Re: [PATCH v5 2/4] gpio: Allow PC/104 devices on X86_64

From: William Breathitt Gray
Date: Fri Jun 17 2016 - 22:50:33 EST


On Fri, Jun 17, 2016 at 07:36:07PM -0700, Greg KH wrote:
>On Fri, Jun 17, 2016 at 12:47:14PM +0200, Thorsten Leemhuis wrote:
>> On 31.05.2016 17:25, Greg KH wrote:
>> > On Tue, May 31, 2016 at 07:23:06AM -0400, William Breathitt Gray wrote:
>> >> On Tue, May 31, 2016 at 10:41:49AM +0200, Linus Walleij wrote:
>> >>> On Sat, May 28, 2016 at 12:08 AM, William Breathitt Gray
>> >>> <vilhelm.gray@xxxxxxxxx> wrote:
>> >>>> With the introduction of the ISA_BUS_API Kconfig option, ISA-style
>> >>>> drivers may be built for X86_64 architectures. This patch changes the
>> >>>> ISA Kconfig option dependency of the PC/104 drivers to ISA_BUS_API, thus
>> >>>> allowing them to build for X86_64 as they are expected to.
>> > [â]
>> >> Greg K-H,
>> >>
>> >> Would you be able to pick up this entire patchset via driver-core; I
>> >> figure that tree is the most appropriate to receive any core ISA bus
>> >> driver changes (unless you see a more fitting path to take).
>> > [â]
>> > Yes, I can take this through the driver core tree as that's where the
>> > original series came from...
>>
>> Was this series merged or did it fell through the cracks? I currently
>> assume the latter, as I can't see it in neither mainline nor linux-next
>> (but maybe I'm missing something). Just wondering, because I have this
>> issue on my regression list for 4.7.
>>
>> For the whole context see:
>> http://thread.gmane.org/gmane.linux.kernel.gpio/17016/
>>
>> Sincerely, your regression tracker for Linux 4.7 (http://bit.ly/28JRmJo)
>
>I don't think this is a regression, I was going to queue these up for
>4.8-rc1. As it is now, 4.7-rc is working just fine in this regards,
>right?
>
>Or am I missing something?

The PC/104 drivers were changed to utilize the ISA bus driver as part of
the original patchset which attempted to decouple the X86_32 dependency
from the ISA Kconfig option; these drivers were updated with the
intention of building on X86_64 in addition to X86_32.

However, the respective patches were merged without the decoupling
changes (since decoupling was the wrong approach), resulting in an
unintentional regression: the PC/104 drivers are now restricted to
X86_32 due to the ISA Kconfig option dependency, while they were capable
of building for X86_64 in previous kernel versions.

This patchset should fix this regression by introducing the ISA_BUS_API
Kconfig option, and the respective Kconfig dependency changes for the
drivers, in order to allow them to build for both X86_64 and X86_32 as
originally capable.

William Breathitt Gray