Boot problem with PDC20269

Boot problem with PDC20269

[Adam Huffman]

When I try to boot a kernel with the latest IDE driver in order to
support my Promise UltraATA133 TX2 card, the system hangs at the point
where it should be listing the two drives attached to the card i.e. it
shows /dev/hdc which is a DVD drive, then stops.  Sometimes I can reboot
with Alt-SysRq, sometimes a hard reset is needed.  This happens with
2.4.17+ide, 2.4.18-rc1, 2.5.5-dj1.

Motherboard is MSI K7T-Turbo 2, K7 1.4, drives attached to the Promise
card are an IBM 60GXP and a Maxtor D740X-6L.  The drives work when
attached to the VIA interface on the m/b.

I've also seen some "Spurious 8259A interrupt IRQ7" messages and some
APIC errors (APIC is currently disabled in the BIOS).

Re: Boot problem with PDC20269

[Andre Hedrick]

Hi Adam,

http://www.tecchannel.de/hardware/817/index.html


We do not put ATAPI devices on such HOSTS.
The driver will not work w/ ATAPI there because it uses a different DMA
engine location and is not supported in Linux.

You will find greater success in putting your ATAPI device on the native
south bridge.

Cheers,

Andre Hedrick
Linux Disk Certification Project                Linux ATA Development




On Sat, 23 Feb 2002, Adam Huffman wrote:

> When I try to boot a kernel with the latest IDE driver in order to
> support my Promise UltraATA133 TX2 card, the system hangs at the point
> where it should be listing the two drives attached to the card i.e. it
> shows /dev/hdc which is a DVD drive, then stops.  Sometimes I can reboot
> with Alt-SysRq, sometimes a hard reset is needed.  This happens with
> 2.4.17+ide, 2.4.18-rc1, 2.5.5-dj1.
> 
> Motherboard is MSI K7T-Turbo 2, K7 1.4, drives attached to the Promise
> card are an IBM 60GXP and a Maxtor D740X-6L.  The drives work when
> attached to the VIA interface on the m/b.

Re: Boot problem with PDC20269

[Gunther Mayer]

Andre Hedrick wrote:

> Hi Adam,
>
> http://www.tecchannel.de/hardware/817/index.html
>
> We do not put ATAPI devices on such HOSTS.

put == support ?

>
> The driver will not work w/ ATAPI there because it uses a different DMA
> engine location and is not supported in Linux.

It is a serious bug in the IDE driver to hang the system (and not the user's
fault).

A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
devices on PDC20269. Ignoring the device.\n");
and continue running.

-
Gunther

Re: Boot problem with PDC20269

[Adam Huffman]

On Sat, 23 Feb 2002, Gunther Mayer wrote:

> Andre Hedrick wrote:
> 
> > Hi Adam,
> >
> > http://www.tecchannel.de/hardware/817/index.html
> >
> > We do not put ATAPI devices on such HOSTS.
> 
> put == support ?
> 

I wasn't clear enough in my original mail - the ATAPI device is on the
VIA southbridge, not the Promise card.  The two drives on the Promise
card are hde and hdg.

> >
> > The driver will not work w/ ATAPI there because it uses a different DMA
> > engine location and is not supported in Linux.
> 
> It is a serious bug in the IDE driver to hang the system (and not the user's
> fault).
> 
> A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
> devices on PDC20269. Ignoring the device.\n");
> and continue running.
> 
> -
> Gunther
> 

At this stage I'm not all that bothered about the performance issues, I
just want to be able to use the two drives.

Thanks for the replies,
Adam

Re: Boot problem with PDC20269

[Mike Fedyk]

On Sat, Feb 23, 2002 at 01:14:15PM +0100, Gunther Mayer wrote:
> Andre Hedrick wrote:
> 
> > Hi Adam,
> >
> > http://www.tecchannel.de/hardware/817/index.html
> >
> > We do not put ATAPI devices on such HOSTS.
> 
> put == support ?
> 
> >
> > The driver will not work w/ ATAPI there because it uses a different DMA
> > engine location and is not supported in Linux.
> 
> It is a serious bug in the IDE driver to hang the system (and not the user's
> fault).
> 
> A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
> devices on PDC20269. Ignoring the device.\n");
> and continue running.
> 

I agree.

