Hi!
The goal of this patch is to add support for the new generation of
MuTIOL southbridges by SiS, namely the 961/961B, 962 and 963. These
integrate the IDE controller, unlike all other SiS chipsets where the
controller is bound to the northbridge.
I had to do quite some research on what SiS IDE controllers exist that
don't use the 96* southbridges, and thus found out that the 633 and 733
controllers were missing. So those were added too.
This patch hovewer integrates a patch from Lionel which adds 630S/ET
UDMA100 support.
And while doing the changes I did also some cleanups, mainly removing a
bunch of debug code that doesn't seem very useful when lspci does the
same job. And removing the config_drive_xfer_rate in favor of functions
from ide-timing.h.
Tested on SiS963, works great.
Patches for current 2.4 and 2.5 attached.
--
Vojtech Pavlik
SuSE Labs, SuSE CR
Vojtech Pavlik wrote:
> Hi!
>
> The goal of this patch is to add support for the new generation of
> MuTIOL southbridges by SiS, namely the 961/961B, 962 and 963. These
> integrate the IDE controller, unlike all other SiS chipsets where the
> controller is bound to the northbridge.
>
> I had to do quite some research on what SiS IDE controllers exist that
> don't use the 96* southbridges, and thus found out that the 633 and 733
> controllers were missing. So those were added too.
>
> This patch hovewer integrates a patch from Lionel which adds 630S/ET
> UDMA100 support.
>
> And while doing the changes I did also some cleanups, mainly removing a
> bunch of debug code that doesn't seem very useful when lspci does the
> same job. And removing the config_drive_xfer_rate in favor of functions
> from ide-timing.h.
>
> Tested on SiS963, works great.
>
> Patches for current 2.4 and 2.5 attached.
I never thought about it, but my SiS963 is not recognized in lspci. The
below is a link which has a diagram of the SiS648/SiS963 combo that is
on my motherboard. Would it enhance performance any or would this
correction just be cosmetic?
Thanks,
David
Link: http://www.sis.com/products/chipsets/oa/pentium4/648.htm
On Fri, May 16, 2003 at 12:07:53PM -0400, David van Hoose wrote:
> > The goal of this patch is to add support for the new generation of
> > MuTIOL southbridges by SiS, namely the 961/961B, 962 and 963. These
> > integrate the IDE controller, unlike all other SiS chipsets where the
> > controller is bound to the northbridge.
> >
> > I had to do quite some research on what SiS IDE controllers exist that
> > don't use the 96* southbridges, and thus found out that the 633 and 733
> > controllers were missing. So those were added too.
> >
> > This patch hovewer integrates a patch from Lionel which adds 630S/ET
> > UDMA100 support.
> >
> > And while doing the changes I did also some cleanups, mainly removing a
> > bunch of debug code that doesn't seem very useful when lspci does the
> > same job. And removing the config_drive_xfer_rate in favor of functions
> > from ide-timing.h.
> >
> > Tested on SiS963, works great.
> >
> > Patches for current 2.4 and 2.5 attached.
>
> I never thought about it, but my SiS963 is not recognized in lspci. The
> below is a link which has a diagram of the SiS648/SiS963 combo that is
> on my motherboard. Would it enhance performance any or would this
> correction just be cosmetic?
There will most likely be no change for you.
--
Vojtech Pavlik
SuSE Labs, SuSE CR
Hi,
Good job, but...
On Fri, 16 May 2003, Vojtech Pavlik wrote:
> And while doing the changes I did also some cleanups, mainly removing a
> bunch of debug code that doesn't seem very useful when lspci does the
> same job. And removing the config_drive_xfer_rate in favor of functions
> from ide-timing.h.
Debug code in fe. SiS IDE driver does the same as lspci given you get to
the point you can run lspci. Dumping of PCI conf regs should be moved to
generic IDE PCI code as it might be useful for other IDE PCI drivers.
Removing config_drive_xfer_rate() is bad,
fe. you don't check for bad drives now.
Regards,
--
Bartlomiej
On Fri, May 16, 2003 at 07:14:36PM +0200, Bartlomiej Zolnierkiewicz wrote:
> Good job, but...
>
> On Fri, 16 May 2003, Vojtech Pavlik wrote:
> > And while doing the changes I did also some cleanups, mainly removing a
> > bunch of debug code that doesn't seem very useful when lspci does the
> > same job. And removing the config_drive_xfer_rate in favor of functions
> > from ide-timing.h.
>
> Debug code in fe. SiS IDE driver does the same as lspci given you get to
> the point you can run lspci. Dumping of PCI conf regs should be moved to
> generic IDE PCI code as it might be useful for other IDE PCI drivers.
Yes, that would be useful.
> Removing config_drive_xfer_rate() is bad,
> fe. you don't check for bad drives now.
True. That needs to be added to ide_find_best_mode in ide-timing.h.
Which needs to be converted to ide-timing.c.
--
Vojtech Pavlik
SuSE Labs, SuSE CR