[parisc-linux] Re: Presicionbook success

[parisc-linux] Re: Presicionbook success

[Gavin Hubbard]

>On Wed, Apr 16, 2003 at 01:07:06PM +0100, M. Grabert wrote:
>> I just wanted you to inform that I successfully installed
>> Debian testing on my Presicionbook.
>
>Sweet!
>
>> - The precompiled kernel from the EISEE team may hang ...
>> 
>>   ... since it has support for the CMD PCI0643 IDE Busmaster chip,
>>   which is obviously used in the Presicionbook, but the driver
>>   doensn't/mustn't be used in Linux, since it causes confusion!
>>   So add "ide=nodma" or something similar to the kernel parameters
>>   (using the IPL) or simply recompile the kernel with no IDE support.
>> 
>>   (The Presicionbook uses 2.5" IDE harddrives, obviously connected
>>    to the CMD chip, but they automagically appear as SCSI drives on LASI
>>    SCSI.)
>
>*blink*.  That's pretty weird.  Could you send the device listing that
>dmesg produces, and the results of lspci?


Before anyone loses too much sleep over this, the Precisionbook has
internal SCSI-IDE bridges that sit between the HBA and the drives. I
believe a similar system was used in some of the earlier Apple Powerbooks.

Regards,

Gavin

Re: [parisc-linux] Re: Presicionbook success

[Matthew Wilcox]

On Thu, Apr 17, 2003 at 01:15:31AM +1200, Gavin Hubbard wrote:
> Before anyone loses too much sleep over this, the Precisionbook has
> internal SCSI-IDE bridges that sit between the HBA and the drives. I
> believe a similar system was used in some of the earlier Apple Powerbooks.

Yep, someone pointed that out in an earlier email.  I'm just trying to
figure out how to handle this for the best.

1) Provide a pci quirk that deletes the IDE device from the bus
2) Hack the IDE driver to not claim that device if it's on a PrecisionBook
3) Ignore the scsi device and let the IDE driver drive the chips
4) Er.. something else.

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk

Re: [parisc-linux] Re: Presicionbook success

[M. Grabert]

On Wed, 16 Apr 2003, Matthew Wilcox wrote:

> On Thu, Apr 17, 2003 at 01:15:31AM +1200, Gavin Hubbard wrote:
> > Before anyone loses too much sleep over this, the Precisionbook has
> > internal SCSI-IDE bridges that sit between the HBA and the drives. I
> > believe a similar system was used in some of the earlier Apple Powerbooks.
>
> Yep, someone pointed that out in an earlier email.  I'm just trying to
> figure out how to handle this for the best.
>
> 1) Provide a pci quirk that deletes the IDE device from the bus
> 2) Hack the IDE driver to not claim that device if it's on a PrecisionBook
> 3) Ignore the scsi device and let the IDE driver drive the chips
> 4) Er.. something else.

The initialization of the CMD PCI0643 IDE controller fails when using
the ESIEE kernel (which has support built-in for this chip).
Something like "failed to get IRQ, will probe later".

So no devices were found on the IDE bus, just on the LASI SCSI.
Probably it's the best just to use SCSI and not to use IDE support at all ...

greetings, max

Re: [parisc-linux] Re: Presicionbook success

[Matthew Wilcox]

On Wed, Apr 16, 2003 at 02:36:25PM +0100, M. Grabert wrote:
> On Wed, 16 Apr 2003, Matthew Wilcox wrote:
> 
> > On Thu, Apr 17, 2003 at 01:15:31AM +1200, Gavin Hubbard wrote:
> > > Before anyone loses too much sleep over this, the Precisionbook has
> > > internal SCSI-IDE bridges that sit between the HBA and the drives. I
> > > believe a similar system was used in some of the earlier Apple Powerbooks.
> >
> > Yep, someone pointed that out in an earlier email.  I'm just trying to
> > figure out how to handle this for the best.
> >
> > 1) Provide a pci quirk that deletes the IDE device from the bus
> > 2) Hack the IDE driver to not claim that device if it's on a PrecisionBook
> > 3) Ignore the scsi device and let the IDE driver drive the chips
> > 4) Er.. something else.
> 
> The initialization of the CMD PCI0643 IDE controller fails when using
> the ESIEE kernel (which has support built-in for this chip).
> Something like "failed to get IRQ, will probe later".

Oh, I thought you said the kernel hung when probing with this driver.
If it just whines about being unable to do anything, that's no problem.

> So no devices were found on the IDE bus, just on the LASI SCSI.
> Probably it's the best just to use SCSI and not to use IDE support at all ...