Andre, what about southbridges that have these chipsets built in?

Re: Boot problem with PDC20269

[Roy Sigurd Karlsbakk]

> > A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
> > devices on PDC20269. Ignoring the device.\n");
> > and continue running.

er .. I'm running a PDC20269 with a few drives. Is that supposed to be
impossible?

roy

--
Roy Sigurd Karlsbakk, Datavaktmester

Computers are like air conditioners.
They stop working when you open Windows.

Re: Boot problem with PDC20269

[Mike Fedyk]

On Sat, Feb 23, 2002 at 09:33:02PM +0100, Roy Sigurd Karlsbakk wrote:
> > > A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
> > > devices on PDC20269. Ignoring the device.\n");
> > > and continue running.
> 
> er .. I'm running a PDC20269 with a few drives. Is that supposed to be
> impossible?
> 

Are they ATAPI drives (cdrom, etc)?

Re: Boot problem with PDC20269

[Adam Huffman]

On Sat, 23 Feb 2002, Roy Sigurd Karlsbakk wrote:

> > > A fix would be to printk("The linux IDE driver does not (yet?)support ATAPI
> > > devices on PDC20269. Ignoring the device.\n");
> > > and continue running.
> 
> er .. I'm running a PDC20269 with a few drives. Is that supposed to be
> impossible?
> 
> roy
> 

May I ask which kernel you are running?

Thanks,
Adam

Re: Boot problem with PDC20269

[Roy Sigurd Karlsbakk]

> > er .. I'm running a PDC20269 with a few drives. Is that supposed to be
> > impossible?
> >
> > roy
> >
>
> May I ask which kernel you are running?

linux-2.4.18pre3 + tux + linuxdiskcert.org's ide_2.4.17 patch
--
Roy Sigurd Karlsbakk, Datavaktmester

Computers are like air conditioners.
They stop working when you open Windows.

Re: Boot problem with PDC20269

[Adam Huffman]

On Sat, 23 Feb 2002, Roy Sigurd Karlsbakk wrote:

> > > er .. I'm running a PDC20269 with a few drives. Is that supposed to be
> > > impossible?
> > >
> > > roy
> > >
> >
> > May I ask which kernel you are running?
> 
> linux-2.4.18pre3 + tux + linuxdiskcert.org's ide_2.4.17 patch
> --
> Roy Sigurd Karlsbakk, Datavaktmester
> 

I've managed to boot successfully now with 2.4.18-rc2-ac2.

I get the following messages when I mount the IBM 60GXP:

hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
PDC202XX: Primary channel reset.
ide2: reset: success

but mounting does proceed.  No such problem with the Maxtor.

Adam

Re: Boot problem with PDC20269

[Holger Lubitz]

Andre Hedrick proclaimed:
> 
> Hi Adam,
> 
> http://www.tecchannel.de/hardware/817/index.html

Interesting article. After the stability issues with VIA that are
believed to be solved, we are now getting to performance issues. Any
chance of the "magic performance booster workaround" done by VIAs 1.04
driver finding its way into linux?

Holger

Re: Boot problem with PDC20269

[Alan Cox]

> > http://www.tecchannel.de/hardware/817/index.html
> 
> Interesting article. After the stability issues with VIA that are
> believed to be solved, we are now getting to performance issues. Any
> chance of the "magic performance booster workaround" done by VIAs 1.04
> driver finding its way into linux?

Only if someone can find out what it actually does, or VIA document it

Alan

Re: Boot problem with PDC20269

[Mike Fedyk]

On Mon, Feb 25, 2002 at 04:39:17PM +0000, Alan Cox wrote:
> > > http://www.tecchannel.de/hardware/817/index.html
> > 
> > Interesting article. After the stability issues with VIA that are
> > believed to be solved, we are now getting to performance issues. Any
> > chance of the "magic performance booster workaround" done by VIAs 1.04
> > driver finding its way into linux?
> 
> Only if someone can find out what it actually does, or VIA document it
> 

The unofficial one is documented, and improves performance greatly on those
chipsets...