i believe this is due to the pmap problem, for which there is no work-around.

Wed, Nov 24, 2004 at 07:04:16PM -0500, John Klos:
> data fault: pc=100aaac addr=0
> kernel trap 30: data access exception
> Stopped in pid 7.1 (ioflush) at netbsd:pseg_get+0x3c:   ldxa 
> [%o2 + %
> g0] 20, %o2
> db>
> db> tr
> genfs_putpages(1, 1, 0, 5dda600, 5df9590, 11) at 
> netbsd:genfs_putpages+0x810
> VOP_PUTPAGES(5de9540, 0, 0, 11, e0017de8, 80) at netbsd:VOP_PUTPAGES+0x30
> ffs_full_fsync(5de9540, 10012, 14973d0, 0, 5de9540, 1892c00) at 
> netbsd:ffs_full_
> fsync+0x244
> ffs_fsync(5df9a60, 10012, 5de9760, 0, 1127840, 6) at netbsd:ffs_fsync+0x44
> VOP_FSYNC(5de9540, 1e4df00, 0, 0, 0, 5dcaea0) at netbsd:VOP_FSYNC+0x38
> ffs_sync(1f43600, 3, 1e4df00, 5dcaea0, 22, 1e) at netbsd:ffs_sync+0x20c
> sync_fsync(5df9d00, 10012, 14973d0, 0, 61f3370, 1845aa8) at 
> netbsd:sync_fsync+0x
> 88
> VOP_FSYNC(61f3370, 1e4df00, 8, 0, 0, 5dcaea0) at netbsd:VOP_FSYNC+0x38
> sched_sync(148bc00, 1817800, 1813400, 1846800, 1846800, 18af400) at 
> netbsd:sched
> _sync+0x1cc
> proc_trampoline(0, 0, 0, 0, 0, 0) at netbsd:proc_trampoline+0x4
> 
> 
> Any ideas?