IDE cable select question

Hello,
Trying to figure out if it the payloads (filo) or coreboots  
resposiblity to detect a 40 or 80 ribbon IDE cable. Reading the ATA  
specifications it looks like a IDENTIFY_DRIVE command query to the  
drive is the way it detects this and then the bios sets this in the  
IDE pci configuration register. Right now, it does not detect this and  
when the Linux IDE driver takes over it only runs at ATA33 not ATA100.

For coreboot to automate this it would have to a IDENTIFY_DRIVE  
command query to the drive and then set IDE pci configuration  
register. Would this be possible to do pre payload or is this  
something that would have to be set manually. If manual could we  
impliment a global setting??

something like:
IDE_CABLE = 40 or 80

Thanks - Joe

[ron minnich]

On Fri, Feb 22, 2008 at 8:15 AM,   wrote:
> For coreboot to automate this it would have to a IDENTIFY_DRIVE
>  command query to the drive and then set IDE pci configuration
>  register. Would this be possible to do pre payload or is this
>  something that would have to be set manually. If manual could we
>  impliment a global setting??

I hate to say this, but I think we need to do this in the code for
that chip. Whicdh chipset is this again?

ron

Quoting ron minnich :
> On Fri, Feb 22, 2008 at 8:15 AM,   wrote:
> > For coreboot to automate this it would have to a IDENTIFY_DRIVE
> >  command query to the drive and then set IDE pci configuration
> >  register. Would this be possible to do pre payload or is this
> >  something that would have to be set manually. If manual could we
> >  impliment a global setting??
>
> I hate to say this, but I think we need to do this in the code for
> that chip. Whicdh chipset is this again?
>
> ron

It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are  
the same on this. So are you saying this can be set in filo, or  
coreboot?

IDE_CONFIG?IDE I/O Configuration Register(IDE?D31:F1)
Address Offset: 54h Attribute: R/W
Default Value: 00h Size: 32 bits

Bit 7 Secondary Slave Channel Cable Reporting ? R/W. BIOS should  
program this bit to tell the IDE
driver which cable is plugged into the channel.
0 = 40 conductor cable is present.
1 = 80 conductor cable is present.
Bit 6 Secondary Master Channel Cable Reporting ? R/W. Same description  
as bit 7
Bit 5 Primary Slave Channel Cable Reporting ? R/W. Same description as bit 7
Bit 4 Primary Master Channel Cable Reporting ? R/W. Same description as bit 7

Thanks - Joe

[ron minnich]

On Fri, Feb 22, 2008 at 9:15 AM,   wrote:
> It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are
>  the same on this. So are you saying this can be set in filo, or
>  coreboot?

I think coreboot has to do it ... if the kernel can not.

ron

Quoting ron minnich :
> On Fri, Feb 22, 2008 at 9:15 AM,   wrote:
> > It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are
> >  the same on this. So are you saying this can be set in filo, or
> >  coreboot?
>
> I think coreboot has to do it ... if the kernel can not.
>
> ron

Ok then, I think it may complicate things too much if we had a  
IDENTIFY_DRIVE command query to the drive (Note: this command query is  
also used for DMA settings). That would be like basicly re-writing an  
IDE driver??

So, I propose a global option (IDE_CABLE = 40 or 80) that can easily  
be adjusted from Config.lb in the targets directory, that sets the  
value for a simple function in  
src/southbridge/intel/i82801xx/i82801xx_ide.c.
Does that make sense?

Thanks - Joe

[Ward Vandewege]

On Fri, Feb 22, 2008 at 01:13:25PM -0500, [e-mail] wrote:
> Ok then, I think it may complicate things too much if we had a  
> IDENTIFY_DRIVE command query to the drive (Note: this command query is  
> also used for DMA settings). That would be like basicly re-writing an  
> IDE driver??
>
> So, I propose a global option (IDE_CABLE = 40 or 80) that can easily  
> be adjusted from Config.lb in the targets directory, that sets the  
> value for a simple function in  
> src/southbridge/intel/i82801xx/i82801xx_ide.c.
> Does that make sense?

I'd really prefer something that's autosensing, if that's possible, so that
changing hard drives does not cause problems or require a bios flash...

Thanks,
Ward.

