Hi,

Just to try it on my Amiga/NetBSD:

ibiscus# time dd if=/dev/rsd1a bs=4096k of=/dev/null
5+1 records in
5+1 records out
24041472 bytes transfered in 38 secs (632670 bytes/sec)
0.0u 2.7s 0:38.27 7.2% 0+0k 6+1io 5pf+0w
ibiscus#

So, I try it on my i386 box:
margo# time dd if=/dev/rwd0a bs=4096k of=/dev/null
6+1 records in
6+1 records out
25804800 bytes transfered in 7 secs (3686400 bytes/sec)
0.0u 7.5s 0:07.70 97.6% 0+0k 7+1io 0pf+0w
margo#

margo have the new UVM kernel (not in the m68k port), and an IDE disk (=>97%) !

> I first detected this problem whilst playing with 1.2G but confirmed it
> is still in 1.3.
> 
> If I go:
> 
> time dd if=/dev/rwd0a bs=4096k of=/devnull
> 
> on the console I get:
> 
> panic: ptdi 5b66063
> 
> pmap_enter
> vm_fault
> vm_fault_wire
> vm_map_pageable
> vslock
> physio
> wdread
> spec_read
> ufsspec_read
> vn_read
> sys_read
> syscall
> 
> I can reproduce this...wd0 is a BSD partition for FreeBSD...the number after
> the ptdi doesn't seem to be constant.  Is anyone interested in this or is
> this just lossage from mixing file system formats ?
> 
> Darren