Re: [PATCH v11 4/7] powerpc/powernv: Override pcibios_default_alignment() to force PCI devices to be page aligned

From: Michael Ellerman <mpe@ellerman.id.au>
To: Bjorn Helgaas <bhelgaas@google.com>, Yongji Xie <elohimes@gmail.com>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>,
	Alexey Kardashevskiy <aik@ozlabs.ru>,
	"linux-pci\@vger.kernel.org" <linux-pci@vger.kernel.org>,
	Gavin Shan <gwshan@linux.vnet.ibm.com>,
	Alex Williamson <alex.williamson@redhat.com>,
	Paul Mackerras <paulus@samba.org>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	linuxppc-dev <linuxppc-dev@lists.ozlabs.org>
Subject: Re: [PATCH v11 4/7] powerpc/powernv: Override pcibios_default_alignment() to force PCI devices to be page aligned
Date: Wed, 19 Apr 2017 11:47:43 +1000	[thread overview]
Message-ID: <87bmrt6udc.fsf@concordia.ellerman.id.au> (raw)
In-Reply-To: <CAErSpo6gtDos6GH5Z7aOoW4a3uS4X-HOOHtPFtyg=F+B6+osrw@mail.gmail.com>

Bjorn Helgaas <bhelgaas@google.com> writes:

> On Mon, Apr 17, 2017 at 4:36 PM, Bjorn Helgaas <bhelgaas@google.com> wrote:
>> From: Yongji Xie <elohimes@gmail.com>
>> diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
>> index 6901a06da2f9..b724487cbd0f 100644
>> --- a/arch/powerpc/platforms/powernv/pci-ioda.c
>> +++ b/arch/powerpc/platforms/powernv/pci-ioda.c
>> @@ -3287,6 +3287,11 @@ static void pnv_pci_setup_bridge(struct pci_bus *bus, unsigned long type)
>>         }
>>  }
>>
>> +static resource_size_t pnv_pci_default_alignment(struct pci_dev *pdev)
>> +{
>> +       return PAGE_SIZE;
>> +}
>
> Is it necessary that pcibios_default_alignment() take a pci_dev
> pointer?

It's not necessary given the current implementation, obviously.

But it did strike me as a good idea to pass it in case we ever want to
do anything device specific in future.

> I'd like this better if it were:
>
>   resource_size_t pcibios_default_alignment(void) { ... }
>
> because the last patch relies on the assumption that all resources of
> *all* devices will be realigned to the same alignment.

But I guess that precludes doing anything device specific, at least
without further changes. So in that case it would be better if the API
didn't include the pci_dev.

Hopefully Yongji can confirm that there were no plans to use the
pci_dev in future patches.

cheers