Quoting Ward Vandewege :
> On Fri, Feb 22, 2008 at 01:13:25PM -0500, [e-mail] wrote:
> > Ok then, I think it may complicate things too much if we had a
> > IDENTIFY_DRIVE command query to the drive (Note: this command query is
> > also used for DMA settings). That would be like basicly re-writing an
> > IDE driver??
> >
> > So, I propose a global option (IDE_CABLE = 40 or 80) that can easily
> > be adjusted from Config.lb in the targets directory, that sets the
> > value for a simple function in
> > src/southbridge/intel/i82801xx/i82801xx_ide.c.
> > Does that make sense?
>
> I'd really prefer something that's autosensing, if that's possible, so that
> changing hard drives does not cause problems or require a bios flash...

It wouldn't cause any issues changing the drive. Were just talking  
about the cable. If you use an ATA33 drive on a 80 pin cable it works  
fine it just runs at ATA33 speed. My problem is I have an ATA100 drive  
and 80 pin cable but Linux thinks I have a 40 pin cable therefore it  
is only running at ATA33 speed.

Thanks - Joe

[Ward Vandewege]

On Fri, Feb 22, 2008 at 01:56:34PM -0500, [e-mail] wrote:
> > I'd really prefer something that's autosensing, if that's possible, so 
> > that
> > changing hard drives does not cause problems or require a bios flash...
>
> It wouldn't cause any issues changing the drive. Were just talking about 
> the cable. If you use an ATA33 drive on a 80 pin cable it works fine it 
> just runs at ATA33 speed. My problem is I have an ATA100 drive and 80 pin 
> cable but Linux thinks I have a 40 pin cable therefore it is only running 
> at ATA33 speed.

So not running at the right speed is a bug, right? I'd prefer if a user
didn't have to flash her bios just to get their drive to run at the correct
speed...

Though, arguably this is perhaps somewhat academic nowadays - IDE is well
past its prime.

Thanks,
Ward.

Quoting Ward Vandewege :
> On Fri, Feb 22, 2008 at 01:56:34PM -0500, [e-mail] wrote:
> > > I'd really prefer something that's autosensing, if that's possible, so
> > > that
> > > changing hard drives does not cause problems or require a bios flash...
> >
> > It wouldn't cause any issues changing the drive. Were just talking about
> > the cable. If you use an ATA33 drive on a 80 pin cable it works fine it
> > just runs at ATA33 speed. My problem is I have an ATA100 drive and 80 pin
> > cable but Linux thinks I have a 40 pin cable therefore it is only running
> > at ATA33 speed.
>
> So not running at the right speed is a bug, right? I'd prefer if a user
> didn't have to flash her bios just to get their drive to run at the correct
> speed...
>
> Though, arguably this is perhaps somewhat academic nowadays - IDE is well
> past its prime.

Maybe it is, but just about every motherboard still has it, and what's  
the point of having it if you can't use it the way it is supposed to  
be used.

Thanks - Joe

[Ward Vandewege]

On Fri, Feb 22, 2008 at 02:23:19PM -0500, [e-mail] wrote:
> > Though, arguably this is perhaps somewhat academic nowadays - IDE is well
> > past its prime.
>
> Maybe it is, but just about every motherboard still has it, and what's  
> the point of having it if you can't use it the way it is supposed to  
> be used.

Yes, agreed. Every component of the board should be fully functional...

Thanks,
Ward.

Quoting [e-mail]:
> Quoting ron minnich :
> > On Fri, Feb 22, 2008 at 9:15 AM,   wrote:
> > > It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are
> > >  the same on this. So are you saying this can be set in filo, or
> > >  coreboot?
> >
> > I think coreboot has to do it ... if the kernel can not.
> >
> > ron
>
> Ok then, I think it may complicate things too much if we had a
> IDENTIFY_DRIVE command query to the drive (Note: this command query is
> also used for DMA settings). That would be like basicly re-writing an
> IDE driver??

After some futher reading, I don't know if this would even be possible  
with coreboot. It would have to be done with filo (file does this  
command query anyways and prints out the drive model number) when the  
drives are fully initialized. Looks like senerio number two...
> So, I propose a global option (IDE_CABLE = 40 or 80) that can easily
> be adjusted from Config.lb in the targets directory, that sets the
> value for a simple function in
> src/southbridge/intel/i82801xx/i82801xx_ide.c.
> Does that make sense?

If there are no objections I will submit a patch. What coreboot system  
files need to be modified for a global option?

