Re: udma66 controller changed devices

[Robert Williams <bob bob usuhs mil>]

"Rich Payne" wrote:

>Hmmmmm.....interesting. Is this the kernel from API or is this your own? I

This is my own build of the HardData kernel...to add HPT366 and Promise UDMA.
This kernel has been patched for udma.
What kernel from API would you recommend?

When I used the kernel on the UP1000 install disk,
boot time found "a filesystem with errors".
I got "UNEXPECTED INCONSISTENCY: RUN fsck MANUALLY."
even though the filesystem has brought down successfully using
the HardData kernel.  The subsequent fsck found
"Inode xxx has imagic flag set", "Illegal block",
and "Inode xxx is in use, but has dtime set" problems
too numerous to fix by hand.  I'm afraid to try it again.


>was just looking at the CONFIG_BLK_DEV_OFFBOARD kernel option, and
>that
>does exactly what you describe, though I'm pretty sure I didn't compile
>the API kernel with that enabled (you can look at the config file on the
>install floppy for this).


__________________________________________________________
Here is the original message:
On Wed, 29 Mar 2000, Robert Williams wrote:

> I inserted a HPT366 UDMA controller into my UP1000
> with no disks attached and the system
> could not boot to completion.
> The boot loader (apb) found and loaded up1000.pal
> and vmlinux.gz on /dev/hda1, but then the system appears to
> have switched the ide devices.
> The onboard ide devices 0 and 1 became 2 and 3,
> (the names of my disks became hde, hdf, and hdg).
> CONFIG_BLK_DEV_OFFBOARD was not set.
> Where normally I would boot to root on /dev/hda5,
> the system was trying to mount /dev/hde5.
_____________________________________________________________

"dennis pantazis" wrote

>yes, the highpoint controller is a bit weird. see this page for help.
>basicly, even if the only drive is attached to the hpt366, its not hda/b it
>goes to hde/f

That's not the behavior I see.  The hpt366 appears to go to hda/b/c/d,
forcing the onboard controllers to e/f/g/h,
with drives attached to both onboard controllers and the hpt366.
This happens even when I use a kernel built without hpt366 support,
as long as the card is inserted.

>Andre
>Hedrick, has kindly provided a patch against the stable 2.2.14 kernel.

Those patches have been applied, I'm fairly sure.
I should probably try a different kernel.

>Many
>people (including myself) have reported success with the driver,

Who's kernel did you use?...What patches?

> http://www.csie.ntu.edu.tw/~b6506063/hpt366/

The problem with the hpt366 howto is that you need to
see /proc/pci to determine the correct boot time option,
right?  I never get that far with the hpt366 in the system.
(I do get this far with the Promise UDMA controller installed,
but there are other problems there.  See below.)

_____________________________________________________

"Vanco, Donald" wrote:

>  If I boot with the no drives hanging off that
> controller my "normal" IDE drives are hda and hdb, etc.  If I hang a drive
> on the 66 controller it bumps them all up.

When a drive is attached to my HPT366 (a SIIG board)
in my UP1000, I see no change in behavior.


>   --
> Bob Williams, http://bob.usuhs.mil/
> office: 301-295-3568
>