*nod*.  It might be fun for someone to play with fixing the IDE driver,
but I don't think that's necessary.

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk

Re: [parisc-linux] Re: Presicionbook success

[M. Grabert]

On Wed, 16 Apr 2003, Matthew Wilcox wrote:

> Date: Wed, 16 Apr 2003 14:44:33 +0100
> From: Matthew Wilcox <willy@debian.org>
> To: M. Grabert <xam@cs.ucc.ie>
> Cc: Matthew Wilcox <willy@debian.org>, Gavin Hubbard <ghub005@xtra.co.nz>,
>      parisc-linux@lists.parisc-linux.org
> Subject: Re: [parisc-linux] Re: Presicionbook success
>
> On Wed, Apr 16, 2003 at 02:36:25PM +0100, M. Grabert wrote:

[...]

> > The initialization of the CMD PCI0643 IDE controller fails when using
> > the ESIEE kernel (which has support built-in for this chip).
> > Something like "failed to get IRQ, will probe later".
>
> Oh, I thought you said the kernel hung when probing with this driver.
> If it just whines about being unable to do anything, that's no problem.

Maybe I was a little bit confusing to you:

1. Using a kernel with CMD IDE support, it will hang with the message
   "will probe IRQs later" ...
2. it will not hang if you pass a ide=nodma, but then it will show
   no devices
3. It will just boot fine without any CMD IDE support in the kernel

> > So no devices were found on the IDE bus, just on the LASI SCSI.
> > Probably it's the best just to use SCSI and not to use IDE support at all ...
>
> *nod*.  It might be fun for someone to play with fixing the IDE driver,
> but I don't think that's necessary.

;)


Slan, Max

RE: [parisc-linux] Re: Presicionbook success

["Beerse, Corné"]

[-- Attachment #1: Type: text/plain, Size: 1870 bytes --]

> -----Original Message-----
> From: M. Grabert [mailto:xam@cs.ucc.ie]
> Sent: woensdag 16 april 2003 15:36
> To: Matthew Wilcox
> Cc: Gavin Hubbard; parisc-linux@lists.parisc-linux.org
> Subject: Re: [parisc-linux] Re: Presicionbook success
> 
> 
> On Wed, 16 Apr 2003, Matthew Wilcox wrote:
> 
> > On Thu, Apr 17, 2003 at 01:15:31AM +1200, Gavin Hubbard wrote:
> > > Before anyone loses too much sleep over this, the 
> Precisionbook has
> > > internal SCSI-IDE bridges that sit between the HBA and 
> the drives. I
> > > believe a similar system was used in some of the earlier 
> Apple Powerbooks.
> >
> > Yep, someone pointed that out in an earlier email.  I'm 
> just trying to
> > figure out how to handle this for the best.
> >
> > 1) Provide a pci quirk that deletes the IDE device from the bus
> > 2) Hack the IDE driver to not claim that device if it's on 
> a PrecisionBook
> > 3) Ignore the scsi device and let the IDE driver drive the chips
> > 4) Er.. something else.
> 
> The initialization of the CMD PCI0643 IDE controller fails when using
> the ESIEE kernel (which has support built-in for this chip).
> Something like "failed to get IRQ, will probe later".
> 
> So no devices were found on the IDE bus, just on the LASI SCSI.
> Probably it's the best just to use SCSI and not to use IDE 
> support at all ...


Is this something like
http://www6.tomshardware.com/storage/20030130/index.html ? Then you
(electronically) cannot find any ide at all, its hidden behind the scsi
interface. If you see part of it, then best to try to handle it as good as
possible (try 'not at all' for a start). I think the 'will probe later' can
be canceled, triggered by a found scsi interface.

For what its worth, I think the IRQ is hooked to the scsi interface, not to
the processor.

btw, what does an `ioscan` from HP-UX report? I guess it only sees scsi.


CBee



[-- Attachment #2: Type: text/html, Size: 3486 bytes --]

RE: [parisc-linux] Re: Presicionbook success

[M. Grabert]

On Wed, 16 Apr 2003, [iso-8859-1] "Beerse, Corn=E9" wrote:

> Is this something like
> http://www6.tomshardware.com/storage/20030130/index.html ? Then you
> (electronically) cannot find any ide at all, its hidden behind the scsi
> interface. If you see part of it, then best to try to handle it as good a=
s
> possible (try 'not at all' for a start). I think the 'will probe later' c=
an
> be canceled, triggered by a found scsi interface.

The hard disks are removeable, but there is no obvious IDE-SCSI bridge
visible. It is build-in inside the laptop (i.e. non-removeable), and it's
obviously visible from the OS (at least linux sees the IDE chip). Whether
you can make any sensible use of it is a different question though ...

