Today was an unhappy morning for our NFS server.  

There were several mb_map full messages on the console, followed by a
drop into the debugger with:

vm_fault(0xf0995900, deadb000, 1, 0) -> 1
kernel: page fault trap, code=0
Stopped at _ip_freef+0x10: movl 0(%ebx), %edi

A trace showed:

ip_freef
ip_drain
m_reclaim
m_clalloc
fxp_add_rfabuf
fxp_intr
Xintr
--- interrupt ---
ip_reass
ipintr
Xsoftnet
--- interrupt ---
Bad frame pointer: 0xf0988e00
0xf0987e20

Continuing from the debugger caused repeated stops at _ip_freef.

Trying to reboot resulted in a panic, and then a sucessful dump.

db>reboot
Syncing disks... done
panic: unmount: dangling vnode


Obviously, I can increase NMBCLUSTERS and hope this never happens
again... any other ideas?