[BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[Benjamin Weber]

Hi all

Since the test2-mm1 sources I get the following error during boot:

VP_IDE: IDE controller at PCI slot 0000:00:11.1
VP_IDE: (ide_setup_pci_device:) Could not enable device.

This results in not being able to use DMA for any devices connected to
my IDE controller. Hdparm says permission denied when I do a hdparm -d1
/dev/hda e.g.

I checked with a vanilla kernel and everything is working fine there.
Going through the broken-out patches from Andrew Morton I found the one
patch that caused the above behavior: nforce2-acpi-fixes.patch

I do not know why it should interfere with my via stuff, but it does. A
vanilla test3 kernel is working fine as well, whereas test3-mm1 shows
the same error as before with test2-mmX.

My board is a ECS K7VTA3 Rev. 5.0. Vanilla Kernel reports the chipset
as:
VP_IDE: VIA vt8235 (rev 00) IDE UDMA133 controller on pci0000:00:11.1

Hope the above information is enough. I can provide more if needed.

--
Benjamin

Re: [BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[Andrew Morton]

Benjamin Weber <shawk@gmx.net> wrote:
>
> Hi all
> 
> Since the test2-mm1 sources I get the following error during boot:
> 
> VP_IDE: IDE controller at PCI slot 0000:00:11.1
> VP_IDE: (ide_setup_pci_device:) Could not enable device.
> 
> This results in not being able to use DMA for any devices connected to
> my IDE controller. Hdparm says permission denied when I do a hdparm -d1
> /dev/hda e.g.
> 
> I checked with a vanilla kernel and everything is working fine there.
> Going through the broken-out patches from Andrew Morton I found the one
> patch that caused the above behavior: nforce2-acpi-fixes.patch

Thanks for working that out.  It must have taken some time.

> I do not know why it should interfere with my via stuff, but it does. A
> vanilla test3 kernel is working fine as well, whereas test3-mm1 shows
> the same error as before with test2-mmX.

Me either.  Unfortunately that patch does five different things so we
cannot easily narrow it down further.

Re: [BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[Andrew de Quincey]

> > Since the test2-mm1 sources I get the following error during boot:
> >
> > VP_IDE: IDE controller at PCI slot 0000:00:11.1
> > VP_IDE: (ide_setup_pci_device:) Could not enable device.
> >
> > This results in not being able to use DMA for any devices connected to
> > my IDE controller. Hdparm says permission denied when I do a hdparm -d1
> > /dev/hda e.g.
> >
> > I checked with a vanilla kernel and everything is working fine there.
> > Going through the broken-out patches from Andrew Morton I found the one
> > patch that caused the above behavior: nforce2-acpi-fixes.patch

Hi, thanks for tracking down the problem

Can you send me: 
a dmesg with the patch (i.e. when it fails (essential))
a dmesg without the patch (i.e. when it succeeds)
the output of /proc/interrupts (when it fails (if possible)) 
the output of /proc/interrupts (when it succeeds) 
the output of /proc/acpi/dsdt
the output of lspci -vvv

(the last two can be either with or without the patch)

Sorry about the long list, but this is what I need to diagnose these problems.


> > I do not know why it should interfere with my via stuff, but it does. A
> > vanilla test3 kernel is working fine as well, whereas test3-mm1 shows
> > the same error as before with test2-mmX.
>
> Me either.  Unfortunately that patch does five different things so we
> cannot easily narrow it down further.

Yeah, I know. My next patch is likely to have to do even more unfortunately. 
Found quite a number more issues with IRQs.

Re: [BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[Bartlomiej Zolnierkiewicz]

On Mon, 11 Aug 2003, Andrew de Quincey wrote:

> > > I do not know why it should interfere with my via stuff, but it does. A
> > > vanilla test3 kernel is working fine as well, whereas test3-mm1 shows
> > > the same error as before with test2-mmX.
> >
> > Me either.  Unfortunately that patch does five different things so we
> > cannot easily narrow it down further.
>
> Yeah, I know. My next patch is likely to have to do even more unfortunately.
> Found quite a number more issues with IRQs.

Split it on logical changes if its possible.

--bartlomiej

Re: [BUG mm-tree of test2/test3] nforce2-acpi-fixes breaks via ide controller

[Andrew de Quincey]

On Monday 11 August 2003 12:28, Bartlomiej Zolnierkiewicz wrote:
> On Mon, 11 Aug 2003, Andrew de Quincey wrote:
> > > > I do not know why it should interfere with my via stuff, but it does.
> > > > A vanilla test3 kernel is working fine as well, whereas test3-mm1
> > > > shows the same error as before with test2-mmX.
> > >
> > > Me either.  Unfortunately that patch does five different things so we
> > > cannot easily narrow it down further.
> >
> > Yeah, I know. My next patch is likely to have to do even more
> > unfortunately. Found quite a number more issues with IRQs.
>
> Split it on logical changes if its possible.

I will attempt to as much as possible, but many of the changes rely on each 
other. Yuck.