Bouyer san.
Thank you for your reply.

> Sigh, we may have to block interrupts for the reset. Please try the
> attached patch.

I try it. And got loops on intr again.
But there are little bit changes.

- normal boot
	-> I got ``bogus intr'' messages.
		messages displayed more than 20 minuts,
		disk access LED is off.

- boot with ddb
	-> wdcebug_mask = 0x7f 
	   	I found LED blinking.
		And kernel found the wd0, wd1 and booting the system
		
  ==========		
  $ egrep -v '^(wdc|__wd)' BOOT-LOG-FILE | tail
  system is cleanwdcstart from wdc_exec_xfer, flags 0x0
  ; not checking
  /dev/rwd0m: filewdc_exec_xfer 0xc1343034 channel 0 drive 0
  system is cleanwdcstart from wdc_exec_xfer, flags 0x0
  ; not checking
  /dev/rwd0n: filewdc_exec_xfer 0xc1343034 channel 0 drive 0
   system is cleanwdcstart from wdc_exec_xfer, flags 0x0
  ; not checking
  Setting tty flagwdc_exec_xfer 0xc1343034 channel 0 drive 0
  s.
  ==========  
I use Serial console that speed is 9600 baud. I think there is a difference
the interrupt timing.

If you are interested in this result, Would you please check the full log at

     http://www.hh.iij4u.or.jp/~nishio/netbsd/1016/
	normal-boot.log			 5.3kB(boot netbsd.p2)	
	ddb-wdcdebug_mask=0x7f.log	99.5kB(boot netbsd.p2 -d,
						    wdcdebug_mask=0x7f)

	ddb-wdcdebug_mask-0x09.log.bz2  10.5kB	(wdcdebug_mask=0x9)
	ddb-wdcdebug_mask-0x0c.log.bz2	 6.6kB	(wdcdebug_mask=0xc)
	ddb-wdcdebug_mask-0x11.log.bz2	 3.6kB	(wdcdebug_mask=0x11)

    netbsd.p2 --- GENERIC config-file + DIAGNOSTIC + DEBUG
	          with Bouyer san's new (spl) patch
		  kernel-tree is current around 2003-10-16 12:00 UTC

> > think that I'll try to test with simple system configuration(VGA and
> > Promise IDE only) on this weekend.
> > 
> > Is this meeninglee? 
> 
> It's worth a try, but I suspect it's more a disk bug.

Okay, I'll try it. I have a another(very old 6GB) disk.
I'll check same PCI card config but exchange the disk, simple config...

Thanks in advance.
-- 
NISHIO Yasuhiro