uma_zalloc_arg panic!

Seems my recent post about the extremely slow probing
of fxp0 during boot (6+ minutes) went unnoticed.   The
slow probing has now manfest itself as a panic.

uma_zalloc_arg+0x12a		lock pop

db> trace
Tracing pid 0 tid 0 td 0xffffffff8062ff20
uma_zalloc_arg at ... +0.12a
vm_map_entry_create at ... +0x23
vm_map_insert at ... +0x258
contigmalloc2 at ... +0xc1
contigmalloc  at ...  +0xe2
alloc_bounce_pages at ... 0x81
bus_dmamap_create at ... 0x145
fxp_attach at ... 0xc64

I'll also note the recent rototilling of mount has rendered
my back up kernels useless because it is no longer possible
to mount my filesystems.  Booting to single user mode is
fine.  At the # prompt:

# fsck -y
# mount -a 
mount: <rw> unknown option
# mount -f -a
mount: <rw> unknown option

-- 
Steve