> For what its worth, I think the IRQ is hooked to the scsi interface, not =
to
> the processor.

This would make sense.

> btw, what does an `ioscan` from HP-UX report? I guess it only sees scsi.

Good question, I don't have the HP-UX root password. The Precisionbook
comes with two 4GB hard disks, both with HP-UX installed, but I have to
ask for the password (or does anybody know how to easily gain root access?)=
,
which I haven't done yet.

Greetings, Max

RE: [parisc-linux] Re: Presicionbook success

["Beerse, Corné"]

[-- Attachment #1: Type: text/plain, Size: 1851 bytes --]

> -----Original Message-----
> From: M. Grabert [mailto:xam@cs.ucc.ie]
> 
> On Wed, 16 Apr 2003, Matthew Wilcox wrote:
> 
> > Date: Wed, 16 Apr 2003 14:44:33 +0100
> >
> > On Wed, Apr 16, 2003 at 02:36:25PM +0100, M. Grabert wrote:
> 
> [...]
> 
> > > The initialization of the CMD PCI0643 IDE controller 
> fails when using
> > > the ESIEE kernel (which has support built-in for this chip).
> > > Something like "failed to get IRQ, will probe later".
> >
> > Oh, I thought you said the kernel hung when probing with 
> this driver.
> > If it just whines about being unable to do anything, that's 
> no problem.
> 
> Maybe I was a little bit confusing to you:
> 
> 1. Using a kernel with CMD IDE support, it will hang with the message
>    "will probe IRQs later" ...
> 2. it will not hang if you pass a ide=nodma, but then it will show
>    no devices
> 3. It will just boot fine without any CMD IDE support in the kernel
> 
> > > So no devices were found on the IDE bus, just on the LASI SCSI.
> > > Probably it's the best just to use SCSI and not to use IDE support at
all ...
> >
> > *nod*.  It might be fun for someone to play with fixing the IDE driver,
> > but I don't think that's necessary.

This gives me more and more the impression it is definitly some predecessor
of http://www6.tomshardware.com/storage/20030130/index.html

I think the ide interface is not perfectly hidden behind the scsi interface,
the cpu can still somehow recognize the ide interface. The ide interface
cannot do dma and irq since that is hooked to the scsi interface, not to the
cpu/memory. Hence, I think the best the ide driver can do is not to try
later but to be disabled at all. Btw, what happens if the ide says to try
later, is there a total system pause (resulting in a halt)? or does the
remainder of the system continue (and is it save to ignore)?


CBee

[-- Attachment #2: Type: text/html, Size: 3449 bytes --]

Re: [parisc-linux] Re: Presicionbook success

[Matthew Wilcox]

On Wed, Apr 16, 2003 at 03:09:58PM +0100, M. Grabert wrote:
> Good question, I don't have the HP-UX root password. The Precisionbook
> comes with two 4GB hard disks, both with HP-UX installed, but I have to
> ask for the password (or does anybody know how to easily gain root access?),
> which I haven't done yet.

Surely you can boot it, interact with IPL and add `single' as a parameter
to boot it in single-user mode?  Or does that ask for the pw too?

(I have to confess to never having used hpux in anger ;-)

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk

RE: [parisc-linux] Re: Presicionbook success

[M. Grabert]

On Wed, 16 Apr 2003, [iso-8859-1] "Beerse, Corn=E9" wrote:

> This gives me more and more the impression it is definitly some predecess=
or
> of http://www6.tomshardware.com/storage/20030130/index.html
>
> I think the ide interface is not perfectly hidden behind the scsi interfa=
ce,
> the cpu can still somehow recognize the ide interface. The ide interface
> cannot do dma and irq since that is hooked to the scsi interface, not to =
the
> cpu/memory. Hence, I think the best the ide driver can do is not to try
> later but to be disabled at all. Btw, what happens if the ide says to try
> later, is there a total system pause (resulting in a halt)? or does the
> remainder of the system continue (and is it save to ignore)?

As stated in my earlier mails, it

- hangs (I waited for quite a while!)
- or continues (if ide=3Dnodma is used)

Slan, Max

Re: [parisc-linux] Re: Presicionbook success

[Bjorn Helgaas]

On Wednesday 16 April 2003 8:15 am, Matthew Wilcox wrote:
> On Wed, Apr 16, 2003 at 03:09:58PM +0100, M. Grabert wrote:
> > Good question, I don't have the HP-UX root password. The Precisionbook
> > comes with two 4GB hard disks, both with HP-UX installed, but I have to
> > ask for the password (or does anybody know how to easily gain root access?),
> > which I haven't done yet.
> 
> Surely you can boot it, interact with IPL and add `single' as a parameter
> to boot it in single-user mode?  Or does that ask for the pw too?

