv_flags = NULL;
               }

Current code:

			FREE(npv, M_VMPVENT);
		} else {
			pv->pv_pmap = NULL;
			pv->pv_flags |= PV_C4M;
			return;
		}

Gdb:ing around in the core, I even found

(gdb) print *pv_table
$4 = {pv_next = 0x0, pv_pmap = 0x0, pv_va = 4028600320, pv_flags = 5}

Does that make any sense? Not to me I have to admit :-)

Harald.