Re: [PATCH] PCI: Clear bridge MEM_64 flag if one child does not support it

[Gavin Shan <gwshan@linux.vnet.ibm.com>]

On Mon, Dec 08, 2014 at 03:59:54PM -0800, Yinghai Lu wrote:
>On Mon, Dec 8, 2014 at 2:56 PM, Benjamin Herrenschmidt
><benh@kernel.crashing.org> wrote:
>> On Mon, 2014-12-08 at 13:52 -0800, Yinghai Lu wrote:
>>> 2. or scan the children resource other than ROM to clear bridge MEM_64
>>>    for mmio pref.
>>>
>>> The patch is using second way so will keep child mmio pref into bridge
>>> mmio pref range.
>>
>> That means that having a single ROM BAR that is 32-bit and prefetchable
>> will downgrade the entire window to 32-bit ? That's not going to work
>> either.
>
>the ROM BAR get skipped during the checking.
>
>+                       if (i != PCI_ROM_RESOURCE)
>+                               mem64_mask &= r->flags & IORESOURCE_MEM_64;
>+               }
>
>>
>> I have GPUs with 16G BARs for example... suddenly they don't fit
>> anaymore because we downgraded the window to 32 bit because somewhere
>> there's a 32-bit pref resource ?
>>
>> That will break more than it fixes...
>
>Please check if this patch break your platform. I tried on my setup on
>x86. and it is still
>working on 64 bit resource allocation.
>

Yinghai, With the patch, the Power8 machine boots up and have resources
assigned properly: 64-bits pref BARs are covered by 64-bits pref windows
and left memory BARs are covered by 32-bits non-pref windows. Note that
I didn't find an 32-bits-pref BAR on the system.

For SRIOV case, Richard still need test.

Thanks,
Gavin

>Thanks
>
>Yinghai
>