I think you want something like:

ISL> hpux -is

RE: [parisc-linux] Re: Presicionbook success

[Alan Cox]

On Mer, 2003-04-16 at 15:09, M. Grabert wrote:
> The hard disks are removeable, but there is no obvious IDE-SCSI bridge
> visible. It is build-in inside the laptop (i.e. non-removeable), and it's
> obviously visible from the OS (at least linux sees the IDE chip). Whether
> you can make any sensible use of it is a different question though ...

That means the system either has seperate IDE and SCSI controllers or
has an IDE controller and an IDE_SCSI bridge for scsi disk. I suspect
the former
> > For what its worth, I think the IRQ is hooked to the scsi interface, not to
> > the processor.
> 
> This would make sense.

It would be common sense for two exclusive items to use the same IRQ,
but if pio works so does the IRQ line

> comes with two 4GB hard disks, both with HP-UX installed, but I have to
> ask for the password (or does anybody know how to easily gain root access?),
> which I haven't done yet.

There are several ways but that would be subverting a rights management
system and I don't wish to get the list owners in trouble.

Re: [parisc-linux] Re: Presicionbook success

[Alan Cox]

On Mer, 2003-04-16 at 14:15, Gavin Hubbard wrote:
> Before anyone loses too much sleep over this, the Precisionbook has
> internal SCSI-IDE bridges that sit between the HBA and the drives. I
> believe a similar system was used in some of the earlier Apple Powerbooks.

So htf do they end up on the IDE controller too ? If they show up on the
scsi bus under Linux then I need the ssvid/ssid for the CMD device (and
which CMD variant is used) and I can vanish it easily enough.

RE: [parisc-linux] Re: Presicionbook success

["Beerse, Corné"]

[-- Attachment #1: Type: text/plain, Size: 1248 bytes --]

> -----Original Message-----
> From: M. Grabert [mailto:xam@cs.ucc.ie]
> Sent: woensdag 16 april 2003 16:18
> 
> On Wed, 16 Apr 2003, [iso-8859-1] "Beerse, Corné" wrote:
> 
> > This gives me more and more the impression it is definitly 
> some predecessor
> > of http://www6.tomshardware.com/storage/20030130/index.html
> >
>>>>>>>>>>>>>>>>>. snip snip <<<<<<<<<<<<<
> 
> As stated in my earlier mails, it
> 
> - hangs (I waited for quite a while!)
> - or continues (if ide=nodma is used)

Then I think the next happens: The ide interface-chip is somehow detected
and the system pauses while nothing can continue. I think the cpu steps
aside to have some dma passing by. It either waits for some dma to happen or
for a note it can continue. Which never happen and never will happen since
the ide interface-chip does this dma to the scsi controler, not to the
cpu-memory.

In this special case, the cpu should not ask the chip if it does dma, the
chip sure does dma but it does it not against core memory but against the
scsi controler. This will be verry hard to detect from the cpu side. I think
the nicest implementation is to use the ide=nodma, either manually or
carefully if a precisionbook is detected.


CBee

[-- Attachment #2: Type: text/html, Size: 2305 bytes --]

Re: [parisc-linux] Re: Presicionbook success

[Stan Sieler]

Re:

> > The hard disks are removeable, but there is no obvious IDE-SCSI bridge
> > visible. It is build-in inside the laptop (i.e. non-removeable), and it's
...

> That means the system either has seperate IDE and SCSI controllers or
> has an IDE controller and an IDE_SCSI bridge for scsi disk. I suspect

The IDE to SCSI converter is a small board inside the "cartridge" 
(or "case") that forms the removable hard drive.   If you remove a hard
drive, and then open its case up, you'll see the converter board clearly.

It is indeed highly similar to the Acard SCSIIDE product (although possibly
from a different vendor).  

BTW, I would not expect either disk to show up as IDE at all.

-- 
Stan Sieler                                           sieler@allegro.com
www.allegro.com/sieler/wanted/index.html                  www.sieler.com

Re: [parisc-linux] Re: Presicionbook success

[Grant Grundler]

On Wed, Apr 16, 2003 at 03:41:33PM +0100, Alan Cox wrote:
> There are several ways but that would be subverting a rights management
> system and I don't wish to get the list owners in trouble.

Alan,
interesting comment.  HPUX/parisc does not have "rights management"
once one has access to the console.

Bjorn already posted the "proper" method (ISL> hpux -is).

grant