Weird ahci init failure possibly caused by PCI layer

Weird ahci init failure possibly caused by PCI layer

[Tejun Heo]

Hello, Jesse.

There is a pretty weird ahci init failure[1] reported by Reinhard Max
against openSUSE 11.3 on Dell Optiplex 960[2].  It looks like
something between the cpu and the controller isn't properly set up and
register reads on the controller return garbage values.  The problem
appeared somewhere between 2.6.33 and 2.6.34-rc3 and is still present
in -rc5.

The following log from -rc3 seemed :-) relevant to me.

 <3>pci 0000:00:1f.2: no compatible bridge window for [mem 0xff970000-0xff9707ff]
 <4>Expanded resource reserved due to conflict with PCI Bus 0000:00
 <7>reserve RAM buffer: 000000000009e400 - 000000000009ffff
 <7>reserve RAM buffer: 00000000cfdffc00 - 00000000cfffffff

The full log from -rc3 is available[3].  Any ideas?

Thanks.

-- 
tejun

[1] https://bugzilla.novell.com/show_bug.cgi?id=598520
[2] https://bugzillafiles.novell.org/attachment.cgi?id=356199
[3] https://bugzillafiles.novell.org/attachment.cgi?id=356197

Re: Weird ahci init failure possibly caused by PCI layer

[Bjorn Helgaas]

On Wednesday 28 April 2010 08:30:53 am Tejun Heo wrote:
> Hello, Jesse.
> 
> There is a pretty weird ahci init failure[1] reported by Reinhard Max
> against openSUSE 11.3 on Dell Optiplex 960[2].  It looks like
> something between the cpu and the controller isn't properly set up and
> register reads on the controller return garbage values.  The problem
> appeared somewhere between 2.6.33 and 2.6.34-rc3 and is still present
> in -rc5.
> 
> The following log from -rc3 seemed :-) relevant to me.
> 
>  <3>pci 0000:00:1f.2: no compatible bridge window for [mem 0xff970000-0xff9707ff]
>  <4>Expanded resource reserved due to conflict with PCI Bus 0000:00
>  <7>reserve RAM buffer: 000000000009e400 - 000000000009ffff
>  <7>reserve RAM buffer: 00000000cfdffc00 - 00000000cfffffff
> 
> The full log from -rc3 is available[3].  Any ideas?

This is the same problem as https://bugzilla.kernel.org/show_bug.cgi?id=15744

The snippet below shows that BIOS left the 00:1f.2 BAR at
[mem 0xff970000-0xff9707ff], which is not in any of the host bridge
MMIO windows, so Linux moved it.  Unfortunately, we moved it to 0xa0000,
which was not a good choice.

The patch here: https://bugzilla.kernel.org/attachment.cgi?id=26149
(attached as comment #11 to 15744) should fix the problem.

Let me know if it doesn't.

Bjorn


  pci_root PNP0A03:00: host bridge window [mem 0x000a0000-0x000bffff]
  pci_root PNP0A03:00: host bridge window [mem 0x000c0000-0x000effff]
  pci_root PNP0A03:00: host bridge window [mem 0x000f0000-0x000fffff]
  pci_root PNP0A03:00: host bridge window [mem 0xf0000000-0xfebfffff]
  pci_root PNP0A03:00: host bridge window [mem 0xcff00000-0xdfffffff]
  pci_root PNP0A03:00: host bridge window [mem 0xff980000-0xff980fff]
  pci_root PNP0A03:00: host bridge window [mem 0xff97c000-0xff97ffff]
  pci_root PNP0A03:00: host bridge window [mem 0xfed20000-0xfed9ffff]
  pci_root PNP0A03:00: host bridge window [mem 0xfeda7000-0xfeda7fff]
  pci 0000:00:1f.2: reg 24: [mem 0xff970000-0xff9707ff]
  pci 0000:00:1f.2: no compatible bridge window for [mem 0xff970000-0xff9707ff]
  pci 0000:00:1f.2: BAR 5: assigned [mem 0x000a0000-0x000a07ff]