[PATCH v8 02/21] acpi: fix acpi_os_ioremap for arm64

[Catalin Marinas]

On Tue, Feb 03, 2015 at 09:08:42AM +0000, Hanjun Guo wrote:
> On 2015年02月03日 06:14, Rafael J. Wysocki wrote:
> > On Monday, February 02, 2015 08:45:30 PM Hanjun Guo wrote:
> >> From: Mark Salter <msalter at redhat.com>
> >>
> >> The acpi_os_ioremap() function may be used to map normal RAM or IO
> >> regions. The current implementation simply uses ioremap_cache(). This
> >> will work for some architectures, but arm64 ioremap_cache() cannot be
> >> used to map IO regions which don't support caching. So for arm64, use
> >> ioremap() for non-RAM regions.
> >>
> >> CC: Rafael J Wysocki <rjw at rjwysocki.net>
> >> Signed-off-by: Mark Salter <msalter at redhat.com>
> >> Signed-off-by: Hanjun Guo <hanjun.guo at linaro.org>
> >> ---
> >>   include/acpi/acpi_io.h | 6 ++++++
> >>   1 file changed, 6 insertions(+)
> >>
> >> diff --git a/include/acpi/acpi_io.h b/include/acpi/acpi_io.h
> >> index 444671e..9d573db 100644
> >> --- a/include/acpi/acpi_io.h
> >> +++ b/include/acpi/acpi_io.h
> >> @@ -1,11 +1,17 @@
> >>   #ifndef _ACPI_IO_H_
> >>   #define _ACPI_IO_H_
> >>
> >> +#include <linux/mm.h>
> >>   #include <linux/io.h>
> >>
> >>   static inline void __iomem *acpi_os_ioremap(acpi_physical_address phys,
> >>   					    acpi_size size)
> >>   {
> >> +#ifdef CONFIG_ARM64
> >> +	if (!page_is_ram(phys >> PAGE_SHIFT))
> >> +		return ioremap(phys, size);
> >> +#endif
> >
> > I don't want to see #ifdef CONFIG_ARM64 in this file.
> >
> > There are multiple examples of how things like this are done.  Generally,
> > the logic is "If the architecture provides its own function for this, use
> > that one, or use the generic one provided here otherwise."
> 
> OK. I think weak function would work.

Probably not in a header file. It's better to define acpi_os_ioremap()
in an arm64 kernel file, together with something like:

#define ARCH_HAS_ACPI_OS_IOREMAP

and the corresponding #ifdef's in the acpi_io.h file.

On arm64 could we make this function call iorema (nocache) all the time?
We need to clarify the contexts where this is used in the core ACPI
code. The acpi_map() function for example checks if the page is ram and
does a kmap(). Do we need to handle the NVS on arm64? AFAICT, we don't
even compile drivers/acpi/sleep.c in.

Are there other cases where acpi_os_ioremap() is called directly and it
needs a cacheable mapping?

-- 
Catalin