Re: Regression: bug 85491: radeon 0000:01:00.0: Fatal error during GPU init

[Bjorn Helgaas]

[+cc linux-kernel for visibility]

For context, this is about a regression caused by 5b28541552ef ("PCI:
Restrict 64-bit prefetchable bridge windows to 64-bit resources"),
which appeared in v3.16.  The bugzilla is at
https://bugzilla.kernel.org/show_bug.cgi?id=85491

The problem is that BIOS programmed an invalid Root Port window
leading to the Radeon device.  The window contains the Radeon device,
so the configuration actually *works* fine, but the window is invalid
because it either overlaps system RAM or starts below the upstream
host bridge window, so Linux discards it:

Zermond's system:
  acpi PNP0A08:00: host bridge window [mem 0xc0000000-0xffffffff] (ignored)
  pci_bus 0000:00: root bus resource [mem 0x00000000-0xfffffffff]
  pci 0000:00:01.0:   bridge window [mem 0xbdf00000-0xddefffff 64bit
pref]  # invalid Root Port window
  pci 0000:00:01.0: can't claim BAR 15 [mem 0xbdf00000-0xddefffff
64bit pref]: address conflict with System RAM [mem
0x00100000-0xbff9ffff]
  pci 0000:01:00.0: can't claim BAR 0 [mem 0xc0000000-0xcfffffff
pref]: no compatible bridge window  # Radeon

Marek's system:
  pci_bus 0000:00: root bus resource [mem 0xc0000000-0xffffffff]   (from _CRS)
  pci 0000:00:01.0:   bridge window [mem 0xbdf00000-0xddefffff 64bit
pref]  # invalid Root Port window
  pci 0000:00:01.0: can't claim BAR 15 [mem 0xbdf00000-0xddefffff
64bit pref]: no compatible bridge window
  pci 0000:01:00.0: can't claim BAR 0 [mem 0xc0000000-0xcfffffff
pref]: no compatible bridge window  # Radeon

The Root Port window had the same problem before 5b28541552ef, of
course, since BIOS set it up.  But before 5b28541552ef, Linux assigned
a valid window big enough for the Radeon:

  pci 0000:00:01.0:   bridge window [mem 0xc0000000-0xcfffffff pref]

After 5b28541552ef, we won't put a 64-bit window below 4GB, so we
assign space above 4GB:

  pci 0000:00:01.0:   bridge window [mem 0x140000000-0x1401fffff 64bit pref]

which is not usable by Radeon, since it only has a 32-bit BAR.

On Mon, Dec 8, 2014 at 5:21 PM, Yinghai Lu <yinghai@xxxxxxxxxx> wrote:
> On Mon, Dec 8, 2014 at 3:38 PM, Gavin Shan <gwshan@xxxxxxxxxxxxxxxxxx> wrote:
>> On Mon, Dec 08, 2014 at 02:46:00PM -0700, Bjorn Helgaas wrote:
>>>Yeah, I didn't word that very clearly.  I meant that after
>>>5b28541552ef, that 64-bit window is wasted unless there's a 64-bit BAR
>>>below it; we can't even place the window below 4GB and use it for
>>>32-bit prefetchable BARs.
>>>
>>
>> Yes, I agree. It's why I suggested to return error from pbus_size_mem()
>> to indicate the case: 64-bits prefetchable window isn't used and it's
>> still available to be used by child 32-bits prefetchable BARs. Please
>> take a look on the attached draft patch, which helps to explain the idea
>> only.
>
> That would not help. The 00:01.0 on Zermond's system support hotplug. so mem
> pref will be used for 64bit pref.

I don't think it is an improvement to restrict use of the 64bit pref
window to a hypothetical future 64-bit capable device.  The
configuration done by BIOS was optimal for the hardware already in the
machine -- the prefetchable BAR is in a prefetchable window.  I don't
think we should change that configuration just because there's a
possibility of a different device in the future.

Bjorn
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/