Re: [RFC PATCH v5 4/8] ACPI, PCI: avoid building pci_slot as module

[Bjorn Helgaas <bhelgaas@google.com>]

On Mon, Jan 28, 2013 at 2:29 PM, Yinghai Lu <yinghai@kernel.org> wrote:
> On Mon, Jan 28, 2013 at 1:09 PM, Bjorn Helgaas <bhelgaas@google.com> wrote:
>> On Sun, Jan 20, 2013 at 5:01 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
>>> On Saturday, January 19, 2013 12:07:42 AM Jiang Liu wrote:
>>>> As discussed in thread at https://patchwork.kernel.org/patch/1946851/,
>>>> there's no value in supporting CONFIG_ACPI_PCI_SLOT=m any more.
>>>> So change Kconfig and code to only support building pci_slot as
>>>> built-in driver.
>>>>
>>>> Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
>>>
>>> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
>>
>> Acked-by: Bjorn Helgaas <bhelgaas@google.com>
>>
>> I think we should eventually get rid of acpi_pci_register_driver() and
>> do this initialization directly from acpi_pci_root_add().  But
>> removing the module option here is a good first step.
>>
>> Rafael, do you want to apply this (and [6/8]) via your tree?  If not,
>> I can take it.
>
> If bios have messed up slot name or idx, we will get strange 1-1....
> other crazy name.
>
> if you really need to put it as built-in, may need to some command
> line that user could switch it off.

It would save us all a lot of time if you gave an example and worked
through the scenario where this is a problem.

We already have the choice of having pci_slot built-in, so if there's
a bug in that config, we already have the bug.  This patch merely
removes a config where the bug might be covered up.

I don't know why "adding a command line switch" appeals to you as the
solution to every problem.  As far as I'm concerned that's not a
solution to ANY problem.  It might be a band-aid to enable users to
limp along while we figure out a correct solution, but it's certainly
not a resolution.

Bjorn