freebsd stable - Jan 2004 - RELENG_4 crash: UDMA100 drive on with old UDMA33 controller

I am having problems with RELENG_4 (cvsup last night) crashing.  I
recently purchased a WDC WD1600JB (160GB) disk drive to attach to my
trusty, old home server:
 ASUS P5A-B Motherboard with integrated Acer UDMA33 controller
 Fujitsu MPC3032AT/6018 (3GB) disk drive (channel 0 master)
 FX322M/w03 CDROM drive (channel 0 slave)

I presently have the WD1600JB was added as the channel 1 master
(no slave).  It is connected to the controller using the 80
conductor/40-pin cable provided with the drive.  (I've also
tried to setups with the WD1600JB replacing the CDROM drive
as slave on channel 0, using both 80/40 and 40/40 cable with
like results.)  I have configured the BIOS (as directed in
the WDC documents) for C/H/S for 1023/16/63 (528M).

RELENG_4 sees that a UDMA100 drive is connected to ata1 and
tries fallback to UDMA33 and then to PIO4.  This is the last
page of messages (which shows the panic).  I'll try to type
in more if this isn't enough (sorry, but I don't a laplink
cable handy):

ad2: trying fallback to PIO mode
ata1: resetting devices .. done

Fatal trap 12: page fault while in kernel mode
fault virtual address	= 0x78007a
fault code		= supervisor read, page not present
instruction pointer	= 0x8:0xc018760e
stack pointer		= 0x10:0xc03b9314
frame pointer		= 0x10:0xc02931c
code segment		= base 0x0, limit 0xfffff, type 0x1b
			= DPL 0, pres 1, def32 1, gran 1
processor eflags	= interrupt enabled, resume, IOPL = 0
current process	= Idle
interrupt mask		= bio
trap number		= 12
panic: page fault

This seems to be related problems others have noted with regards
to Acer UDMA33 controller and RELENG_4.  It does seem specific
to 4.x as tests against 4.8 failed same as above yet tests against
FreeBSD 5.2 and NetBSD 1.6 were okay.

Kurt

On Mon, 12 Jan 2004, Kurt D. Zeilenga wrote:
> I am having problems with RELENG_4 (cvsup last night) crashing.  I
> recently purchased a WDC WD1600JB (160GB) disk drive to attach to my
> trusty, old home server:
Try running this in loader(8) before booting the kernel:

set hw.ata.ata_dma=0

This disables DMA support. If your system runs OK with this set then you
might want to check your cabling, or try isolating your disks to their own
channels.
>  ASUS P5A-B Motherboard with integrated Acer UDMA33 controller
>  Fujitsu MPC3032AT/6018 (3GB) disk drive (channel 0 master)
>  FX322M/w03 CDROM drive (channel 0 slave)
> I presently have the WD1600JB was added as the channel 1 master
> (no slave).  It is connected to the controller using the 80
> conductor/40-pin cable provided with the drive.  (I've also
> tried to setups with the WD1600JB replacing the CDROM drive
> as slave on channel 0, using both 80/40 and 40/40 cable with
> like results.)  I have configured the BIOS (as directed in
> the WDC documents) for C/H/S for 1023/16/63 (528M).
-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite@gumbysoft.com          |  www.FreeBSD.org

On Mon, 12 Jan 2004, Kurt D. Zeilenga wrote:
> I am having problems with RELENG_4 (cvsup last night) crashing.  I
> recently purchased a WDC WD1600JB (160GB) disk drive to attach to my
> trusty, old home server:
>  ASUS P5A-B Motherboard with integrated Acer UDMA33 controller
>  Fujitsu MPC3032AT/6018 (3GB) disk drive (channel 0 master)
>  FX322M/w03 CDROM drive (channel 0 slave)
>
> I presently have the WD1600JB was added as the channel 1 master
> (no slave).  It is connected to the controller using the 80
> conductor/40-pin cable provided with the drive.  (I've also
> tried to setups with the WD1600JB replacing the CDROM drive
> as slave on channel 0, using both 80/40 and 40/40 cable with
> like results.)  I have configured the BIOS (as directed in
> the WDC documents) for C/H/S for 1023/16/63 (528M).
>
> RELENG_4 sees that a UDMA100 drive is connected to ata1 and
> tries fallback to UDMA33 and then to PIO4.  This is the last
> page of messages (which shows the panic).  I'll try to type
> in more if this isn't enough (sorry, but I don't a laplink
> cable handy):
>
> ad2: trying fallback to PIO mode
> ata1: resetting devices .. done
>
> Fatal trap 12: page fault while in kernel mode
> fault virtual address	= 0x78007a
> fault code		= supervisor read, page not present
> instruction pointer	= 0x8:0xc018760e
> stack pointer		= 0x10:0xc03b9314
> frame pointer		= 0x10:0xc02931c
> code segment		= base 0x0, limit 0xfffff, type 0x1b
> 			= DPL 0, pres 1, def32 1, gran 1
> processor eflags	= interrupt enabled, resume, IOPL = 0
> current process	= Idle
> interrupt mask		= bio
> trap number		= 12
> panic: page fault
>
> This seems to be related problems others have noted with regards
> to Acer UDMA33 controller and RELENG_4.  It does seem specific
> to 4.x as tests against 4.8 failed same as above yet tests against
> FreeBSD 5.2 and NetBSD 1.6 were okay.
I haven't really noticed a problem with crashing or panics, but I do
have the resetting devices errors show up in my logs quite a bit.
I had thought it was just related to the [expletive deleted] IBM drives,
but it would seem that you are using the same motherboard I am.
I keep meaning to just purchase a UDMA-100 controller card to stick in
there...but now my server's on the other side of the continent from me.
Ah well.
Me too, almost.
								-Tenebrae.
---
The sending of any unsolicited email advertising messages to this domain
may result in the imposition of civil liability against you in accordance
with Cal. Bus. & Prof. Code Section 17538.45.