Re: [PATCH v3 3/3] gpio: add Intel SCH GPIO controller driver

[Mike Rapoport <mike@compulab.co.il>]

David Brownell wrote:
> On Sunday 21 February 2010, Denis Turischev wrote:
>> v2: there is no acpi_check_region, it will be implemented in mfd-core
>> v3: patch refreshed against the latest Linus tree
> 
> Could such call really address the GPIO conflict issue I mentioned?
> 
> The AML bytecodes I looked at were writing directly to Southbridge
> GPIO registers (or reading them), or relying on ACPI to mediate the
> GPIO interrupts.  ISTR that button drivers, and code to switch into
> or out of low power states, were good sources of such bad examples.

I'm really not an ACPI expert, but as far as I understand possibility of 
such conflicts largely depends on particular board/BIOS implementation. 
On the hardware we have such conflict cannot happen, unless there are 
bugs in ACPI we are not yet aware of. :)

> Calls like that should clearly be able to handle cases where ACPI
> has a "Real" Driver (tm) ... e.g. for SMBus hardware.
> 
> I'm not sure what a good solution for this would be, short of just
> not using ACPI ... which may not be practical, given the limited
> degree of x86 board/system support for Linux.
> 
> I mention this mostly because when I looked at the issue in the
> context of an ICHx GPIO driver, I didn't see a good solution to
> the problem then ... and nothing seems to have changed meanwhile.

I've looked at two x86 drivers in drivers/gpiolib (cs5535 and langwell) 
and there's no treatment of ACPI in either of them. Since SCH is defined 
by Intel as "embedded" product, having a GPIO driver for it seems 
logical even despite problems you mention.

> - Dave
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/


-- 
Sincerely yours,
Mike.