Thanks - Joe

[Carl-Daniel Hailfinger]

On 22.02.2008 21:03, [e-mail] wrote:
> Quoting [e-mail]:
> > Quoting ron minnich :
> > > On Fri, Feb 22, 2008 at 9:15 AM,   wrote:
> > > > It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are
> > > >  the same on this. So are you saying this can be set in filo, or
> > > >  coreboot?
> > >
> > > I think coreboot has to do it ... if the kernel can not.
> > >
> > > ron
> >
> > Ok then, I think it may complicate things too much if we had a
> > IDENTIFY_DRIVE command query to the drive (Note: this command query is
> > also used for DMA settings). That would be like basicly re-writing an
> > IDE driver??
>
> After some futher reading, I don't know if this would even be possible  
> with coreboot. It would have to be done with filo (file does this  
> command query anyways and prints out the drive model number) when the  
> drives are fully initialized. Looks like senerio number two...
> > So, I propose a global option (IDE_CABLE = 40 or 80) that can easily
> > be adjusted from Config.lb in the targets directory, that sets the
> > value for a simple function in
> > src/southbridge/intel/i82801xx/i82801xx_ide.c.
> > Does that make sense?
>
> If there are no objections I will submit a patch. What coreboot system  
> files need to be modified for a global option?

Can you please check with the Linux IDE guys? The matter is horribly 
complicated because you have to check whether the 80-wire cable was 
detected by the controller and check the same for the drive. 
[e-mail] is the address you want to write to and the 
experts there can tell you exactly what the BIOS is expected to do.

Regards,
Carl-Daniel

Quoting Carl-Daniel Hailfinger :
> On 22.02.2008 21:03, [e-mail] wrote:
> > Quoting [e-mail]:
> > > Quoting ron minnich :
> > > > On Fri, Feb 22, 2008 at 9:15 AM,   wrote:
> > > > > It is the Intel 82801DB ICH4 chipset. I think most of the ICH's are
> > > > > the same on this. So are you saying this can be set in filo, or
> > > > > coreboot?
> > > >
> > > > I think coreboot has to do it ... if the kernel can not.
> > > >
> > > > ron
> > >
> > > Ok then, I think it may complicate things too much if we had a
> > > IDENTIFY_DRIVE command query to the drive (Note: this command query is
> > > also used for DMA settings). That would be like basicly re-writing an
> > > IDE driver??
> >
> > After some futher reading, I don't know if this would even be   
> > possible with coreboot. It would have to be done with filo (file   
> > does this  command query anyways and prints out the drive model   
> > number) when the  drives are fully initialized. Looks like senerio   
> > number two...
> > > So, I propose a global option (IDE_CABLE = 40 or 80) that can easily
> > > be adjusted from Config.lb in the targets directory, that sets the
> > > value for a simple function in
> > > src/southbridge/intel/i82801xx/i82801xx_ide.c.
> > > Does that make sense?
> >
> > If there are no objections I will submit a patch. What coreboot   
> > system files need to be modified for a global option?
>
> Can you please check with the Linux IDE guys? The matter is horribly
> complicated because you have to check whether the 80-wire cable was
> detected by the controller and check the same for the drive.
> [e-mail] is the address you want to write to and the
> experts there can tell you exactly what the BIOS is expected to do.
>
> Regards,
> Carl-Daniel

OK, but amazingly intel was nice enough to provide this doc:

http://www.intel.com/design/chipsets/manuals/298600.htm

Intel® 82801EB (ICH5), 82801ER (ICH5R), 82801DB (ICH4), 82801CA  
(ICH3), 82801BA (ICH2), 82801AA (ICH), and 82801AB (ICH0) IDE  
Controller Programmer?s Reference Manual

I tells us alot about what the BIOS is expected to do. I suppose I  
could take chunks of code out of filo and put it in coreboot, but how  
good would that be with a bunch of duplicated code. Well, I will try  
the Linux IDE guys, maybe they know a simple way to determine the  
presence of drives and their capabilities (IDENTIFY_DRIVE Command).

Thanks - Joe

Well,
Much to my suprise it looks like pin 34 of the IDE connetor does not  
assert a signal if the cable is 40 pin and does assert a signal if the  
cable is 80 pin. This is a standard for all motherboards that support  
ATA66/ATA100. Now I just have to figure out where this signal is  
asserted too (maybe a GPIO -> in). Once I figure that out this will be  
easy. I can just tell coreboot to: read ide cable signal, set ide  
cable select register bit to 1.

Thanks - Joe

> Well,
> Much to my suprise it looks like pin 34 of the IDE connetor does not
> assert a signal if the cable is 40

**Correction** conductors
> and does assert a signal if the
> cable is 80

**Correction** conductors
> . This is a standard for all motherboards that support
> ATA66/ATA100. Now I just have to figure out where this signal is
> asserted too (maybe a GPIO -> in). Once I figure that out this will be
> easy. I can just tell coreboot to: read ide cable signal, set ide
> cable select register bit to 1.

Thanks - Joe