Subject: Re: IDE Breakage in 1.5: false alarm?
To: None <bouyer@antioche.lip6.fr>
From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
List: current-users
Date: 02/09/2001 00:24:20
In <20010208160003.D9671@antioche.lip6.fr>
bouyer@antioche.lip6.fr wrote:

> > > The machine I'm typing this has a P5A, and never noticed any problems other
> > > than not being able to do UDMA mode 2 with Ultra/66 or Ultra/100 drives
> > > (works fine with Ultra/33 drives).

> Yes, there's something with this chipset that causes this. I don't know what.
> I've not been able to track it down; maybe it's a silicon bug (the way the
> BIOS set it up doesn't work as well).

Do you check Linux driver for M5229 on the ALi site?
http://www.ali.com.tw/eng/support/drvlnx.htm

The Linux driver initialize "setup time" at register offset 0x58 or 0x5c,
but acer_setup_channel() in pciide.c does not.
The Linux driver also supports Ultra/66 on newer chips and
seems to handle many other quirks.

My CATS uses the M5229 pciide chipset and I also got UDMA errors.
I have made some patches but -current CATS kernel cannot boot because of
"panic: ubc_init: failed to map ubc_object." Sigh..
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp