lkml.org 
[lkml]   [2013]   [Feb]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v3 00/22] PCI: Iterate pci host bridge instead of pci root bus
From
On Wed, Feb 6, 2013 at 3:02 PM, Bjorn Helgaas <bhelgaas@google.com> wrote:
> On Wed, Feb 6, 2013 at 3:05 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
>> On Wednesday, February 06, 2013 01:53:50 PM Yinghai Lu wrote:
>>> On Wed, Feb 6, 2013 at 1:43 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
>>> > On Wednesday, February 06, 2013 01:28:27 PM Yinghai Lu wrote:
>>> >> so you still did not answer you want 1 or 2 yet:
>>> >>
>>> >> for sgi_hotplug,
>>> >>
>>> >> 1. still keep the module support, and register acpi_pci_driver later.
>>> >> 2. built-in support only, and need to register acpi_pci_driver early.
>>> >
>>> > Please work with the assumption that acpi_pci_driver is not going to be there
>>> > any more.
>>> >
>>>
>>> I think I could change ioapic and iommu hotplug to weak add/remove because they
>>> should be built-in by nature.
>>>
>>> but how about others like sgi_hotplug etc?
>
> I think that could be handled with pcibios_root_bridge_prepare() or
> something similar -- something that happens in the "add host bridge"
> path. I'm not saying we necessarily have all the right hooks in place
> yet. It's just that I'd rather figure out what the right hooks *are*
> and add them if necessary, than just convert pci_root_buses to
> for_each_pci_host_bridge().

so we will have all built-in for apci_pci_driver and call add/remove directly?

BTW, looks like sgi_hotplug may need some time to convert ...
Also not sure who can test it.

>
>> I'd really prefer to wait for the patchset removing acpi_pci_driver (from
>> Myron) before proceeding with more changes in that area.
>>
>> Myron, do you have a prototype based on the current linux-next?
>
> I think it's really the acpiphp/pciehp issue that's holding up the
> removal of acpi_pci_driver. I'm not sure if anybody's working on
> that.

Myron's last Dec's version would split all acpi_pci_driver's
add/remove into acpi_pci_root_add/remove.

other way is use host-bridge nodifier, looks like that is not favorable.

BTW, looks like I could put add/remove calling directly in
acpi_pci_root_add/remove
for ioapic and iommu.

Thanks

Yinghai


\
 
 \ /
  Last update: 2013-02-07 01:22    [W:0.049 / U:0.480 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site