Hi,
I've got this problem while stress-testing the ste driver I've back-ported
to 1.5.3_RC2. It's not in the tree yet so it's why I didn't send a PR. 
however the problem looks generic enouth to either be present in -current,
and/or be present in 1.5.x with other ethernet drivers.

After several hours of tests I got the panic:
pool ipqepl: reentrancy atr file .../tcp_subr.c line 1022
previous entry at file ....ip_input.c line 688

The ddb trace is:
panic()
pool_put()
tcp_freeq()
tcp_drain()
m_reclaim()
ste_add_rxbuf()
ste_rxintr()
ste_intr()
Xintr 5
--- interrupt ---
_pool_get()
ip_input()
ipintr()


Should pool_* operations on ipqepl be protected by splnet() ?

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--